diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32227 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 5352,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 10.9943,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 10.9584,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 11.4022,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 10.9768,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 11.0889,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 11.0464,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 11.2208,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 11.0702,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 11.4155,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 11.0631,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 11.197,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 11.056,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 11.1092,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 10.8994,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 11.103,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 11.1372,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 11.1003,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 10.8299,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 10.2585,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 10.3535,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6e-06,
+      "loss": 10.0379,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 10.0975,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 9.8731,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 9.2341,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 9.0927,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-06,
+      "loss": 8.9826,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8e-06,
+      "loss": 8.9649,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 8.6652,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 8.7757,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4e-06,
+      "loss": 8.5096,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6e-06,
+      "loss": 8.6883,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 8.2975,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 8.285,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2e-06,
+      "loss": 7.9462,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4e-06,
+      "loss": 8.0024,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 8.2322,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8e-06,
+      "loss": 7.8988,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-06,
+      "loss": 8.1555,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2e-06,
+      "loss": 7.9796,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 7.8011,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6e-06,
+      "loss": 7.8403,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 7.6365,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 7.8086,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2e-06,
+      "loss": 7.5116,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4e-06,
+      "loss": 7.8187,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.6e-06,
+      "loss": 7.5173,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.8e-06,
+      "loss": 7.5973,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 7.8332,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 7.487,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 7.462,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 7.4334,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.8e-06,
+      "loss": 7.4941,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9e-06,
+      "loss": 7.2823,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.2e-06,
+      "loss": 7.4285,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.4e-06,
+      "loss": 7.5708,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 7.306,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 7.2824,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 7.1456,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.02e-05,
+      "loss": 7.1856,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.04e-05,
+      "loss": 7.2322,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.06e-05,
+      "loss": 7.3762,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.08e-05,
+      "loss": 7.3175,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 7.358,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 7.2088,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 7.1944,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.16e-05,
+      "loss": 7.1452,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.18e-05,
+      "loss": 7.2423,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2e-05,
+      "loss": 7.2724,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.22e-05,
+      "loss": 7.2636,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.24e-05,
+      "loss": 7.1627,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 7.2538,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 7.1301,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 7.2295,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.32e-05,
+      "loss": 7.0985,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 7.3331,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 7.2283,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 7.2221,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 7.1114,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.42e-05,
+      "loss": 7.0985,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.44e-05,
+      "loss": 7.1191,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 7.2704,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.48e-05,
+      "loss": 7.3029,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5e-05,
+      "loss": 7.0338,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.52e-05,
+      "loss": 7.1617,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.54e-05,
+      "loss": 7.1962,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.56e-05,
+      "loss": 7.2343,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.58e-05,
+      "loss": 7.2556,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 7.2534,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.62e-05,
+      "loss": 7.2853,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 7.0127,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.66e-05,
+      "loss": 7.329,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 6.9507,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 7.0706,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 7.3435,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.74e-05,
+      "loss": 7.3304,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.76e-05,
+      "loss": 7.1647,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.78e-05,
+      "loss": 7.0002,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8e-05,
+      "loss": 6.8276,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 7.1761,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.84e-05,
+      "loss": 7.0169,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.86e-05,
+      "loss": 7.3488,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.88e-05,
+      "loss": 7.2829,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9e-05,
+      "loss": 7.2905,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 7.176,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.94e-05,
+      "loss": 7.1516,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 7.0835,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 7.573,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 7.1754,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 7.0634,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.04e-05,
+      "loss": 7.1904,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.06e-05,
+      "loss": 7.0143,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.08e-05,
+      "loss": 6.9704,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1e-05,
+      "loss": 7.1983,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.12e-05,
+      "loss": 7.2388,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 7.1558,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.16e-05,
+      "loss": 7.1881,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.18e-05,
+      "loss": 7.3307,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 7.0756,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.22e-05,
+      "loss": 7.1239,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 7.1285,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.26e-05,
+      "loss": 7.3267,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 7.2647,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 7.2048,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.32e-05,
+      "loss": 7.3158,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 6.9004,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.36e-05,
+      "loss": 7.3811,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.38e-05,
+      "loss": 7.1474,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4e-05,
+      "loss": 7.1854,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 7.0567,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.44e-05,
+      "loss": 7.0879,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.46e-05,
+      "loss": 7.3049,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.48e-05,
+      "loss": 7.0845,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5e-05,
+      "loss": 7.2161,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 7.0246,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.54e-05,
+      "loss": 7.0001,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 7.2634,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.58e-05,
+      "loss": 7.1607,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 7.0718,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 7.2687,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.64e-05,
+      "loss": 7.1654,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 7.3757,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 7.3598,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 6.98,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 7.1408,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 7.2928,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 7.1745,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 7.2032,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 7.1541,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 6.9952,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.84e-05,
+      "loss": 6.344,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.86e-05,
+      "loss": 7.1416,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.88e-05,
+      "loss": 7.2178,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9e-05,
+      "loss": 6.9869,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 7.2288,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.94e-05,
+      "loss": 6.9245,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.96e-05,
+      "loss": 7.2831,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.98e-05,
+      "loss": 7.0794,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3e-05,
+      "loss": 7.1408,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.02e-05,
+      "loss": 7.0602,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.04e-05,
+      "loss": 7.2424,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.06e-05,
+      "loss": 7.2148,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.08e-05,
+      "loss": 7.3888,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1e-05,
+      "loss": 7.2205,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.12e-05,
+      "loss": 7.1783,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 7.2127,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.16e-05,
+      "loss": 7.201,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.18e-05,
+      "loss": 7.1274,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 6.9574,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 6.9177,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.24e-05,
+      "loss": 7.2344,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.26e-05,
+      "loss": 7.1829,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 7.106,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3e-05,
+      "loss": 7.1077,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.32e-05,
+      "loss": 6.979,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 7.1942,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 6.9882,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.38e-05,
+      "loss": 7.0006,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 7.077,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 6.9483,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 7.1153,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.46e-05,
+      "loss": 6.9924,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.48e-05,
+      "loss": 6.9844,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.5e-05,
+      "loss": 7.2299,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.52e-05,
+      "loss": 6.8925,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.54e-05,
+      "loss": 7.1434,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.56e-05,
+      "loss": 6.9794,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.58e-05,
+      "loss": 6.8796,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6e-05,
+      "loss": 7.0778,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.62e-05,
+      "loss": 7.0181,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 7.0357,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.66e-05,
+      "loss": 7.1621,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.68e-05,
+      "loss": 6.8678,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7e-05,
+      "loss": 7.0031,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.72e-05,
+      "loss": 7.0492,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.74e-05,
+      "loss": 7.0843,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.76e-05,
+      "loss": 7.044,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 7.0957,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8e-05,
+      "loss": 6.8913,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.82e-05,
+      "loss": 6.7978,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 6.443,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.86e-05,
+      "loss": 7.2658,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.88e-05,
+      "loss": 7.0871,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 7.3244,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 6.966,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.94e-05,
+      "loss": 6.9823,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 7.1579,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 7.1491,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4e-05,
+      "loss": 7.0796,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.02e-05,
+      "loss": 7.118,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 7.2473,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 7.0947,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.08e-05,
+      "loss": 7.2109,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1e-05,
+      "loss": 7.1369,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.12e-05,
+      "loss": 6.9616,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.14e-05,
+      "loss": 7.2364,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.16e-05,
+      "loss": 6.8616,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.18e-05,
+      "loss": 7.0692,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.2e-05,
+      "loss": 6.9306,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.22e-05,
+      "loss": 7.0661,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.24e-05,
+      "loss": 7.1495,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.26e-05,
+      "loss": 7.1552,
+      "step": 221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 7.2373,
+      "step": 222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3e-05,
+      "loss": 6.8844,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.32e-05,
+      "loss": 7.1065,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 7.201,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.36e-05,
+      "loss": 7.4243,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.38e-05,
+      "loss": 7.2527,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 7.2063,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 7.0557,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.44e-05,
+      "loss": 7.0629,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.46e-05,
+      "loss": 7.0912,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 7.1477,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5e-05,
+      "loss": 7.3119,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.52e-05,
+      "loss": 7.0351,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 6.9773,
+      "step": 235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 7.3856,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.58e-05,
+      "loss": 7.1324,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 7.0901,
+      "step": 238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 7.0031,
+      "step": 239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.64e-05,
+      "loss": 7.06,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 7.0717,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 7.0482,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7e-05,
+      "loss": 6.9197,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.72e-05,
+      "loss": 7.1217,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.74e-05,
+      "loss": 7.0378,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76e-05,
+      "loss": 6.9547,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.78e-05,
+      "loss": 6.61,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8e-05,
+      "loss": 6.9574,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.82e-05,
+      "loss": 6.6961,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 6.4984,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.86e-05,
+      "loss": 7.2069,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.88e-05,
+      "loss": 7.4438,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9e-05,
+      "loss": 7.1281,
+      "step": 253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.92e-05,
+      "loss": 7.264,
+      "step": 254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.94e-05,
+      "loss": 7.1284,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.96e-05,
+      "loss": 7.1746,
+      "step": 256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 7.1866,
+      "step": 257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5e-05,
+      "loss": 7.1018,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.02e-05,
+      "loss": 7.3555,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 6.7875,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 6.9759,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.08e-05,
+      "loss": 7.0293,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 7.2576,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 7.1699,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.14e-05,
+      "loss": 6.9988,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.16e-05,
+      "loss": 6.9783,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 7.1326,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 7.0865,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.22e-05,
+      "loss": 7.1328,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 6.9455,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 6.9803,
+      "step": 271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.28e-05,
+      "loss": 7.0873,
+      "step": 272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 6.9735,
+      "step": 273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 7.1154,
+      "step": 274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 7.2227,
+      "step": 275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 7.2743,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 7.1936,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 7.2583,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.420000000000001e-05,
+      "loss": 7.0363,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 7.0358,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 7.2046,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 7.2714,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 7.2108,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 6.8612,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 7.1636,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 6.923,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 6.9685,
+      "step": 287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 7.2593,
+      "step": 288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.620000000000001e-05,
+      "loss": 6.9878,
+      "step": 289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 7.1322,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.66e-05,
+      "loss": 7.0868,
+      "step": 291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.68e-05,
+      "loss": 7.2568,
+      "step": 292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 7.0662,
+      "step": 293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.72e-05,
+      "loss": 7.0929,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.74e-05,
+      "loss": 6.9213,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.76e-05,
+      "loss": 6.7823,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 6.8251,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.8e-05,
+      "loss": 7.1662,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.82e-05,
+      "loss": 6.7,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 6.5886,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.86e-05,
+      "loss": 7.1817,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.88e-05,
+      "loss": 7.0857,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9e-05,
+      "loss": 7.0472,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.92e-05,
+      "loss": 6.9869,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.94e-05,
+      "loss": 7.0928,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.96e-05,
+      "loss": 7.1116,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 7.17,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6e-05,
+      "loss": 7.2483,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.02e-05,
+      "loss": 7.1166,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.04e-05,
+      "loss": 7.3166,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.06e-05,
+      "loss": 7.0532,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.08e-05,
+      "loss": 7.1509,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.1e-05,
+      "loss": 7.0713,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.12e-05,
+      "loss": 7.1442,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.14e-05,
+      "loss": 7.1455,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.16e-05,
+      "loss": 7.0402,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.18e-05,
+      "loss": 6.969,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.2e-05,
+      "loss": 7.0087,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.220000000000001e-05,
+      "loss": 7.1762,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.24e-05,
+      "loss": 7.1858,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.26e-05,
+      "loss": 7.3863,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 7.2937,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.3e-05,
+      "loss": 7.2723,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.32e-05,
+      "loss": 7.0623,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.340000000000001e-05,
+      "loss": 6.9822,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.36e-05,
+      "loss": 7.1546,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.38e-05,
+      "loss": 7.0314,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 7.133,
+      "step": 328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.42e-05,
+      "loss": 7.0376,
+      "step": 329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 7.1706,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.460000000000001e-05,
+      "loss": 7.1314,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.48e-05,
+      "loss": 7.1215,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 7.2526,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.52e-05,
+      "loss": 7.2246,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.54e-05,
+      "loss": 7.0868,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 7.0976,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.58e-05,
+      "loss": 7.1152,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.6e-05,
+      "loss": 6.9849,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.620000000000001e-05,
+      "loss": 6.9359,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.64e-05,
+      "loss": 7.0532,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.66e-05,
+      "loss": 6.955,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 7.2644,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.7e-05,
+      "loss": 7.1705,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 7.1836,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.740000000000001e-05,
+      "loss": 6.9061,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.76e-05,
+      "loss": 6.9669,
+      "step": 346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 7.4457,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 7.0827,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.82e-05,
+      "loss": 6.5416,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 6.1589,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.860000000000001e-05,
+      "loss": 7.2102,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 7.2278,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.9e-05,
+      "loss": 7.1306,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.92e-05,
+      "loss": 7.0525,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 7.0724,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.96e-05,
+      "loss": 7.0648,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.98e-05,
+      "loss": 7.2456,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7e-05,
+      "loss": 7.1056,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.02e-05,
+      "loss": 7.2209,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.04e-05,
+      "loss": 7.0958,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.06e-05,
+      "loss": 6.8967,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.08e-05,
+      "loss": 6.7813,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.1e-05,
+      "loss": 7.0357,
+      "step": 363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.12e-05,
+      "loss": 6.924,
+      "step": 364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.14e-05,
+      "loss": 7.1054,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.16e-05,
+      "loss": 7.1807,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.18e-05,
+      "loss": 6.9746,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.2e-05,
+      "loss": 7.0519,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.22e-05,
+      "loss": 7.0763,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.24e-05,
+      "loss": 6.9779,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.26e-05,
+      "loss": 6.9394,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 7.2521,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.3e-05,
+      "loss": 7.136,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.32e-05,
+      "loss": 6.9745,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.340000000000001e-05,
+      "loss": 6.7593,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.36e-05,
+      "loss": 7.0613,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.38e-05,
+      "loss": 7.0142,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.4e-05,
+      "loss": 6.9591,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.42e-05,
+      "loss": 6.9696,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.44e-05,
+      "loss": 7.2282,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.46e-05,
+      "loss": 6.9052,
+      "step": 381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.48e-05,
+      "loss": 6.8876,
+      "step": 382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 7.0237,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.52e-05,
+      "loss": 7.0708,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.54e-05,
+      "loss": 7.1279,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 7.2487,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.58e-05,
+      "loss": 7.0603,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.6e-05,
+      "loss": 6.8195,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.620000000000001e-05,
+      "loss": 7.0025,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.64e-05,
+      "loss": 7.0806,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.66e-05,
+      "loss": 6.8648,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 7.4388,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.7e-05,
+      "loss": 7.0427,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.72e-05,
+      "loss": 6.9174,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.740000000000001e-05,
+      "loss": 6.9422,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.76e-05,
+      "loss": 6.7755,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 6.5908,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 6.7942,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.82e-05,
+      "loss": 6.9399,
+      "step": 399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 6.5178,
+      "step": 400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 7.860000000000001e-05,
+      "loss": 7.6662,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.88e-05,
+      "loss": 7.4403,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 7.1939,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 7.3696,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.94e-05,
+      "loss": 6.994,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 7.3248,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 7.98e-05,
+      "loss": 7.0381,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8e-05,
+      "loss": 7.0852,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.020000000000001e-05,
+      "loss": 7.0731,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.04e-05,
+      "loss": 7.0279,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.060000000000001e-05,
+      "loss": 7.0357,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 6.9296,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.1e-05,
+      "loss": 7.1563,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 6.9103,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.14e-05,
+      "loss": 7.1143,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.16e-05,
+      "loss": 7.0693,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.18e-05,
+      "loss": 7.042,
+      "step": 417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.2e-05,
+      "loss": 6.9777,
+      "step": 418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.22e-05,
+      "loss": 6.9689,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.24e-05,
+      "loss": 7.1131,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.26e-05,
+      "loss": 7.0906,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.28e-05,
+      "loss": 7.0861,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.3e-05,
+      "loss": 6.953,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.32e-05,
+      "loss": 7.2407,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.34e-05,
+      "loss": 7.1928,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.36e-05,
+      "loss": 7.0813,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.38e-05,
+      "loss": 7.0063,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.4e-05,
+      "loss": 6.9934,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.42e-05,
+      "loss": 7.0596,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.44e-05,
+      "loss": 7.0651,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.46e-05,
+      "loss": 6.9085,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.48e-05,
+      "loss": 7.0998,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.5e-05,
+      "loss": 6.9893,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.52e-05,
+      "loss": 7.0977,
+      "step": 434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.54e-05,
+      "loss": 6.9914,
+      "step": 435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 6.9715,
+      "step": 436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.58e-05,
+      "loss": 6.9418,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.6e-05,
+      "loss": 7.0783,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 6.8824,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.64e-05,
+      "loss": 7.0227,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.66e-05,
+      "loss": 6.9784,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 6.9761,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7e-05,
+      "loss": 6.7901,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.72e-05,
+      "loss": 6.8864,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 6.8672,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76e-05,
+      "loss": 7.1435,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78e-05,
+      "loss": 6.8382,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 6.9071,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82e-05,
+      "loss": 6.9615,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 6.3838,
+      "step": 450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.86e-05,
+      "loss": 7.0146,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.88e-05,
+      "loss": 7.0833,
+      "step": 452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 7.2134,
+      "step": 453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.92e-05,
+      "loss": 7.107,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.94e-05,
+      "loss": 6.9886,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 7.0828,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.98e-05,
+      "loss": 6.8274,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9e-05,
+      "loss": 7.0311,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.020000000000001e-05,
+      "loss": 6.8049,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.04e-05,
+      "loss": 7.1272,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.06e-05,
+      "loss": 6.8705,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 6.8506,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.1e-05,
+      "loss": 6.8942,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 7.0391,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.140000000000001e-05,
+      "loss": 6.8398,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.16e-05,
+      "loss": 7.1005,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 6.9834,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 7.0702,
+      "step": 468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.22e-05,
+      "loss": 6.7883,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.240000000000001e-05,
+      "loss": 7.0088,
+      "step": 470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.260000000000001e-05,
+      "loss": 7.033,
+      "step": 471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.28e-05,
+      "loss": 6.848,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 6.8449,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.320000000000002e-05,
+      "loss": 7.0345,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.340000000000001e-05,
+      "loss": 6.8402,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.360000000000001e-05,
+      "loss": 6.9558,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.38e-05,
+      "loss": 7.0038,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.4e-05,
+      "loss": 6.9085,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.42e-05,
+      "loss": 7.0282,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.44e-05,
+      "loss": 7.0454,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.46e-05,
+      "loss": 7.155,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.48e-05,
+      "loss": 6.991,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.5e-05,
+      "loss": 7.3644,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.52e-05,
+      "loss": 6.9936,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.54e-05,
+      "loss": 7.0514,
+      "step": 485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.56e-05,
+      "loss": 7.0196,
+      "step": 486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.58e-05,
+      "loss": 6.9094,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.6e-05,
+      "loss": 7.1765,
+      "step": 488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.620000000000001e-05,
+      "loss": 6.9902,
+      "step": 489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.64e-05,
+      "loss": 6.9647,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.66e-05,
+      "loss": 6.8088,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 7.1804,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.7e-05,
+      "loss": 7.0074,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.72e-05,
+      "loss": 7.0812,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.74e-05,
+      "loss": 6.8391,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.76e-05,
+      "loss": 6.9229,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.78e-05,
+      "loss": 6.7293,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.8e-05,
+      "loss": 6.917,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.82e-05,
+      "loss": 6.6942,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.84e-05,
+      "loss": 6.4796,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 10.769026756286621,
+      "eval_runtime": 1321.3649,
+      "eval_samples_per_second": 1.999,
+      "eval_steps_per_second": 0.25,
+      "eval_wer": 1.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.86e-05,
+      "loss": 7.1292,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.88e-05,
+      "loss": 7.1513,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 6.953,
+      "step": 503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.92e-05,
+      "loss": 7.0816,
+      "step": 504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.94e-05,
+      "loss": 6.9108,
+      "step": 505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 7.0561,
+      "step": 506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.98e-05,
+      "loss": 6.8647,
+      "step": 507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001,
+      "loss": 7.1634,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.997938994229185e-05,
+      "loss": 7.1713,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.995877988458369e-05,
+      "loss": 7.1024,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.993816982687552e-05,
+      "loss": 7.0799,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.991755976916735e-05,
+      "loss": 6.9018,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.98969497114592e-05,
+      "loss": 6.9022,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.987633965375104e-05,
+      "loss": 6.8078,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.985572959604287e-05,
+      "loss": 7.0772,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.983511953833471e-05,
+      "loss": 7.0942,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.981450948062655e-05,
+      "loss": 7.0663,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.979389942291838e-05,
+      "loss": 7.1295,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.977328936521023e-05,
+      "loss": 6.983,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.975267930750206e-05,
+      "loss": 6.9278,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.97320692497939e-05,
+      "loss": 6.9215,
+      "step": 521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.971145919208574e-05,
+      "loss": 7.0512,
+      "step": 522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.969084913437759e-05,
+      "loss": 6.8983,
+      "step": 523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.967023907666943e-05,
+      "loss": 6.9866,
+      "step": 524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.964962901896126e-05,
+      "loss": 7.0045,
+      "step": 525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.962901896125309e-05,
+      "loss": 7.0079,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.960840890354493e-05,
+      "loss": 7.2007,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.958779884583678e-05,
+      "loss": 6.841,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.956718878812861e-05,
+      "loss": 6.9953,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.954657873042045e-05,
+      "loss": 7.1213,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.95259686727123e-05,
+      "loss": 6.7855,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.950535861500412e-05,
+      "loss": 6.8574,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.948474855729597e-05,
+      "loss": 6.9793,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.94641384995878e-05,
+      "loss": 6.82,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.944352844187964e-05,
+      "loss": 7.2106,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.942291838417148e-05,
+      "loss": 7.0077,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.940230832646333e-05,
+      "loss": 6.8816,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.938169826875516e-05,
+      "loss": 6.896,
+      "step": 538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.936108821104699e-05,
+      "loss": 6.9621,
+      "step": 539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.934047815333883e-05,
+      "loss": 7.1627,
+      "step": 540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.931986809563067e-05,
+      "loss": 6.9173,
+      "step": 541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.929925803792252e-05,
+      "loss": 7.1056,
+      "step": 542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.927864798021435e-05,
+      "loss": 6.9196,
+      "step": 543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.925803792250619e-05,
+      "loss": 6.9014,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.923742786479803e-05,
+      "loss": 6.6504,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.921681780708986e-05,
+      "loss": 6.7571,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.91962077493817e-05,
+      "loss": 7.0316,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.917559769167354e-05,
+      "loss": 6.7477,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.915498763396538e-05,
+      "loss": 6.3523,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.913437757625722e-05,
+      "loss": 6.1208,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.911376751854907e-05,
+      "loss": 7.3993,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.90931574608409e-05,
+      "loss": 7.4291,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.907254740313273e-05,
+      "loss": 7.1349,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.905193734542457e-05,
+      "loss": 7.1203,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.903132728771641e-05,
+      "loss": 7.0658,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.901071723000826e-05,
+      "loss": 7.1187,
+      "step": 556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.899010717230009e-05,
+      "loss": 7.1719,
+      "step": 557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.896949711459193e-05,
+      "loss": 7.1313,
+      "step": 558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.894888705688376e-05,
+      "loss": 6.8195,
+      "step": 559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.89282769991756e-05,
+      "loss": 7.0463,
+      "step": 560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.890766694146743e-05,
+      "loss": 7.0163,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.888705688375928e-05,
+      "loss": 6.9171,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.886644682605112e-05,
+      "loss": 7.0382,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.884583676834296e-05,
+      "loss": 7.2258,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.882522671063479e-05,
+      "loss": 7.0302,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.880461665292662e-05,
+      "loss": 6.9996,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.878400659521847e-05,
+      "loss": 7.2168,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.876339653751031e-05,
+      "loss": 7.0449,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.874278647980215e-05,
+      "loss": 7.135,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.872217642209398e-05,
+      "loss": 7.0256,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.870156636438583e-05,
+      "loss": 6.946,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.868095630667767e-05,
+      "loss": 6.9494,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.86603462489695e-05,
+      "loss": 6.9072,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.863973619126134e-05,
+      "loss": 6.8277,
+      "step": 574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.861912613355317e-05,
+      "loss": 6.9807,
+      "step": 575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.859851607584502e-05,
+      "loss": 6.9837,
+      "step": 576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.857790601813686e-05,
+      "loss": 7.0911,
+      "step": 577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.85572959604287e-05,
+      "loss": 6.8185,
+      "step": 578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.853668590272053e-05,
+      "loss": 7.0473,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.851607584501236e-05,
+      "loss": 7.2018,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.84954657873042e-05,
+      "loss": 6.9656,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.847485572959605e-05,
+      "loss": 6.9458,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.845424567188789e-05,
+      "loss": 6.8463,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.843363561417972e-05,
+      "loss": 7.2183,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.841302555647157e-05,
+      "loss": 6.992,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.83924154987634e-05,
+      "loss": 6.8293,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.837180544105524e-05,
+      "loss": 7.114,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.835119538334708e-05,
+      "loss": 6.9653,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.833058532563891e-05,
+      "loss": 7.1877,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.830997526793076e-05,
+      "loss": 7.2924,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.82893652102226e-05,
+      "loss": 6.9767,
+      "step": 591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.826875515251444e-05,
+      "loss": 6.9955,
+      "step": 592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.824814509480627e-05,
+      "loss": 6.7293,
+      "step": 593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.82275350370981e-05,
+      "loss": 6.9697,
+      "step": 594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.820692497938995e-05,
+      "loss": 6.7856,
+      "step": 595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.818631492168179e-05,
+      "loss": 6.6475,
+      "step": 596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.816570486397363e-05,
+      "loss": 6.6353,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.814509480626546e-05,
+      "loss": 6.6394,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.81244847485573e-05,
+      "loss": 6.7021,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.810387469084913e-05,
+      "loss": 6.5174,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.808326463314098e-05,
+      "loss": 7.0698,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.806265457543281e-05,
+      "loss": 7.1515,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.804204451772465e-05,
+      "loss": 6.932,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.80214344600165e-05,
+      "loss": 7.0276,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.800082440230834e-05,
+      "loss": 6.8156,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.798021434460017e-05,
+      "loss": 7.0082,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.7959604286892e-05,
+      "loss": 7.1666,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.793899422918384e-05,
+      "loss": 6.9138,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.791838417147568e-05,
+      "loss": 7.0705,
+      "step": 609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.789777411376753e-05,
+      "loss": 6.8852,
+      "step": 610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.787716405605936e-05,
+      "loss": 7.1828,
+      "step": 611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.78565539983512e-05,
+      "loss": 7.0118,
+      "step": 612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.783594394064303e-05,
+      "loss": 7.047,
+      "step": 613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.781533388293487e-05,
+      "loss": 7.1401,
+      "step": 614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.779472382522672e-05,
+      "loss": 7.0227,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.777411376751855e-05,
+      "loss": 7.0281,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.775350370981039e-05,
+      "loss": 7.0143,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.773289365210223e-05,
+      "loss": 6.7438,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.771228359439408e-05,
+      "loss": 7.0274,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.769167353668591e-05,
+      "loss": 6.9697,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.767106347897774e-05,
+      "loss": 6.991,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.765045342126958e-05,
+      "loss": 6.9967,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.762984336356142e-05,
+      "loss": 7.0565,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.760923330585327e-05,
+      "loss": 7.2098,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.75886232481451e-05,
+      "loss": 6.9764,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.756801319043694e-05,
+      "loss": 7.0403,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.754740313272877e-05,
+      "loss": 6.8805,
+      "step": 627
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.752679307502061e-05,
+      "loss": 7.0653,
+      "step": 628
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.750618301731246e-05,
+      "loss": 7.1583,
+      "step": 629
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.748557295960429e-05,
+      "loss": 6.8649,
+      "step": 630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.746496290189613e-05,
+      "loss": 6.8806,
+      "step": 631
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.744435284418797e-05,
+      "loss": 6.8198,
+      "step": 632
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.74237427864798e-05,
+      "loss": 6.8546,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.740313272877163e-05,
+      "loss": 7.0996,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.738252267106348e-05,
+      "loss": 6.9703,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.736191261335532e-05,
+      "loss": 6.8964,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.734130255564716e-05,
+      "loss": 6.9181,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.732069249793901e-05,
+      "loss": 6.7736,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.730008244023084e-05,
+      "loss": 7.0206,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.727947238252268e-05,
+      "loss": 7.0819,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.725886232481451e-05,
+      "loss": 6.859,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.723825226710635e-05,
+      "loss": 7.0651,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.72176422093982e-05,
+      "loss": 6.8661,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.719703215169003e-05,
+      "loss": 6.8842,
+      "step": 644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.717642209398187e-05,
+      "loss": 6.4939,
+      "step": 645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.715581203627371e-05,
+      "loss": 6.7665,
+      "step": 646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.713520197856554e-05,
+      "loss": 6.7676,
+      "step": 647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.711459192085737e-05,
+      "loss": 6.934,
+      "step": 648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.709398186314922e-05,
+      "loss": 7.0966,
+      "step": 649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.707337180544106e-05,
+      "loss": 6.5484,
+      "step": 650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.70527617477329e-05,
+      "loss": 7.1918,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.703215169002475e-05,
+      "loss": 7.0346,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.701154163231658e-05,
+      "loss": 7.1564,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.69909315746084e-05,
+      "loss": 7.1137,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.697032151690025e-05,
+      "loss": 6.9553,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.694971145919209e-05,
+      "loss": 6.9581,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.692910140148392e-05,
+      "loss": 6.8835,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.690849134377577e-05,
+      "loss": 7.0785,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.688788128606761e-05,
+      "loss": 7.0144,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.686727122835945e-05,
+      "loss": 6.9955,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.684666117065128e-05,
+      "loss": 7.0173,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.682605111294311e-05,
+      "loss": 6.9422,
+      "step": 662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.680544105523496e-05,
+      "loss": 6.9396,
+      "step": 663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.67848309975268e-05,
+      "loss": 7.0642,
+      "step": 664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.676422093981864e-05,
+      "loss": 6.9206,
+      "step": 665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.674361088211047e-05,
+      "loss": 6.9842,
+      "step": 666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.672300082440232e-05,
+      "loss": 6.9922,
+      "step": 667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.670239076669415e-05,
+      "loss": 6.932,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.668178070898599e-05,
+      "loss": 7.0071,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.666117065127783e-05,
+      "loss": 6.8581,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.664056059356966e-05,
+      "loss": 7.2524,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.66199505358615e-05,
+      "loss": 7.0018,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.659934047815335e-05,
+      "loss": 6.9941,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.657873042044518e-05,
+      "loss": 6.9293,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.655812036273702e-05,
+      "loss": 6.9698,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.653751030502885e-05,
+      "loss": 7.0254,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.65169002473207e-05,
+      "loss": 6.7698,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.649629018961254e-05,
+      "loss": 6.978,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.647568013190438e-05,
+      "loss": 6.9172,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.645507007419621e-05,
+      "loss": 7.0378,
+      "step": 680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.643446001648804e-05,
+      "loss": 7.114,
+      "step": 681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.641384995877989e-05,
+      "loss": 6.9028,
+      "step": 682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.639323990107173e-05,
+      "loss": 6.9735,
+      "step": 683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.637262984336357e-05,
+      "loss": 6.9267,
+      "step": 684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.63520197856554e-05,
+      "loss": 7.1803,
+      "step": 685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.633140972794725e-05,
+      "loss": 6.9781,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.631079967023909e-05,
+      "loss": 6.8014,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.629018961253092e-05,
+      "loss": 6.9893,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.626957955482275e-05,
+      "loss": 7.2205,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.624896949711459e-05,
+      "loss": 6.9157,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.622835943940644e-05,
+      "loss": 6.9495,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.620774938169828e-05,
+      "loss": 7.0654,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.618713932399012e-05,
+      "loss": 6.9128,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.616652926628195e-05,
+      "loss": 6.9559,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.614591920857378e-05,
+      "loss": 6.8798,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.612530915086563e-05,
+      "loss": 6.9558,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.610469909315747e-05,
+      "loss": 6.8029,
+      "step": 697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.60840890354493e-05,
+      "loss": 6.8157,
+      "step": 698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.606347897774114e-05,
+      "loss": 6.6476,
+      "step": 699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.604286892003299e-05,
+      "loss": 6.6171,
+      "step": 700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.602225886232481e-05,
+      "loss": 6.9615,
+      "step": 701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.600164880461666e-05,
+      "loss": 7.0158,
+      "step": 702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.598103874690849e-05,
+      "loss": 7.0117,
+      "step": 703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.596042868920033e-05,
+      "loss": 7.2157,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.593981863149217e-05,
+      "loss": 6.9422,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.591920857378402e-05,
+      "loss": 7.1864,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.589859851607586e-05,
+      "loss": 7.0463,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.587798845836769e-05,
+      "loss": 7.1544,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.585737840065952e-05,
+      "loss": 6.8326,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.583676834295136e-05,
+      "loss": 7.0521,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.581615828524321e-05,
+      "loss": 6.981,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.579554822753504e-05,
+      "loss": 6.9879,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.577493816982688e-05,
+      "loss": 6.909,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.575432811211872e-05,
+      "loss": 7.1222,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.573371805441055e-05,
+      "loss": 7.0187,
+      "step": 715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.57131079967024e-05,
+      "loss": 6.8798,
+      "step": 716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.569249793899423e-05,
+      "loss": 6.8435,
+      "step": 717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.567188788128607e-05,
+      "loss": 7.0416,
+      "step": 718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.565127782357791e-05,
+      "loss": 6.789,
+      "step": 719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.563066776586976e-05,
+      "loss": 6.9611,
+      "step": 720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.561005770816159e-05,
+      "loss": 7.0588,
+      "step": 721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.558944765045342e-05,
+      "loss": 7.1861,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.556883759274526e-05,
+      "loss": 7.0159,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.55482275350371e-05,
+      "loss": 6.9765,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.552761747732895e-05,
+      "loss": 6.8697,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.550700741962078e-05,
+      "loss": 6.7586,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.548639736191262e-05,
+      "loss": 6.5953,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.546578730420445e-05,
+      "loss": 6.9291,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.54451772464963e-05,
+      "loss": 6.7768,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.542456718878814e-05,
+      "loss": 7.0097,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.540395713107997e-05,
+      "loss": 6.8871,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.538334707337181e-05,
+      "loss": 6.7709,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.536273701566365e-05,
+      "loss": 6.9573,
+      "step": 733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.53421269579555e-05,
+      "loss": 6.9672,
+      "step": 734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.532151690024733e-05,
+      "loss": 6.8999,
+      "step": 735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.530090684253916e-05,
+      "loss": 6.8073,
+      "step": 736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.5280296784831e-05,
+      "loss": 6.8856,
+      "step": 737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.525968672712284e-05,
+      "loss": 6.8825,
+      "step": 738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.523907666941469e-05,
+      "loss": 6.9071,
+      "step": 739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.521846661170652e-05,
+      "loss": 6.9184,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.519785655399836e-05,
+      "loss": 7.0114,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.517724649629019e-05,
+      "loss": 6.982,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.515663643858203e-05,
+      "loss": 6.963,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.513602638087386e-05,
+      "loss": 7.0866,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.51154163231657e-05,
+      "loss": 6.9062,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.509480626545755e-05,
+      "loss": 6.5833,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.50741962077494e-05,
+      "loss": 6.6821,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.505358615004122e-05,
+      "loss": 6.6342,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.503297609233305e-05,
+      "loss": 6.4835,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.50123660346249e-05,
+      "loss": 6.368,
+      "step": 750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.499175597691674e-05,
+      "loss": 7.1185,
+      "step": 751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.497114591920858e-05,
+      "loss": 7.0435,
+      "step": 752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.495053586150041e-05,
+      "loss": 6.9648,
+      "step": 753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.492992580379226e-05,
+      "loss": 6.9231,
+      "step": 754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.49093157460841e-05,
+      "loss": 6.9011,
+      "step": 755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.488870568837593e-05,
+      "loss": 7.0139,
+      "step": 756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.486809563066777e-05,
+      "loss": 6.9894,
+      "step": 757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.48474855729596e-05,
+      "loss": 7.069,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.482687551525145e-05,
+      "loss": 6.68,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.480626545754329e-05,
+      "loss": 6.7537,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.478565539983513e-05,
+      "loss": 6.8721,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.476504534212696e-05,
+      "loss": 6.8874,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.474443528441879e-05,
+      "loss": 6.9849,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.472382522671064e-05,
+      "loss": 6.8239,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.470321516900248e-05,
+      "loss": 6.8274,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.468260511129432e-05,
+      "loss": 6.6282,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.466199505358615e-05,
+      "loss": 6.9672,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.4641384995878e-05,
+      "loss": 6.8185,
+      "step": 768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.462077493816983e-05,
+      "loss": 7.0341,
+      "step": 769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.460016488046167e-05,
+      "loss": 6.7661,
+      "step": 770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.457955482275351e-05,
+      "loss": 6.9149,
+      "step": 771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.455894476504534e-05,
+      "loss": 6.9028,
+      "step": 772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.453833470733719e-05,
+      "loss": 6.7691,
+      "step": 773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.451772464962903e-05,
+      "loss": 6.9149,
+      "step": 774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.449711459192086e-05,
+      "loss": 6.7891,
+      "step": 775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.447650453421269e-05,
+      "loss": 6.9535,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.445589447650453e-05,
+      "loss": 6.809,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.443528441879638e-05,
+      "loss": 6.8816,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.441467436108822e-05,
+      "loss": 6.9042,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.439406430338006e-05,
+      "loss": 6.8693,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.437345424567189e-05,
+      "loss": 6.9413,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.435284418796374e-05,
+      "loss": 6.8283,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.433223413025557e-05,
+      "loss": 6.9375,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.431162407254741e-05,
+      "loss": 6.936,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.429101401483924e-05,
+      "loss": 6.7873,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.427040395713108e-05,
+      "loss": 6.8551,
+      "step": 786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.424979389942293e-05,
+      "loss": 6.6501,
+      "step": 787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.422918384171477e-05,
+      "loss": 7.0524,
+      "step": 788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.42085737840066e-05,
+      "loss": 6.8138,
+      "step": 789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.418796372629843e-05,
+      "loss": 6.9682,
+      "step": 790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.416735366859027e-05,
+      "loss": 6.8771,
+      "step": 791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.414674361088212e-05,
+      "loss": 6.5622,
+      "step": 792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.412613355317396e-05,
+      "loss": 6.9002,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.41055234954658e-05,
+      "loss": 6.9118,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.408491343775763e-05,
+      "loss": 6.7161,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.406430338004946e-05,
+      "loss": 6.7147,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.40436933223413e-05,
+      "loss": 6.3663,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.402308326463315e-05,
+      "loss": 6.6257,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.400247320692498e-05,
+      "loss": 6.6646,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.398186314921682e-05,
+      "loss": 6.5045,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.396125309150866e-05,
+      "loss": 7.2057,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.394064303380051e-05,
+      "loss": 7.0751,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.392003297609234e-05,
+      "loss": 7.1096,
+      "step": 803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.389942291838417e-05,
+      "loss": 6.9131,
+      "step": 804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.387881286067601e-05,
+      "loss": 7.1441,
+      "step": 805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.385820280296785e-05,
+      "loss": 7.019,
+      "step": 806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.38375927452597e-05,
+      "loss": 6.9014,
+      "step": 807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.381698268755153e-05,
+      "loss": 6.8237,
+      "step": 808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.379637262984337e-05,
+      "loss": 7.0874,
+      "step": 809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.37757625721352e-05,
+      "loss": 6.9981,
+      "step": 810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.375515251442704e-05,
+      "loss": 6.9141,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.373454245671889e-05,
+      "loss": 7.0097,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.371393239901072e-05,
+      "loss": 6.8919,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.369332234130256e-05,
+      "loss": 6.8607,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.36727122835944e-05,
+      "loss": 6.8775,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.365210222588623e-05,
+      "loss": 6.9519,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.363149216817806e-05,
+      "loss": 6.7872,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.361088211046991e-05,
+      "loss": 6.9194,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.359027205276175e-05,
+      "loss": 7.0224,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.35696619950536e-05,
+      "loss": 6.8667,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.354905193734544e-05,
+      "loss": 6.9883,
+      "step": 821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.352844187963727e-05,
+      "loss": 6.875,
+      "step": 822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.35078318219291e-05,
+      "loss": 7.054,
+      "step": 823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.348722176422094e-05,
+      "loss": 6.9536,
+      "step": 824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.346661170651278e-05,
+      "loss": 6.953,
+      "step": 825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.344600164880463e-05,
+      "loss": 6.7752,
+      "step": 826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.342539159109646e-05,
+      "loss": 6.9207,
+      "step": 827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.34047815333883e-05,
+      "loss": 6.76,
+      "step": 828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.338417147568014e-05,
+      "loss": 6.9268,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.336356141797197e-05,
+      "loss": 6.906,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.33429513602638e-05,
+      "loss": 7.0418,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.332234130255565e-05,
+      "loss": 6.9929,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.330173124484749e-05,
+      "loss": 6.874,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.328112118713933e-05,
+      "loss": 6.5568,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.326051112943118e-05,
+      "loss": 6.6961,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.323990107172301e-05,
+      "loss": 6.851,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.321929101401484e-05,
+      "loss": 6.9095,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.319868095630668e-05,
+      "loss": 6.7053,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.317807089859852e-05,
+      "loss": 6.711,
+      "step": 839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.315746084089035e-05,
+      "loss": 6.8332,
+      "step": 840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.31368507831822e-05,
+      "loss": 7.0099,
+      "step": 841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.311624072547404e-05,
+      "loss": 6.9028,
+      "step": 842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.309563066776587e-05,
+      "loss": 6.6767,
+      "step": 843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.307502061005771e-05,
+      "loss": 7.0714,
+      "step": 844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.305441055234954e-05,
+      "loss": 6.5427,
+      "step": 845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.303380049464139e-05,
+      "loss": 6.5091,
+      "step": 846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.301319043693323e-05,
+      "loss": 6.7675,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.299258037922507e-05,
+      "loss": 6.8377,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.297197032151692e-05,
+      "loss": 6.4041,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.295136026380875e-05,
+      "loss": 6.3575,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.293075020610058e-05,
+      "loss": 6.9923,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.291014014839242e-05,
+      "loss": 7.0345,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.288953009068426e-05,
+      "loss": 6.8224,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.286892003297609e-05,
+      "loss": 7.0084,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.284830997526794e-05,
+      "loss": 6.9961,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.282769991755978e-05,
+      "loss": 6.8788,
+      "step": 856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.280708985985161e-05,
+      "loss": 6.8649,
+      "step": 857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.278647980214345e-05,
+      "loss": 6.946,
+      "step": 858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.276586974443528e-05,
+      "loss": 6.8718,
+      "step": 859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.274525968672713e-05,
+      "loss": 7.0052,
+      "step": 860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.272464962901897e-05,
+      "loss": 6.7953,
+      "step": 861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.270403957131081e-05,
+      "loss": 6.7643,
+      "step": 862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.268342951360264e-05,
+      "loss": 6.6812,
+      "step": 863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.266281945589447e-05,
+      "loss": 6.7101,
+      "step": 864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.264220939818632e-05,
+      "loss": 6.8645,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.262159934047816e-05,
+      "loss": 6.9544,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.260098928277e-05,
+      "loss": 6.7718,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.258037922506183e-05,
+      "loss": 6.7881,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.255976916735368e-05,
+      "loss": 6.852,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.25391591096455e-05,
+      "loss": 6.8457,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.251854905193735e-05,
+      "loss": 6.8808,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.249793899422918e-05,
+      "loss": 6.8655,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.247732893652102e-05,
+      "loss": 6.8937,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.245671887881287e-05,
+      "loss": 6.8309,
+      "step": 874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.243610882110471e-05,
+      "loss": 7.1575,
+      "step": 875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.241549876339655e-05,
+      "loss": 6.8417,
+      "step": 876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.239488870568838e-05,
+      "loss": 7.055,
+      "step": 877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.237427864798021e-05,
+      "loss": 6.9065,
+      "step": 878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.235366859027206e-05,
+      "loss": 6.7328,
+      "step": 879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.23330585325639e-05,
+      "loss": 6.912,
+      "step": 880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.231244847485574e-05,
+      "loss": 6.945,
+      "step": 881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.229183841714757e-05,
+      "loss": 6.6611,
+      "step": 882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.227122835943942e-05,
+      "loss": 6.911,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.225061830173125e-05,
+      "loss": 6.8797,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.223000824402309e-05,
+      "loss": 6.9586,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.220939818631492e-05,
+      "loss": 7.1094,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.218878812860676e-05,
+      "loss": 6.9738,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.21681780708986e-05,
+      "loss": 6.7878,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.214756801319045e-05,
+      "loss": 6.7673,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.212695795548228e-05,
+      "loss": 6.7735,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.210634789777411e-05,
+      "loss": 6.7932,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.208573784006595e-05,
+      "loss": 6.9499,
+      "step": 892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.20651277823578e-05,
+      "loss": 6.8664,
+      "step": 893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.204451772464964e-05,
+      "loss": 6.6964,
+      "step": 894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.202390766694147e-05,
+      "loss": 6.6728,
+      "step": 895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.200329760923331e-05,
+      "loss": 6.5146,
+      "step": 896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.198268755152516e-05,
+      "loss": 6.437,
+      "step": 897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.196207749381698e-05,
+      "loss": 6.8636,
+      "step": 898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.194146743610883e-05,
+      "loss": 6.2736,
+      "step": 899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.192085737840066e-05,
+      "loss": 6.2022,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.19002473206925e-05,
+      "loss": 6.8252,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.187963726298434e-05,
+      "loss": 7.0397,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.185902720527619e-05,
+      "loss": 7.0932,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.183841714756802e-05,
+      "loss": 6.8398,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.181780708985985e-05,
+      "loss": 7.1522,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.179719703215169e-05,
+      "loss": 7.1035,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.177658697444353e-05,
+      "loss": 6.8955,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.175597691673538e-05,
+      "loss": 6.775,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.173536685902721e-05,
+      "loss": 6.939,
+      "step": 909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.171475680131905e-05,
+      "loss": 6.9065,
+      "step": 910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.169414674361088e-05,
+      "loss": 7.0268,
+      "step": 911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.167353668590272e-05,
+      "loss": 6.8254,
+      "step": 912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.165292662819457e-05,
+      "loss": 6.7227,
+      "step": 913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.16323165704864e-05,
+      "loss": 7.0315,
+      "step": 914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.161170651277824e-05,
+      "loss": 6.8389,
+      "step": 915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.159109645507008e-05,
+      "loss": 6.7625,
+      "step": 916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.157048639736191e-05,
+      "loss": 6.7653,
+      "step": 917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.154987633965374e-05,
+      "loss": 6.8669,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.152926628194559e-05,
+      "loss": 6.8483,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.150865622423743e-05,
+      "loss": 6.6915,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.148804616652927e-05,
+      "loss": 6.8683,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.146743610882112e-05,
+      "loss": 6.6229,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.144682605111295e-05,
+      "loss": 6.7383,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.142621599340479e-05,
+      "loss": 6.9092,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.140560593569662e-05,
+      "loss": 6.907,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.138499587798846e-05,
+      "loss": 6.9491,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.13643858202803e-05,
+      "loss": 6.7509,
+      "step": 927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.134377576257214e-05,
+      "loss": 6.7511,
+      "step": 928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.132316570486398e-05,
+      "loss": 6.7765,
+      "step": 929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.130255564715582e-05,
+      "loss": 6.8292,
+      "step": 930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.128194558944765e-05,
+      "loss": 6.9093,
+      "step": 931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.126133553173948e-05,
+      "loss": 7.0032,
+      "step": 932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.124072547403133e-05,
+      "loss": 6.6126,
+      "step": 933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.122011541632317e-05,
+      "loss": 6.9268,
+      "step": 934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.119950535861501e-05,
+      "loss": 6.8153,
+      "step": 935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.117889530090684e-05,
+      "loss": 6.9395,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.115828524319869e-05,
+      "loss": 6.8741,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.113767518549052e-05,
+      "loss": 6.7412,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.111706512778236e-05,
+      "loss": 6.539,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.10964550700742e-05,
+      "loss": 6.819,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.107584501236603e-05,
+      "loss": 6.8212,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.105523495465788e-05,
+      "loss": 6.6822,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.103462489694972e-05,
+      "loss": 6.9584,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.101401483924156e-05,
+      "loss": 6.6507,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.09934047815334e-05,
+      "loss": 6.934,
+      "step": 945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.097279472382522e-05,
+      "loss": 6.602,
+      "step": 946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.095218466611707e-05,
+      "loss": 6.2367,
+      "step": 947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.093157460840891e-05,
+      "loss": 6.725,
+      "step": 948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.091096455070075e-05,
+      "loss": 6.7638,
+      "step": 949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.089035449299258e-05,
+      "loss": 5.8387,
+      "step": 950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.086974443528443e-05,
+      "loss": 6.9744,
+      "step": 951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.084913437757626e-05,
+      "loss": 7.032,
+      "step": 952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.08285243198681e-05,
+      "loss": 7.0119,
+      "step": 953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.080791426215994e-05,
+      "loss": 6.8657,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078730420445177e-05,
+      "loss": 6.8281,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076669414674362e-05,
+      "loss": 6.8987,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074608408903546e-05,
+      "loss": 6.9349,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072547403132729e-05,
+      "loss": 6.7392,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070486397361912e-05,
+      "loss": 6.9836,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068425391591096e-05,
+      "loss": 6.7774,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06636438582028e-05,
+      "loss": 6.6995,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064303380049465e-05,
+      "loss": 6.8346,
+      "step": 962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062242374278649e-05,
+      "loss": 6.7985,
+      "step": 963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.060181368507832e-05,
+      "loss": 6.7641,
+      "step": 964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.058120362737015e-05,
+      "loss": 6.7869,
+      "step": 965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0560593569662e-05,
+      "loss": 6.6657,
+      "step": 966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.053998351195384e-05,
+      "loss": 6.8386,
+      "step": 967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051937345424568e-05,
+      "loss": 6.7846,
+      "step": 968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049876339653751e-05,
+      "loss": 6.9469,
+      "step": 969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047815333882936e-05,
+      "loss": 6.6813,
+      "step": 970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04575432811212e-05,
+      "loss": 6.8754,
+      "step": 971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043693322341303e-05,
+      "loss": 6.8385,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.041632316570486e-05,
+      "loss": 6.8949,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.03957131079967e-05,
+      "loss": 7.0168,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.037510305028855e-05,
+      "loss": 6.8606,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.035449299258039e-05,
+      "loss": 6.7158,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.033388293487223e-05,
+      "loss": 7.1046,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.031327287716406e-05,
+      "loss": 6.7823,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029266281945589e-05,
+      "loss": 6.7202,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027205276174774e-05,
+      "loss": 6.6599,
+      "step": 980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025144270403958e-05,
+      "loss": 6.6146,
+      "step": 981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023083264633141e-05,
+      "loss": 6.6924,
+      "step": 982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021022258862325e-05,
+      "loss": 6.9321,
+      "step": 983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01896125309151e-05,
+      "loss": 6.8577,
+      "step": 984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016900247320693e-05,
+      "loss": 6.695,
+      "step": 985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014839241549877e-05,
+      "loss": 6.8493,
+      "step": 986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01277823577906e-05,
+      "loss": 6.8749,
+      "step": 987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010717230008244e-05,
+      "loss": 6.7436,
+      "step": 988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008656224237429e-05,
+      "loss": 7.0578,
+      "step": 989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006595218466613e-05,
+      "loss": 6.6429,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.004534212695796e-05,
+      "loss": 6.5682,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.00247320692498e-05,
+      "loss": 6.9671,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.000412201154163e-05,
+      "loss": 6.6607,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.998351195383347e-05,
+      "loss": 6.6917,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.996290189612532e-05,
+      "loss": 6.6738,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.994229183841715e-05,
+      "loss": 6.2442,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.992168178070899e-05,
+      "loss": 6.3304,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.990107172300083e-05,
+      "loss": 6.9551,
+      "step": 998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.988046166529266e-05,
+      "loss": 6.5172,
+      "step": 999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.985985160758451e-05,
+      "loss": 6.2294,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 10.509584426879883,
+      "eval_runtime": 1319.6205,
+      "eval_samples_per_second": 2.002,
+      "eval_steps_per_second": 0.251,
+      "eval_wer": 1.0,
+      "step": 1000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.983924154987634e-05,
+      "loss": 7.1895,
+      "step": 1001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.981863149216818e-05,
+      "loss": 6.9751,
+      "step": 1002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.979802143446002e-05,
+      "loss": 6.9608,
+      "step": 1003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.977741137675187e-05,
+      "loss": 7.0492,
+      "step": 1004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.97568013190437e-05,
+      "loss": 7.1122,
+      "step": 1005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.973619126133553e-05,
+      "loss": 6.9864,
+      "step": 1006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.971558120362737e-05,
+      "loss": 6.8662,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.969497114591921e-05,
+      "loss": 6.8473,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.967436108821106e-05,
+      "loss": 6.8182,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.965375103050289e-05,
+      "loss": 6.7838,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.963314097279473e-05,
+      "loss": 6.8163,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.961253091508657e-05,
+      "loss": 6.8971,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.95919208573784e-05,
+      "loss": 6.7517,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.957131079967023e-05,
+      "loss": 7.0059,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.955070074196208e-05,
+      "loss": 6.8824,
+      "step": 1015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.953009068425392e-05,
+      "loss": 6.7466,
+      "step": 1016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.950948062654576e-05,
+      "loss": 6.8158,
+      "step": 1017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.948887056883761e-05,
+      "loss": 6.7501,
+      "step": 1018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.946826051112944e-05,
+      "loss": 7.0113,
+      "step": 1019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.944765045342127e-05,
+      "loss": 7.001,
+      "step": 1020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.942704039571311e-05,
+      "loss": 6.8119,
+      "step": 1021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.940643033800495e-05,
+      "loss": 6.6904,
+      "step": 1022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.938582028029678e-05,
+      "loss": 6.638,
+      "step": 1023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.936521022258863e-05,
+      "loss": 6.8542,
+      "step": 1024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.934460016488047e-05,
+      "loss": 6.8893,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.93239901071723e-05,
+      "loss": 6.7111,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.930338004946414e-05,
+      "loss": 6.8946,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.928276999175597e-05,
+      "loss": 6.906,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.926215993404782e-05,
+      "loss": 6.7917,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.924154987633966e-05,
+      "loss": 6.8436,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.92209398186315e-05,
+      "loss": 6.7664,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.920032976092333e-05,
+      "loss": 6.8412,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.917971970321516e-05,
+      "loss": 6.7491,
+      "step": 1033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.915910964550701e-05,
+      "loss": 6.5337,
+      "step": 1034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.913849958779885e-05,
+      "loss": 6.7558,
+      "step": 1035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.91178895300907e-05,
+      "loss": 6.8304,
+      "step": 1036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.909727947238252e-05,
+      "loss": 6.5428,
+      "step": 1037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.907666941467437e-05,
+      "loss": 6.7771,
+      "step": 1038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.905605935696621e-05,
+      "loss": 6.7969,
+      "step": 1039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.903544929925804e-05,
+      "loss": 6.7782,
+      "step": 1040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.901483924154988e-05,
+      "loss": 6.7926,
+      "step": 1041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.899422918384171e-05,
+      "loss": 6.7056,
+      "step": 1042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.897361912613356e-05,
+      "loss": 6.7019,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.89530090684254e-05,
+      "loss": 6.6115,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.893239901071724e-05,
+      "loss": 6.5894,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.891178895300907e-05,
+      "loss": 6.7535,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.88911788953009e-05,
+      "loss": 6.6846,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.887056883759275e-05,
+      "loss": 6.1387,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.884995877988459e-05,
+      "loss": 6.2811,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.882934872217643e-05,
+      "loss": 6.5854,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.880873866446826e-05,
+      "loss": 7.3416,
+      "step": 1051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.87881286067601e-05,
+      "loss": 6.8758,
+      "step": 1052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.876751854905194e-05,
+      "loss": 6.9227,
+      "step": 1053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.874690849134378e-05,
+      "loss": 6.8716,
+      "step": 1054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.872629843363561e-05,
+      "loss": 6.9722,
+      "step": 1055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.870568837592745e-05,
+      "loss": 6.9472,
+      "step": 1056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.86850783182193e-05,
+      "loss": 6.993,
+      "step": 1057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.866446826051114e-05,
+      "loss": 6.8096,
+      "step": 1058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.864385820280298e-05,
+      "loss": 6.8049,
+      "step": 1059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.862324814509481e-05,
+      "loss": 6.8168,
+      "step": 1060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.860263808738664e-05,
+      "loss": 6.8581,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.858202802967849e-05,
+      "loss": 6.8689,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.856141797197033e-05,
+      "loss": 7.0835,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.854080791426217e-05,
+      "loss": 6.7236,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.8520197856554e-05,
+      "loss": 6.9696,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.849958779884585e-05,
+      "loss": 6.5983,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.847897774113768e-05,
+      "loss": 7.0441,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.845836768342952e-05,
+      "loss": 6.803,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.843775762572135e-05,
+      "loss": 6.7793,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.841714756801319e-05,
+      "loss": 6.8664,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.839653751030504e-05,
+      "loss": 6.7362,
+      "step": 1071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.837592745259688e-05,
+      "loss": 6.6317,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.835531739488871e-05,
+      "loss": 6.8396,
+      "step": 1073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.833470733718054e-05,
+      "loss": 6.7941,
+      "step": 1074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.831409727947238e-05,
+      "loss": 6.7505,
+      "step": 1075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.829348722176423e-05,
+      "loss": 6.8142,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.827287716405607e-05,
+      "loss": 6.7486,
+      "step": 1077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.82522671063479e-05,
+      "loss": 6.6813,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.823165704863974e-05,
+      "loss": 6.9219,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.821104699093157e-05,
+      "loss": 6.8199,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.819043693322342e-05,
+      "loss": 6.8333,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.816982687551526e-05,
+      "loss": 6.7589,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.814921681780709e-05,
+      "loss": 6.6839,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.812860676009893e-05,
+      "loss": 6.8829,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.810799670239078e-05,
+      "loss": 6.6602,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.808738664468262e-05,
+      "loss": 6.8833,
+      "step": 1086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.806677658697445e-05,
+      "loss": 6.9797,
+      "step": 1087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.804616652926628e-05,
+      "loss": 6.8678,
+      "step": 1088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.802555647155812e-05,
+      "loss": 6.8566,
+      "step": 1089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.800494641384997e-05,
+      "loss": 6.6994,
+      "step": 1090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.798433635614181e-05,
+      "loss": 6.8069,
+      "step": 1091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.796372629843364e-05,
+      "loss": 6.8855,
+      "step": 1092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.794311624072548e-05,
+      "loss": 6.6551,
+      "step": 1093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.792250618301731e-05,
+      "loss": 6.3406,
+      "step": 1094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.790189612530915e-05,
+      "loss": 6.644,
+      "step": 1095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.7881286067601e-05,
+      "loss": 6.4605,
+      "step": 1096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.786067600989283e-05,
+      "loss": 6.5818,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.784006595218467e-05,
+      "loss": 6.6784,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.781945589447651e-05,
+      "loss": 6.4825,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.779884583676834e-05,
+      "loss": 6.1243,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.777823577906017e-05,
+      "loss": 6.7527,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.775762572135202e-05,
+      "loss": 7.0371,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.773701566364386e-05,
+      "loss": 6.9179,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.77164056059357e-05,
+      "loss": 6.8311,
+      "step": 1104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.769579554822755e-05,
+      "loss": 6.7603,
+      "step": 1105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.767518549051938e-05,
+      "loss": 6.7055,
+      "step": 1106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.765457543281122e-05,
+      "loss": 6.8738,
+      "step": 1107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.763396537510305e-05,
+      "loss": 6.7932,
+      "step": 1108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.76133553173949e-05,
+      "loss": 6.922,
+      "step": 1109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.759274525968672e-05,
+      "loss": 6.8378,
+      "step": 1110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.757213520197857e-05,
+      "loss": 6.6728,
+      "step": 1111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.755152514427041e-05,
+      "loss": 6.87,
+      "step": 1112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.753091508656225e-05,
+      "loss": 6.7664,
+      "step": 1113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.751030502885408e-05,
+      "loss": 6.786,
+      "step": 1114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.748969497114591e-05,
+      "loss": 6.834,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.746908491343776e-05,
+      "loss": 6.5574,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.74484748557296e-05,
+      "loss": 6.6804,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.742786479802144e-05,
+      "loss": 6.871,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.740725474031329e-05,
+      "loss": 6.8164,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.738664468260512e-05,
+      "loss": 6.6891,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.736603462489695e-05,
+      "loss": 6.7652,
+      "step": 1121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.734542456718879e-05,
+      "loss": 6.7018,
+      "step": 1122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.732481450948063e-05,
+      "loss": 6.5966,
+      "step": 1123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.730420445177246e-05,
+      "loss": 6.7043,
+      "step": 1124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.728359439406431e-05,
+      "loss": 6.6884,
+      "step": 1125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.726298433635615e-05,
+      "loss": 6.5879,
+      "step": 1126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.724237427864798e-05,
+      "loss": 6.6834,
+      "step": 1127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.722176422093982e-05,
+      "loss": 7.0349,
+      "step": 1128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.720115416323165e-05,
+      "loss": 6.7407,
+      "step": 1129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.71805441055235e-05,
+      "loss": 6.8786,
+      "step": 1130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.715993404781534e-05,
+      "loss": 6.799,
+      "step": 1131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.713932399010718e-05,
+      "loss": 6.7843,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.711871393239901e-05,
+      "loss": 6.6275,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.709810387469086e-05,
+      "loss": 6.6047,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.707749381698269e-05,
+      "loss": 6.7882,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.705688375927453e-05,
+      "loss": 6.7289,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.703627370156637e-05,
+      "loss": 6.7386,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.70156636438582e-05,
+      "loss": 6.7633,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.699505358615005e-05,
+      "loss": 6.505,
+      "step": 1139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.697444352844189e-05,
+      "loss": 6.7294,
+      "step": 1140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.695383347073372e-05,
+      "loss": 6.7089,
+      "step": 1141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.693322341302555e-05,
+      "loss": 7.0219,
+      "step": 1142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.691261335531739e-05,
+      "loss": 6.78,
+      "step": 1143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.689200329760924e-05,
+      "loss": 6.7707,
+      "step": 1144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.687139323990108e-05,
+      "loss": 6.5602,
+      "step": 1145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.685078318219292e-05,
+      "loss": 6.5457,
+      "step": 1146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.683017312448475e-05,
+      "loss": 6.5214,
+      "step": 1147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.680956306677658e-05,
+      "loss": 6.4071,
+      "step": 1148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.678895300906843e-05,
+      "loss": 6.2183,
+      "step": 1149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.676834295136027e-05,
+      "loss": 5.9782,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.674773289365211e-05,
+      "loss": 6.8632,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.672712283594394e-05,
+      "loss": 7.0484,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.670651277823579e-05,
+      "loss": 6.7587,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.668590272052763e-05,
+      "loss": 6.9431,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.666529266281946e-05,
+      "loss": 6.9344,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.664468260511129e-05,
+      "loss": 6.8539,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.662407254740313e-05,
+      "loss": 6.7283,
+      "step": 1157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.660346248969498e-05,
+      "loss": 6.9349,
+      "step": 1158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.658285243198682e-05,
+      "loss": 6.5559,
+      "step": 1159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.656224237427866e-05,
+      "loss": 6.6893,
+      "step": 1160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.654163231657049e-05,
+      "loss": 6.8265,
+      "step": 1161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.652102225886232e-05,
+      "loss": 6.6181,
+      "step": 1162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.650041220115417e-05,
+      "loss": 6.6978,
+      "step": 1163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.647980214344601e-05,
+      "loss": 6.6792,
+      "step": 1164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.645919208573784e-05,
+      "loss": 6.8079,
+      "step": 1165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.643858202802968e-05,
+      "loss": 6.804,
+      "step": 1166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.641797197032153e-05,
+      "loss": 6.8039,
+      "step": 1167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 8.639736191261336e-05,
+      "loss": 6.4179,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.63767518549052e-05,
+      "loss": 6.7071,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.635614179719703e-05,
+      "loss": 6.784,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.633553173948887e-05,
+      "loss": 6.7009,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.631492168178072e-05,
+      "loss": 6.5875,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.629431162407256e-05,
+      "loss": 6.9206,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.627370156636439e-05,
+      "loss": 6.6054,
+      "step": 1174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.625309150865622e-05,
+      "loss": 6.5818,
+      "step": 1175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.623248145094806e-05,
+      "loss": 6.6984,
+      "step": 1176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.62118713932399e-05,
+      "loss": 6.6868,
+      "step": 1177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.619126133553175e-05,
+      "loss": 6.6141,
+      "step": 1178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.617065127782358e-05,
+      "loss": 6.4755,
+      "step": 1179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.615004122011542e-05,
+      "loss": 6.8743,
+      "step": 1180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.612943116240727e-05,
+      "loss": 6.6367,
+      "step": 1181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.61088211046991e-05,
+      "loss": 6.6419,
+      "step": 1182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.608821104699094e-05,
+      "loss": 6.6622,
+      "step": 1183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.606760098928277e-05,
+      "loss": 6.5273,
+      "step": 1184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.604699093157461e-05,
+      "loss": 7.0137,
+      "step": 1185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 8.602638087386646e-05,
+      "loss": 6.62,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.60057708161583e-05,
+      "loss": 6.7784,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.598516075845013e-05,
+      "loss": 6.6826,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.596455070074196e-05,
+      "loss": 6.6295,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.59439406430338e-05,
+      "loss": 6.615,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.592333058532564e-05,
+      "loss": 6.7134,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.590272052761749e-05,
+      "loss": 6.4659,
+      "step": 1192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.588211046990932e-05,
+      "loss": 6.4737,
+      "step": 1193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.586150041220116e-05,
+      "loss": 6.565,
+      "step": 1194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.584089035449299e-05,
+      "loss": 6.2785,
+      "step": 1195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.582028029678483e-05,
+      "loss": 6.4619,
+      "step": 1196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.579967023907666e-05,
+      "loss": 6.2607,
+      "step": 1197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.577906018136851e-05,
+      "loss": 6.2957,
+      "step": 1198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.575845012366035e-05,
+      "loss": 5.925,
+      "step": 1199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.57378400659522e-05,
+      "loss": 5.5422,
+      "step": 1200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.571723000824404e-05,
+      "loss": 6.846,
+      "step": 1201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.569661995053587e-05,
+      "loss": 6.6924,
+      "step": 1202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.56760098928277e-05,
+      "loss": 6.8836,
+      "step": 1203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.565539983511954e-05,
+      "loss": 6.7017,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.563478977741138e-05,
+      "loss": 6.7779,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.561417971970323e-05,
+      "loss": 6.6439,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.559356966199506e-05,
+      "loss": 6.8293,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.55729596042869e-05,
+      "loss": 6.7646,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.555234954657873e-05,
+      "loss": 6.6797,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.553173948887057e-05,
+      "loss": 6.688,
+      "step": 1210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.55111294311624e-05,
+      "loss": 6.7483,
+      "step": 1211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.549051937345425e-05,
+      "loss": 6.4949,
+      "step": 1212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.546990931574609e-05,
+      "loss": 6.7115,
+      "step": 1213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.544929925803793e-05,
+      "loss": 6.9018,
+      "step": 1214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.542868920032976e-05,
+      "loss": 6.6845,
+      "step": 1215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.54080791426216e-05,
+      "loss": 6.6071,
+      "step": 1216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.538746908491344e-05,
+      "loss": 6.6471,
+      "step": 1217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.536685902720528e-05,
+      "loss": 6.373,
+      "step": 1218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.534624896949712e-05,
+      "loss": 6.6219,
+      "step": 1219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.532563891178895e-05,
+      "loss": 6.7948,
+      "step": 1220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.53050288540808e-05,
+      "loss": 6.7108,
+      "step": 1221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.528441879637263e-05,
+      "loss": 6.7589,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.526380873866447e-05,
+      "loss": 6.8153,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.524319868095631e-05,
+      "loss": 6.6347,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.522258862324814e-05,
+      "loss": 6.5844,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.520197856553999e-05,
+      "loss": 6.6931,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.518136850783183e-05,
+      "loss": 6.6448,
+      "step": 1227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.516075845012367e-05,
+      "loss": 6.6149,
+      "step": 1228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.51401483924155e-05,
+      "loss": 6.6538,
+      "step": 1229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.511953833470733e-05,
+      "loss": 6.966,
+      "step": 1230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.509892827699918e-05,
+      "loss": 6.7125,
+      "step": 1231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.507831821929102e-05,
+      "loss": 6.4244,
+      "step": 1232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.505770816158286e-05,
+      "loss": 6.6848,
+      "step": 1233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.50370981038747e-05,
+      "loss": 6.5554,
+      "step": 1234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.501648804616654e-05,
+      "loss": 6.7319,
+      "step": 1235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.499587798845837e-05,
+      "loss": 6.6032,
+      "step": 1236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.497526793075021e-05,
+      "loss": 6.7213,
+      "step": 1237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.495465787304205e-05,
+      "loss": 6.4107,
+      "step": 1238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.493404781533388e-05,
+      "loss": 6.7667,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.491343775762573e-05,
+      "loss": 6.6309,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.489282769991757e-05,
+      "loss": 6.6216,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.48722176422094e-05,
+      "loss": 6.7474,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.485160758450123e-05,
+      "loss": 6.5319,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.483099752679307e-05,
+      "loss": 6.4562,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.481038746908492e-05,
+      "loss": 6.3578,
+      "step": 1245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.478977741137676e-05,
+      "loss": 6.4926,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.47691673536686e-05,
+      "loss": 6.4763,
+      "step": 1247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.474855729596043e-05,
+      "loss": 6.183,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.472794723825228e-05,
+      "loss": 6.0288,
+      "step": 1249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.47073371805441e-05,
+      "loss": 6.1746,
+      "step": 1250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.468672712283595e-05,
+      "loss": 6.8434,
+      "step": 1251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.466611706512778e-05,
+      "loss": 6.6436,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.464550700741962e-05,
+      "loss": 6.7352,
+      "step": 1253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.462489694971147e-05,
+      "loss": 6.7205,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.460428689200331e-05,
+      "loss": 6.9031,
+      "step": 1255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.458367683429514e-05,
+      "loss": 6.6338,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.456306677658697e-05,
+      "loss": 6.6341,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.454245671887881e-05,
+      "loss": 6.7685,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.452184666117066e-05,
+      "loss": 6.6786,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.45012366034625e-05,
+      "loss": 6.8482,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.448062654575433e-05,
+      "loss": 6.7403,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.446001648804617e-05,
+      "loss": 6.6734,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.4439406430338e-05,
+      "loss": 6.6826,
+      "step": 1263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.441879637262985e-05,
+      "loss": 6.5286,
+      "step": 1264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.439818631492169e-05,
+      "loss": 6.7194,
+      "step": 1265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.437757625721352e-05,
+      "loss": 6.7763,
+      "step": 1266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.435696619950536e-05,
+      "loss": 6.5541,
+      "step": 1267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.43363561417972e-05,
+      "loss": 6.6596,
+      "step": 1268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.431574608408904e-05,
+      "loss": 6.5588,
+      "step": 1269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.429513602638088e-05,
+      "loss": 6.5542,
+      "step": 1270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.427452596867271e-05,
+      "loss": 6.4311,
+      "step": 1271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.425391591096455e-05,
+      "loss": 6.4367,
+      "step": 1272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.42333058532564e-05,
+      "loss": 6.7051,
+      "step": 1273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.421269579554824e-05,
+      "loss": 6.5136,
+      "step": 1274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.419208573784007e-05,
+      "loss": 6.6791,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.417147568013191e-05,
+      "loss": 6.772,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.415086562242374e-05,
+      "loss": 6.7937,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.413025556471559e-05,
+      "loss": 6.6365,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.410964550700743e-05,
+      "loss": 6.6913,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.408903544929926e-05,
+      "loss": 6.498,
+      "step": 1280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.40684253915911e-05,
+      "loss": 6.4998,
+      "step": 1281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.404781533388295e-05,
+      "loss": 6.5633,
+      "step": 1282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.402720527617478e-05,
+      "loss": 6.6948,
+      "step": 1283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.40065952184666e-05,
+      "loss": 6.5341,
+      "step": 1284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.398598516075845e-05,
+      "loss": 6.5084,
+      "step": 1285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.396537510305029e-05,
+      "loss": 6.69,
+      "step": 1286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.394476504534214e-05,
+      "loss": 6.6173,
+      "step": 1287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.392415498763398e-05,
+      "loss": 6.4869,
+      "step": 1288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.390354492992581e-05,
+      "loss": 6.6442,
+      "step": 1289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.388293487221764e-05,
+      "loss": 6.3683,
+      "step": 1290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.386232481450948e-05,
+      "loss": 6.5191,
+      "step": 1291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.384171475680132e-05,
+      "loss": 6.5717,
+      "step": 1292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.382110469909317e-05,
+      "loss": 6.5682,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.3800494641385e-05,
+      "loss": 6.3313,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.377988458367684e-05,
+      "loss": 6.3508,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.375927452596868e-05,
+      "loss": 6.1617,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.373866446826051e-05,
+      "loss": 6.1414,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.371805441055234e-05,
+      "loss": 5.7855,
+      "step": 1298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.369744435284419e-05,
+      "loss": 6.3681,
+      "step": 1299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.367683429513603e-05,
+      "loss": 5.561,
+      "step": 1300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.365622423742787e-05,
+      "loss": 6.8089,
+      "step": 1301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.363561417971972e-05,
+      "loss": 6.8368,
+      "step": 1302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.361500412201155e-05,
+      "loss": 6.8361,
+      "step": 1303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.359439406430338e-05,
+      "loss": 6.7604,
+      "step": 1304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.357378400659522e-05,
+      "loss": 6.7128,
+      "step": 1305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.355317394888706e-05,
+      "loss": 6.7931,
+      "step": 1306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.35325638911789e-05,
+      "loss": 6.459,
+      "step": 1307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.351195383347074e-05,
+      "loss": 6.66,
+      "step": 1308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.349134377576258e-05,
+      "loss": 6.4563,
+      "step": 1309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.347073371805441e-05,
+      "loss": 6.4806,
+      "step": 1310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.345012366034625e-05,
+      "loss": 6.65,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.342951360263808e-05,
+      "loss": 6.6799,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.340890354492993e-05,
+      "loss": 6.6662,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.338829348722177e-05,
+      "loss": 6.4375,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.336768342951361e-05,
+      "loss": 6.9393,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.334707337180544e-05,
+      "loss": 6.5135,
+      "step": 1316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.332646331409727e-05,
+      "loss": 6.3682,
+      "step": 1317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.330585325638912e-05,
+      "loss": 6.6544,
+      "step": 1318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.328524319868096e-05,
+      "loss": 6.5103,
+      "step": 1319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.32646331409728e-05,
+      "loss": 6.5362,
+      "step": 1320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.324402308326463e-05,
+      "loss": 6.3246,
+      "step": 1321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.322341302555648e-05,
+      "loss": 6.5399,
+      "step": 1322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.320280296784832e-05,
+      "loss": 6.6631,
+      "step": 1323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.318219291014015e-05,
+      "loss": 6.681,
+      "step": 1324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.3161582852432e-05,
+      "loss": 6.6926,
+      "step": 1325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.314097279472382e-05,
+      "loss": 6.4671,
+      "step": 1326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.312036273701567e-05,
+      "loss": 6.3436,
+      "step": 1327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.309975267930751e-05,
+      "loss": 6.5232,
+      "step": 1328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.307914262159935e-05,
+      "loss": 6.7289,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.305853256389118e-05,
+      "loss": 6.5073,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.303792250618301e-05,
+      "loss": 6.4028,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.301731244847486e-05,
+      "loss": 6.4198,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.29967023907667e-05,
+      "loss": 6.6198,
+      "step": 1333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.297609233305854e-05,
+      "loss": 6.6072,
+      "step": 1334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.295548227535037e-05,
+      "loss": 6.3152,
+      "step": 1335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.293487221764222e-05,
+      "loss": 6.4827,
+      "step": 1336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.291426215993405e-05,
+      "loss": 6.553,
+      "step": 1337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.289365210222589e-05,
+      "loss": 6.5539,
+      "step": 1338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.287304204451772e-05,
+      "loss": 6.5596,
+      "step": 1339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.285243198680956e-05,
+      "loss": 6.545,
+      "step": 1340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.28318219291014e-05,
+      "loss": 6.6858,
+      "step": 1341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.281121187139325e-05,
+      "loss": 6.3996,
+      "step": 1342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.27906018136851e-05,
+      "loss": 6.6223,
+      "step": 1343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.276999175597692e-05,
+      "loss": 6.2708,
+      "step": 1344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.274938169826875e-05,
+      "loss": 6.2858,
+      "step": 1345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.27287716405606e-05,
+      "loss": 6.4188,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.270816158285244e-05,
+      "loss": 6.2983,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.268755152514427e-05,
+      "loss": 5.8298,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.266694146743611e-05,
+      "loss": 6.1361,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.264633140972796e-05,
+      "loss": 5.3409,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.262572135201979e-05,
+      "loss": 6.8958,
+      "step": 1351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.260511129431163e-05,
+      "loss": 6.9611,
+      "step": 1352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.258450123660346e-05,
+      "loss": 6.8644,
+      "step": 1353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.25638911788953e-05,
+      "loss": 6.7143,
+      "step": 1354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.254328112118715e-05,
+      "loss": 6.7114,
+      "step": 1355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.252267106347899e-05,
+      "loss": 6.8363,
+      "step": 1356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.250206100577082e-05,
+      "loss": 6.5921,
+      "step": 1357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.248145094806265e-05,
+      "loss": 6.8289,
+      "step": 1358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.246084089035449e-05,
+      "loss": 6.6884,
+      "step": 1359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.244023083264634e-05,
+      "loss": 6.8144,
+      "step": 1360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.241962077493818e-05,
+      "loss": 6.6957,
+      "step": 1361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.239901071723001e-05,
+      "loss": 6.6271,
+      "step": 1362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.237840065952185e-05,
+      "loss": 6.5578,
+      "step": 1363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.23577906018137e-05,
+      "loss": 6.8864,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.233718054410553e-05,
+      "loss": 6.536,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.231657048639737e-05,
+      "loss": 6.5373,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.22959604286892e-05,
+      "loss": 6.5517,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.227535037098104e-05,
+      "loss": 6.4646,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.225474031327289e-05,
+      "loss": 6.6211,
+      "step": 1369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.223413025556473e-05,
+      "loss": 6.6103,
+      "step": 1370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.221352019785656e-05,
+      "loss": 6.7559,
+      "step": 1371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.219291014014839e-05,
+      "loss": 6.4042,
+      "step": 1372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.217230008244023e-05,
+      "loss": 6.6626,
+      "step": 1373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.215169002473208e-05,
+      "loss": 6.6922,
+      "step": 1374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.213107996702392e-05,
+      "loss": 6.604,
+      "step": 1375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.211046990931575e-05,
+      "loss": 6.3454,
+      "step": 1376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.208985985160759e-05,
+      "loss": 6.5629,
+      "step": 1377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.206924979389942e-05,
+      "loss": 6.5585,
+      "step": 1378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.204863973619127e-05,
+      "loss": 6.4784,
+      "step": 1379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.20280296784831e-05,
+      "loss": 6.6428,
+      "step": 1380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.200741962077494e-05,
+      "loss": 6.4031,
+      "step": 1381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.198680956306678e-05,
+      "loss": 6.6332,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.196619950535863e-05,
+      "loss": 6.488,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.194558944765046e-05,
+      "loss": 6.6822,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.192497938994228e-05,
+      "loss": 6.5656,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.190436933223413e-05,
+      "loss": 6.5676,
+      "step": 1386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.188375927452597e-05,
+      "loss": 6.5405,
+      "step": 1387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.186314921681781e-05,
+      "loss": 6.5056,
+      "step": 1388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.184253915910966e-05,
+      "loss": 6.5807,
+      "step": 1389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.182192910140149e-05,
+      "loss": 6.5002,
+      "step": 1390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.180131904369333e-05,
+      "loss": 6.1892,
+      "step": 1391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.178070898598516e-05,
+      "loss": 6.4367,
+      "step": 1392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.1760098928277e-05,
+      "loss": 6.1806,
+      "step": 1393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.173948887056883e-05,
+      "loss": 6.3434,
+      "step": 1394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.171887881286068e-05,
+      "loss": 6.3628,
+      "step": 1395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.169826875515252e-05,
+      "loss": 6.3835,
+      "step": 1396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.167765869744436e-05,
+      "loss": 6.1255,
+      "step": 1397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.16570486397362e-05,
+      "loss": 5.9603,
+      "step": 1398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.163643858202802e-05,
+      "loss": 5.651,
+      "step": 1399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.161582852431987e-05,
+      "loss": 5.6158,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.159521846661171e-05,
+      "loss": 6.8158,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.157460840890355e-05,
+      "loss": 6.6695,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.155399835119538e-05,
+      "loss": 6.609,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.153338829348723e-05,
+      "loss": 6.5046,
+      "step": 1404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.151277823577906e-05,
+      "loss": 6.5663,
+      "step": 1405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.14921681780709e-05,
+      "loss": 6.8443,
+      "step": 1406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.147155812036274e-05,
+      "loss": 6.557,
+      "step": 1407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.145094806265457e-05,
+      "loss": 6.519,
+      "step": 1408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.143033800494642e-05,
+      "loss": 6.5501,
+      "step": 1409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.140972794723826e-05,
+      "loss": 6.4199,
+      "step": 1410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.13891178895301e-05,
+      "loss": 6.4742,
+      "step": 1411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.136850783182193e-05,
+      "loss": 6.4243,
+      "step": 1412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.134789777411376e-05,
+      "loss": 6.6315,
+      "step": 1413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.132728771640561e-05,
+      "loss": 6.4863,
+      "step": 1414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.130667765869745e-05,
+      "loss": 6.6131,
+      "step": 1415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.12860676009893e-05,
+      "loss": 6.7752,
+      "step": 1416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.126545754328112e-05,
+      "loss": 6.6978,
+      "step": 1417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.124484748557297e-05,
+      "loss": 6.632,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.12242374278648e-05,
+      "loss": 6.4805,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.120362737015664e-05,
+      "loss": 6.4896,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.118301731244848e-05,
+      "loss": 6.6703,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.116240725474031e-05,
+      "loss": 6.5425,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.114179719703216e-05,
+      "loss": 6.4224,
+      "step": 1423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.1121187139324e-05,
+      "loss": 6.6223,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.110057708161583e-05,
+      "loss": 6.5126,
+      "step": 1425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.107996702390766e-05,
+      "loss": 6.6098,
+      "step": 1426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.10593569661995e-05,
+      "loss": 6.4336,
+      "step": 1427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.103874690849135e-05,
+      "loss": 6.469,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.101813685078319e-05,
+      "loss": 6.5586,
+      "step": 1429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.099752679307503e-05,
+      "loss": 6.6164,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.097691673536686e-05,
+      "loss": 6.6007,
+      "step": 1431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.09563066776587e-05,
+      "loss": 6.6084,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.093569661995054e-05,
+      "loss": 6.6805,
+      "step": 1433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.091508656224238e-05,
+      "loss": 6.4296,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.089447650453421e-05,
+      "loss": 6.5496,
+      "step": 1435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.087386644682605e-05,
+      "loss": 6.3642,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.08532563891179e-05,
+      "loss": 6.5239,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.083264633140974e-05,
+      "loss": 6.4075,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.081203627370157e-05,
+      "loss": 6.317,
+      "step": 1439
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.07914262159934e-05,
+      "loss": 6.4798,
+      "step": 1440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.077081615828524e-05,
+      "loss": 6.5507,
+      "step": 1441
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.075020610057709e-05,
+      "loss": 6.3364,
+      "step": 1442
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.072959604286893e-05,
+      "loss": 6.2062,
+      "step": 1443
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.070898598516077e-05,
+      "loss": 6.2246,
+      "step": 1444
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.06883759274526e-05,
+      "loss": 6.1041,
+      "step": 1445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.066776586974443e-05,
+      "loss": 6.2448,
+      "step": 1446
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.064715581203628e-05,
+      "loss": 5.947,
+      "step": 1447
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.062654575432812e-05,
+      "loss": 5.7385,
+      "step": 1448
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.060593569661995e-05,
+      "loss": 5.7783,
+      "step": 1449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.058532563891179e-05,
+      "loss": 5.4841,
+      "step": 1450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.056471558120364e-05,
+      "loss": 6.9039,
+      "step": 1451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.054410552349547e-05,
+      "loss": 6.9061,
+      "step": 1452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.052349546578731e-05,
+      "loss": 6.5211,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.050288540807914e-05,
+      "loss": 6.6925,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.048227535037098e-05,
+      "loss": 6.6307,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.046166529266283e-05,
+      "loss": 6.6721,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.044105523495467e-05,
+      "loss": 6.7929,
+      "step": 1457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.04204451772465e-05,
+      "loss": 6.6712,
+      "step": 1458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.039983511953834e-05,
+      "loss": 6.502,
+      "step": 1459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.037922506183017e-05,
+      "loss": 6.6736,
+      "step": 1460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.035861500412202e-05,
+      "loss": 6.5812,
+      "step": 1461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.033800494641386e-05,
+      "loss": 6.5349,
+      "step": 1462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.031739488870569e-05,
+      "loss": 6.3841,
+      "step": 1463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.029678483099753e-05,
+      "loss": 6.4254,
+      "step": 1464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.027617477328938e-05,
+      "loss": 6.6366,
+      "step": 1465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.02555647155812e-05,
+      "loss": 6.4399,
+      "step": 1466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.023495465787304e-05,
+      "loss": 6.5797,
+      "step": 1467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.021434460016488e-05,
+      "loss": 6.4913,
+      "step": 1468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.019373454245672e-05,
+      "loss": 6.2793,
+      "step": 1469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.017312448474857e-05,
+      "loss": 6.3523,
+      "step": 1470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.015251442704041e-05,
+      "loss": 6.4952,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.013190436933224e-05,
+      "loss": 6.2869,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.011129431162407e-05,
+      "loss": 6.4113,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.009068425391591e-05,
+      "loss": 6.2499,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.007007419620776e-05,
+      "loss": 6.5783,
+      "step": 1475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.00494641384996e-05,
+      "loss": 6.5573,
+      "step": 1476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.002885408079143e-05,
+      "loss": 6.667,
+      "step": 1477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.000824402308327e-05,
+      "loss": 6.2801,
+      "step": 1478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.99876339653751e-05,
+      "loss": 6.4518,
+      "step": 1479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.996702390766695e-05,
+      "loss": 6.2863,
+      "step": 1480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.994641384995877e-05,
+      "loss": 6.5258,
+      "step": 1481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.992580379225062e-05,
+      "loss": 6.5281,
+      "step": 1482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.990519373454246e-05,
+      "loss": 6.5639,
+      "step": 1483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.98845836768343e-05,
+      "loss": 6.1624,
+      "step": 1484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.986397361912615e-05,
+      "loss": 6.5808,
+      "step": 1485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.984336356141798e-05,
+      "loss": 6.4646,
+      "step": 1486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.982275350370981e-05,
+      "loss": 6.3639,
+      "step": 1487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.980214344600165e-05,
+      "loss": 6.4067,
+      "step": 1488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.97815333882935e-05,
+      "loss": 6.4335,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.976092333058532e-05,
+      "loss": 6.1266,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.974031327287717e-05,
+      "loss": 6.3239,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.971970321516901e-05,
+      "loss": 6.4637,
+      "step": 1492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.969909315746084e-05,
+      "loss": 6.1175,
+      "step": 1493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.967848309975268e-05,
+      "loss": 6.4221,
+      "step": 1494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.965787304204451e-05,
+      "loss": 6.1656,
+      "step": 1495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.963726298433636e-05,
+      "loss": 6.4285,
+      "step": 1496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.96166529266282e-05,
+      "loss": 5.9578,
+      "step": 1497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.959604286892004e-05,
+      "loss": 5.9881,
+      "step": 1498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.957543281121187e-05,
+      "loss": 5.338,
+      "step": 1499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.95548227535037e-05,
+      "loss": 5.7859,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 13.754671096801758,
+      "eval_runtime": 1322.0968,
+      "eval_samples_per_second": 1.998,
+      "eval_steps_per_second": 0.25,
+      "eval_wer": 1.0017453391511304,
+      "step": 1500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.953421269579555e-05,
+      "loss": 6.6723,
+      "step": 1501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.951360263808739e-05,
+      "loss": 6.8885,
+      "step": 1502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.949299258037923e-05,
+      "loss": 6.7283,
+      "step": 1503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.947238252267106e-05,
+      "loss": 6.6297,
+      "step": 1504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.945177246496291e-05,
+      "loss": 6.6581,
+      "step": 1505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.943116240725475e-05,
+      "loss": 6.6324,
+      "step": 1506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.941055234954658e-05,
+      "loss": 6.5038,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.938994229183842e-05,
+      "loss": 6.5113,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.936933223413025e-05,
+      "loss": 6.6097,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.93487221764221e-05,
+      "loss": 6.4359,
+      "step": 1510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.932811211871394e-05,
+      "loss": 6.3489,
+      "step": 1511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.930750206100578e-05,
+      "loss": 6.4349,
+      "step": 1512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.928689200329761e-05,
+      "loss": 6.307,
+      "step": 1513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.926628194558944e-05,
+      "loss": 6.4262,
+      "step": 1514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.924567188788129e-05,
+      "loss": 6.3284,
+      "step": 1515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.922506183017313e-05,
+      "loss": 6.5092,
+      "step": 1516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.920445177246497e-05,
+      "loss": 6.3617,
+      "step": 1517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.91838417147568e-05,
+      "loss": 6.4496,
+      "step": 1518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.916323165704865e-05,
+      "loss": 6.3697,
+      "step": 1519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.914262159934048e-05,
+      "loss": 6.3962,
+      "step": 1520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.912201154163232e-05,
+      "loss": 6.4277,
+      "step": 1521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.910140148392415e-05,
+      "loss": 6.4263,
+      "step": 1522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.9080791426216e-05,
+      "loss": 6.5048,
+      "step": 1523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.906018136850784e-05,
+      "loss": 6.4714,
+      "step": 1524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.903957131079968e-05,
+      "loss": 6.2625,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.901896125309151e-05,
+      "loss": 6.2747,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.899835119538334e-05,
+      "loss": 6.38,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.897774113767518e-05,
+      "loss": 6.2112,
+      "step": 1528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.895713107996703e-05,
+      "loss": 6.375,
+      "step": 1529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.893652102225887e-05,
+      "loss": 6.3159,
+      "step": 1530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.891591096455071e-05,
+      "loss": 6.2258,
+      "step": 1531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.889530090684254e-05,
+      "loss": 6.1093,
+      "step": 1532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.887469084913439e-05,
+      "loss": 6.4539,
+      "step": 1533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.885408079142622e-05,
+      "loss": 6.3002,
+      "step": 1534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.883347073371806e-05,
+      "loss": 6.3641,
+      "step": 1535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.881286067600989e-05,
+      "loss": 6.2894,
+      "step": 1536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.879225061830173e-05,
+      "loss": 6.2724,
+      "step": 1537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.877164056059358e-05,
+      "loss": 6.5049,
+      "step": 1538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.875103050288542e-05,
+      "loss": 6.2737,
+      "step": 1539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.873042044517725e-05,
+      "loss": 6.25,
+      "step": 1540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.870981038746908e-05,
+      "loss": 6.2985,
+      "step": 1541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.868920032976092e-05,
+      "loss": 6.2135,
+      "step": 1542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.866859027205277e-05,
+      "loss": 6.1809,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.864798021434461e-05,
+      "loss": 6.1791,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.862737015663644e-05,
+      "loss": 6.2036,
+      "step": 1545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.860676009892828e-05,
+      "loss": 5.8358,
+      "step": 1546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.858615004122011e-05,
+      "loss": 5.9255,
+      "step": 1547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.856553998351196e-05,
+      "loss": 5.5548,
+      "step": 1548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.85449299258038e-05,
+      "loss": 5.6425,
+      "step": 1549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.852431986809563e-05,
+      "loss": 5.5332,
+      "step": 1550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.850370981038747e-05,
+      "loss": 6.7436,
+      "step": 1551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.848309975267932e-05,
+      "loss": 6.7697,
+      "step": 1552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.846248969497116e-05,
+      "loss": 6.5713,
+      "step": 1553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.844187963726299e-05,
+      "loss": 6.3137,
+      "step": 1554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.842126957955482e-05,
+      "loss": 6.5256,
+      "step": 1555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.840065952184666e-05,
+      "loss": 6.6239,
+      "step": 1556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.83800494641385e-05,
+      "loss": 6.3652,
+      "step": 1557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.835943940643035e-05,
+      "loss": 6.3865,
+      "step": 1558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.833882934872218e-05,
+      "loss": 6.3597,
+      "step": 1559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.831821929101402e-05,
+      "loss": 6.4912,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.829760923330585e-05,
+      "loss": 6.5519,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82769991755977e-05,
+      "loss": 6.4939,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.825638911788954e-05,
+      "loss": 6.3752,
+      "step": 1563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.823577906018137e-05,
+      "loss": 6.2081,
+      "step": 1564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.821516900247321e-05,
+      "loss": 6.3505,
+      "step": 1565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.819455894476506e-05,
+      "loss": 6.3314,
+      "step": 1566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.817394888705689e-05,
+      "loss": 6.2007,
+      "step": 1567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.815333882934872e-05,
+      "loss": 6.3803,
+      "step": 1568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.813272877164056e-05,
+      "loss": 6.4836,
+      "step": 1569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.81121187139324e-05,
+      "loss": 6.3539,
+      "step": 1570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.809150865622425e-05,
+      "loss": 6.4897,
+      "step": 1571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.807089859851609e-05,
+      "loss": 6.581,
+      "step": 1572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.805028854080792e-05,
+      "loss": 6.3875,
+      "step": 1573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.802967848309975e-05,
+      "loss": 6.1355,
+      "step": 1574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.800906842539159e-05,
+      "loss": 6.0615,
+      "step": 1575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.798845836768344e-05,
+      "loss": 6.1832,
+      "step": 1576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.796784830997527e-05,
+      "loss": 6.125,
+      "step": 1577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.794723825226711e-05,
+      "loss": 6.3613,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.792662819455895e-05,
+      "loss": 6.3064,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.79060181368508e-05,
+      "loss": 6.3335,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.788540807914263e-05,
+      "loss": 6.4822,
+      "step": 1581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.786479802143445e-05,
+      "loss": 6.5168,
+      "step": 1582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.78441879637263e-05,
+      "loss": 6.3182,
+      "step": 1583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.782357790601814e-05,
+      "loss": 6.1804,
+      "step": 1584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.780296784830998e-05,
+      "loss": 6.3152,
+      "step": 1585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.778235779060181e-05,
+      "loss": 6.2435,
+      "step": 1586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.776174773289366e-05,
+      "loss": 6.3649,
+      "step": 1587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.774113767518549e-05,
+      "loss": 6.2956,
+      "step": 1588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.772052761747733e-05,
+      "loss": 6.2906,
+      "step": 1589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.769991755976917e-05,
+      "loss": 6.2203,
+      "step": 1590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.7679307502061e-05,
+      "loss": 6.1462,
+      "step": 1591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.765869744435285e-05,
+      "loss": 6.1384,
+      "step": 1592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.763808738664469e-05,
+      "loss": 5.9677,
+      "step": 1593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.761747732893652e-05,
+      "loss": 5.8599,
+      "step": 1594
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.759686727122836e-05,
+      "loss": 6.1369,
+      "step": 1595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.75762572135202e-05,
+      "loss": 5.9377,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.755564715581204e-05,
+      "loss": 5.7642,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.753503709810388e-05,
+      "loss": 5.8439,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.751442704039572e-05,
+      "loss": 5.9778,
+      "step": 1599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.749381698268755e-05,
+      "loss": 5.4054,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.74732069249794e-05,
+      "loss": 6.8132,
+      "step": 1601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.745259686727123e-05,
+      "loss": 6.4352,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.743198680956307e-05,
+      "loss": 6.663,
+      "step": 1603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.741137675185491e-05,
+      "loss": 6.6237,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.739076669414674e-05,
+      "loss": 6.2351,
+      "step": 1605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.737015663643859e-05,
+      "loss": 6.4498,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.734954657873043e-05,
+      "loss": 6.6253,
+      "step": 1607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.732893652102226e-05,
+      "loss": 6.4016,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.730832646331409e-05,
+      "loss": 6.6199,
+      "step": 1609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.728771640560593e-05,
+      "loss": 6.3524,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.726710634789778e-05,
+      "loss": 6.5455,
+      "step": 1611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.724649629018962e-05,
+      "loss": 6.5292,
+      "step": 1612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.722588623248146e-05,
+      "loss": 6.5911,
+      "step": 1613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.72052761747733e-05,
+      "loss": 6.3152,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.718466611706512e-05,
+      "loss": 6.3507,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.716405605935697e-05,
+      "loss": 6.2601,
+      "step": 1616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.714344600164881e-05,
+      "loss": 6.358,
+      "step": 1617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.712283594394065e-05,
+      "loss": 6.2729,
+      "step": 1618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.710222588623248e-05,
+      "loss": 6.25,
+      "step": 1619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.708161582852433e-05,
+      "loss": 6.4666,
+      "step": 1620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.706100577081616e-05,
+      "loss": 6.496,
+      "step": 1621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.7040395713108e-05,
+      "loss": 6.2441,
+      "step": 1622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.701978565539983e-05,
+      "loss": 6.433,
+      "step": 1623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.699917559769167e-05,
+      "loss": 6.1261,
+      "step": 1624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.697856553998352e-05,
+      "loss": 6.445,
+      "step": 1625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.695795548227536e-05,
+      "loss": 6.487,
+      "step": 1626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.69373454245672e-05,
+      "loss": 6.3512,
+      "step": 1627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.691673536685903e-05,
+      "loss": 6.3751,
+      "step": 1628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.689612530915086e-05,
+      "loss": 6.2509,
+      "step": 1629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.68755152514427e-05,
+      "loss": 6.1044,
+      "step": 1630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.685490519373455e-05,
+      "loss": 6.1718,
+      "step": 1631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.683429513602638e-05,
+      "loss": 6.4578,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.681368507831822e-05,
+      "loss": 6.1984,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.679307502061007e-05,
+      "loss": 6.3056,
+      "step": 1634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.67724649629019e-05,
+      "loss": 6.1774,
+      "step": 1635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.675185490519374e-05,
+      "loss": 6.2339,
+      "step": 1636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.673124484748557e-05,
+      "loss": 6.0308,
+      "step": 1637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.671063478977741e-05,
+      "loss": 6.3409,
+      "step": 1638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.669002473206926e-05,
+      "loss": 6.0673,
+      "step": 1639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.66694146743611e-05,
+      "loss": 5.9582,
+      "step": 1640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.664880461665293e-05,
+      "loss": 6.1056,
+      "step": 1641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.662819455894476e-05,
+      "loss": 6.0144,
+      "step": 1642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.66075845012366e-05,
+      "loss": 6.0707,
+      "step": 1643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.658697444352845e-05,
+      "loss": 5.8554,
+      "step": 1644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.656636438582029e-05,
+      "loss": 5.969,
+      "step": 1645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.654575432811212e-05,
+      "loss": 5.866,
+      "step": 1646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.652514427040396e-05,
+      "loss": 5.3577,
+      "step": 1647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.65045342126958e-05,
+      "loss": 5.5293,
+      "step": 1648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.648392415498764e-05,
+      "loss": 5.8241,
+      "step": 1649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.646331409727948e-05,
+      "loss": 4.7029,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.644270403957131e-05,
+      "loss": 6.8336,
+      "step": 1651
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.642209398186315e-05,
+      "loss": 6.8522,
+      "step": 1652
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.6401483924155e-05,
+      "loss": 6.8629,
+      "step": 1653
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.638087386644684e-05,
+      "loss": 6.4754,
+      "step": 1654
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.636026380873867e-05,
+      "loss": 6.3516,
+      "step": 1655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.63396537510305e-05,
+      "loss": 6.5487,
+      "step": 1656
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.631904369332234e-05,
+      "loss": 6.4746,
+      "step": 1657
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.629843363561419e-05,
+      "loss": 6.4064,
+      "step": 1658
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.627782357790603e-05,
+      "loss": 6.18,
+      "step": 1659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.625721352019786e-05,
+      "loss": 6.3152,
+      "step": 1660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.62366034624897e-05,
+      "loss": 6.3586,
+      "step": 1661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.621599340478153e-05,
+      "loss": 6.2372,
+      "step": 1662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.619538334707338e-05,
+      "loss": 6.1655,
+      "step": 1663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.61747732893652e-05,
+      "loss": 6.4477,
+      "step": 1664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.615416323165705e-05,
+      "loss": 6.4334,
+      "step": 1665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.613355317394889e-05,
+      "loss": 6.3551,
+      "step": 1666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.611294311624074e-05,
+      "loss": 6.3373,
+      "step": 1667
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 7.609233305853258e-05,
+      "loss": 6.3359,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.60717230008244e-05,
+      "loss": 6.2578,
+      "step": 1669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.605111294311624e-05,
+      "loss": 6.6616,
+      "step": 1670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.603050288540808e-05,
+      "loss": 6.2783,
+      "step": 1671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.600989282769993e-05,
+      "loss": 6.4077,
+      "step": 1672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.598928276999176e-05,
+      "loss": 6.3968,
+      "step": 1673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.59686727122836e-05,
+      "loss": 6.5875,
+      "step": 1674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.594806265457544e-05,
+      "loss": 6.578,
+      "step": 1675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.592745259686727e-05,
+      "loss": 6.086,
+      "step": 1676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.590684253915912e-05,
+      "loss": 6.1493,
+      "step": 1677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.588623248145094e-05,
+      "loss": 6.2071,
+      "step": 1678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.586562242374279e-05,
+      "loss": 6.4358,
+      "step": 1679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.584501236603463e-05,
+      "loss": 6.4295,
+      "step": 1680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.582440230832648e-05,
+      "loss": 6.1906,
+      "step": 1681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.58037922506183e-05,
+      "loss": 6.1454,
+      "step": 1682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.578318219291013e-05,
+      "loss": 6.303,
+      "step": 1683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.576257213520198e-05,
+      "loss": 6.2885,
+      "step": 1684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.574196207749382e-05,
+      "loss": 6.3411,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.572135201978566e-05,
+      "loss": 6.2932,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.57007419620775e-05,
+      "loss": 6.2879,
+      "step": 1687
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.568013190436934e-05,
+      "loss": 6.3244,
+      "step": 1688
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.565952184666117e-05,
+      "loss": 6.3288,
+      "step": 1689
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.563891178895301e-05,
+      "loss": 5.8205,
+      "step": 1690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.561830173124485e-05,
+      "loss": 6.1915,
+      "step": 1691
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.559769167353668e-05,
+      "loss": 6.1144,
+      "step": 1692
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.557708161582853e-05,
+      "loss": 6.2232,
+      "step": 1693
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.555647155812037e-05,
+      "loss": 6.1406,
+      "step": 1694
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.553586150041221e-05,
+      "loss": 6.0955,
+      "step": 1695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.551525144270404e-05,
+      "loss": 5.8895,
+      "step": 1696
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.549464138499587e-05,
+      "loss": 5.6026,
+      "step": 1697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.547403132728772e-05,
+      "loss": 5.9436,
+      "step": 1698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.545342126957956e-05,
+      "loss": 5.3007,
+      "step": 1699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.54328112118714e-05,
+      "loss": 4.936,
+      "step": 1700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.541220115416323e-05,
+      "loss": 6.8987,
+      "step": 1701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.539159109645508e-05,
+      "loss": 6.6929,
+      "step": 1702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.537098103874691e-05,
+      "loss": 6.8099,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.535037098103875e-05,
+      "loss": 6.8334,
+      "step": 1704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.532976092333058e-05,
+      "loss": 6.7801,
+      "step": 1705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.530915086562242e-05,
+      "loss": 6.567,
+      "step": 1706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.528854080791427e-05,
+      "loss": 6.5278,
+      "step": 1707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.526793075020611e-05,
+      "loss": 6.5903,
+      "step": 1708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.524732069249794e-05,
+      "loss": 6.6181,
+      "step": 1709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.522671063478977e-05,
+      "loss": 6.7529,
+      "step": 1710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.520610057708161e-05,
+      "loss": 6.3619,
+      "step": 1711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.518549051937346e-05,
+      "loss": 6.5322,
+      "step": 1712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.51648804616653e-05,
+      "loss": 6.6858,
+      "step": 1713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.514427040395714e-05,
+      "loss": 6.5605,
+      "step": 1714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.512366034624897e-05,
+      "loss": 6.3312,
+      "step": 1715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.510305028854082e-05,
+      "loss": 6.4678,
+      "step": 1716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.508244023083265e-05,
+      "loss": 6.6303,
+      "step": 1717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.506183017312449e-05,
+      "loss": 6.5718,
+      "step": 1718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.504122011541632e-05,
+      "loss": 6.2367,
+      "step": 1719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.502061005770816e-05,
+      "loss": 6.5411,
+      "step": 1720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 6.2289,
+      "step": 1721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.497938994229185e-05,
+      "loss": 6.2666,
+      "step": 1722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.495877988458368e-05,
+      "loss": 6.3734,
+      "step": 1723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.493816982687551e-05,
+      "loss": 6.348,
+      "step": 1724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.491755976916735e-05,
+      "loss": 6.3644,
+      "step": 1725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.48969497114592e-05,
+      "loss": 6.202,
+      "step": 1726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.487633965375104e-05,
+      "loss": 6.3468,
+      "step": 1727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.485572959604287e-05,
+      "loss": 6.4894,
+      "step": 1728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.483511953833471e-05,
+      "loss": 6.4731,
+      "step": 1729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.481450948062654e-05,
+      "loss": 6.3239,
+      "step": 1730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.479389942291839e-05,
+      "loss": 6.0349,
+      "step": 1731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.477328936521023e-05,
+      "loss": 6.2273,
+      "step": 1732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.475267930750206e-05,
+      "loss": 6.1722,
+      "step": 1733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.47320692497939e-05,
+      "loss": 6.1955,
+      "step": 1734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.471145919208575e-05,
+      "loss": 6.4348,
+      "step": 1735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.469084913437758e-05,
+      "loss": 6.4883,
+      "step": 1736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.467023907666942e-05,
+      "loss": 5.9311,
+      "step": 1737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.464962901896125e-05,
+      "loss": 6.1114,
+      "step": 1738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.462901896125309e-05,
+      "loss": 6.0823,
+      "step": 1739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.460840890354494e-05,
+      "loss": 6.1456,
+      "step": 1740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.458779884583678e-05,
+      "loss": 6.0059,
+      "step": 1741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.456718878812861e-05,
+      "loss": 6.1632,
+      "step": 1742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.454657873042045e-05,
+      "loss": 6.14,
+      "step": 1743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.452596867271228e-05,
+      "loss": 5.7737,
+      "step": 1744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.450535861500413e-05,
+      "loss": 5.8,
+      "step": 1745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.448474855729597e-05,
+      "loss": 5.9718,
+      "step": 1746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.44641384995878e-05,
+      "loss": 5.899,
+      "step": 1747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.444352844187964e-05,
+      "loss": 5.3356,
+      "step": 1748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.442291838417149e-05,
+      "loss": 5.0416,
+      "step": 1749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.440230832646332e-05,
+      "loss": 5.1395,
+      "step": 1750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.438169826875515e-05,
+      "loss": 6.5064,
+      "step": 1751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.436108821104699e-05,
+      "loss": 6.4211,
+      "step": 1752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.434047815333883e-05,
+      "loss": 6.7651,
+      "step": 1753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.431986809563068e-05,
+      "loss": 6.5074,
+      "step": 1754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.429925803792252e-05,
+      "loss": 6.7467,
+      "step": 1755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.427864798021435e-05,
+      "loss": 6.2585,
+      "step": 1756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.425803792250618e-05,
+      "loss": 6.2969,
+      "step": 1757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.423742786479802e-05,
+      "loss": 6.3445,
+      "step": 1758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.421681780708987e-05,
+      "loss": 6.4834,
+      "step": 1759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.41962077493817e-05,
+      "loss": 6.3534,
+      "step": 1760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.417559769167354e-05,
+      "loss": 6.3592,
+      "step": 1761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.415498763396538e-05,
+      "loss": 6.2128,
+      "step": 1762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.413437757625723e-05,
+      "loss": 6.2277,
+      "step": 1763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.411376751854906e-05,
+      "loss": 6.4297,
+      "step": 1764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.409315746084089e-05,
+      "loss": 6.1905,
+      "step": 1765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.407254740313273e-05,
+      "loss": 6.353,
+      "step": 1766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.405193734542457e-05,
+      "loss": 6.0997,
+      "step": 1767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.403132728771642e-05,
+      "loss": 6.0389,
+      "step": 1768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.401071723000826e-05,
+      "loss": 6.1909,
+      "step": 1769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.399010717230009e-05,
+      "loss": 6.1645,
+      "step": 1770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.396949711459192e-05,
+      "loss": 6.3143,
+      "step": 1771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.394888705688376e-05,
+      "loss": 6.1018,
+      "step": 1772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.39282769991756e-05,
+      "loss": 6.2156,
+      "step": 1773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.390766694146744e-05,
+      "loss": 6.1875,
+      "step": 1774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.388705688375928e-05,
+      "loss": 6.091,
+      "step": 1775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.386644682605112e-05,
+      "loss": 6.1959,
+      "step": 1776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.384583676834295e-05,
+      "loss": 5.9469,
+      "step": 1777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.38252267106348e-05,
+      "loss": 6.0417,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.380461665292662e-05,
+      "loss": 6.1808,
+      "step": 1779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.378400659521847e-05,
+      "loss": 5.9332,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.376339653751031e-05,
+      "loss": 5.8296,
+      "step": 1781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.374278647980215e-05,
+      "loss": 5.8377,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.372217642209398e-05,
+      "loss": 5.4688,
+      "step": 1783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.370156636438581e-05,
+      "loss": 5.7211,
+      "step": 1784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.368095630667766e-05,
+      "loss": 6.722,
+      "step": 1785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.36603462489695e-05,
+      "loss": 6.8508,
+      "step": 1786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.363973619126134e-05,
+      "loss": 6.4532,
+      "step": 1787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.361912613355317e-05,
+      "loss": 6.3318,
+      "step": 1788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.359851607584502e-05,
+      "loss": 6.3874,
+      "step": 1789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.357790601813686e-05,
+      "loss": 6.3401,
+      "step": 1790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.355729596042869e-05,
+      "loss": 6.1114,
+      "step": 1791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.353668590272052e-05,
+      "loss": 6.2306,
+      "step": 1792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.351607584501236e-05,
+      "loss": 6.25,
+      "step": 1793
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.349546578730421e-05,
+      "loss": 6.356,
+      "step": 1794
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.347485572959605e-05,
+      "loss": 6.6261,
+      "step": 1795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.34542456718879e-05,
+      "loss": 6.3501,
+      "step": 1796
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.343363561417972e-05,
+      "loss": 6.0522,
+      "step": 1797
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.341302555647155e-05,
+      "loss": 6.3529,
+      "step": 1798
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.33924154987634e-05,
+      "loss": 6.238,
+      "step": 1799
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.337180544105524e-05,
+      "loss": 6.3849,
+      "step": 1800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.335119538334708e-05,
+      "loss": 6.1521,
+      "step": 1801
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.333058532563891e-05,
+      "loss": 6.0876,
+      "step": 1802
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.330997526793076e-05,
+      "loss": 6.1947,
+      "step": 1803
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.328936521022259e-05,
+      "loss": 6.2976,
+      "step": 1804
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.326875515251443e-05,
+      "loss": 6.0193,
+      "step": 1805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.324814509480626e-05,
+      "loss": 6.1926,
+      "step": 1806
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.32275350370981e-05,
+      "loss": 5.8249,
+      "step": 1807
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.320692497938995e-05,
+      "loss": 6.082,
+      "step": 1808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.318631492168179e-05,
+      "loss": 6.2665,
+      "step": 1809
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 7.316570486397363e-05,
+      "loss": 6.1055,
+      "step": 1810
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.314509480626546e-05,
+      "loss": 6.1897,
+      "step": 1811
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.31244847485573e-05,
+      "loss": 6.4533,
+      "step": 1812
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.310387469084914e-05,
+      "loss": 6.2788,
+      "step": 1813
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.308326463314098e-05,
+      "loss": 6.0038,
+      "step": 1814
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.306265457543281e-05,
+      "loss": 6.5261,
+      "step": 1815
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.304204451772465e-05,
+      "loss": 6.5334,
+      "step": 1816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.30214344600165e-05,
+      "loss": 6.3832,
+      "step": 1817
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.300082440230833e-05,
+      "loss": 6.3488,
+      "step": 1818
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.298021434460017e-05,
+      "loss": 6.3527,
+      "step": 1819
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.2959604286892e-05,
+      "loss": 6.3372,
+      "step": 1820
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.293899422918384e-05,
+      "loss": 6.2826,
+      "step": 1821
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.291838417147569e-05,
+      "loss": 6.0329,
+      "step": 1822
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.289777411376753e-05,
+      "loss": 5.9251,
+      "step": 1823
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.287716405605936e-05,
+      "loss": 6.2476,
+      "step": 1824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.285655399835119e-05,
+      "loss": 5.8438,
+      "step": 1825
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.283594394064303e-05,
+      "loss": 6.2469,
+      "step": 1826
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.281533388293488e-05,
+      "loss": 5.6287,
+      "step": 1827
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.279472382522672e-05,
+      "loss": 5.8789,
+      "step": 1828
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.277411376751855e-05,
+      "loss": 5.8134,
+      "step": 1829
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.27535037098104e-05,
+      "loss": 5.5021,
+      "step": 1830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.273289365210222e-05,
+      "loss": 5.2052,
+      "step": 1831
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.271228359439407e-05,
+      "loss": 5.6184,
+      "step": 1832
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.269167353668591e-05,
+      "loss": 5.1316,
+      "step": 1833
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.267106347897774e-05,
+      "loss": 5.4363,
+      "step": 1834
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.265045342126958e-05,
+      "loss": 6.8016,
+      "step": 1835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.262984336356143e-05,
+      "loss": 6.6736,
+      "step": 1836
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.260923330585327e-05,
+      "loss": 6.539,
+      "step": 1837
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.25886232481451e-05,
+      "loss": 6.5137,
+      "step": 1838
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.256801319043693e-05,
+      "loss": 6.417,
+      "step": 1839
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.254740313272877e-05,
+      "loss": 6.5209,
+      "step": 1840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.252679307502062e-05,
+      "loss": 6.3556,
+      "step": 1841
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.250618301731246e-05,
+      "loss": 6.4017,
+      "step": 1842
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.248557295960429e-05,
+      "loss": 6.4154,
+      "step": 1843
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.246496290189613e-05,
+      "loss": 6.655,
+      "step": 1844
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.244435284418796e-05,
+      "loss": 6.2909,
+      "step": 1845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 7.24237427864798e-05,
+      "loss": 6.2227,
+      "step": 1846
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.240313272877164e-05,
+      "loss": 6.3727,
+      "step": 1847
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.238252267106348e-05,
+      "loss": 6.0336,
+      "step": 1848
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.236191261335532e-05,
+      "loss": 6.3156,
+      "step": 1849
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.234130255564717e-05,
+      "loss": 6.4039,
+      "step": 1850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.2320692497939e-05,
+      "loss": 6.2369,
+      "step": 1851
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.230008244023083e-05,
+      "loss": 6.2632,
+      "step": 1852
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.227947238252267e-05,
+      "loss": 6.1932,
+      "step": 1853
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.225886232481451e-05,
+      "loss": 5.9555,
+      "step": 1854
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.223825226710636e-05,
+      "loss": 6.291,
+      "step": 1855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.22176422093982e-05,
+      "loss": 6.1569,
+      "step": 1856
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.219703215169003e-05,
+      "loss": 6.2376,
+      "step": 1857
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.217642209398187e-05,
+      "loss": 6.0347,
+      "step": 1858
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.21558120362737e-05,
+      "loss": 6.0034,
+      "step": 1859
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.213520197856555e-05,
+      "loss": 6.1119,
+      "step": 1860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.211459192085738e-05,
+      "loss": 5.9701,
+      "step": 1861
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.209398186314922e-05,
+      "loss": 6.3005,
+      "step": 1862
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.207337180544106e-05,
+      "loss": 6.1879,
+      "step": 1863
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 7.20527617477329e-05,
+      "loss": 6.1382,
+      "step": 1864
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.203215169002474e-05,
+      "loss": 6.4226,
+      "step": 1865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.201154163231657e-05,
+      "loss": 5.8622,
+      "step": 1866
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.199093157460841e-05,
+      "loss": 6.1435,
+      "step": 1867
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.197032151690025e-05,
+      "loss": 6.0543,
+      "step": 1868
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.19497114591921e-05,
+      "loss": 6.0013,
+      "step": 1869
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.192910140148393e-05,
+      "loss": 6.0608,
+      "step": 1870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.190849134377577e-05,
+      "loss": 6.2417,
+      "step": 1871
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.18878812860676e-05,
+      "loss": 5.9665,
+      "step": 1872
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.186727122835944e-05,
+      "loss": 5.9812,
+      "step": 1873
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.184666117065129e-05,
+      "loss": 5.9866,
+      "step": 1874
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.182605111294311e-05,
+      "loss": 5.9281,
+      "step": 1875
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.180544105523496e-05,
+      "loss": 5.9735,
+      "step": 1876
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.17848309975268e-05,
+      "loss": 5.907,
+      "step": 1877
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.176422093981863e-05,
+      "loss": 5.7226,
+      "step": 1878
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.174361088211046e-05,
+      "loss": 5.9319,
+      "step": 1879
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.17230008244023e-05,
+      "loss": 5.2742,
+      "step": 1880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.170239076669415e-05,
+      "loss": 5.2245,
+      "step": 1881
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 7.168178070898599e-05,
+      "loss": 5.6733,
+      "step": 1882
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.166117065127783e-05,
+      "loss": 4.9064,
+      "step": 1883
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.164056059356966e-05,
+      "loss": 5.2621,
+      "step": 1884
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.161995053586151e-05,
+      "loss": 6.3769,
+      "step": 1885
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.159934047815334e-05,
+      "loss": 6.5522,
+      "step": 1886
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.157873042044518e-05,
+      "loss": 6.7445,
+      "step": 1887
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.155812036273702e-05,
+      "loss": 6.741,
+      "step": 1888
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.153751030502885e-05,
+      "loss": 6.4579,
+      "step": 1889
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.15169002473207e-05,
+      "loss": 6.5868,
+      "step": 1890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.149629018961254e-05,
+      "loss": 6.5091,
+      "step": 1891
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.147568013190437e-05,
+      "loss": 6.3756,
+      "step": 1892
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.14550700741962e-05,
+      "loss": 6.2766,
+      "step": 1893
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.143446001648804e-05,
+      "loss": 6.1053,
+      "step": 1894
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.141384995877989e-05,
+      "loss": 6.3209,
+      "step": 1895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.139323990107173e-05,
+      "loss": 6.3257,
+      "step": 1896
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.137262984336357e-05,
+      "loss": 6.3316,
+      "step": 1897
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.13520197856554e-05,
+      "loss": 6.6602,
+      "step": 1898
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 7.133140972794723e-05,
+      "loss": 6.095,
+      "step": 1899
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.131079967023908e-05,
+      "loss": 6.2617,
+      "step": 1900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.129018961253092e-05,
+      "loss": 6.2745,
+      "step": 1901
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.126957955482275e-05,
+      "loss": 6.0751,
+      "step": 1902
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.12489694971146e-05,
+      "loss": 6.0871,
+      "step": 1903
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.122835943940644e-05,
+      "loss": 6.429,
+      "step": 1904
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.120774938169828e-05,
+      "loss": 6.3095,
+      "step": 1905
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.118713932399011e-05,
+      "loss": 6.2037,
+      "step": 1906
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.116652926628194e-05,
+      "loss": 6.0297,
+      "step": 1907
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.114591920857378e-05,
+      "loss": 6.2163,
+      "step": 1908
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.112530915086563e-05,
+      "loss": 6.0086,
+      "step": 1909
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.110469909315747e-05,
+      "loss": 6.082,
+      "step": 1910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.10840890354493e-05,
+      "loss": 6.339,
+      "step": 1911
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.106347897774114e-05,
+      "loss": 6.3119,
+      "step": 1912
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.104286892003297e-05,
+      "loss": 6.1296,
+      "step": 1913
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.102225886232482e-05,
+      "loss": 6.3032,
+      "step": 1914
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.100164880461666e-05,
+      "loss": 6.2159,
+      "step": 1915
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.098103874690849e-05,
+      "loss": 6.0998,
+      "step": 1916
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 7.096042868920033e-05,
+      "loss": 6.06,
+      "step": 1917
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.093981863149218e-05,
+      "loss": 6.3286,
+      "step": 1918
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.091920857378401e-05,
+      "loss": 6.1292,
+      "step": 1919
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.089859851607585e-05,
+      "loss": 5.8884,
+      "step": 1920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.087798845836768e-05,
+      "loss": 5.8613,
+      "step": 1921
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.085737840065952e-05,
+      "loss": 6.0918,
+      "step": 1922
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.083676834295137e-05,
+      "loss": 6.0741,
+      "step": 1923
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.081615828524321e-05,
+      "loss": 6.1614,
+      "step": 1924
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.079554822753504e-05,
+      "loss": 6.2435,
+      "step": 1925
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.077493816982687e-05,
+      "loss": 5.7098,
+      "step": 1926
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.075432811211871e-05,
+      "loss": 5.7991,
+      "step": 1927
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.073371805441056e-05,
+      "loss": 5.7895,
+      "step": 1928
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.07131079967024e-05,
+      "loss": 5.4833,
+      "step": 1929
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.069249793899423e-05,
+      "loss": 5.221,
+      "step": 1930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.067188788128607e-05,
+      "loss": 4.8789,
+      "step": 1931
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.065127782357792e-05,
+      "loss": 5.4228,
+      "step": 1932
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.063066776586975e-05,
+      "loss": 4.1895,
+      "step": 1933
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.061005770816158e-05,
+      "loss": 4.2875,
+      "step": 1934
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 7.058944765045342e-05,
+      "loss": 6.7407,
+      "step": 1935
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.056883759274526e-05,
+      "loss": 6.4043,
+      "step": 1936
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.05482275350371e-05,
+      "loss": 6.6195,
+      "step": 1937
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.052761747732895e-05,
+      "loss": 6.6489,
+      "step": 1938
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.050700741962078e-05,
+      "loss": 6.6773,
+      "step": 1939
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.048639736191261e-05,
+      "loss": 6.6056,
+      "step": 1940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.046578730420445e-05,
+      "loss": 6.4539,
+      "step": 1941
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.04451772464963e-05,
+      "loss": 6.2917,
+      "step": 1942
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.042456718878814e-05,
+      "loss": 6.441,
+      "step": 1943
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.040395713107997e-05,
+      "loss": 6.2151,
+      "step": 1944
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.038334707337181e-05,
+      "loss": 6.6105,
+      "step": 1945
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.036273701566364e-05,
+      "loss": 6.119,
+      "step": 1946
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.034212695795549e-05,
+      "loss": 6.59,
+      "step": 1947
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.032151690024732e-05,
+      "loss": 6.0529,
+      "step": 1948
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.030090684253916e-05,
+      "loss": 6.4011,
+      "step": 1949
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.0280296784831e-05,
+      "loss": 6.0184,
+      "step": 1950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.025968672712285e-05,
+      "loss": 6.382,
+      "step": 1951
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.023907666941469e-05,
+      "loss": 6.1324,
+      "step": 1952
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 7.021846661170652e-05,
+      "loss": 6.2615,
+      "step": 1953
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.019785655399835e-05,
+      "loss": 6.2009,
+      "step": 1954
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.017724649629019e-05,
+      "loss": 6.2041,
+      "step": 1955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.015663643858204e-05,
+      "loss": 6.1434,
+      "step": 1956
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.013602638087387e-05,
+      "loss": 6.161,
+      "step": 1957
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.011541632316571e-05,
+      "loss": 6.2077,
+      "step": 1958
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.009480626545755e-05,
+      "loss": 6.068,
+      "step": 1959
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.007419620774938e-05,
+      "loss": 5.8918,
+      "step": 1960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.005358615004123e-05,
+      "loss": 6.1657,
+      "step": 1961
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.003297609233306e-05,
+      "loss": 6.0497,
+      "step": 1962
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 7.00123660346249e-05,
+      "loss": 6.0809,
+      "step": 1963
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.999175597691674e-05,
+      "loss": 6.1268,
+      "step": 1964
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.997114591920859e-05,
+      "loss": 6.162,
+      "step": 1965
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.995053586150042e-05,
+      "loss": 6.0088,
+      "step": 1966
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.992992580379225e-05,
+      "loss": 6.0618,
+      "step": 1967
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.990931574608409e-05,
+      "loss": 5.7682,
+      "step": 1968
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.988870568837593e-05,
+      "loss": 6.016,
+      "step": 1969
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.986809563066778e-05,
+      "loss": 6.1438,
+      "step": 1970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.98474855729596e-05,
+      "loss": 5.9619,
+      "step": 1971
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.982687551525145e-05,
+      "loss": 6.2298,
+      "step": 1972
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.980626545754329e-05,
+      "loss": 6.0031,
+      "step": 1973
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.978565539983512e-05,
+      "loss": 5.9033,
+      "step": 1974
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.976504534212697e-05,
+      "loss": 6.3249,
+      "step": 1975
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.97444352844188e-05,
+      "loss": 5.9185,
+      "step": 1976
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.972382522671064e-05,
+      "loss": 5.7083,
+      "step": 1977
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.970321516900248e-05,
+      "loss": 5.6635,
+      "step": 1978
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.968260511129432e-05,
+      "loss": 5.7517,
+      "step": 1979
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.966199505358615e-05,
+      "loss": 5.657,
+      "step": 1980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.964138499587798e-05,
+      "loss": 5.9614,
+      "step": 1981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.962077493816983e-05,
+      "loss": 5.2092,
+      "step": 1982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.960016488046167e-05,
+      "loss": 5.1481,
+      "step": 1983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.957955482275351e-05,
+      "loss": 4.6062,
+      "step": 1984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.955894476504534e-05,
+      "loss": 6.3583,
+      "step": 1985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.953833470733719e-05,
+      "loss": 6.4402,
+      "step": 1986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.951772464962902e-05,
+      "loss": 6.3579,
+      "step": 1987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.949711459192086e-05,
+      "loss": 6.3863,
+      "step": 1988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.947650453421269e-05,
+      "loss": 6.2665,
+      "step": 1989
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.945589447650453e-05,
+      "loss": 6.167,
+      "step": 1990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.943528441879638e-05,
+      "loss": 6.2893,
+      "step": 1991
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.941467436108822e-05,
+      "loss": 5.9272,
+      "step": 1992
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.939406430338005e-05,
+      "loss": 6.3382,
+      "step": 1993
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.937345424567188e-05,
+      "loss": 6.2742,
+      "step": 1994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.935284418796372e-05,
+      "loss": 6.3302,
+      "step": 1995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.933223413025557e-05,
+      "loss": 6.3073,
+      "step": 1996
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.931162407254741e-05,
+      "loss": 6.2554,
+      "step": 1997
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.929101401483924e-05,
+      "loss": 6.1768,
+      "step": 1998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.927040395713108e-05,
+      "loss": 6.2346,
+      "step": 1999
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.924979389942293e-05,
+      "loss": 6.0219,
+      "step": 2000
+    },
+    {
+      "epoch": 1.12,
+      "eval_loss": 15.496562004089355,
+      "eval_runtime": 1323.7007,
+      "eval_samples_per_second": 1.996,
+      "eval_steps_per_second": 0.25,
+      "eval_wer": 1.000654502181674,
+      "step": 2000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.922918384171476e-05,
+      "loss": 6.1631,
+      "step": 2001
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.92085737840066e-05,
+      "loss": 6.1246,
+      "step": 2002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.918796372629843e-05,
+      "loss": 6.188,
+      "step": 2003
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.916735366859027e-05,
+      "loss": 6.1743,
+      "step": 2004
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.914674361088212e-05,
+      "loss": 5.9699,
+      "step": 2005
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.912613355317396e-05,
+      "loss": 5.9532,
+      "step": 2006
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.910552349546579e-05,
+      "loss": 6.0209,
+      "step": 2007
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.908491343775762e-05,
+      "loss": 6.053,
+      "step": 2008
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.906430338004946e-05,
+      "loss": 6.0931,
+      "step": 2009
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.904369332234131e-05,
+      "loss": 5.7504,
+      "step": 2010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.902308326463315e-05,
+      "loss": 6.0419,
+      "step": 2011
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.900247320692498e-05,
+      "loss": 6.0502,
+      "step": 2012
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.898186314921682e-05,
+      "loss": 6.0082,
+      "step": 2013
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.896125309150865e-05,
+      "loss": 6.2437,
+      "step": 2014
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.89406430338005e-05,
+      "loss": 5.982,
+      "step": 2015
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.892003297609234e-05,
+      "loss": 5.9837,
+      "step": 2016
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.889942291838417e-05,
+      "loss": 5.8709,
+      "step": 2017
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.887881286067601e-05,
+      "loss": 6.1067,
+      "step": 2018
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.885820280296786e-05,
+      "loss": 5.9583,
+      "step": 2019
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.88375927452597e-05,
+      "loss": 6.0019,
+      "step": 2020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.881698268755153e-05,
+      "loss": 5.8361,
+      "step": 2021
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.879637262984336e-05,
+      "loss": 5.7651,
+      "step": 2022
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.87757625721352e-05,
+      "loss": 5.7909,
+      "step": 2023
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.875515251442705e-05,
+      "loss": 6.0283,
+      "step": 2024
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.873454245671889e-05,
+      "loss": 5.8875,
+      "step": 2025
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.871393239901072e-05,
+      "loss": 5.7122,
+      "step": 2026
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.869332234130256e-05,
+      "loss": 5.9138,
+      "step": 2027
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.867271228359439e-05,
+      "loss": 5.6584,
+      "step": 2028
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.865210222588624e-05,
+      "loss": 5.7417,
+      "step": 2029
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.863149216817807e-05,
+      "loss": 5.7285,
+      "step": 2030
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.861088211046991e-05,
+      "loss": 5.6083,
+      "step": 2031
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.859027205276175e-05,
+      "loss": 5.0971,
+      "step": 2032
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.85696619950536e-05,
+      "loss": 5.2789,
+      "step": 2033
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.854905193734543e-05,
+      "loss": 5.0661,
+      "step": 2034
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.852844187963726e-05,
+      "loss": 6.5277,
+      "step": 2035
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.85078318219291e-05,
+      "loss": 6.5096,
+      "step": 2036
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.848722176422094e-05,
+      "loss": 6.4087,
+      "step": 2037
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.846661170651279e-05,
+      "loss": 6.3021,
+      "step": 2038
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.844600164880463e-05,
+      "loss": 6.3638,
+      "step": 2039
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.842539159109646e-05,
+      "loss": 6.2208,
+      "step": 2040
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.840478153338829e-05,
+      "loss": 5.873,
+      "step": 2041
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.838417147568013e-05,
+      "loss": 6.3203,
+      "step": 2042
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.836356141797198e-05,
+      "loss": 6.3828,
+      "step": 2043
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.83429513602638e-05,
+      "loss": 6.1402,
+      "step": 2044
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.832234130255565e-05,
+      "loss": 6.3322,
+      "step": 2045
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.830173124484749e-05,
+      "loss": 6.213,
+      "step": 2046
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.828112118713934e-05,
+      "loss": 6.1765,
+      "step": 2047
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.826051112943117e-05,
+      "loss": 6.1591,
+      "step": 2048
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.8239901071723e-05,
+      "loss": 6.2926,
+      "step": 2049
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.821929101401484e-05,
+      "loss": 6.3418,
+      "step": 2050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.819868095630668e-05,
+      "loss": 6.0468,
+      "step": 2051
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.817807089859853e-05,
+      "loss": 6.0098,
+      "step": 2052
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.815746084089036e-05,
+      "loss": 6.3538,
+      "step": 2053
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.81368507831822e-05,
+      "loss": 6.0953,
+      "step": 2054
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.811624072547403e-05,
+      "loss": 6.2692,
+      "step": 2055
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.809563066776587e-05,
+      "loss": 6.1488,
+      "step": 2056
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.807502061005772e-05,
+      "loss": 5.7936,
+      "step": 2057
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.805441055234955e-05,
+      "loss": 6.1277,
+      "step": 2058
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.803380049464139e-05,
+      "loss": 6.2566,
+      "step": 2059
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.801319043693323e-05,
+      "loss": 6.0533,
+      "step": 2060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.799258037922506e-05,
+      "loss": 6.0881,
+      "step": 2061
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.79719703215169e-05,
+      "loss": 5.9229,
+      "step": 2062
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.795136026380874e-05,
+      "loss": 5.9916,
+      "step": 2063
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.793075020610058e-05,
+      "loss": 5.7237,
+      "step": 2064
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.791014014839242e-05,
+      "loss": 6.0816,
+      "step": 2065
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.788953009068427e-05,
+      "loss": 6.0755,
+      "step": 2066
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.78689200329761e-05,
+      "loss": 5.8085,
+      "step": 2067
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.784830997526794e-05,
+      "loss": 6.0001,
+      "step": 2068
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.782769991755977e-05,
+      "loss": 5.9467,
+      "step": 2069
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.780708985985161e-05,
+      "loss": 5.9096,
+      "step": 2070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.778647980214346e-05,
+      "loss": 6.0679,
+      "step": 2071
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.776586974443528e-05,
+      "loss": 5.8684,
+      "step": 2072
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.774525968672713e-05,
+      "loss": 5.7592,
+      "step": 2073
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.772464962901897e-05,
+      "loss": 5.6765,
+      "step": 2074
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.77040395713108e-05,
+      "loss": 5.9594,
+      "step": 2075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.768342951360263e-05,
+      "loss": 5.6854,
+      "step": 2076
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.766281945589447e-05,
+      "loss": 5.665,
+      "step": 2077
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.764220939818632e-05,
+      "loss": 5.699,
+      "step": 2078
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.762159934047816e-05,
+      "loss": 5.453,
+      "step": 2079
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.760098928277e-05,
+      "loss": 5.6179,
+      "step": 2080
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.758037922506183e-05,
+      "loss": 5.4101,
+      "step": 2081
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.755976916735366e-05,
+      "loss": 4.7626,
+      "step": 2082
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.753915910964551e-05,
+      "loss": 4.6147,
+      "step": 2083
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.751854905193735e-05,
+      "loss": 4.2746,
+      "step": 2084
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.749793899422918e-05,
+      "loss": 6.6981,
+      "step": 2085
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.747732893652102e-05,
+      "loss": 6.2172,
+      "step": 2086
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.745671887881287e-05,
+      "loss": 6.583,
+      "step": 2087
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.74361088211047e-05,
+      "loss": 6.5243,
+      "step": 2088
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.741549876339654e-05,
+      "loss": 6.3215,
+      "step": 2089
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.739488870568837e-05,
+      "loss": 6.3071,
+      "step": 2090
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.737427864798021e-05,
+      "loss": 6.1755,
+      "step": 2091
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.735366859027206e-05,
+      "loss": 6.2807,
+      "step": 2092
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.73330585325639e-05,
+      "loss": 6.1846,
+      "step": 2093
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.731244847485574e-05,
+      "loss": 6.3009,
+      "step": 2094
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.729183841714757e-05,
+      "loss": 6.3237,
+      "step": 2095
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.72712283594394e-05,
+      "loss": 6.2201,
+      "step": 2096
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.725061830173125e-05,
+      "loss": 6.2391,
+      "step": 2097
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.723000824402309e-05,
+      "loss": 6.0313,
+      "step": 2098
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.720939818631492e-05,
+      "loss": 6.2093,
+      "step": 2099
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.718878812860676e-05,
+      "loss": 6.0364,
+      "step": 2100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.716817807089861e-05,
+      "loss": 6.3371,
+      "step": 2101
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.714756801319044e-05,
+      "loss": 5.9394,
+      "step": 2102
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.712695795548228e-05,
+      "loss": 5.8904,
+      "step": 2103
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.710634789777411e-05,
+      "loss": 6.0285,
+      "step": 2104
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.708573784006595e-05,
+      "loss": 6.0394,
+      "step": 2105
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.70651277823578e-05,
+      "loss": 6.0823,
+      "step": 2106
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.704451772464964e-05,
+      "loss": 6.2314,
+      "step": 2107
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.702390766694147e-05,
+      "loss": 5.9968,
+      "step": 2108
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.70032976092333e-05,
+      "loss": 5.9124,
+      "step": 2109
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.698268755152514e-05,
+      "loss": 5.8474,
+      "step": 2110
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.696207749381699e-05,
+      "loss": 6.0957,
+      "step": 2111
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.694146743610883e-05,
+      "loss": 6.0672,
+      "step": 2112
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.692085737840066e-05,
+      "loss": 6.0414,
+      "step": 2113
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.69002473206925e-05,
+      "loss": 6.1328,
+      "step": 2114
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.687963726298435e-05,
+      "loss": 5.8443,
+      "step": 2115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.685902720527618e-05,
+      "loss": 5.8522,
+      "step": 2116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.6838417147568e-05,
+      "loss": 5.7923,
+      "step": 2117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.681780708985985e-05,
+      "loss": 6.144,
+      "step": 2118
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.67971970321517e-05,
+      "loss": 5.9532,
+      "step": 2119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.677658697444354e-05,
+      "loss": 5.7745,
+      "step": 2120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.675597691673538e-05,
+      "loss": 6.0354,
+      "step": 2121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.673536685902721e-05,
+      "loss": 6.0307,
+      "step": 2122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.671475680131904e-05,
+      "loss": 5.6876,
+      "step": 2123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.669414674361088e-05,
+      "loss": 5.6976,
+      "step": 2124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.667353668590273e-05,
+      "loss": 5.7321,
+      "step": 2125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.665292662819457e-05,
+      "loss": 5.6091,
+      "step": 2126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.66323165704864e-05,
+      "loss": 5.8132,
+      "step": 2127
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.661170651277824e-05,
+      "loss": 5.8644,
+      "step": 2128
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.659109645507007e-05,
+      "loss": 5.3233,
+      "step": 2129
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.657048639736192e-05,
+      "loss": 5.6143,
+      "step": 2130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.654987633965375e-05,
+      "loss": 5.1681,
+      "step": 2131
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.652926628194559e-05,
+      "loss": 5.383,
+      "step": 2132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.650865622423743e-05,
+      "loss": 4.8962,
+      "step": 2133
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.648804616652928e-05,
+      "loss": 4.9056,
+      "step": 2134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.64674361088211e-05,
+      "loss": 6.7507,
+      "step": 2135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.644682605111294e-05,
+      "loss": 6.6432,
+      "step": 2136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.642621599340478e-05,
+      "loss": 6.5765,
+      "step": 2137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.640560593569662e-05,
+      "loss": 6.3653,
+      "step": 2138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.638499587798847e-05,
+      "loss": 6.356,
+      "step": 2139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.63643858202803e-05,
+      "loss": 6.5398,
+      "step": 2140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.634377576257214e-05,
+      "loss": 6.3054,
+      "step": 2141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.632316570486398e-05,
+      "loss": 6.1053,
+      "step": 2142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.630255564715581e-05,
+      "loss": 6.2217,
+      "step": 2143
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.628194558944766e-05,
+      "loss": 6.1067,
+      "step": 2144
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.626133553173949e-05,
+      "loss": 6.5636,
+      "step": 2145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.624072547403133e-05,
+      "loss": 6.2459,
+      "step": 2146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.622011541632317e-05,
+      "loss": 6.1126,
+      "step": 2147
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.619950535861502e-05,
+      "loss": 6.3532,
+      "step": 2148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.617889530090685e-05,
+      "loss": 6.1415,
+      "step": 2149
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.615828524319868e-05,
+      "loss": 6.3646,
+      "step": 2150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.613767518549052e-05,
+      "loss": 6.1164,
+      "step": 2151
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.611706512778236e-05,
+      "loss": 6.0621,
+      "step": 2152
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.60964550700742e-05,
+      "loss": 6.1596,
+      "step": 2153
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.607584501236604e-05,
+      "loss": 5.9848,
+      "step": 2154
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.605523495465788e-05,
+      "loss": 6.0085,
+      "step": 2155
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.603462489694971e-05,
+      "loss": 5.943,
+      "step": 2156
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.601401483924155e-05,
+      "loss": 6.1705,
+      "step": 2157
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.59934047815334e-05,
+      "loss": 5.9881,
+      "step": 2158
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.597279472382523e-05,
+      "loss": 6.1217,
+      "step": 2159
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.595218466611707e-05,
+      "loss": 5.9979,
+      "step": 2160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.593157460840891e-05,
+      "loss": 5.9699,
+      "step": 2161
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.591096455070076e-05,
+      "loss": 6.0975,
+      "step": 2162
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.589035449299259e-05,
+      "loss": 5.9272,
+      "step": 2163
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.586974443528442e-05,
+      "loss": 5.9836,
+      "step": 2164
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.584913437757626e-05,
+      "loss": 5.896,
+      "step": 2165
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.58285243198681e-05,
+      "loss": 5.9604,
+      "step": 2166
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 6.580791426215995e-05,
+      "loss": 5.9916,
+      "step": 2167
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.578730420445178e-05,
+      "loss": 5.8871,
+      "step": 2168
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.576669414674362e-05,
+      "loss": 5.8517,
+      "step": 2169
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.574608408903545e-05,
+      "loss": 6.022,
+      "step": 2170
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.572547403132729e-05,
+      "loss": 5.8726,
+      "step": 2171
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.570486397361912e-05,
+      "loss": 6.0097,
+      "step": 2172
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.568425391591096e-05,
+      "loss": 5.8198,
+      "step": 2173
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.566364385820281e-05,
+      "loss": 5.8286,
+      "step": 2174
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.564303380049465e-05,
+      "loss": 5.7201,
+      "step": 2175
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.562242374278648e-05,
+      "loss": 5.5982,
+      "step": 2176
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.560181368507831e-05,
+      "loss": 5.4879,
+      "step": 2177
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.558120362737015e-05,
+      "loss": 5.4083,
+      "step": 2178
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.5560593569662e-05,
+      "loss": 5.066,
+      "step": 2179
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.553998351195384e-05,
+      "loss": 5.6788,
+      "step": 2180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.551937345424568e-05,
+      "loss": 5.1659,
+      "step": 2181
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.549876339653751e-05,
+      "loss": 4.432,
+      "step": 2182
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.547815333882934e-05,
+      "loss": 4.7796,
+      "step": 2183
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.545754328112119e-05,
+      "loss": 4.5003,
+      "step": 2184
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.543693322341303e-05,
+      "loss": 6.5239,
+      "step": 2185
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.541632316570486e-05,
+      "loss": 6.4874,
+      "step": 2186
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.53957131079967e-05,
+      "loss": 6.5089,
+      "step": 2187
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.537510305028855e-05,
+      "loss": 6.6392,
+      "step": 2188
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.535449299258039e-05,
+      "loss": 6.5906,
+      "step": 2189
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.533388293487222e-05,
+      "loss": 6.2606,
+      "step": 2190
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.531327287716405e-05,
+      "loss": 6.2502,
+      "step": 2191
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.52926628194559e-05,
+      "loss": 6.459,
+      "step": 2192
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.527205276174774e-05,
+      "loss": 6.287,
+      "step": 2193
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.525144270403958e-05,
+      "loss": 6.0641,
+      "step": 2194
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.523083264633141e-05,
+      "loss": 6.341,
+      "step": 2195
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.521022258862325e-05,
+      "loss": 6.0879,
+      "step": 2196
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.518961253091508e-05,
+      "loss": 6.3201,
+      "step": 2197
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.516900247320693e-05,
+      "loss": 6.2408,
+      "step": 2198
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.514839241549877e-05,
+      "loss": 6.1589,
+      "step": 2199
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.51277823577906e-05,
+      "loss": 6.1206,
+      "step": 2200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.510717230008244e-05,
+      "loss": 6.1978,
+      "step": 2201
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.508656224237429e-05,
+      "loss": 6.1662,
+      "step": 2202
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.506595218466612e-05,
+      "loss": 6.119,
+      "step": 2203
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.504534212695795e-05,
+      "loss": 6.2899,
+      "step": 2204
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.502473206924979e-05,
+      "loss": 6.1891,
+      "step": 2205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.500412201154163e-05,
+      "loss": 6.1601,
+      "step": 2206
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.498351195383348e-05,
+      "loss": 6.1654,
+      "step": 2207
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.496290189612532e-05,
+      "loss": 5.8683,
+      "step": 2208
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.494229183841715e-05,
+      "loss": 5.7457,
+      "step": 2209
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.4921681780709e-05,
+      "loss": 6.0387,
+      "step": 2210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.490107172300082e-05,
+      "loss": 5.8191,
+      "step": 2211
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.488046166529267e-05,
+      "loss": 5.9898,
+      "step": 2212
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.485985160758451e-05,
+      "loss": 5.8303,
+      "step": 2213
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.483924154987634e-05,
+      "loss": 6.0946,
+      "step": 2214
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.481863149216818e-05,
+      "loss": 5.8074,
+      "step": 2215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.479802143446003e-05,
+      "loss": 5.9847,
+      "step": 2216
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.477741137675186e-05,
+      "loss": 6.0669,
+      "step": 2217
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.475680131904369e-05,
+      "loss": 6.1585,
+      "step": 2218
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.473619126133553e-05,
+      "loss": 6.0244,
+      "step": 2219
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.471558120362737e-05,
+      "loss": 6.0474,
+      "step": 2220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.469497114591922e-05,
+      "loss": 6.0693,
+      "step": 2221
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.467436108821106e-05,
+      "loss": 5.6556,
+      "step": 2222
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.465375103050289e-05,
+      "loss": 5.7469,
+      "step": 2223
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.463314097279472e-05,
+      "loss": 5.7649,
+      "step": 2224
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.461253091508656e-05,
+      "loss": 5.8245,
+      "step": 2225
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.45919208573784e-05,
+      "loss": 5.5172,
+      "step": 2226
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.457131079967024e-05,
+      "loss": 5.7024,
+      "step": 2227
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.455070074196208e-05,
+      "loss": 5.5549,
+      "step": 2228
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.453009068425392e-05,
+      "loss": 5.5432,
+      "step": 2229
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.450948062654575e-05,
+      "loss": 5.3941,
+      "step": 2230
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.44888705688376e-05,
+      "loss": 5.11,
+      "step": 2231
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.446826051112943e-05,
+      "loss": 5.0231,
+      "step": 2232
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.444765045342127e-05,
+      "loss": 4.5619,
+      "step": 2233
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.442704039571311e-05,
+      "loss": 4.7537,
+      "step": 2234
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.440643033800496e-05,
+      "loss": 6.4782,
+      "step": 2235
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.438582028029679e-05,
+      "loss": 6.4069,
+      "step": 2236
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.436521022258863e-05,
+      "loss": 6.267,
+      "step": 2237
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.434460016488046e-05,
+      "loss": 6.4899,
+      "step": 2238
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.43239901071723e-05,
+      "loss": 6.1803,
+      "step": 2239
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.430338004946415e-05,
+      "loss": 6.2265,
+      "step": 2240
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.428276999175598e-05,
+      "loss": 6.2026,
+      "step": 2241
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.426215993404782e-05,
+      "loss": 6.2286,
+      "step": 2242
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.424154987633966e-05,
+      "loss": 6.0059,
+      "step": 2243
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.422093981863149e-05,
+      "loss": 5.9502,
+      "step": 2244
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.420032976092334e-05,
+      "loss": 6.2846,
+      "step": 2245
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.417971970321517e-05,
+      "loss": 6.0352,
+      "step": 2246
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.415910964550701e-05,
+      "loss": 6.0844,
+      "step": 2247
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.413849958779885e-05,
+      "loss": 5.9324,
+      "step": 2248
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.41178895300907e-05,
+      "loss": 5.8395,
+      "step": 2249
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.409727947238253e-05,
+      "loss": 5.7373,
+      "step": 2250
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.407666941467436e-05,
+      "loss": 6.1256,
+      "step": 2251
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.40560593569662e-05,
+      "loss": 6.0227,
+      "step": 2252
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.403544929925804e-05,
+      "loss": 6.0676,
+      "step": 2253
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.401483924154989e-05,
+      "loss": 5.9807,
+      "step": 2254
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.399422918384172e-05,
+      "loss": 6.009,
+      "step": 2255
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.397361912613356e-05,
+      "loss": 6.1252,
+      "step": 2256
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.39530090684254e-05,
+      "loss": 5.7307,
+      "step": 2257
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.393239901071723e-05,
+      "loss": 6.0216,
+      "step": 2258
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.391178895300906e-05,
+      "loss": 5.9037,
+      "step": 2259
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.38911788953009e-05,
+      "loss": 6.1228,
+      "step": 2260
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.387056883759275e-05,
+      "loss": 6.0367,
+      "step": 2261
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.384995877988459e-05,
+      "loss": 5.8727,
+      "step": 2262
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.382934872217644e-05,
+      "loss": 6.0839,
+      "step": 2263
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.380873866446827e-05,
+      "loss": 6.0069,
+      "step": 2264
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.37881286067601e-05,
+      "loss": 6.0504,
+      "step": 2265
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.376751854905194e-05,
+      "loss": 5.7871,
+      "step": 2266
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.374690849134378e-05,
+      "loss": 5.8572,
+      "step": 2267
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.372629843363561e-05,
+      "loss": 5.8462,
+      "step": 2268
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.370568837592745e-05,
+      "loss": 5.9735,
+      "step": 2269
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.36850783182193e-05,
+      "loss": 5.6834,
+      "step": 2270
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.366446826051113e-05,
+      "loss": 6.12,
+      "step": 2271
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.364385820280297e-05,
+      "loss": 5.8128,
+      "step": 2272
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.36232481450948e-05,
+      "loss": 5.8983,
+      "step": 2273
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.360263808738664e-05,
+      "loss": 5.6224,
+      "step": 2274
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.358202802967849e-05,
+      "loss": 5.9593,
+      "step": 2275
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.356141797197033e-05,
+      "loss": 5.32,
+      "step": 2276
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.354080791426217e-05,
+      "loss": 5.1954,
+      "step": 2277
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.352019785655399e-05,
+      "loss": 5.2401,
+      "step": 2278
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.349958779884583e-05,
+      "loss": 5.5798,
+      "step": 2279
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.347897774113768e-05,
+      "loss": 4.8629,
+      "step": 2280
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.345836768342952e-05,
+      "loss": 4.9714,
+      "step": 2281
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.343775762572135e-05,
+      "loss": 4.8109,
+      "step": 2282
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.34171475680132e-05,
+      "loss": 4.9339,
+      "step": 2283
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.339653751030504e-05,
+      "loss": 4.9987,
+      "step": 2284
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.337592745259687e-05,
+      "loss": 6.771,
+      "step": 2285
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.335531739488871e-05,
+      "loss": 6.2792,
+      "step": 2286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.333470733718054e-05,
+      "loss": 6.2168,
+      "step": 2287
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.331409727947238e-05,
+      "loss": 6.2771,
+      "step": 2288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.329348722176423e-05,
+      "loss": 6.2932,
+      "step": 2289
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.327287716405607e-05,
+      "loss": 6.0278,
+      "step": 2290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.32522671063479e-05,
+      "loss": 5.9444,
+      "step": 2291
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.323165704863973e-05,
+      "loss": 6.3625,
+      "step": 2292
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.321104699093157e-05,
+      "loss": 6.0556,
+      "step": 2293
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.319043693322342e-05,
+      "loss": 6.0384,
+      "step": 2294
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.316982687551526e-05,
+      "loss": 6.1691,
+      "step": 2295
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.314921681780709e-05,
+      "loss": 5.9597,
+      "step": 2296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.312860676009893e-05,
+      "loss": 5.9306,
+      "step": 2297
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.310799670239076e-05,
+      "loss": 5.9257,
+      "step": 2298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.308738664468261e-05,
+      "loss": 6.0685,
+      "step": 2299
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.306677658697445e-05,
+      "loss": 6.0871,
+      "step": 2300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.304616652926628e-05,
+      "loss": 6.1666,
+      "step": 2301
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.302555647155812e-05,
+      "loss": 5.9999,
+      "step": 2302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.300494641384997e-05,
+      "loss": 5.8489,
+      "step": 2303
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.298433635614181e-05,
+      "loss": 6.0053,
+      "step": 2304
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.296372629843364e-05,
+      "loss": 6.0093,
+      "step": 2305
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.294311624072547e-05,
+      "loss": 6.0211,
+      "step": 2306
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.292250618301731e-05,
+      "loss": 6.057,
+      "step": 2307
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.290189612530916e-05,
+      "loss": 5.8718,
+      "step": 2308
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.2881286067601e-05,
+      "loss": 5.8148,
+      "step": 2309
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 6.286067600989283e-05,
+      "loss": 5.8424,
+      "step": 2310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.284006595218467e-05,
+      "loss": 5.9733,
+      "step": 2311
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.28194558944765e-05,
+      "loss": 5.9925,
+      "step": 2312
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.279884583676835e-05,
+      "loss": 6.0461,
+      "step": 2313
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.277823577906018e-05,
+      "loss": 5.8378,
+      "step": 2314
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.275762572135202e-05,
+      "loss": 5.9278,
+      "step": 2315
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.273701566364386e-05,
+      "loss": 5.8177,
+      "step": 2316
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.271640560593571e-05,
+      "loss": 5.7545,
+      "step": 2317
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.269579554822754e-05,
+      "loss": 6.1442,
+      "step": 2318
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.267518549051937e-05,
+      "loss": 5.6581,
+      "step": 2319
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.265457543281121e-05,
+      "loss": 5.9769,
+      "step": 2320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.263396537510305e-05,
+      "loss": 6.0064,
+      "step": 2321
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.26133553173949e-05,
+      "loss": 5.6422,
+      "step": 2322
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.259274525968673e-05,
+      "loss": 5.7914,
+      "step": 2323
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.257213520197857e-05,
+      "loss": 5.5305,
+      "step": 2324
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.255152514427041e-05,
+      "loss": 5.5658,
+      "step": 2325
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.253091508656224e-05,
+      "loss": 5.3647,
+      "step": 2326
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.251030502885409e-05,
+      "loss": 5.2408,
+      "step": 2327
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 6.248969497114592e-05,
+      "loss": 5.1506,
+      "step": 2328
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.246908491343776e-05,
+      "loss": 4.9608,
+      "step": 2329
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.24484748557296e-05,
+      "loss": 4.8251,
+      "step": 2330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.242786479802145e-05,
+      "loss": 4.4761,
+      "step": 2331
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.240725474031328e-05,
+      "loss": 4.3094,
+      "step": 2332
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.23866446826051e-05,
+      "loss": 4.5418,
+      "step": 2333
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.236603462489695e-05,
+      "loss": 4.4399,
+      "step": 2334
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.234542456718879e-05,
+      "loss": 6.6193,
+      "step": 2335
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.232481450948064e-05,
+      "loss": 6.4067,
+      "step": 2336
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.230420445177247e-05,
+      "loss": 6.4091,
+      "step": 2337
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.228359439406431e-05,
+      "loss": 6.4824,
+      "step": 2338
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.226298433635614e-05,
+      "loss": 6.4644,
+      "step": 2339
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.224237427864798e-05,
+      "loss": 6.3168,
+      "step": 2340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.222176422093983e-05,
+      "loss": 6.1637,
+      "step": 2341
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.220115416323166e-05,
+      "loss": 5.93,
+      "step": 2342
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.21805441055235e-05,
+      "loss": 5.9956,
+      "step": 2343
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.215993404781534e-05,
+      "loss": 6.0299,
+      "step": 2344
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 6.213932399010717e-05,
+      "loss": 6.2086,
+      "step": 2345
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.2118713932399e-05,
+      "loss": 6.2814,
+      "step": 2346
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.209810387469085e-05,
+      "loss": 6.1956,
+      "step": 2347
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.207749381698269e-05,
+      "loss": 6.1036,
+      "step": 2348
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.205688375927453e-05,
+      "loss": 5.8926,
+      "step": 2349
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.203627370156638e-05,
+      "loss": 6.1973,
+      "step": 2350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.20156636438582e-05,
+      "loss": 5.9136,
+      "step": 2351
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.199505358615005e-05,
+      "loss": 5.8949,
+      "step": 2352
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.197444352844188e-05,
+      "loss": 6.0556,
+      "step": 2353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.195383347073372e-05,
+      "loss": 5.8397,
+      "step": 2354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.193322341302555e-05,
+      "loss": 5.8506,
+      "step": 2355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.19126133553174e-05,
+      "loss": 5.8701,
+      "step": 2356
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.189200329760924e-05,
+      "loss": 6.087,
+      "step": 2357
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.187139323990108e-05,
+      "loss": 5.6798,
+      "step": 2358
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.185078318219291e-05,
+      "loss": 6.1065,
+      "step": 2359
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.183017312448474e-05,
+      "loss": 6.1676,
+      "step": 2360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.180956306677659e-05,
+      "loss": 5.9871,
+      "step": 2361
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.178895300906843e-05,
+      "loss": 5.875,
+      "step": 2362
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 6.176834295136027e-05,
+      "loss": 5.9861,
+      "step": 2363
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.174773289365212e-05,
+      "loss": 5.9745,
+      "step": 2364
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.172712283594395e-05,
+      "loss": 5.9864,
+      "step": 2365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.170651277823577e-05,
+      "loss": 5.6138,
+      "step": 2366
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.168590272052762e-05,
+      "loss": 5.8488,
+      "step": 2367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.166529266281946e-05,
+      "loss": 5.7086,
+      "step": 2368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.164468260511129e-05,
+      "loss": 5.9017,
+      "step": 2369
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.162407254740313e-05,
+      "loss": 5.8135,
+      "step": 2370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.160346248969498e-05,
+      "loss": 5.6579,
+      "step": 2371
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.158285243198682e-05,
+      "loss": 5.6759,
+      "step": 2372
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.156224237427865e-05,
+      "loss": 5.7245,
+      "step": 2373
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.154163231657048e-05,
+      "loss": 5.5857,
+      "step": 2374
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.152102225886232e-05,
+      "loss": 5.9084,
+      "step": 2375
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.150041220115417e-05,
+      "loss": 5.7678,
+      "step": 2376
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.147980214344601e-05,
+      "loss": 5.27,
+      "step": 2377
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.145919208573784e-05,
+      "loss": 5.2836,
+      "step": 2378
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.143858202802968e-05,
+      "loss": 5.2527,
+      "step": 2379
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.141797197032151e-05,
+      "loss": 5.0726,
+      "step": 2380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 6.139736191261336e-05,
+      "loss": 4.596,
+      "step": 2381
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.13767518549052e-05,
+      "loss": 4.4636,
+      "step": 2382
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.135614179719703e-05,
+      "loss": 4.6327,
+      "step": 2383
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.133553173948887e-05,
+      "loss": 4.6107,
+      "step": 2384
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.131492168178072e-05,
+      "loss": 6.5827,
+      "step": 2385
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.129431162407255e-05,
+      "loss": 6.6261,
+      "step": 2386
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.127370156636439e-05,
+      "loss": 6.6341,
+      "step": 2387
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.125309150865622e-05,
+      "loss": 6.4606,
+      "step": 2388
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.123248145094806e-05,
+      "loss": 6.5439,
+      "step": 2389
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.121187139323991e-05,
+      "loss": 6.5022,
+      "step": 2390
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.119126133553175e-05,
+      "loss": 6.1396,
+      "step": 2391
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.117065127782358e-05,
+      "loss": 6.1479,
+      "step": 2392
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.115004122011541e-05,
+      "loss": 6.0012,
+      "step": 2393
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.112943116240725e-05,
+      "loss": 6.0761,
+      "step": 2394
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.11088211046991e-05,
+      "loss": 5.92,
+      "step": 2395
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.108821104699094e-05,
+      "loss": 6.0262,
+      "step": 2396
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.106760098928277e-05,
+      "loss": 6.0072,
+      "step": 2397
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.104699093157461e-05,
+      "loss": 6.1731,
+      "step": 2398
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 6.102638087386645e-05,
+      "loss": 6.075,
+      "step": 2399
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.1005770816158294e-05,
+      "loss": 5.9694,
+      "step": 2400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.0985160758450124e-05,
+      "loss": 6.2467,
+      "step": 2401
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.096455070074196e-05,
+      "loss": 5.9527,
+      "step": 2402
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.0943940643033804e-05,
+      "loss": 6.2593,
+      "step": 2403
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.092333058532565e-05,
+      "loss": 5.8669,
+      "step": 2404
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.0902720527617484e-05,
+      "loss": 6.0725,
+      "step": 2405
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.0882110469909314e-05,
+      "loss": 6.1165,
+      "step": 2406
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.086150041220116e-05,
+      "loss": 5.7195,
+      "step": 2407
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.0840890354492994e-05,
+      "loss": 5.8946,
+      "step": 2408
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.082028029678484e-05,
+      "loss": 5.839,
+      "step": 2409
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.079967023907667e-05,
+      "loss": 5.9338,
+      "step": 2410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.077906018136851e-05,
+      "loss": 5.8549,
+      "step": 2411
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.075845012366035e-05,
+      "loss": 5.6487,
+      "step": 2412
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.073784006595219e-05,
+      "loss": 5.8746,
+      "step": 2413
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.0717230008244034e-05,
+      "loss": 5.7437,
+      "step": 2414
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.069661995053586e-05,
+      "loss": 5.869,
+      "step": 2415
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.06760098928277e-05,
+      "loss": 6.062,
+      "step": 2416
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 6.065539983511954e-05,
+      "loss": 5.8543,
+      "step": 2417
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.063478977741138e-05,
+      "loss": 5.904,
+      "step": 2418
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.061417971970322e-05,
+      "loss": 6.0218,
+      "step": 2419
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.059356966199505e-05,
+      "loss": 5.8784,
+      "step": 2420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.0572959604286897e-05,
+      "loss": 6.022,
+      "step": 2421
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.055234954657873e-05,
+      "loss": 5.7566,
+      "step": 2422
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.0531739488870576e-05,
+      "loss": 5.7456,
+      "step": 2423
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.0511129431162406e-05,
+      "loss": 5.6459,
+      "step": 2424
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.049051937345425e-05,
+      "loss": 5.7706,
+      "step": 2425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.0469909315746086e-05,
+      "loss": 5.7006,
+      "step": 2426
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.044929925803793e-05,
+      "loss": 5.5467,
+      "step": 2427
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.0428689200329766e-05,
+      "loss": 5.2735,
+      "step": 2428
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.0408079142621596e-05,
+      "loss": 5.5859,
+      "step": 2429
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.038746908491344e-05,
+      "loss": 5.2278,
+      "step": 2430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.036685902720528e-05,
+      "loss": 5.1114,
+      "step": 2431
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.034624896949712e-05,
+      "loss": 4.7538,
+      "step": 2432
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.032563891178895e-05,
+      "loss": 4.1802,
+      "step": 2433
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.030502885408079e-05,
+      "loss": 4.4042,
+      "step": 2434
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 6.028441879637263e-05,
+      "loss": 6.5203,
+      "step": 2435
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.026380873866447e-05,
+      "loss": 6.4601,
+      "step": 2436
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.0243198680956316e-05,
+      "loss": 6.4345,
+      "step": 2437
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.0222588623248146e-05,
+      "loss": 6.2445,
+      "step": 2438
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.020197856553998e-05,
+      "loss": 6.1898,
+      "step": 2439
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.0181368507831826e-05,
+      "loss": 6.0689,
+      "step": 2440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.016075845012367e-05,
+      "loss": 6.1299,
+      "step": 2441
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.01401483924155e-05,
+      "loss": 6.1717,
+      "step": 2442
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.0119538334707336e-05,
+      "loss": 6.2791,
+      "step": 2443
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.009892827699918e-05,
+      "loss": 6.2394,
+      "step": 2444
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.0078318219291016e-05,
+      "loss": 6.1911,
+      "step": 2445
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.005770816158286e-05,
+      "loss": 6.2473,
+      "step": 2446
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.003709810387469e-05,
+      "loss": 6.1123,
+      "step": 2447
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 6.001648804616653e-05,
+      "loss": 5.8622,
+      "step": 2448
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.999587798845837e-05,
+      "loss": 5.7588,
+      "step": 2449
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.997526793075021e-05,
+      "loss": 5.8624,
+      "step": 2450
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.9954657873042056e-05,
+      "loss": 5.8168,
+      "step": 2451
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.9934047815333885e-05,
+      "loss": 5.9534,
+      "step": 2452
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.991343775762572e-05,
+      "loss": 6.191,
+      "step": 2453
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.9892827699917565e-05,
+      "loss": 5.9854,
+      "step": 2454
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.98722176422094e-05,
+      "loss": 5.9814,
+      "step": 2455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.985160758450123e-05,
+      "loss": 5.878,
+      "step": 2456
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.9830997526793075e-05,
+      "loss": 5.8879,
+      "step": 2457
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.981038746908492e-05,
+      "loss": 5.9156,
+      "step": 2458
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.9789777411376755e-05,
+      "loss": 6.0119,
+      "step": 2459
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.97691673536686e-05,
+      "loss": 5.7436,
+      "step": 2460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.974855729596043e-05,
+      "loss": 6.0177,
+      "step": 2461
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.972794723825227e-05,
+      "loss": 5.8751,
+      "step": 2462
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.970733718054411e-05,
+      "loss": 5.9553,
+      "step": 2463
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.968672712283595e-05,
+      "loss": 5.8383,
+      "step": 2464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.966611706512778e-05,
+      "loss": 5.9689,
+      "step": 2465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.964550700741962e-05,
+      "loss": 5.6316,
+      "step": 2466
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.962489694971146e-05,
+      "loss": 5.8278,
+      "step": 2467
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.9604286892003305e-05,
+      "loss": 5.668,
+      "step": 2468
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.958367683429514e-05,
+      "loss": 5.8898,
+      "step": 2469
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.956306677658697e-05,
+      "loss": 5.9813,
+      "step": 2470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9542456718878815e-05,
+      "loss": 5.596,
+      "step": 2471
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.952184666117066e-05,
+      "loss": 5.6197,
+      "step": 2472
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9501236603462495e-05,
+      "loss": 5.7545,
+      "step": 2473
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9480626545754325e-05,
+      "loss": 5.4748,
+      "step": 2474
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.946001648804617e-05,
+      "loss": 5.6478,
+      "step": 2475
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9439406430338005e-05,
+      "loss": 5.909,
+      "step": 2476
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.941879637262985e-05,
+      "loss": 5.5243,
+      "step": 2477
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.939818631492169e-05,
+      "loss": 5.891,
+      "step": 2478
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.937757625721352e-05,
+      "loss": 5.0999,
+      "step": 2479
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.935696619950536e-05,
+      "loss": 5.254,
+      "step": 2480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.93363561417972e-05,
+      "loss": 4.7389,
+      "step": 2481
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.931574608408904e-05,
+      "loss": 4.785,
+      "step": 2482
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.929513602638088e-05,
+      "loss": 4.5273,
+      "step": 2483
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.927452596867271e-05,
+      "loss": 4.6437,
+      "step": 2484
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9253915910964554e-05,
+      "loss": 6.5951,
+      "step": 2485
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.923330585325639e-05,
+      "loss": 6.4603,
+      "step": 2486
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9212695795548234e-05,
+      "loss": 6.3344,
+      "step": 2487
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.9192085737840064e-05,
+      "loss": 6.2191,
+      "step": 2488
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.917147568013191e-05,
+      "loss": 6.1617,
+      "step": 2489
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.9150865622423744e-05,
+      "loss": 6.133,
+      "step": 2490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.913025556471559e-05,
+      "loss": 5.931,
+      "step": 2491
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.9109645507007424e-05,
+      "loss": 6.122,
+      "step": 2492
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.9089035449299254e-05,
+      "loss": 5.9028,
+      "step": 2493
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.90684253915911e-05,
+      "loss": 5.8767,
+      "step": 2494
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.904781533388294e-05,
+      "loss": 5.8472,
+      "step": 2495
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.902720527617478e-05,
+      "loss": 5.7917,
+      "step": 2496
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.900659521846661e-05,
+      "loss": 5.7144,
+      "step": 2497
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.898598516075845e-05,
+      "loss": 6.0156,
+      "step": 2498
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8965375103050294e-05,
+      "loss": 5.8232,
+      "step": 2499
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.894476504534213e-05,
+      "loss": 5.9142,
+      "step": 2500
+    },
+    {
+      "epoch": 1.4,
+      "eval_loss": 18.59187889099121,
+      "eval_runtime": 1322.6566,
+      "eval_samples_per_second": 1.997,
+      "eval_steps_per_second": 0.25,
+      "eval_wer": 1.0,
+      "step": 2500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8924154987633974e-05,
+      "loss": 5.7901,
+      "step": 2501
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8903544929925804e-05,
+      "loss": 5.7028,
+      "step": 2502
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.888293487221764e-05,
+      "loss": 5.9474,
+      "step": 2503
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8862324814509484e-05,
+      "loss": 5.8746,
+      "step": 2504
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.884171475680133e-05,
+      "loss": 5.6878,
+      "step": 2505
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8821104699093164e-05,
+      "loss": 5.8153,
+      "step": 2506
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8800494641384993e-05,
+      "loss": 5.697,
+      "step": 2507
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.877988458367684e-05,
+      "loss": 5.9145,
+      "step": 2508
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.875927452596868e-05,
+      "loss": 5.721,
+      "step": 2509
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.873866446826052e-05,
+      "loss": 5.8829,
+      "step": 2510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.871805441055235e-05,
+      "loss": 5.9887,
+      "step": 2511
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.869744435284419e-05,
+      "loss": 5.7313,
+      "step": 2512
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.867683429513603e-05,
+      "loss": 5.5539,
+      "step": 2513
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.865622423742787e-05,
+      "loss": 5.5933,
+      "step": 2514
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8635614179719713e-05,
+      "loss": 5.71,
+      "step": 2515
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.861500412201154e-05,
+      "loss": 5.5441,
+      "step": 2516
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.859439406430338e-05,
+      "loss": 5.5995,
+      "step": 2517
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.857378400659522e-05,
+      "loss": 5.7763,
+      "step": 2518
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8553173948887067e-05,
+      "loss": 5.894,
+      "step": 2519
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8532563891178896e-05,
+      "loss": 5.6307,
+      "step": 2520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.851195383347073e-05,
+      "loss": 5.759,
+      "step": 2521
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8491343775762576e-05,
+      "loss": 5.6364,
+      "step": 2522
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.847073371805441e-05,
+      "loss": 5.417,
+      "step": 2523
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8450123660346256e-05,
+      "loss": 5.3691,
+      "step": 2524
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8429513602638086e-05,
+      "loss": 5.408,
+      "step": 2525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.840890354492993e-05,
+      "loss": 5.6791,
+      "step": 2526
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8388293487221766e-05,
+      "loss": 5.6272,
+      "step": 2527
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.836768342951361e-05,
+      "loss": 5.2763,
+      "step": 2528
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.834707337180544e-05,
+      "loss": 5.0643,
+      "step": 2529
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8326463314097276e-05,
+      "loss": 5.203,
+      "step": 2530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.830585325638912e-05,
+      "loss": 5.1267,
+      "step": 2531
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.828524319868096e-05,
+      "loss": 4.807,
+      "step": 2532
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.82646331409728e-05,
+      "loss": 4.9342,
+      "step": 2533
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.824402308326463e-05,
+      "loss": 4.8368,
+      "step": 2534
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.822341302555647e-05,
+      "loss": 6.4036,
+      "step": 2535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8202802967848316e-05,
+      "loss": 6.4512,
+      "step": 2536
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.818219291014015e-05,
+      "loss": 6.4055,
+      "step": 2537
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8161582852431996e-05,
+      "loss": 6.328,
+      "step": 2538
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8140972794723826e-05,
+      "loss": 5.9897,
+      "step": 2539
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.812036273701566e-05,
+      "loss": 6.1129,
+      "step": 2540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8099752679307506e-05,
+      "loss": 6.0076,
+      "step": 2541
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.807914262159935e-05,
+      "loss": 6.029,
+      "step": 2542
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.805853256389118e-05,
+      "loss": 6.0714,
+      "step": 2543
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8037922506183016e-05,
+      "loss": 5.9973,
+      "step": 2544
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.801731244847486e-05,
+      "loss": 5.8884,
+      "step": 2545
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.79967023907667e-05,
+      "loss": 5.9896,
+      "step": 2546
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.797609233305854e-05,
+      "loss": 5.8941,
+      "step": 2547
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.795548227535037e-05,
+      "loss": 5.7984,
+      "step": 2548
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.793487221764221e-05,
+      "loss": 5.7139,
+      "step": 2549
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.791426215993405e-05,
+      "loss": 5.741,
+      "step": 2550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.789365210222589e-05,
+      "loss": 5.8849,
+      "step": 2551
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.787304204451772e-05,
+      "loss": 5.9284,
+      "step": 2552
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.7852431986809565e-05,
+      "loss": 5.8631,
+      "step": 2553
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.78318219291014e-05,
+      "loss": 5.8491,
+      "step": 2554
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.7811211871393245e-05,
+      "loss": 5.9435,
+      "step": 2555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.779060181368509e-05,
+      "loss": 5.7345,
+      "step": 2556
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.776999175597692e-05,
+      "loss": 5.8416,
+      "step": 2557
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.7749381698268755e-05,
+      "loss": 5.7709,
+      "step": 2558
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.77287716405606e-05,
+      "loss": 5.7495,
+      "step": 2559
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.7708161582852435e-05,
+      "loss": 5.9021,
+      "step": 2560
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7687551525144265e-05,
+      "loss": 5.6159,
+      "step": 2561
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.766694146743611e-05,
+      "loss": 5.788,
+      "step": 2562
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.764633140972795e-05,
+      "loss": 5.7431,
+      "step": 2563
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.762572135201979e-05,
+      "loss": 5.8135,
+      "step": 2564
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.760511129431163e-05,
+      "loss": 5.7594,
+      "step": 2565
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.758450123660346e-05,
+      "loss": 6.0838,
+      "step": 2566
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7563891178895305e-05,
+      "loss": 5.7919,
+      "step": 2567
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.754328112118714e-05,
+      "loss": 5.9469,
+      "step": 2568
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7522671063478985e-05,
+      "loss": 5.9776,
+      "step": 2569
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.750206100577082e-05,
+      "loss": 5.6395,
+      "step": 2570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.748145094806265e-05,
+      "loss": 5.8936,
+      "step": 2571
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7460840890354495e-05,
+      "loss": 5.5559,
+      "step": 2572
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.744023083264634e-05,
+      "loss": 5.8402,
+      "step": 2573
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7419620774938175e-05,
+      "loss": 5.7583,
+      "step": 2574
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.7399010717230004e-05,
+      "loss": 5.4738,
+      "step": 2575
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.737840065952185e-05,
+      "loss": 5.3861,
+      "step": 2576
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.735779060181369e-05,
+      "loss": 5.6683,
+      "step": 2577
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.733718054410553e-05,
+      "loss": 5.2936,
+      "step": 2578
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.731657048639737e-05,
+      "loss": 4.868,
+      "step": 2579
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.72959604286892e-05,
+      "loss": 5.2662,
+      "step": 2580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.727535037098104e-05,
+      "loss": 4.9541,
+      "step": 2581
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.725474031327288e-05,
+      "loss": 4.8811,
+      "step": 2582
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7234130255564724e-05,
+      "loss": 4.5947,
+      "step": 2583
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7213520197856554e-05,
+      "loss": 3.9571,
+      "step": 2584
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.719291014014839e-05,
+      "loss": 6.2165,
+      "step": 2585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7172300082440234e-05,
+      "loss": 6.5307,
+      "step": 2586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.715169002473207e-05,
+      "loss": 6.2618,
+      "step": 2587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7131079967023914e-05,
+      "loss": 6.2119,
+      "step": 2588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7110469909315744e-05,
+      "loss": 5.9386,
+      "step": 2589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.708985985160759e-05,
+      "loss": 6.1379,
+      "step": 2590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.7069249793899424e-05,
+      "loss": 5.861,
+      "step": 2591
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.704863973619127e-05,
+      "loss": 6.1214,
+      "step": 2592
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.70280296784831e-05,
+      "loss": 6.1466,
+      "step": 2593
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.700741962077494e-05,
+      "loss": 6.215,
+      "step": 2594
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.698680956306678e-05,
+      "loss": 6.0027,
+      "step": 2595
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.696619950535862e-05,
+      "loss": 5.8485,
+      "step": 2596
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.694558944765046e-05,
+      "loss": 6.0161,
+      "step": 2597
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.692497938994229e-05,
+      "loss": 5.7492,
+      "step": 2598
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.690436933223413e-05,
+      "loss": 5.7642,
+      "step": 2599
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.6883759274525974e-05,
+      "loss": 5.9603,
+      "step": 2600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.686314921681781e-05,
+      "loss": 5.9208,
+      "step": 2601
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.6842539159109654e-05,
+      "loss": 5.9048,
+      "step": 2602
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.6821929101401484e-05,
+      "loss": 5.7918,
+      "step": 2603
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.680131904369333e-05,
+      "loss": 5.9366,
+      "step": 2604
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.6780708985985164e-05,
+      "loss": 5.7445,
+      "step": 2605
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.676009892827701e-05,
+      "loss": 5.7203,
+      "step": 2606
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.673948887056884e-05,
+      "loss": 5.4464,
+      "step": 2607
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.671887881286067e-05,
+      "loss": 5.8416,
+      "step": 2608
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.669826875515252e-05,
+      "loss": 5.453,
+      "step": 2609
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.667765869744436e-05,
+      "loss": 5.7977,
+      "step": 2610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.66570486397362e-05,
+      "loss": 5.7749,
+      "step": 2611
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.6636438582028027e-05,
+      "loss": 5.8265,
+      "step": 2612
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.661582852431987e-05,
+      "loss": 5.7769,
+      "step": 2613
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.659521846661171e-05,
+      "loss": 5.7212,
+      "step": 2614
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.657460840890355e-05,
+      "loss": 5.6724,
+      "step": 2615
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.655399835119538e-05,
+      "loss": 5.6212,
+      "step": 2616
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.653338829348722e-05,
+      "loss": 5.9557,
+      "step": 2617
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.651277823577906e-05,
+      "loss": 5.4544,
+      "step": 2618
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.64921681780709e-05,
+      "loss": 5.7835,
+      "step": 2619
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6471558120362747e-05,
+      "loss": 5.6501,
+      "step": 2620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6450948062654576e-05,
+      "loss": 5.6136,
+      "step": 2621
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.643033800494641e-05,
+      "loss": 5.592,
+      "step": 2622
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6409727947238256e-05,
+      "loss": 5.6509,
+      "step": 2623
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.63891178895301e-05,
+      "loss": 5.6131,
+      "step": 2624
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6368507831821936e-05,
+      "loss": 5.5366,
+      "step": 2625
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6347897774113766e-05,
+      "loss": 5.3244,
+      "step": 2626
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.632728771640561e-05,
+      "loss": 5.4092,
+      "step": 2627
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.6306677658697446e-05,
+      "loss": 5.6433,
+      "step": 2628
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.628606760098929e-05,
+      "loss": 4.909,
+      "step": 2629
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.626545754328112e-05,
+      "loss": 5.1848,
+      "step": 2630
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.624484748557296e-05,
+      "loss": 5.0638,
+      "step": 2631
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.62242374278648e-05,
+      "loss": 5.1416,
+      "step": 2632
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.620362737015664e-05,
+      "loss": 4.3338,
+      "step": 2633
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.618301731244848e-05,
+      "loss": 4.0624,
+      "step": 2634
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.616240725474031e-05,
+      "loss": 6.3927,
+      "step": 2635
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.614179719703215e-05,
+      "loss": 6.468,
+      "step": 2636
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.6121187139323996e-05,
+      "loss": 6.4229,
+      "step": 2637
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.610057708161583e-05,
+      "loss": 5.96,
+      "step": 2638
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.607996702390766e-05,
+      "loss": 6.009,
+      "step": 2639
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.6059356966199506e-05,
+      "loss": 6.0452,
+      "step": 2640
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.603874690849135e-05,
+      "loss": 5.8857,
+      "step": 2641
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.6018136850783186e-05,
+      "loss": 5.8823,
+      "step": 2642
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.599752679307503e-05,
+      "loss": 6.3062,
+      "step": 2643
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.597691673536686e-05,
+      "loss": 5.9064,
+      "step": 2644
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.5956306677658695e-05,
+      "loss": 6.1042,
+      "step": 2645
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.593569661995054e-05,
+      "loss": 5.9905,
+      "step": 2646
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.591508656224238e-05,
+      "loss": 6.0662,
+      "step": 2647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.589447650453421e-05,
+      "loss": 5.8334,
+      "step": 2648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.587386644682605e-05,
+      "loss": 6.0055,
+      "step": 2649
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.585325638911789e-05,
+      "loss": 5.6764,
+      "step": 2650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.5832646331409735e-05,
+      "loss": 5.7396,
+      "step": 2651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.581203627370157e-05,
+      "loss": 5.7029,
+      "step": 2652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.57914262159934e-05,
+      "loss": 5.8036,
+      "step": 2653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.5770816158285245e-05,
+      "loss": 5.9269,
+      "step": 2654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.575020610057708e-05,
+      "loss": 5.5839,
+      "step": 2655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.5729596042868925e-05,
+      "loss": 5.6285,
+      "step": 2656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.570898598516077e-05,
+      "loss": 5.8291,
+      "step": 2657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.56883759274526e-05,
+      "loss": 5.6403,
+      "step": 2658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.5667765869744435e-05,
+      "loss": 5.9204,
+      "step": 2659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.564715581203628e-05,
+      "loss": 5.9101,
+      "step": 2660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.562654575432812e-05,
+      "loss": 5.8895,
+      "step": 2661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.560593569661995e-05,
+      "loss": 5.679,
+      "step": 2662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.558532563891179e-05,
+      "loss": 5.8258,
+      "step": 2663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.556471558120363e-05,
+      "loss": 5.7111,
+      "step": 2664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.554410552349547e-05,
+      "loss": 5.9546,
+      "step": 2665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.552349546578731e-05,
+      "loss": 5.7587,
+      "step": 2666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.550288540807914e-05,
+      "loss": 5.7617,
+      "step": 2667
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.5482275350370985e-05,
+      "loss": 5.6652,
+      "step": 2668
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.546166529266282e-05,
+      "loss": 5.9374,
+      "step": 2669
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.5441055234954665e-05,
+      "loss": 5.692,
+      "step": 2670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.5420445177246495e-05,
+      "loss": 5.7827,
+      "step": 2671
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.539983511953834e-05,
+      "loss": 5.7402,
+      "step": 2672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.5379225061830175e-05,
+      "loss": 5.6595,
+      "step": 2673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.535861500412202e-05,
+      "loss": 5.495,
+      "step": 2674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.5338004946413855e-05,
+      "loss": 5.5096,
+      "step": 2675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.5317394888705684e-05,
+      "loss": 5.2512,
+      "step": 2676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.529678483099753e-05,
+      "loss": 5.3943,
+      "step": 2677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.527617477328937e-05,
+      "loss": 5.4625,
+      "step": 2678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.525556471558121e-05,
+      "loss": 4.8919,
+      "step": 2679
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.523495465787304e-05,
+      "loss": 4.6155,
+      "step": 2680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.521434460016488e-05,
+      "loss": 4.9008,
+      "step": 2681
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.519373454245672e-05,
+      "loss": 4.2268,
+      "step": 2682
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.517312448474856e-05,
+      "loss": 4.2809,
+      "step": 2683
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.5152514427040404e-05,
+      "loss": 3.8784,
+      "step": 2684
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.5131904369332234e-05,
+      "loss": 6.6671,
+      "step": 2685
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.511129431162407e-05,
+      "loss": 6.4423,
+      "step": 2686
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.5090684253915914e-05,
+      "loss": 6.4163,
+      "step": 2687
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.507007419620776e-05,
+      "loss": 6.4371,
+      "step": 2688
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.5049464138499594e-05,
+      "loss": 6.3983,
+      "step": 2689
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.5028854080791424e-05,
+      "loss": 6.0475,
+      "step": 2690
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.500824402308327e-05,
+      "loss": 5.9444,
+      "step": 2691
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.4987633965375104e-05,
+      "loss": 5.8132,
+      "step": 2692
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.496702390766695e-05,
+      "loss": 5.9174,
+      "step": 2693
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.494641384995878e-05,
+      "loss": 5.7761,
+      "step": 2694
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.492580379225062e-05,
+      "loss": 6.2468,
+      "step": 2695
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.490519373454246e-05,
+      "loss": 6.0611,
+      "step": 2696
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.48845836768343e-05,
+      "loss": 5.9891,
+      "step": 2697
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.4863973619126144e-05,
+      "loss": 5.7021,
+      "step": 2698
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.4843363561417974e-05,
+      "loss": 5.8497,
+      "step": 2699
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.482275350370981e-05,
+      "loss": 5.8544,
+      "step": 2700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.4802143446001654e-05,
+      "loss": 5.6569,
+      "step": 2701
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.478153338829349e-05,
+      "loss": 5.5629,
+      "step": 2702
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.476092333058532e-05,
+      "loss": 5.7124,
+      "step": 2703
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.4740313272877163e-05,
+      "loss": 6.0113,
+      "step": 2704
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.471970321516901e-05,
+      "loss": 5.4811,
+      "step": 2705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.4699093157460843e-05,
+      "loss": 5.8539,
+      "step": 2706
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.467848309975269e-05,
+      "loss": 5.6219,
+      "step": 2707
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.465787304204452e-05,
+      "loss": 5.6332,
+      "step": 2708
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.463726298433636e-05,
+      "loss": 5.7932,
+      "step": 2709
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.46166529266282e-05,
+      "loss": 5.7133,
+      "step": 2710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.459604286892004e-05,
+      "loss": 5.581,
+      "step": 2711
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.457543281121187e-05,
+      "loss": 5.7376,
+      "step": 2712
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.4554822753503706e-05,
+      "loss": 5.6705,
+      "step": 2713
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.453421269579555e-05,
+      "loss": 6.0672,
+      "step": 2714
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.451360263808739e-05,
+      "loss": 5.727,
+      "step": 2715
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.449299258037923e-05,
+      "loss": 5.6685,
+      "step": 2716
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.447238252267106e-05,
+      "loss": 5.7941,
+      "step": 2717
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.44517724649629e-05,
+      "loss": 5.4307,
+      "step": 2718
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.4431162407254746e-05,
+      "loss": 5.4934,
+      "step": 2719
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.441055234954658e-05,
+      "loss": 5.4979,
+      "step": 2720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.4389942291838426e-05,
+      "loss": 5.6014,
+      "step": 2721
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.4369332234130256e-05,
+      "loss": 5.6167,
+      "step": 2722
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.434872217642209e-05,
+      "loss": 5.7636,
+      "step": 2723
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.4328112118713936e-05,
+      "loss": 5.4219,
+      "step": 2724
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.430750206100578e-05,
+      "loss": 5.6091,
+      "step": 2725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.428689200329761e-05,
+      "loss": 5.5347,
+      "step": 2726
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.4266281945589446e-05,
+      "loss": 5.509,
+      "step": 2727
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.424567188788129e-05,
+      "loss": 5.0722,
+      "step": 2728
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.422506183017313e-05,
+      "loss": 5.1858,
+      "step": 2729
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.420445177246497e-05,
+      "loss": 5.5228,
+      "step": 2730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.41838417147568e-05,
+      "loss": 4.7512,
+      "step": 2731
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.416323165704864e-05,
+      "loss": 4.3598,
+      "step": 2732
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.414262159934048e-05,
+      "loss": 5.0303,
+      "step": 2733
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.412201154163232e-05,
+      "loss": 3.8957,
+      "step": 2734
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.410140148392415e-05,
+      "loss": 6.7927,
+      "step": 2735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.4080791426215996e-05,
+      "loss": 6.3645,
+      "step": 2736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.406018136850783e-05,
+      "loss": 6.502,
+      "step": 2737
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.4039571310799676e-05,
+      "loss": 6.3437,
+      "step": 2738
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.401896125309151e-05,
+      "loss": 6.0587,
+      "step": 2739
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.399835119538334e-05,
+      "loss": 6.2409,
+      "step": 2740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.3977741137675186e-05,
+      "loss": 6.0423,
+      "step": 2741
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.395713107996703e-05,
+      "loss": 5.9563,
+      "step": 2742
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.3936521022258866e-05,
+      "loss": 5.8911,
+      "step": 2743
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.391591096455071e-05,
+      "loss": 6.0492,
+      "step": 2744
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.389530090684254e-05,
+      "loss": 6.0198,
+      "step": 2745
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.387469084913438e-05,
+      "loss": 6.1499,
+      "step": 2746
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.385408079142622e-05,
+      "loss": 6.0223,
+      "step": 2747
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.383347073371806e-05,
+      "loss": 6.0197,
+      "step": 2748
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.381286067600989e-05,
+      "loss": 5.9114,
+      "step": 2749
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.379225061830173e-05,
+      "loss": 6.0084,
+      "step": 2750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.377164056059357e-05,
+      "loss": 5.764,
+      "step": 2751
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.3751030502885415e-05,
+      "loss": 5.6981,
+      "step": 2752
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.373042044517725e-05,
+      "loss": 5.5797,
+      "step": 2753
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.370981038746908e-05,
+      "loss": 5.8573,
+      "step": 2754
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.3689200329760925e-05,
+      "loss": 6.0034,
+      "step": 2755
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.366859027205277e-05,
+      "loss": 5.8674,
+      "step": 2756
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.3647980214344605e-05,
+      "loss": 6.0442,
+      "step": 2757
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.3627370156636435e-05,
+      "loss": 5.8236,
+      "step": 2758
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.360676009892828e-05,
+      "loss": 5.836,
+      "step": 2759
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.3586150041220115e-05,
+      "loss": 5.7103,
+      "step": 2760
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.356553998351196e-05,
+      "loss": 5.834,
+      "step": 2761
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.35449299258038e-05,
+      "loss": 5.8321,
+      "step": 2762
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.352431986809563e-05,
+      "loss": 5.715,
+      "step": 2763
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.350370981038747e-05,
+      "loss": 5.7562,
+      "step": 2764
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.348309975267931e-05,
+      "loss": 5.4303,
+      "step": 2765
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.3462489694971155e-05,
+      "loss": 5.542,
+      "step": 2766
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.3441879637262985e-05,
+      "loss": 5.7416,
+      "step": 2767
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.342126957955482e-05,
+      "loss": 5.7627,
+      "step": 2768
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.3400659521846665e-05,
+      "loss": 5.2022,
+      "step": 2769
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.33800494641385e-05,
+      "loss": 5.6851,
+      "step": 2770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.3359439406430345e-05,
+      "loss": 5.558,
+      "step": 2771
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.3338829348722174e-05,
+      "loss": 5.5541,
+      "step": 2772
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.331821929101402e-05,
+      "loss": 5.4567,
+      "step": 2773
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.3297609233305854e-05,
+      "loss": 5.4195,
+      "step": 2774
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.32769991755977e-05,
+      "loss": 5.2493,
+      "step": 2775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.325638911788954e-05,
+      "loss": 5.3501,
+      "step": 2776
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.323577906018137e-05,
+      "loss": 5.0934,
+      "step": 2777
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.321516900247321e-05,
+      "loss": 5.4446,
+      "step": 2778
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.319455894476505e-05,
+      "loss": 4.9355,
+      "step": 2779
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.317394888705689e-05,
+      "loss": 4.7122,
+      "step": 2780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.315333882934872e-05,
+      "loss": 4.3588,
+      "step": 2781
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.313272877164056e-05,
+      "loss": 4.4043,
+      "step": 2782
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.3112118713932404e-05,
+      "loss": 4.6048,
+      "step": 2783
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.309150865622424e-05,
+      "loss": 3.5184,
+      "step": 2784
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.3070898598516084e-05,
+      "loss": 6.2321,
+      "step": 2785
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.3050288540807914e-05,
+      "loss": 6.6032,
+      "step": 2786
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.302967848309975e-05,
+      "loss": 6.4747,
+      "step": 2787
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.3009068425391594e-05,
+      "loss": 6.0285,
+      "step": 2788
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.298845836768344e-05,
+      "loss": 6.0427,
+      "step": 2789
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.296784830997527e-05,
+      "loss": 5.832,
+      "step": 2790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.2947238252267104e-05,
+      "loss": 6.0142,
+      "step": 2791
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.292662819455895e-05,
+      "loss": 5.7314,
+      "step": 2792
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.290601813685079e-05,
+      "loss": 5.7732,
+      "step": 2793
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.288540807914263e-05,
+      "loss": 5.9518,
+      "step": 2794
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.286479802143446e-05,
+      "loss": 5.9541,
+      "step": 2795
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.28441879637263e-05,
+      "loss": 5.9807,
+      "step": 2796
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.282357790601814e-05,
+      "loss": 6.0053,
+      "step": 2797
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.280296784830998e-05,
+      "loss": 6.1149,
+      "step": 2798
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.278235779060181e-05,
+      "loss": 5.8564,
+      "step": 2799
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.2761747732893654e-05,
+      "loss": 5.6504,
+      "step": 2800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.274113767518549e-05,
+      "loss": 5.7547,
+      "step": 2801
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.2720527617477334e-05,
+      "loss": 5.7427,
+      "step": 2802
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.269991755976918e-05,
+      "loss": 5.8852,
+      "step": 2803
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.267930750206101e-05,
+      "loss": 5.5738,
+      "step": 2804
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.2658697444352843e-05,
+      "loss": 5.6673,
+      "step": 2805
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.263808738664469e-05,
+      "loss": 5.9246,
+      "step": 2806
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.261747732893652e-05,
+      "loss": 5.7039,
+      "step": 2807
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.259686727122837e-05,
+      "loss": 5.9319,
+      "step": 2808
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.2576257213520197e-05,
+      "loss": 5.5924,
+      "step": 2809
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.255564715581204e-05,
+      "loss": 5.3909,
+      "step": 2810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.2535037098103877e-05,
+      "loss": 5.713,
+      "step": 2811
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.251442704039572e-05,
+      "loss": 5.6611,
+      "step": 2812
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.249381698268755e-05,
+      "loss": 5.9764,
+      "step": 2813
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.247320692497939e-05,
+      "loss": 5.757,
+      "step": 2814
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.245259686727123e-05,
+      "loss": 5.6443,
+      "step": 2815
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.243198680956307e-05,
+      "loss": 5.4124,
+      "step": 2816
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.241137675185491e-05,
+      "loss": 5.4078,
+      "step": 2817
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.239076669414674e-05,
+      "loss": 5.3505,
+      "step": 2818
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.237015663643858e-05,
+      "loss": 5.7052,
+      "step": 2819
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.2349546578730426e-05,
+      "loss": 5.4948,
+      "step": 2820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.232893652102226e-05,
+      "loss": 5.4401,
+      "step": 2821
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.230832646331409e-05,
+      "loss": 5.3929,
+      "step": 2822
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.2287716405605936e-05,
+      "loss": 5.4454,
+      "step": 2823
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.226710634789778e-05,
+      "loss": 5.4855,
+      "step": 2824
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.2246496290189616e-05,
+      "loss": 5.4984,
+      "step": 2825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.222588623248146e-05,
+      "loss": 5.3734,
+      "step": 2826
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.220527617477329e-05,
+      "loss": 5.2363,
+      "step": 2827
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.2184666117065126e-05,
+      "loss": 5.4789,
+      "step": 2828
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.216405605935697e-05,
+      "loss": 4.969,
+      "step": 2829
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.214344600164881e-05,
+      "loss": 4.8523,
+      "step": 2830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.212283594394065e-05,
+      "loss": 4.2979,
+      "step": 2831
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.210222588623248e-05,
+      "loss": 4.2262,
+      "step": 2832
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.208161582852432e-05,
+      "loss": 4.138,
+      "step": 2833
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.206100577081616e-05,
+      "loss": 4.2949,
+      "step": 2834
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.2040395713108e-05,
+      "loss": 6.6696,
+      "step": 2835
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.201978565539983e-05,
+      "loss": 6.254,
+      "step": 2836
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.1999175597691676e-05,
+      "loss": 6.2813,
+      "step": 2837
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.197856553998351e-05,
+      "loss": 6.3864,
+      "step": 2838
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.1957955482275356e-05,
+      "loss": 6.3677,
+      "step": 2839
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.19373454245672e-05,
+      "loss": 6.0178,
+      "step": 2840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.191673536685903e-05,
+      "loss": 6.0931,
+      "step": 2841
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.1896125309150865e-05,
+      "loss": 5.9587,
+      "step": 2842
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.187551525144271e-05,
+      "loss": 6.1064,
+      "step": 2843
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.1854905193734545e-05,
+      "loss": 6.1668,
+      "step": 2844
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.1834295136026375e-05,
+      "loss": 6.0073,
+      "step": 2845
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.181368507831822e-05,
+      "loss": 6.0163,
+      "step": 2846
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.179307502061006e-05,
+      "loss": 6.0154,
+      "step": 2847
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.17724649629019e-05,
+      "loss": 5.9107,
+      "step": 2848
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.175185490519374e-05,
+      "loss": 5.7459,
+      "step": 2849
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.173124484748557e-05,
+      "loss": 5.8149,
+      "step": 2850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.1710634789777415e-05,
+      "loss": 5.7108,
+      "step": 2851
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.169002473206925e-05,
+      "loss": 5.7573,
+      "step": 2852
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.1669414674361095e-05,
+      "loss": 5.9029,
+      "step": 2853
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.1648804616652925e-05,
+      "loss": 5.6344,
+      "step": 2854
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.162819455894476e-05,
+      "loss": 5.6151,
+      "step": 2855
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.1607584501236605e-05,
+      "loss": 5.7319,
+      "step": 2856
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.158697444352845e-05,
+      "loss": 5.5918,
+      "step": 2857
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.1566364385820285e-05,
+      "loss": 5.6666,
+      "step": 2858
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.1545754328112115e-05,
+      "loss": 5.4391,
+      "step": 2859
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.152514427040396e-05,
+      "loss": 5.4475,
+      "step": 2860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.15045342126958e-05,
+      "loss": 5.8912,
+      "step": 2861
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.148392415498764e-05,
+      "loss": 5.6641,
+      "step": 2862
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.146331409727948e-05,
+      "loss": 5.8244,
+      "step": 2863
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.144270403957131e-05,
+      "loss": 5.8281,
+      "step": 2864
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.142209398186315e-05,
+      "loss": 5.2527,
+      "step": 2865
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.140148392415499e-05,
+      "loss": 5.5685,
+      "step": 2866
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.1380873866446835e-05,
+      "loss": 5.3888,
+      "step": 2867
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.1360263808738665e-05,
+      "loss": 5.4446,
+      "step": 2868
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.13396537510305e-05,
+      "loss": 5.5593,
+      "step": 2869
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.1319043693322345e-05,
+      "loss": 5.226,
+      "step": 2870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.129843363561419e-05,
+      "loss": 5.473,
+      "step": 2871
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.1277823577906025e-05,
+      "loss": 5.33,
+      "step": 2872
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.1257213520197854e-05,
+      "loss": 5.5023,
+      "step": 2873
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.12366034624897e-05,
+      "loss": 5.4893,
+      "step": 2874
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.1215993404781534e-05,
+      "loss": 5.4934,
+      "step": 2875
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.119538334707338e-05,
+      "loss": 5.2252,
+      "step": 2876
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.117477328936521e-05,
+      "loss": 4.9892,
+      "step": 2877
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.115416323165705e-05,
+      "loss": 5.4179,
+      "step": 2878
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.113355317394889e-05,
+      "loss": 5.015,
+      "step": 2879
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.111294311624073e-05,
+      "loss": 5.0428,
+      "step": 2880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.1092333058532574e-05,
+      "loss": 5.0233,
+      "step": 2881
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.10717230008244e-05,
+      "loss": 4.7935,
+      "step": 2882
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.105111294311624e-05,
+      "loss": 4.1623,
+      "step": 2883
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.1030502885408084e-05,
+      "loss": 4.1555,
+      "step": 2884
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.100989282769992e-05,
+      "loss": 6.483,
+      "step": 2885
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.098928276999175e-05,
+      "loss": 6.3773,
+      "step": 2886
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.0968672712283594e-05,
+      "loss": 6.3269,
+      "step": 2887
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.094806265457544e-05,
+      "loss": 6.1417,
+      "step": 2888
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.0927452596867274e-05,
+      "loss": 6.2228,
+      "step": 2889
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.090684253915912e-05,
+      "loss": 5.6857,
+      "step": 2890
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.088623248145095e-05,
+      "loss": 5.8776,
+      "step": 2891
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.0865622423742784e-05,
+      "loss": 5.8407,
+      "step": 2892
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.084501236603463e-05,
+      "loss": 5.8435,
+      "step": 2893
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.082440230832647e-05,
+      "loss": 5.824,
+      "step": 2894
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.080379225061831e-05,
+      "loss": 5.789,
+      "step": 2895
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.078318219291014e-05,
+      "loss": 5.8227,
+      "step": 2896
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.076257213520198e-05,
+      "loss": 5.9495,
+      "step": 2897
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.0741962077493824e-05,
+      "loss": 5.7963,
+      "step": 2898
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.072135201978566e-05,
+      "loss": 5.5862,
+      "step": 2899
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.070074196207749e-05,
+      "loss": 5.7162,
+      "step": 2900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.0680131904369333e-05,
+      "loss": 5.6098,
+      "step": 2901
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.065952184666117e-05,
+      "loss": 5.694,
+      "step": 2902
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.0638911788953013e-05,
+      "loss": 5.7095,
+      "step": 2903
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.061830173124486e-05,
+      "loss": 5.596,
+      "step": 2904
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.059769167353669e-05,
+      "loss": 5.7468,
+      "step": 2905
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.057708161582852e-05,
+      "loss": 5.7125,
+      "step": 2906
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.055647155812037e-05,
+      "loss": 5.8804,
+      "step": 2907
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.053586150041221e-05,
+      "loss": 5.6151,
+      "step": 2908
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.051525144270404e-05,
+      "loss": 5.4165,
+      "step": 2909
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.0494641384995876e-05,
+      "loss": 5.4888,
+      "step": 2910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.047403132728772e-05,
+      "loss": 5.6692,
+      "step": 2911
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.0453421269579556e-05,
+      "loss": 5.6111,
+      "step": 2912
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.04328112118714e-05,
+      "loss": 5.8472,
+      "step": 2913
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.041220115416323e-05,
+      "loss": 5.6902,
+      "step": 2914
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.039159109645507e-05,
+      "loss": 5.5513,
+      "step": 2915
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.037098103874691e-05,
+      "loss": 5.3788,
+      "step": 2916
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.035037098103875e-05,
+      "loss": 5.4353,
+      "step": 2917
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.032976092333058e-05,
+      "loss": 5.5274,
+      "step": 2918
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.0309150865622426e-05,
+      "loss": 5.5732,
+      "step": 2919
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.028854080791426e-05,
+      "loss": 5.472,
+      "step": 2920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.0267930750206106e-05,
+      "loss": 5.4856,
+      "step": 2921
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.024732069249794e-05,
+      "loss": 5.5026,
+      "step": 2922
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.022671063478977e-05,
+      "loss": 5.4506,
+      "step": 2923
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.0206100577081616e-05,
+      "loss": 5.1028,
+      "step": 2924
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.018549051937346e-05,
+      "loss": 5.3179,
+      "step": 2925
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.0164880461665296e-05,
+      "loss": 5.4887,
+      "step": 2926
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.014427040395714e-05,
+      "loss": 4.9979,
+      "step": 2927
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.012366034624897e-05,
+      "loss": 5.532,
+      "step": 2928
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.010305028854081e-05,
+      "loss": 4.8991,
+      "step": 2929
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.008244023083265e-05,
+      "loss": 4.9945,
+      "step": 2930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.006183017312449e-05,
+      "loss": 4.5562,
+      "step": 2931
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.004122011541632e-05,
+      "loss": 4.7683,
+      "step": 2932
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.002061005770816e-05,
+      "loss": 4.4515,
+      "step": 2933
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5e-05,
+      "loss": 3.6295,
+      "step": 2934
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9979389942291846e-05,
+      "loss": 6.0934,
+      "step": 2935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9958779884583676e-05,
+      "loss": 6.2427,
+      "step": 2936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.993816982687552e-05,
+      "loss": 6.2501,
+      "step": 2937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9917559769167356e-05,
+      "loss": 5.9979,
+      "step": 2938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.989694971145919e-05,
+      "loss": 5.8803,
+      "step": 2939
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.987633965375103e-05,
+      "loss": 5.6942,
+      "step": 2940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.985572959604287e-05,
+      "loss": 5.8034,
+      "step": 2941
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9835119538334716e-05,
+      "loss": 5.641,
+      "step": 2942
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9814509480626545e-05,
+      "loss": 6.0814,
+      "step": 2943
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.979389942291839e-05,
+      "loss": 5.8509,
+      "step": 2944
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9773289365210225e-05,
+      "loss": 5.931,
+      "step": 2945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.975267930750206e-05,
+      "loss": 5.9978,
+      "step": 2946
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.97320692497939e-05,
+      "loss": 6.0963,
+      "step": 2947
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.971145919208574e-05,
+      "loss": 5.8914,
+      "step": 2948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.969084913437758e-05,
+      "loss": 5.5398,
+      "step": 2949
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9670239076669415e-05,
+      "loss": 5.7136,
+      "step": 2950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.964962901896126e-05,
+      "loss": 5.7492,
+      "step": 2951
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.9629018961253095e-05,
+      "loss": 5.4765,
+      "step": 2952
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.960840890354493e-05,
+      "loss": 5.5388,
+      "step": 2953
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.958779884583677e-05,
+      "loss": 5.7643,
+      "step": 2954
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.956718878812861e-05,
+      "loss": 5.7582,
+      "step": 2955
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.954657873042045e-05,
+      "loss": 5.9383,
+      "step": 2956
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9525968672712285e-05,
+      "loss": 5.3801,
+      "step": 2957
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.950535861500413e-05,
+      "loss": 5.7217,
+      "step": 2958
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9484748557295965e-05,
+      "loss": 5.5787,
+      "step": 2959
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.94641384995878e-05,
+      "loss": 5.7074,
+      "step": 2960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.944352844187964e-05,
+      "loss": 5.6082,
+      "step": 2961
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.942291838417148e-05,
+      "loss": 5.4113,
+      "step": 2962
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.940230832646331e-05,
+      "loss": 5.6214,
+      "step": 2963
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9381698268755155e-05,
+      "loss": 5.695,
+      "step": 2964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.936108821104699e-05,
+      "loss": 5.6469,
+      "step": 2965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.9340478153338835e-05,
+      "loss": 5.3937,
+      "step": 2966
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.931986809563067e-05,
+      "loss": 5.3793,
+      "step": 2967
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.929925803792251e-05,
+      "loss": 5.4278,
+      "step": 2968
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.927864798021435e-05,
+      "loss": 5.3329,
+      "step": 2969
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.925803792250618e-05,
+      "loss": 5.4046,
+      "step": 2970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9237427864798024e-05,
+      "loss": 5.3621,
+      "step": 2971
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.921681780708986e-05,
+      "loss": 5.5643,
+      "step": 2972
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.91962077493817e-05,
+      "loss": 5.7094,
+      "step": 2973
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.917559769167354e-05,
+      "loss": 5.7099,
+      "step": 2974
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.915498763396538e-05,
+      "loss": 5.5654,
+      "step": 2975
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.913437757625722e-05,
+      "loss": 5.2909,
+      "step": 2976
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.911376751854905e-05,
+      "loss": 5.4886,
+      "step": 2977
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9093157460840894e-05,
+      "loss": 5.1702,
+      "step": 2978
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.907254740313273e-05,
+      "loss": 5.166,
+      "step": 2979
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.905193734542457e-05,
+      "loss": 4.7005,
+      "step": 2980
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9031327287716404e-05,
+      "loss": 4.609,
+      "step": 2981
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.901071723000825e-05,
+      "loss": 4.7536,
+      "step": 2982
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.8990107172300084e-05,
+      "loss": 4.2948,
+      "step": 2983
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.896949711459192e-05,
+      "loss": 4.3511,
+      "step": 2984
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.8948887056883764e-05,
+      "loss": 6.6435,
+      "step": 2985
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.89282769991756e-05,
+      "loss": 6.4827,
+      "step": 2986
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.890766694146744e-05,
+      "loss": 6.445,
+      "step": 2987
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.8887056883759274e-05,
+      "loss": 6.3789,
+      "step": 2988
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.886644682605112e-05,
+      "loss": 6.3487,
+      "step": 2989
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8845836768342954e-05,
+      "loss": 6.1719,
+      "step": 2990
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.882522671063479e-05,
+      "loss": 6.3097,
+      "step": 2991
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8804616652926634e-05,
+      "loss": 5.851,
+      "step": 2992
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.878400659521847e-05,
+      "loss": 5.9957,
+      "step": 2993
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.876339653751031e-05,
+      "loss": 5.6974,
+      "step": 2994
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8742786479802144e-05,
+      "loss": 5.7307,
+      "step": 2995
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.872217642209399e-05,
+      "loss": 5.8638,
+      "step": 2996
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.870156636438582e-05,
+      "loss": 5.9128,
+      "step": 2997
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.868095630667766e-05,
+      "loss": 5.9028,
+      "step": 2998
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8660346248969504e-05,
+      "loss": 5.7087,
+      "step": 2999
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.863973619126134e-05,
+      "loss": 5.6761,
+      "step": 3000
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 16.960098266601562,
+      "eval_runtime": 1334.4635,
+      "eval_samples_per_second": 1.98,
+      "eval_steps_per_second": 0.248,
+      "eval_wer": 1.0,
+      "step": 3000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.861912613355318e-05,
+      "loss": 5.5905,
+      "step": 3001
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8598516075845013e-05,
+      "loss": 5.5754,
+      "step": 3002
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.857790601813686e-05,
+      "loss": 5.5713,
+      "step": 3003
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8557295960428687e-05,
+      "loss": 5.9027,
+      "step": 3004
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.853668590272053e-05,
+      "loss": 5.711,
+      "step": 3005
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.851607584501237e-05,
+      "loss": 5.7977,
+      "step": 3006
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.84954657873042e-05,
+      "loss": 5.8576,
+      "step": 3007
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8474855729596047e-05,
+      "loss": 5.4058,
+      "step": 3008
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.845424567188788e-05,
+      "loss": 5.5432,
+      "step": 3009
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8433635614179727e-05,
+      "loss": 5.6276,
+      "step": 3010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8413025556471556e-05,
+      "loss": 5.7842,
+      "step": 3011
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.83924154987634e-05,
+      "loss": 5.6691,
+      "step": 3012
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8371805441055236e-05,
+      "loss": 5.6965,
+      "step": 3013
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.835119538334707e-05,
+      "loss": 5.5238,
+      "step": 3014
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8330585325638916e-05,
+      "loss": 5.4294,
+      "step": 3015
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.830997526793075e-05,
+      "loss": 5.5859,
+      "step": 3016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.828936521022259e-05,
+      "loss": 5.8723,
+      "step": 3017
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8268755152514426e-05,
+      "loss": 5.4862,
+      "step": 3018
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.824814509480627e-05,
+      "loss": 5.6382,
+      "step": 3019
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8227535037098106e-05,
+      "loss": 5.3826,
+      "step": 3020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.820692497938994e-05,
+      "loss": 5.5957,
+      "step": 3021
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.8186314921681786e-05,
+      "loss": 5.3667,
+      "step": 3022
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.816570486397362e-05,
+      "loss": 5.3344,
+      "step": 3023
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.814509480626546e-05,
+      "loss": 5.4663,
+      "step": 3024
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.8124484748557296e-05,
+      "loss": 5.6224,
+      "step": 3025
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.810387469084914e-05,
+      "loss": 5.6679,
+      "step": 3026
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.8083264633140976e-05,
+      "loss": 5.3889,
+      "step": 3027
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.806265457543281e-05,
+      "loss": 5.1358,
+      "step": 3028
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.804204451772465e-05,
+      "loss": 4.8834,
+      "step": 3029
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.802143446001649e-05,
+      "loss": 5.2398,
+      "step": 3030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.800082440230833e-05,
+      "loss": 4.9875,
+      "step": 3031
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7980214344600166e-05,
+      "loss": 4.5499,
+      "step": 3032
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.795960428689201e-05,
+      "loss": 4.5488,
+      "step": 3033
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7938994229183846e-05,
+      "loss": 3.7851,
+      "step": 3034
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.791838417147568e-05,
+      "loss": 6.247,
+      "step": 3035
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.789777411376752e-05,
+      "loss": 6.1313,
+      "step": 3036
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.787716405605936e-05,
+      "loss": 6.2583,
+      "step": 3037
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.78565539983512e-05,
+      "loss": 5.9628,
+      "step": 3038
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.7835943940643035e-05,
+      "loss": 6.0476,
+      "step": 3039
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.781533388293488e-05,
+      "loss": 5.7625,
+      "step": 3040
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.779472382522671e-05,
+      "loss": 6.0996,
+      "step": 3041
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.777411376751855e-05,
+      "loss": 5.4776,
+      "step": 3042
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.775350370981039e-05,
+      "loss": 5.6714,
+      "step": 3043
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7732893652102225e-05,
+      "loss": 5.6129,
+      "step": 3044
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.771228359439407e-05,
+      "loss": 5.7005,
+      "step": 3045
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7691673536685905e-05,
+      "loss": 5.845,
+      "step": 3046
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.767106347897775e-05,
+      "loss": 5.7577,
+      "step": 3047
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.765045342126958e-05,
+      "loss": 5.7636,
+      "step": 3048
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.762984336356142e-05,
+      "loss": 5.4162,
+      "step": 3049
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.760923330585326e-05,
+      "loss": 5.4843,
+      "step": 3050
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7588623248145095e-05,
+      "loss": 5.5686,
+      "step": 3051
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.756801319043693e-05,
+      "loss": 5.6714,
+      "step": 3052
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7547403132728775e-05,
+      "loss": 5.795,
+      "step": 3053
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.752679307502061e-05,
+      "loss": 5.5836,
+      "step": 3054
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.750618301731245e-05,
+      "loss": 5.866,
+      "step": 3055
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.748557295960429e-05,
+      "loss": 5.4484,
+      "step": 3056
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.746496290189613e-05,
+      "loss": 5.8637,
+      "step": 3057
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.7444352844187965e-05,
+      "loss": 5.6725,
+      "step": 3058
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.74237427864798e-05,
+      "loss": 5.719,
+      "step": 3059
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7403132728771645e-05,
+      "loss": 5.6675,
+      "step": 3060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.738252267106348e-05,
+      "loss": 5.4754,
+      "step": 3061
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.736191261335532e-05,
+      "loss": 5.7349,
+      "step": 3062
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.734130255564716e-05,
+      "loss": 5.4839,
+      "step": 3063
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7320692497939e-05,
+      "loss": 5.559,
+      "step": 3064
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7300082440230835e-05,
+      "loss": 5.3717,
+      "step": 3065
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.727947238252267e-05,
+      "loss": 5.4572,
+      "step": 3066
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7258862324814515e-05,
+      "loss": 5.326,
+      "step": 3067
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7238252267106344e-05,
+      "loss": 5.7818,
+      "step": 3068
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.721764220939819e-05,
+      "loss": 5.5257,
+      "step": 3069
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.719703215169003e-05,
+      "loss": 5.6525,
+      "step": 3070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.717642209398187e-05,
+      "loss": 5.6279,
+      "step": 3071
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7155812036273704e-05,
+      "loss": 5.3997,
+      "step": 3072
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.713520197856554e-05,
+      "loss": 5.551,
+      "step": 3073
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7114591920857384e-05,
+      "loss": 5.4583,
+      "step": 3074
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.7093981863149214e-05,
+      "loss": 5.3976,
+      "step": 3075
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.707337180544106e-05,
+      "loss": 5.2285,
+      "step": 3076
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.70527617477329e-05,
+      "loss": 5.3089,
+      "step": 3077
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.703215169002473e-05,
+      "loss": 4.9581,
+      "step": 3078
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7011541632316574e-05,
+      "loss": 4.9583,
+      "step": 3079
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.699093157460841e-05,
+      "loss": 4.75,
+      "step": 3080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6970321516900254e-05,
+      "loss": 4.5465,
+      "step": 3081
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6949711459192084e-05,
+      "loss": 4.6335,
+      "step": 3082
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.692910140148393e-05,
+      "loss": 4.6033,
+      "step": 3083
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6908491343775764e-05,
+      "loss": 3.4964,
+      "step": 3084
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.68878812860676e-05,
+      "loss": 6.3402,
+      "step": 3085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6867271228359444e-05,
+      "loss": 6.3707,
+      "step": 3086
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.684666117065128e-05,
+      "loss": 6.14,
+      "step": 3087
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.682605111294312e-05,
+      "loss": 6.2036,
+      "step": 3088
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6805441055234954e-05,
+      "loss": 6.0157,
+      "step": 3089
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.67848309975268e-05,
+      "loss": 5.775,
+      "step": 3090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6764220939818634e-05,
+      "loss": 5.8102,
+      "step": 3091
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.674361088211047e-05,
+      "loss": 5.578,
+      "step": 3092
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.6723000824402314e-05,
+      "loss": 5.547,
+      "step": 3093
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.670239076669415e-05,
+      "loss": 6.0101,
+      "step": 3094
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.668178070898599e-05,
+      "loss": 6.1514,
+      "step": 3095
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.6661170651277824e-05,
+      "loss": 5.9794,
+      "step": 3096
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.664056059356967e-05,
+      "loss": 5.8762,
+      "step": 3097
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.6619950535861504e-05,
+      "loss": 5.8051,
+      "step": 3098
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.659934047815334e-05,
+      "loss": 5.685,
+      "step": 3099
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.657873042044518e-05,
+      "loss": 5.6357,
+      "step": 3100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.655812036273702e-05,
+      "loss": 5.4024,
+      "step": 3101
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.653751030502886e-05,
+      "loss": 5.7835,
+      "step": 3102
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.651690024732069e-05,
+      "loss": 5.7811,
+      "step": 3103
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.649629018961254e-05,
+      "loss": 5.7703,
+      "step": 3104
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.647568013190437e-05,
+      "loss": 5.6608,
+      "step": 3105
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.645507007419621e-05,
+      "loss": 5.985,
+      "step": 3106
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.6434460016488046e-05,
+      "loss": 5.4905,
+      "step": 3107
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.641384995877989e-05,
+      "loss": 5.5153,
+      "step": 3108
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.6393239901071726e-05,
+      "loss": 5.626,
+      "step": 3109
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.637262984336356e-05,
+      "loss": 5.7921,
+      "step": 3110
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.6352019785655406e-05,
+      "loss": 5.5655,
+      "step": 3111
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.6331409727947236e-05,
+      "loss": 5.7692,
+      "step": 3112
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.631079967023908e-05,
+      "loss": 5.3995,
+      "step": 3113
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.6290189612530916e-05,
+      "loss": 5.6182,
+      "step": 3114
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.626957955482275e-05,
+      "loss": 5.5781,
+      "step": 3115
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.624896949711459e-05,
+      "loss": 5.4869,
+      "step": 3116
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.622835943940643e-05,
+      "loss": 5.562,
+      "step": 3117
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.6207749381698276e-05,
+      "loss": 5.4102,
+      "step": 3118
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.6187139323990106e-05,
+      "loss": 5.3611,
+      "step": 3119
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.616652926628195e-05,
+      "loss": 5.6127,
+      "step": 3120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.6145919208573786e-05,
+      "loss": 5.5359,
+      "step": 3121
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.612530915086562e-05,
+      "loss": 5.2957,
+      "step": 3122
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.610469909315746e-05,
+      "loss": 5.2826,
+      "step": 3123
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.60840890354493e-05,
+      "loss": 5.1736,
+      "step": 3124
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.606347897774114e-05,
+      "loss": 5.1955,
+      "step": 3125
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.6042868920032976e-05,
+      "loss": 5.2258,
+      "step": 3126
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.602225886232482e-05,
+      "loss": 5.1112,
+      "step": 3127
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.6001648804616656e-05,
+      "loss": 4.5242,
+      "step": 3128
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.598103874690849e-05,
+      "loss": 5.0299,
+      "step": 3129
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.596042868920033e-05,
+      "loss": 4.7415,
+      "step": 3130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.593981863149217e-05,
+      "loss": 4.2763,
+      "step": 3131
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.591920857378401e-05,
+      "loss": 4.6881,
+      "step": 3132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.5898598516075846e-05,
+      "loss": 4.4063,
+      "step": 3133
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.587798845836769e-05,
+      "loss": 3.8932,
+      "step": 3134
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.5857378400659526e-05,
+      "loss": 6.5079,
+      "step": 3135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.583676834295136e-05,
+      "loss": 6.6432,
+      "step": 3136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.58161582852432e-05,
+      "loss": 6.4147,
+      "step": 3137
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.579554822753504e-05,
+      "loss": 6.3245,
+      "step": 3138
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.577493816982687e-05,
+      "loss": 6.1681,
+      "step": 3139
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.5754328112118715e-05,
+      "loss": 6.1698,
+      "step": 3140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.573371805441056e-05,
+      "loss": 6.1969,
+      "step": 3141
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.5713107996702395e-05,
+      "loss": 5.8063,
+      "step": 3142
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.569249793899423e-05,
+      "loss": 5.907,
+      "step": 3143
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.567188788128607e-05,
+      "loss": 5.8898,
+      "step": 3144
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.565127782357791e-05,
+      "loss": 5.5589,
+      "step": 3145
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.563066776586974e-05,
+      "loss": 5.8336,
+      "step": 3146
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.5610057708161585e-05,
+      "loss": 6.1281,
+      "step": 3147
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.558944765045342e-05,
+      "loss": 5.8023,
+      "step": 3148
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.556883759274526e-05,
+      "loss": 5.8118,
+      "step": 3149
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.55482275350371e-05,
+      "loss": 5.7816,
+      "step": 3150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.552761747732894e-05,
+      "loss": 5.5165,
+      "step": 3151
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.550700741962078e-05,
+      "loss": 5.6958,
+      "step": 3152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.548639736191261e-05,
+      "loss": 5.5323,
+      "step": 3153
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.5465787304204455e-05,
+      "loss": 5.8422,
+      "step": 3154
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.544517724649629e-05,
+      "loss": 5.7006,
+      "step": 3155
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.542456718878813e-05,
+      "loss": 5.4958,
+      "step": 3156
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.540395713107997e-05,
+      "loss": 5.6166,
+      "step": 3157
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.538334707337181e-05,
+      "loss": 5.4958,
+      "step": 3158
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.5362737015663645e-05,
+      "loss": 5.3287,
+      "step": 3159
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.534212695795548e-05,
+      "loss": 5.3935,
+      "step": 3160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.5321516900247325e-05,
+      "loss": 5.1473,
+      "step": 3161
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.530090684253916e-05,
+      "loss": 5.4204,
+      "step": 3162
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.5280296784831e-05,
+      "loss": 5.3813,
+      "step": 3163
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.525968672712284e-05,
+      "loss": 5.7133,
+      "step": 3164
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.523907666941468e-05,
+      "loss": 5.677,
+      "step": 3165
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.5218466611706515e-05,
+      "loss": 5.5851,
+      "step": 3166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.519785655399835e-05,
+      "loss": 5.5578,
+      "step": 3167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.5177246496290194e-05,
+      "loss": 5.5437,
+      "step": 3168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.515663643858203e-05,
+      "loss": 5.2055,
+      "step": 3169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.513602638087387e-05,
+      "loss": 5.1145,
+      "step": 3170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.5115416323165704e-05,
+      "loss": 5.6634,
+      "step": 3171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.509480626545755e-05,
+      "loss": 5.192,
+      "step": 3172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.5074196207749384e-05,
+      "loss": 5.1495,
+      "step": 3173
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.505358615004122e-05,
+      "loss": 5.2917,
+      "step": 3174
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.5032976092333064e-05,
+      "loss": 5.1774,
+      "step": 3175
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.50123660346249e-05,
+      "loss": 5.016,
+      "step": 3176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.499175597691674e-05,
+      "loss": 5.0582,
+      "step": 3177
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4971145919208574e-05,
+      "loss": 4.6815,
+      "step": 3178
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.495053586150042e-05,
+      "loss": 4.575,
+      "step": 3179
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4929925803792254e-05,
+      "loss": 4.3785,
+      "step": 3180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.490931574608409e-05,
+      "loss": 4.6115,
+      "step": 3181
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4888705688375934e-05,
+      "loss": 4.397,
+      "step": 3182
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4868095630667764e-05,
+      "loss": 4.5611,
+      "step": 3183
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.484748557295961e-05,
+      "loss": 3.9241,
+      "step": 3184
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.4826875515251444e-05,
+      "loss": 6.2305,
+      "step": 3185
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.480626545754329e-05,
+      "loss": 6.4099,
+      "step": 3186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.478565539983512e-05,
+      "loss": 6.3702,
+      "step": 3187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.476504534212696e-05,
+      "loss": 6.2495,
+      "step": 3188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.4744435284418804e-05,
+      "loss": 6.1973,
+      "step": 3189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.4723825226710634e-05,
+      "loss": 6.2641,
+      "step": 3190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.470321516900248e-05,
+      "loss": 5.9749,
+      "step": 3191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.4682605111294314e-05,
+      "loss": 5.7804,
+      "step": 3192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.466199505358615e-05,
+      "loss": 5.5762,
+      "step": 3193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.464138499587799e-05,
+      "loss": 5.759,
+      "step": 3194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.462077493816983e-05,
+      "loss": 6.156,
+      "step": 3195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.460016488046167e-05,
+      "loss": 5.6865,
+      "step": 3196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.4579554822753503e-05,
+      "loss": 5.827,
+      "step": 3197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.455894476504535e-05,
+      "loss": 5.8325,
+      "step": 3198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.4538334707337183e-05,
+      "loss": 5.7281,
+      "step": 3199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.451772464962902e-05,
+      "loss": 5.8727,
+      "step": 3200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.449711459192086e-05,
+      "loss": 5.7423,
+      "step": 3201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.44765045342127e-05,
+      "loss": 5.5457,
+      "step": 3202
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.4455894476504537e-05,
+      "loss": 5.6537,
+      "step": 3203
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.443528441879637e-05,
+      "loss": 5.7254,
+      "step": 3204
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.4414674361088217e-05,
+      "loss": 5.6592,
+      "step": 3205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.439406430338005e-05,
+      "loss": 5.5766,
+      "step": 3206
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.437345424567189e-05,
+      "loss": 5.53,
+      "step": 3207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.4352844187963726e-05,
+      "loss": 5.4371,
+      "step": 3208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.433223413025557e-05,
+      "loss": 5.5417,
+      "step": 3209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.4311624072547406e-05,
+      "loss": 5.8157,
+      "step": 3210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.429101401483924e-05,
+      "loss": 5.4279,
+      "step": 3211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.4270403957131086e-05,
+      "loss": 5.3381,
+      "step": 3212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.424979389942292e-05,
+      "loss": 5.5063,
+      "step": 3213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.422918384171476e-05,
+      "loss": 5.5456,
+      "step": 3214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.4208573784006596e-05,
+      "loss": 5.5231,
+      "step": 3215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.418796372629844e-05,
+      "loss": 5.3868,
+      "step": 3216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.416735366859027e-05,
+      "loss": 5.3985,
+      "step": 3217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.414674361088211e-05,
+      "loss": 5.3855,
+      "step": 3218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.412613355317395e-05,
+      "loss": 5.4029,
+      "step": 3219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.4105523495465786e-05,
+      "loss": 5.5219,
+      "step": 3220
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.408491343775763e-05,
+      "loss": 5.5083,
+      "step": 3221
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4064303380049466e-05,
+      "loss": 5.5048,
+      "step": 3222
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.404369332234131e-05,
+      "loss": 5.0737,
+      "step": 3223
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.402308326463314e-05,
+      "loss": 5.1574,
+      "step": 3224
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.400247320692498e-05,
+      "loss": 5.5569,
+      "step": 3225
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.398186314921682e-05,
+      "loss": 5.6752,
+      "step": 3226
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3961253091508656e-05,
+      "loss": 5.4583,
+      "step": 3227
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.39406430338005e-05,
+      "loss": 5.4408,
+      "step": 3228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3920032976092336e-05,
+      "loss": 4.8424,
+      "step": 3229
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.389942291838417e-05,
+      "loss": 4.6179,
+      "step": 3230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.387881286067601e-05,
+      "loss": 4.2164,
+      "step": 3231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.385820280296785e-05,
+      "loss": 4.4672,
+      "step": 3232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.383759274525969e-05,
+      "loss": 4.1475,
+      "step": 3233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3816982687551526e-05,
+      "loss": 4.0489,
+      "step": 3234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.379637262984336e-05,
+      "loss": 6.2986,
+      "step": 3235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3775762572135205e-05,
+      "loss": 6.12,
+      "step": 3236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.375515251442704e-05,
+      "loss": 5.937,
+      "step": 3237
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.373454245671888e-05,
+      "loss": 5.7298,
+      "step": 3238
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.371393239901072e-05,
+      "loss": 5.5001,
+      "step": 3239
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.369332234130256e-05,
+      "loss": 5.6899,
+      "step": 3240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3672712283594395e-05,
+      "loss": 5.709,
+      "step": 3241
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.365210222588623e-05,
+      "loss": 5.6731,
+      "step": 3242
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3631492168178075e-05,
+      "loss": 5.8171,
+      "step": 3243
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.361088211046991e-05,
+      "loss": 5.8501,
+      "step": 3244
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.359027205276175e-05,
+      "loss": 5.5596,
+      "step": 3245
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.356966199505359e-05,
+      "loss": 5.6844,
+      "step": 3246
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.354905193734543e-05,
+      "loss": 5.671,
+      "step": 3247
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3528441879637265e-05,
+      "loss": 5.4187,
+      "step": 3248
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.35078318219291e-05,
+      "loss": 5.3147,
+      "step": 3249
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3487221764220945e-05,
+      "loss": 5.5356,
+      "step": 3250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3466611706512775e-05,
+      "loss": 5.3978,
+      "step": 3251
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.344600164880462e-05,
+      "loss": 5.6248,
+      "step": 3252
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.342539159109646e-05,
+      "loss": 5.5547,
+      "step": 3253
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.340478153338829e-05,
+      "loss": 5.4911,
+      "step": 3254
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3384171475680135e-05,
+      "loss": 5.8001,
+      "step": 3255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.336356141797197e-05,
+      "loss": 5.4475,
+      "step": 3256
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3342951360263815e-05,
+      "loss": 5.4783,
+      "step": 3257
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3322341302555645e-05,
+      "loss": 5.7017,
+      "step": 3258
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.330173124484749e-05,
+      "loss": 5.5363,
+      "step": 3259
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.328112118713933e-05,
+      "loss": 5.7294,
+      "step": 3260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.326051112943116e-05,
+      "loss": 5.5237,
+      "step": 3261
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3239901071723005e-05,
+      "loss": 5.4946,
+      "step": 3262
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.321929101401484e-05,
+      "loss": 5.5063,
+      "step": 3263
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.319868095630668e-05,
+      "loss": 5.2962,
+      "step": 3264
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3178070898598514e-05,
+      "loss": 5.2847,
+      "step": 3265
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.315746084089036e-05,
+      "loss": 5.2221,
+      "step": 3266
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3136850783182194e-05,
+      "loss": 5.2168,
+      "step": 3267
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.311624072547403e-05,
+      "loss": 5.3174,
+      "step": 3268
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3095630667765874e-05,
+      "loss": 5.402,
+      "step": 3269
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.307502061005771e-05,
+      "loss": 5.5608,
+      "step": 3270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.305441055234955e-05,
+      "loss": 5.263,
+      "step": 3271
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3033800494641384e-05,
+      "loss": 5.2796,
+      "step": 3272
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.301319043693323e-05,
+      "loss": 4.9098,
+      "step": 3273
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.2992580379225064e-05,
+      "loss": 5.4077,
+      "step": 3274
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.29719703215169e-05,
+      "loss": 5.3724,
+      "step": 3275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.2951360263808744e-05,
+      "loss": 5.1926,
+      "step": 3276
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.293075020610058e-05,
+      "loss": 4.8654,
+      "step": 3277
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.291014014839242e-05,
+      "loss": 4.7527,
+      "step": 3278
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.2889530090684254e-05,
+      "loss": 4.9271,
+      "step": 3279
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.28689200329761e-05,
+      "loss": 4.5649,
+      "step": 3280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.2848309975267934e-05,
+      "loss": 5.0093,
+      "step": 3281
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.282769991755977e-05,
+      "loss": 4.381,
+      "step": 3282
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.2807089859851614e-05,
+      "loss": 4.2148,
+      "step": 3283
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.278647980214345e-05,
+      "loss": 3.9339,
+      "step": 3284
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.276586974443529e-05,
+      "loss": 6.1312,
+      "step": 3285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.2745259686727124e-05,
+      "loss": 6.5256,
+      "step": 3286
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.272464962901897e-05,
+      "loss": 6.3133,
+      "step": 3287
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.27040395713108e-05,
+      "loss": 6.0509,
+      "step": 3288
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.268342951360264e-05,
+      "loss": 6.3869,
+      "step": 3289
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.266281945589448e-05,
+      "loss": 6.0576,
+      "step": 3290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.2642209398186314e-05,
+      "loss": 5.9814,
+      "step": 3291
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.262159934047816e-05,
+      "loss": 5.9296,
+      "step": 3292
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2600989282769994e-05,
+      "loss": 5.7038,
+      "step": 3293
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.258037922506184e-05,
+      "loss": 5.3696,
+      "step": 3294
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.255976916735367e-05,
+      "loss": 5.8631,
+      "step": 3295
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.253915910964551e-05,
+      "loss": 5.7216,
+      "step": 3296
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.251854905193735e-05,
+      "loss": 5.769,
+      "step": 3297
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.249793899422918e-05,
+      "loss": 6.0181,
+      "step": 3298
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.247732893652103e-05,
+      "loss": 5.9639,
+      "step": 3299
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.245671887881286e-05,
+      "loss": 5.8763,
+      "step": 3300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.24361088211047e-05,
+      "loss": 5.4752,
+      "step": 3301
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2415498763396537e-05,
+      "loss": 5.6417,
+      "step": 3302
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.239488870568838e-05,
+      "loss": 5.7145,
+      "step": 3303
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2374278647980216e-05,
+      "loss": 5.595,
+      "step": 3304
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.235366859027205e-05,
+      "loss": 5.6083,
+      "step": 3305
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.233305853256389e-05,
+      "loss": 5.5918,
+      "step": 3306
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.231244847485573e-05,
+      "loss": 5.4666,
+      "step": 3307
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.229183841714757e-05,
+      "loss": 5.4354,
+      "step": 3308
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.2271228359439406e-05,
+      "loss": 5.3463,
+      "step": 3309
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.225061830173125e-05,
+      "loss": 5.7034,
+      "step": 3310
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.2230008244023086e-05,
+      "loss": 5.6103,
+      "step": 3311
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.220939818631492e-05,
+      "loss": 5.5189,
+      "step": 3312
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.218878812860676e-05,
+      "loss": 5.3411,
+      "step": 3313
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.21681780708986e-05,
+      "loss": 5.8469,
+      "step": 3314
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.214756801319044e-05,
+      "loss": 5.1504,
+      "step": 3315
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.2126957955482276e-05,
+      "loss": 5.587,
+      "step": 3316
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.210634789777412e-05,
+      "loss": 5.3031,
+      "step": 3317
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.2085737840065956e-05,
+      "loss": 5.379,
+      "step": 3318
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.206512778235779e-05,
+      "loss": 5.1322,
+      "step": 3319
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.204451772464963e-05,
+      "loss": 5.2721,
+      "step": 3320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.202390766694147e-05,
+      "loss": 5.3163,
+      "step": 3321
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.20032976092333e-05,
+      "loss": 5.3586,
+      "step": 3322
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.1982687551525146e-05,
+      "loss": 5.2086,
+      "step": 3323
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.196207749381699e-05,
+      "loss": 5.403,
+      "step": 3324
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.194146743610882e-05,
+      "loss": 4.9347,
+      "step": 3325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.192085737840066e-05,
+      "loss": 4.8861,
+      "step": 3326
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.19002473206925e-05,
+      "loss": 4.9488,
+      "step": 3327
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.187963726298434e-05,
+      "loss": 4.7301,
+      "step": 3328
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.185902720527617e-05,
+      "loss": 4.8365,
+      "step": 3329
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.1838417147568016e-05,
+      "loss": 4.8502,
+      "step": 3330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.181780708985986e-05,
+      "loss": 4.8659,
+      "step": 3331
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.179719703215169e-05,
+      "loss": 4.4798,
+      "step": 3332
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.177658697444353e-05,
+      "loss": 3.9956,
+      "step": 3333
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.175597691673537e-05,
+      "loss": 3.7229,
+      "step": 3334
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.1735366859027205e-05,
+      "loss": 6.0699,
+      "step": 3335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.171475680131904e-05,
+      "loss": 6.0446,
+      "step": 3336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.1694146743610885e-05,
+      "loss": 5.776,
+      "step": 3337
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.167353668590272e-05,
+      "loss": 5.8673,
+      "step": 3338
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.165292662819456e-05,
+      "loss": 5.7402,
+      "step": 3339
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.16323165704864e-05,
+      "loss": 5.5649,
+      "step": 3340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.161170651277824e-05,
+      "loss": 5.6119,
+      "step": 3341
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.1591096455070075e-05,
+      "loss": 5.6371,
+      "step": 3342
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.157048639736191e-05,
+      "loss": 5.6995,
+      "step": 3343
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.1549876339653755e-05,
+      "loss": 5.7293,
+      "step": 3344
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.152926628194559e-05,
+      "loss": 5.9956,
+      "step": 3345
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.150865622423743e-05,
+      "loss": 5.5443,
+      "step": 3346
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.148804616652927e-05,
+      "loss": 5.5409,
+      "step": 3347
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.146743610882111e-05,
+      "loss": 5.6532,
+      "step": 3348
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.1446826051112945e-05,
+      "loss": 5.4659,
+      "step": 3349
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.142621599340478e-05,
+      "loss": 5.4347,
+      "step": 3350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.1405605935696625e-05,
+      "loss": 5.7067,
+      "step": 3351
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.138499587798846e-05,
+      "loss": 5.4366,
+      "step": 3352
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.13643858202803e-05,
+      "loss": 5.5702,
+      "step": 3353
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.1343775762572135e-05,
+      "loss": 5.6123,
+      "step": 3354
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.132316570486398e-05,
+      "loss": 5.8279,
+      "step": 3355
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.1302555647155815e-05,
+      "loss": 5.3833,
+      "step": 3356
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.128194558944765e-05,
+      "loss": 5.6998,
+      "step": 3357
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.1261335531739495e-05,
+      "loss": 5.5354,
+      "step": 3358
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.1240725474031325e-05,
+      "loss": 5.4652,
+      "step": 3359
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.122011541632317e-05,
+      "loss": 5.313,
+      "step": 3360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.1199505358615005e-05,
+      "loss": 5.7661,
+      "step": 3361
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.117889530090685e-05,
+      "loss": 5.4502,
+      "step": 3362
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.1158285243198685e-05,
+      "loss": 5.6141,
+      "step": 3363
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.113767518549052e-05,
+      "loss": 5.3409,
+      "step": 3364
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.1117065127782365e-05,
+      "loss": 5.5752,
+      "step": 3365
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.1096455070074194e-05,
+      "loss": 5.3241,
+      "step": 3366
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.107584501236604e-05,
+      "loss": 5.6452,
+      "step": 3367
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.1055234954657874e-05,
+      "loss": 5.6178,
+      "step": 3368
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.103462489694971e-05,
+      "loss": 5.4709,
+      "step": 3369
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.101401483924155e-05,
+      "loss": 5.1829,
+      "step": 3370
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.099340478153339e-05,
+      "loss": 5.489,
+      "step": 3371
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.097279472382523e-05,
+      "loss": 5.3858,
+      "step": 3372
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0952184666117064e-05,
+      "loss": 5.0831,
+      "step": 3373
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.093157460840891e-05,
+      "loss": 5.3024,
+      "step": 3374
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0910964550700744e-05,
+      "loss": 5.3974,
+      "step": 3375
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.089035449299258e-05,
+      "loss": 5.1107,
+      "step": 3376
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.086974443528442e-05,
+      "loss": 4.9754,
+      "step": 3377
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.084913437757626e-05,
+      "loss": 4.6871,
+      "step": 3378
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.08285243198681e-05,
+      "loss": 4.8203,
+      "step": 3379
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0807914262159934e-05,
+      "loss": 4.2289,
+      "step": 3380
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.078730420445178e-05,
+      "loss": 4.5062,
+      "step": 3381
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0766694146743614e-05,
+      "loss": 4.8398,
+      "step": 3382
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.074608408903545e-05,
+      "loss": 4.6227,
+      "step": 3383
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.072547403132729e-05,
+      "loss": 3.7491,
+      "step": 3384
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.070486397361913e-05,
+      "loss": 6.2525,
+      "step": 3385
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.068425391591097e-05,
+      "loss": 6.0752,
+      "step": 3386
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0663643858202804e-05,
+      "loss": 6.1343,
+      "step": 3387
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.064303380049465e-05,
+      "loss": 6.1431,
+      "step": 3388
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0622423742786484e-05,
+      "loss": 5.9386,
+      "step": 3389
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.060181368507832e-05,
+      "loss": 5.9311,
+      "step": 3390
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.058120362737016e-05,
+      "loss": 5.8697,
+      "step": 3391
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0560593569662e-05,
+      "loss": 5.3034,
+      "step": 3392
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.053998351195383e-05,
+      "loss": 5.6131,
+      "step": 3393
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0519373454245673e-05,
+      "loss": 5.6489,
+      "step": 3394
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.049876339653752e-05,
+      "loss": 5.6986,
+      "step": 3395
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.047815333882935e-05,
+      "loss": 5.8621,
+      "step": 3396
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.045754328112119e-05,
+      "loss": 5.6194,
+      "step": 3397
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.043693322341303e-05,
+      "loss": 5.5137,
+      "step": 3398
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.041632316570487e-05,
+      "loss": 5.503,
+      "step": 3399
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.03957131079967e-05,
+      "loss": 5.5253,
+      "step": 3400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.037510305028854e-05,
+      "loss": 5.6719,
+      "step": 3401
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0354492992580387e-05,
+      "loss": 5.2867,
+      "step": 3402
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0333882934872216e-05,
+      "loss": 5.6767,
+      "step": 3403
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.031327287716406e-05,
+      "loss": 5.4734,
+      "step": 3404
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0292662819455896e-05,
+      "loss": 5.3024,
+      "step": 3405
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.027205276174773e-05,
+      "loss": 5.6079,
+      "step": 3406
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.025144270403957e-05,
+      "loss": 5.3155,
+      "step": 3407
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.023083264633141e-05,
+      "loss": 5.3534,
+      "step": 3408
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.021022258862325e-05,
+      "loss": 5.5757,
+      "step": 3409
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0189612530915086e-05,
+      "loss": 5.6358,
+      "step": 3410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.016900247320693e-05,
+      "loss": 5.3268,
+      "step": 3411
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0148392415498766e-05,
+      "loss": 5.4113,
+      "step": 3412
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.01277823577906e-05,
+      "loss": 5.3119,
+      "step": 3413
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.010717230008244e-05,
+      "loss": 5.4045,
+      "step": 3414
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.008656224237428e-05,
+      "loss": 5.3918,
+      "step": 3415
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.006595218466612e-05,
+      "loss": 5.2643,
+      "step": 3416
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0045342126957956e-05,
+      "loss": 5.6131,
+      "step": 3417
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.00247320692498e-05,
+      "loss": 5.2872,
+      "step": 3418
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0004122011541636e-05,
+      "loss": 5.2959,
+      "step": 3419
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.998351195383347e-05,
+      "loss": 5.2153,
+      "step": 3420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.996290189612531e-05,
+      "loss": 5.3143,
+      "step": 3421
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.994229183841715e-05,
+      "loss": 5.2001,
+      "step": 3422
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.992168178070899e-05,
+      "loss": 5.0732,
+      "step": 3423
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.9901071723000826e-05,
+      "loss": 5.1598,
+      "step": 3424
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.988046166529266e-05,
+      "loss": 5.0808,
+      "step": 3425
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.9859851607584506e-05,
+      "loss": 4.8424,
+      "step": 3426
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.983924154987634e-05,
+      "loss": 4.7635,
+      "step": 3427
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.981863149216818e-05,
+      "loss": 5.0744,
+      "step": 3428
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.979802143446002e-05,
+      "loss": 4.4816,
+      "step": 3429
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.977741137675185e-05,
+      "loss": 4.8335,
+      "step": 3430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.9756801319043696e-05,
+      "loss": 4.186,
+      "step": 3431
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.973619126133553e-05,
+      "loss": 4.4019,
+      "step": 3432
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.9715581203627376e-05,
+      "loss": 3.7319,
+      "step": 3433
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.969497114591921e-05,
+      "loss": 4.0751,
+      "step": 3434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.967436108821105e-05,
+      "loss": 5.9855,
+      "step": 3435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.965375103050289e-05,
+      "loss": 5.953,
+      "step": 3436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.963314097279472e-05,
+      "loss": 5.7745,
+      "step": 3437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.9612530915086565e-05,
+      "loss": 5.8006,
+      "step": 3438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.95919208573784e-05,
+      "loss": 5.6913,
+      "step": 3439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.957131079967024e-05,
+      "loss": 5.5104,
+      "step": 3440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.9550700741962075e-05,
+      "loss": 5.6275,
+      "step": 3441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.953009068425392e-05,
+      "loss": 5.5625,
+      "step": 3442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.9509480626545755e-05,
+      "loss": 5.5603,
+      "step": 3443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.948887056883759e-05,
+      "loss": 5.7096,
+      "step": 3444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.9468260511129435e-05,
+      "loss": 5.5242,
+      "step": 3445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.944765045342127e-05,
+      "loss": 5.5425,
+      "step": 3446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.942704039571311e-05,
+      "loss": 5.4492,
+      "step": 3447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.9406430338004945e-05,
+      "loss": 5.4982,
+      "step": 3448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.938582028029679e-05,
+      "loss": 5.559,
+      "step": 3449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.9365210222588625e-05,
+      "loss": 5.5101,
+      "step": 3450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.934460016488046e-05,
+      "loss": 5.6185,
+      "step": 3451
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.9323990107172305e-05,
+      "loss": 5.7636,
+      "step": 3452
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.930338004946414e-05,
+      "loss": 5.4604,
+      "step": 3453
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.928276999175598e-05,
+      "loss": 5.4974,
+      "step": 3454
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.9262159934047815e-05,
+      "loss": 5.5244,
+      "step": 3455
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.924154987633966e-05,
+      "loss": 5.0976,
+      "step": 3456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.9220939818631495e-05,
+      "loss": 5.6602,
+      "step": 3457
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.920032976092333e-05,
+      "loss": 5.4898,
+      "step": 3458
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.9179719703215175e-05,
+      "loss": 5.3281,
+      "step": 3459
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.915910964550701e-05,
+      "loss": 5.4928,
+      "step": 3460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.913849958779885e-05,
+      "loss": 5.2609,
+      "step": 3461
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.9117889530090684e-05,
+      "loss": 5.0064,
+      "step": 3462
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.909727947238253e-05,
+      "loss": 5.5969,
+      "step": 3463
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.907666941467436e-05,
+      "loss": 5.4982,
+      "step": 3464
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.90560593569662e-05,
+      "loss": 5.3903,
+      "step": 3465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.9035449299258044e-05,
+      "loss": 5.4929,
+      "step": 3466
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.9014839241549874e-05,
+      "loss": 5.4419,
+      "step": 3467
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.899422918384172e-05,
+      "loss": 5.0764,
+      "step": 3468
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.8973619126133554e-05,
+      "loss": 5.1773,
+      "step": 3469
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.89530090684254e-05,
+      "loss": 5.4131,
+      "step": 3470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.893239901071723e-05,
+      "loss": 4.9462,
+      "step": 3471
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.891178895300907e-05,
+      "loss": 5.1302,
+      "step": 3472
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.889117889530091e-05,
+      "loss": 5.3741,
+      "step": 3473
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8870568837592744e-05,
+      "loss": 5.7522,
+      "step": 3474
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.884995877988459e-05,
+      "loss": 5.1779,
+      "step": 3475
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8829348722176424e-05,
+      "loss": 4.596,
+      "step": 3476
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.880873866446826e-05,
+      "loss": 5.3504,
+      "step": 3477
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.87881286067601e-05,
+      "loss": 4.9981,
+      "step": 3478
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.876751854905194e-05,
+      "loss": 5.2599,
+      "step": 3479
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.874690849134378e-05,
+      "loss": 4.8636,
+      "step": 3480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8726298433635614e-05,
+      "loss": 4.3372,
+      "step": 3481
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.870568837592746e-05,
+      "loss": 4.3269,
+      "step": 3482
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.8685078318219294e-05,
+      "loss": 3.8859,
+      "step": 3483
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.866446826051113e-05,
+      "loss": 4.1612,
+      "step": 3484
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.864385820280297e-05,
+      "loss": 6.6559,
+      "step": 3485
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.862324814509481e-05,
+      "loss": 6.3815,
+      "step": 3486
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.860263808738665e-05,
+      "loss": 6.2367,
+      "step": 3487
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.8582028029678484e-05,
+      "loss": 6.2754,
+      "step": 3488
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.856141797197033e-05,
+      "loss": 6.1662,
+      "step": 3489
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.8540807914262164e-05,
+      "loss": 6.1569,
+      "step": 3490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.8520197856554e-05,
+      "loss": 5.7571,
+      "step": 3491
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.849958779884584e-05,
+      "loss": 5.6761,
+      "step": 3492
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.847897774113768e-05,
+      "loss": 5.7094,
+      "step": 3493
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.845836768342952e-05,
+      "loss": 5.7374,
+      "step": 3494
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.843775762572135e-05,
+      "loss": 5.858,
+      "step": 3495
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.841714756801319e-05,
+      "loss": 6.0393,
+      "step": 3496
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.839653751030503e-05,
+      "loss": 6.157,
+      "step": 3497
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.837592745259687e-05,
+      "loss": 6.1623,
+      "step": 3498
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.8355317394888707e-05,
+      "loss": 5.7901,
+      "step": 3499
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.833470733718055e-05,
+      "loss": 5.73,
+      "step": 3500
+    },
+    {
+      "epoch": 1.96,
+      "eval_loss": 18.985713958740234,
+      "eval_runtime": 1332.9397,
+      "eval_samples_per_second": 1.982,
+      "eval_steps_per_second": 0.248,
+      "eval_wer": 1.0003966679888934,
+      "step": 3500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.831409727947238e-05,
+      "loss": 5.8867,
+      "step": 3501
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.829348722176422e-05,
+      "loss": 5.4725,
+      "step": 3502
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.827287716405606e-05,
+      "loss": 5.4952,
+      "step": 3503
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.82522671063479e-05,
+      "loss": 5.4118,
+      "step": 3504
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.823165704863974e-05,
+      "loss": 5.2445,
+      "step": 3505
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.8211046990931576e-05,
+      "loss": 5.486,
+      "step": 3506
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.819043693322342e-05,
+      "loss": 5.1649,
+      "step": 3507
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.816982687551525e-05,
+      "loss": 5.6425,
+      "step": 3508
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.814921681780709e-05,
+      "loss": 5.3597,
+      "step": 3509
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.812860676009893e-05,
+      "loss": 5.3209,
+      "step": 3510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.8107996702390766e-05,
+      "loss": 5.1255,
+      "step": 3511
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.80873866446826e-05,
+      "loss": 5.2896,
+      "step": 3512
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.8066776586974446e-05,
+      "loss": 5.2099,
+      "step": 3513
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.804616652926629e-05,
+      "loss": 5.3256,
+      "step": 3514
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.802555647155812e-05,
+      "loss": 5.2819,
+      "step": 3515
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.800494641384996e-05,
+      "loss": 4.9691,
+      "step": 3516
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.79843363561418e-05,
+      "loss": 5.4512,
+      "step": 3517
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7963726298433636e-05,
+      "loss": 5.4823,
+      "step": 3518
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.794311624072547e-05,
+      "loss": 5.3446,
+      "step": 3519
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7922506183017316e-05,
+      "loss": 5.2289,
+      "step": 3520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.790189612530915e-05,
+      "loss": 5.3252,
+      "step": 3521
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.788128606760099e-05,
+      "loss": 5.4441,
+      "step": 3522
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.786067600989283e-05,
+      "loss": 5.253,
+      "step": 3523
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.784006595218467e-05,
+      "loss": 5.012,
+      "step": 3524
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7819455894476506e-05,
+      "loss": 5.1485,
+      "step": 3525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.779884583676834e-05,
+      "loss": 5.2651,
+      "step": 3526
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7778235779060186e-05,
+      "loss": 4.9457,
+      "step": 3527
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.775762572135202e-05,
+      "loss": 5.3019,
+      "step": 3528
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.773701566364386e-05,
+      "loss": 4.9211,
+      "step": 3529
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.77164056059357e-05,
+      "loss": 4.4118,
+      "step": 3530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.769579554822754e-05,
+      "loss": 4.6741,
+      "step": 3531
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7675185490519375e-05,
+      "loss": 4.5633,
+      "step": 3532
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.765457543281121e-05,
+      "loss": 4.0043,
+      "step": 3533
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7633965375103055e-05,
+      "loss": 3.8232,
+      "step": 3534
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7613355317394885e-05,
+      "loss": 5.9907,
+      "step": 3535
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.759274525968673e-05,
+      "loss": 6.2248,
+      "step": 3536
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.757213520197857e-05,
+      "loss": 6.2092,
+      "step": 3537
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.755152514427041e-05,
+      "loss": 5.8029,
+      "step": 3538
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7530915086562245e-05,
+      "loss": 5.7797,
+      "step": 3539
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.751030502885408e-05,
+      "loss": 5.593,
+      "step": 3540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7489694971145925e-05,
+      "loss": 5.8408,
+      "step": 3541
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7469084913437755e-05,
+      "loss": 5.4401,
+      "step": 3542
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.74484748557296e-05,
+      "loss": 5.7795,
+      "step": 3543
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7427864798021435e-05,
+      "loss": 5.4889,
+      "step": 3544
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.740725474031327e-05,
+      "loss": 5.367,
+      "step": 3545
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7386644682605115e-05,
+      "loss": 5.5118,
+      "step": 3546
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.736603462489695e-05,
+      "loss": 5.577,
+      "step": 3547
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.734542456718879e-05,
+      "loss": 5.8672,
+      "step": 3548
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7324814509480625e-05,
+      "loss": 5.2326,
+      "step": 3549
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.730420445177247e-05,
+      "loss": 5.3383,
+      "step": 3550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7283594394064305e-05,
+      "loss": 5.6227,
+      "step": 3551
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.726298433635614e-05,
+      "loss": 5.2739,
+      "step": 3552
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7242374278647985e-05,
+      "loss": 5.5119,
+      "step": 3553
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.722176422093982e-05,
+      "loss": 5.8369,
+      "step": 3554
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.720115416323166e-05,
+      "loss": 5.4672,
+      "step": 3555
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7180544105523495e-05,
+      "loss": 5.7119,
+      "step": 3556
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.715993404781534e-05,
+      "loss": 5.1124,
+      "step": 3557
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7139323990107175e-05,
+      "loss": 5.4202,
+      "step": 3558
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.711871393239901e-05,
+      "loss": 5.1007,
+      "step": 3559
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.709810387469085e-05,
+      "loss": 5.051,
+      "step": 3560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.707749381698269e-05,
+      "loss": 5.4202,
+      "step": 3561
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.705688375927453e-05,
+      "loss": 5.1877,
+      "step": 3562
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7036273701566364e-05,
+      "loss": 5.1967,
+      "step": 3563
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.701566364385821e-05,
+      "loss": 5.206,
+      "step": 3564
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.6995053586150044e-05,
+      "loss": 4.4709,
+      "step": 3565
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.697444352844188e-05,
+      "loss": 4.7975,
+      "step": 3566
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.695383347073372e-05,
+      "loss": 5.2066,
+      "step": 3567
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.693322341302556e-05,
+      "loss": 3.105,
+      "step": 3568
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.69126133553174e-05,
+      "loss": 6.127,
+      "step": 3569
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.6892003297609234e-05,
+      "loss": 5.883,
+      "step": 3570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.687139323990108e-05,
+      "loss": 6.1041,
+      "step": 3571
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.685078318219291e-05,
+      "loss": 6.0245,
+      "step": 3572
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.683017312448475e-05,
+      "loss": 6.0736,
+      "step": 3573
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.680956306677659e-05,
+      "loss": 5.984,
+      "step": 3574
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.678895300906843e-05,
+      "loss": 5.8976,
+      "step": 3575
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.676834295136026e-05,
+      "loss": 5.6771,
+      "step": 3576
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.6747732893652104e-05,
+      "loss": 5.5385,
+      "step": 3577
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.672712283594395e-05,
+      "loss": 5.5155,
+      "step": 3578
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.670651277823578e-05,
+      "loss": 5.556,
+      "step": 3579
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.668590272052762e-05,
+      "loss": 5.4327,
+      "step": 3580
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.666529266281946e-05,
+      "loss": 5.6209,
+      "step": 3581
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.6644682605111294e-05,
+      "loss": 5.4614,
+      "step": 3582
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.662407254740313e-05,
+      "loss": 5.565,
+      "step": 3583
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.6603462489694974e-05,
+      "loss": 5.4472,
+      "step": 3584
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.658285243198682e-05,
+      "loss": 5.5381,
+      "step": 3585
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.656224237427865e-05,
+      "loss": 5.4924,
+      "step": 3586
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.654163231657049e-05,
+      "loss": 5.3181,
+      "step": 3587
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.652102225886233e-05,
+      "loss": 5.1983,
+      "step": 3588
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.6500412201154163e-05,
+      "loss": 5.2494,
+      "step": 3589
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.6479802143446e-05,
+      "loss": 5.4013,
+      "step": 3590
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.6459192085737843e-05,
+      "loss": 5.228,
+      "step": 3591
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.643858202802968e-05,
+      "loss": 5.4497,
+      "step": 3592
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.641797197032152e-05,
+      "loss": 5.3885,
+      "step": 3593
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.639736191261336e-05,
+      "loss": 5.2425,
+      "step": 3594
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.63767518549052e-05,
+      "loss": 5.2752,
+      "step": 3595
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.635614179719703e-05,
+      "loss": 5.0694,
+      "step": 3596
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.633553173948887e-05,
+      "loss": 5.3092,
+      "step": 3597
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.631492168178071e-05,
+      "loss": 5.2803,
+      "step": 3598
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.629431162407255e-05,
+      "loss": 5.5403,
+      "step": 3599
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6273701566364386e-05,
+      "loss": 5.3674,
+      "step": 3600
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.625309150865623e-05,
+      "loss": 5.1751,
+      "step": 3601
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6232481450948066e-05,
+      "loss": 4.9461,
+      "step": 3602
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.62118713932399e-05,
+      "loss": 4.881,
+      "step": 3603
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.619126133553174e-05,
+      "loss": 4.9013,
+      "step": 3604
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.617065127782358e-05,
+      "loss": 5.384,
+      "step": 3605
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.615004122011541e-05,
+      "loss": 5.2077,
+      "step": 3606
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6129431162407256e-05,
+      "loss": 4.933,
+      "step": 3607
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.61088211046991e-05,
+      "loss": 5.1936,
+      "step": 3608
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6088211046990936e-05,
+      "loss": 5.3849,
+      "step": 3609
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.606760098928277e-05,
+      "loss": 4.9283,
+      "step": 3610
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.604699093157461e-05,
+      "loss": 4.5964,
+      "step": 3611
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.602638087386645e-05,
+      "loss": 4.8913,
+      "step": 3612
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.600577081615828e-05,
+      "loss": 4.4793,
+      "step": 3613
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.5985160758450126e-05,
+      "loss": 4.0519,
+      "step": 3614
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.596455070074196e-05,
+      "loss": 3.8786,
+      "step": 3615
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.59439406430338e-05,
+      "loss": 4.2617,
+      "step": 3616
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.592333058532564e-05,
+      "loss": 3.8635,
+      "step": 3617
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.590272052761748e-05,
+      "loss": 3.0257,
+      "step": 3618
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.5882110469909316e-05,
+      "loss": 6.024,
+      "step": 3619
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.586150041220115e-05,
+      "loss": 5.4739,
+      "step": 3620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.5840890354492996e-05,
+      "loss": 5.881,
+      "step": 3621
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.582028029678483e-05,
+      "loss": 5.7769,
+      "step": 3622
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.579967023907667e-05,
+      "loss": 5.5308,
+      "step": 3623
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.577906018136851e-05,
+      "loss": 5.5062,
+      "step": 3624
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.575845012366035e-05,
+      "loss": 5.8769,
+      "step": 3625
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.5737840065952186e-05,
+      "loss": 5.63,
+      "step": 3626
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.571723000824402e-05,
+      "loss": 5.4859,
+      "step": 3627
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.5696619950535866e-05,
+      "loss": 5.8555,
+      "step": 3628
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.56760098928277e-05,
+      "loss": 5.5083,
+      "step": 3629
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.565539983511954e-05,
+      "loss": 5.3481,
+      "step": 3630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.5634789777411375e-05,
+      "loss": 5.424,
+      "step": 3631
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.561417971970322e-05,
+      "loss": 5.3468,
+      "step": 3632
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.5593569661995055e-05,
+      "loss": 5.6152,
+      "step": 3633
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.557295960428689e-05,
+      "loss": 5.4798,
+      "step": 3634
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.5552349546578735e-05,
+      "loss": 5.524,
+      "step": 3635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.553173948887057e-05,
+      "loss": 5.7306,
+      "step": 3636
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.551112943116241e-05,
+      "loss": 5.2901,
+      "step": 3637
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.5490519373454245e-05,
+      "loss": 5.4819,
+      "step": 3638
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.546990931574609e-05,
+      "loss": 5.206,
+      "step": 3639
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.5449299258037925e-05,
+      "loss": 5.3303,
+      "step": 3640
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.542868920032976e-05,
+      "loss": 5.5385,
+      "step": 3641
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.5408079142621605e-05,
+      "loss": 5.344,
+      "step": 3642
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.5387469084913435e-05,
+      "loss": 5.3553,
+      "step": 3643
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.536685902720528e-05,
+      "loss": 5.2297,
+      "step": 3644
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.5346248969497115e-05,
+      "loss": 5.7073,
+      "step": 3645
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.532563891178896e-05,
+      "loss": 5.1669,
+      "step": 3646
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.530502885408079e-05,
+      "loss": 5.4904,
+      "step": 3647
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.528441879637263e-05,
+      "loss": 5.3136,
+      "step": 3648
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.5263808738664475e-05,
+      "loss": 5.1276,
+      "step": 3649
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.5243198680956305e-05,
+      "loss": 4.8803,
+      "step": 3650
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.522258862324815e-05,
+      "loss": 5.3768,
+      "step": 3651
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.5201978565539985e-05,
+      "loss": 5.1056,
+      "step": 3652
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.518136850783182e-05,
+      "loss": 5.1066,
+      "step": 3653
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.516075845012366e-05,
+      "loss": 5.0517,
+      "step": 3654
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.51401483924155e-05,
+      "loss": 5.0335,
+      "step": 3655
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.5119538334707345e-05,
+      "loss": 5.0176,
+      "step": 3656
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.5098928276999174e-05,
+      "loss": 4.9724,
+      "step": 3657
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.507831821929102e-05,
+      "loss": 5.1814,
+      "step": 3658
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.5057708161582854e-05,
+      "loss": 4.8925,
+      "step": 3659
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.503709810387469e-05,
+      "loss": 5.175,
+      "step": 3660
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.501648804616653e-05,
+      "loss": 4.674,
+      "step": 3661
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.499587798845837e-05,
+      "loss": 4.6165,
+      "step": 3662
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.497526793075021e-05,
+      "loss": 4.652,
+      "step": 3663
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.4954657873042044e-05,
+      "loss": 4.423,
+      "step": 3664
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.493404781533389e-05,
+      "loss": 4.4095,
+      "step": 3665
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.4913437757625724e-05,
+      "loss": 4.3349,
+      "step": 3666
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.489282769991756e-05,
+      "loss": 4.0616,
+      "step": 3667
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.48722176422094e-05,
+      "loss": 2.8314,
+      "step": 3668
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.485160758450124e-05,
+      "loss": 5.9588,
+      "step": 3669
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.483099752679308e-05,
+      "loss": 6.396,
+      "step": 3670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.4810387469084914e-05,
+      "loss": 5.8509,
+      "step": 3671
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.478977741137676e-05,
+      "loss": 5.8099,
+      "step": 3672
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.4769167353668594e-05,
+      "loss": 5.6465,
+      "step": 3673
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.474855729596043e-05,
+      "loss": 5.1601,
+      "step": 3674
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.472794723825227e-05,
+      "loss": 5.57,
+      "step": 3675
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.470733718054411e-05,
+      "loss": 5.3305,
+      "step": 3676
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.468672712283594e-05,
+      "loss": 5.6513,
+      "step": 3677
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.4666117065127784e-05,
+      "loss": 5.5599,
+      "step": 3678
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.464550700741962e-05,
+      "loss": 5.6709,
+      "step": 3679
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.4624896949711464e-05,
+      "loss": 5.627,
+      "step": 3680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.46042868920033e-05,
+      "loss": 5.3791,
+      "step": 3681
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.458367683429514e-05,
+      "loss": 5.453,
+      "step": 3682
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.456306677658698e-05,
+      "loss": 5.4888,
+      "step": 3683
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.454245671887881e-05,
+      "loss": 5.4331,
+      "step": 3684
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.4521846661170654e-05,
+      "loss": 5.2723,
+      "step": 3685
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.450123660346249e-05,
+      "loss": 5.3436,
+      "step": 3686
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.448062654575433e-05,
+      "loss": 5.2062,
+      "step": 3687
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.446001648804617e-05,
+      "loss": 5.4847,
+      "step": 3688
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.443940643033801e-05,
+      "loss": 5.1766,
+      "step": 3689
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.441879637262985e-05,
+      "loss": 5.2443,
+      "step": 3690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.439818631492168e-05,
+      "loss": 5.3263,
+      "step": 3691
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.437757625721352e-05,
+      "loss": 5.2855,
+      "step": 3692
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.435696619950536e-05,
+      "loss": 5.0108,
+      "step": 3693
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.4336356141797197e-05,
+      "loss": 5.3299,
+      "step": 3694
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.431574608408903e-05,
+      "loss": 5.5701,
+      "step": 3695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.4295136026380877e-05,
+      "loss": 5.1422,
+      "step": 3696
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.427452596867271e-05,
+      "loss": 5.1912,
+      "step": 3697
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.425391591096455e-05,
+      "loss": 5.0496,
+      "step": 3698
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.423330585325639e-05,
+      "loss": 5.188,
+      "step": 3699
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.421269579554823e-05,
+      "loss": 5.132,
+      "step": 3700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.4192085737840066e-05,
+      "loss": 5.2242,
+      "step": 3701
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.41714756801319e-05,
+      "loss": 5.0612,
+      "step": 3702
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.4150865622423746e-05,
+      "loss": 4.8958,
+      "step": 3703
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.413025556471558e-05,
+      "loss": 5.1424,
+      "step": 3704
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.410964550700742e-05,
+      "loss": 5.1985,
+      "step": 3705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.408903544929926e-05,
+      "loss": 5.1018,
+      "step": 3706
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.40684253915911e-05,
+      "loss": 5.1781,
+      "step": 3707
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.4047815333882936e-05,
+      "loss": 5.0455,
+      "step": 3708
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.402720527617477e-05,
+      "loss": 4.8446,
+      "step": 3709
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.4006595218466616e-05,
+      "loss": 5.2548,
+      "step": 3710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.398598516075845e-05,
+      "loss": 4.7819,
+      "step": 3711
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.396537510305029e-05,
+      "loss": 4.3011,
+      "step": 3712
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.394476504534213e-05,
+      "loss": 4.3467,
+      "step": 3713
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.392415498763397e-05,
+      "loss": 4.6985,
+      "step": 3714
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.3903544929925806e-05,
+      "loss": 3.9453,
+      "step": 3715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.388293487221764e-05,
+      "loss": 3.9671,
+      "step": 3716
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.3862324814509486e-05,
+      "loss": 3.5639,
+      "step": 3717
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.3841714756801316e-05,
+      "loss": 3.3643,
+      "step": 3718
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.382110469909316e-05,
+      "loss": 6.1308,
+      "step": 3719
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3800494641385e-05,
+      "loss": 6.3318,
+      "step": 3720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.377988458367683e-05,
+      "loss": 5.7818,
+      "step": 3721
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3759274525968676e-05,
+      "loss": 6.0105,
+      "step": 3722
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.373866446826051e-05,
+      "loss": 5.9744,
+      "step": 3723
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.371805441055235e-05,
+      "loss": 5.8121,
+      "step": 3724
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3697444352844185e-05,
+      "loss": 5.4698,
+      "step": 3725
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.367683429513603e-05,
+      "loss": 5.4952,
+      "step": 3726
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.365622423742787e-05,
+      "loss": 5.1861,
+      "step": 3727
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.36356141797197e-05,
+      "loss": 5.471,
+      "step": 3728
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3615004122011545e-05,
+      "loss": 5.804,
+      "step": 3729
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.359439406430338e-05,
+      "loss": 5.7317,
+      "step": 3730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.357378400659522e-05,
+      "loss": 5.5114,
+      "step": 3731
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3553173948887055e-05,
+      "loss": 5.5317,
+      "step": 3732
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.35325638911789e-05,
+      "loss": 6.036,
+      "step": 3733
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3511953833470735e-05,
+      "loss": 5.978,
+      "step": 3734
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.349134377576257e-05,
+      "loss": 5.3144,
+      "step": 3735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.3470733718054415e-05,
+      "loss": 5.3651,
+      "step": 3736
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.345012366034625e-05,
+      "loss": 5.3305,
+      "step": 3737
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.342951360263809e-05,
+      "loss": 5.2955,
+      "step": 3738
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3408903544929925e-05,
+      "loss": 5.5505,
+      "step": 3739
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.338829348722177e-05,
+      "loss": 5.3005,
+      "step": 3740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3367683429513605e-05,
+      "loss": 5.4158,
+      "step": 3741
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.334707337180544e-05,
+      "loss": 5.4036,
+      "step": 3742
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3326463314097285e-05,
+      "loss": 5.1308,
+      "step": 3743
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.330585325638912e-05,
+      "loss": 5.7601,
+      "step": 3744
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.328524319868096e-05,
+      "loss": 5.3097,
+      "step": 3745
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3264633140972795e-05,
+      "loss": 5.2356,
+      "step": 3746
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.324402308326464e-05,
+      "loss": 5.1494,
+      "step": 3747
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.322341302555647e-05,
+      "loss": 5.1521,
+      "step": 3748
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.320280296784831e-05,
+      "loss": 5.2293,
+      "step": 3749
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.318219291014015e-05,
+      "loss": 5.34,
+      "step": 3750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.316158285243199e-05,
+      "loss": 5.191,
+      "step": 3751
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.314097279472383e-05,
+      "loss": 5.3764,
+      "step": 3752
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.3120362737015665e-05,
+      "loss": 4.9735,
+      "step": 3753
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.309975267930751e-05,
+      "loss": 5.4452,
+      "step": 3754
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.307914262159934e-05,
+      "loss": 5.3847,
+      "step": 3755
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.305853256389118e-05,
+      "loss": 5.1075,
+      "step": 3756
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.303792250618302e-05,
+      "loss": 4.8739,
+      "step": 3757
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.3017312448474854e-05,
+      "loss": 5.2622,
+      "step": 3758
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.29967023907667e-05,
+      "loss": 4.9111,
+      "step": 3759
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2976092333058534e-05,
+      "loss": 4.9147,
+      "step": 3760
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.295548227535038e-05,
+      "loss": 4.9901,
+      "step": 3761
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.293487221764221e-05,
+      "loss": 4.8524,
+      "step": 3762
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.291426215993405e-05,
+      "loss": 4.7337,
+      "step": 3763
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.289365210222589e-05,
+      "loss": 4.3966,
+      "step": 3764
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2873042044517724e-05,
+      "loss": 4.3567,
+      "step": 3765
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.285243198680956e-05,
+      "loss": 4.3291,
+      "step": 3766
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2831821929101404e-05,
+      "loss": 3.6785,
+      "step": 3767
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.281121187139324e-05,
+      "loss": 4.2036,
+      "step": 3768
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.279060181368508e-05,
+      "loss": 6.0295,
+      "step": 3769
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.276999175597692e-05,
+      "loss": 5.6817,
+      "step": 3770
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.274938169826876e-05,
+      "loss": 5.7622,
+      "step": 3771
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2728771640560594e-05,
+      "loss": 5.7484,
+      "step": 3772
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.270816158285243e-05,
+      "loss": 5.8812,
+      "step": 3773
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2687551525144274e-05,
+      "loss": 5.3487,
+      "step": 3774
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.266694146743611e-05,
+      "loss": 5.3863,
+      "step": 3775
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.264633140972795e-05,
+      "loss": 5.8218,
+      "step": 3776
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.262572135201979e-05,
+      "loss": 5.5258,
+      "step": 3777
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.260511129431163e-05,
+      "loss": 5.6251,
+      "step": 3778
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2584501236603464e-05,
+      "loss": 5.4199,
+      "step": 3779
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.25638911788953e-05,
+      "loss": 5.4895,
+      "step": 3780
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2543281121187144e-05,
+      "loss": 5.4357,
+      "step": 3781
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2522671063478974e-05,
+      "loss": 5.561,
+      "step": 3782
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.250206100577082e-05,
+      "loss": 4.9942,
+      "step": 3783
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.248145094806266e-05,
+      "loss": 5.3038,
+      "step": 3784
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.24608408903545e-05,
+      "loss": 5.3436,
+      "step": 3785
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2440230832646333e-05,
+      "loss": 5.3524,
+      "step": 3786
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.241962077493817e-05,
+      "loss": 5.5513,
+      "step": 3787
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2399010717230013e-05,
+      "loss": 5.1925,
+      "step": 3788
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.237840065952184e-05,
+      "loss": 5.2149,
+      "step": 3789
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.235779060181369e-05,
+      "loss": 5.3527,
+      "step": 3790
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.233718054410553e-05,
+      "loss": 5.3624,
+      "step": 3791
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.231657048639736e-05,
+      "loss": 5.4406,
+      "step": 3792
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.22959604286892e-05,
+      "loss": 5.356,
+      "step": 3793
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.227535037098104e-05,
+      "loss": 5.3362,
+      "step": 3794
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2254740313272876e-05,
+      "loss": 5.3934,
+      "step": 3795
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.223413025556471e-05,
+      "loss": 5.5152,
+      "step": 3796
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2213520197856556e-05,
+      "loss": 5.1382,
+      "step": 3797
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.219291014014839e-05,
+      "loss": 5.1583,
+      "step": 3798
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.217230008244023e-05,
+      "loss": 5.0157,
+      "step": 3799
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.215169002473207e-05,
+      "loss": 5.317,
+      "step": 3800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.213107996702391e-05,
+      "loss": 4.9174,
+      "step": 3801
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2110469909315746e-05,
+      "loss": 5.2968,
+      "step": 3802
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.208985985160758e-05,
+      "loss": 5.0331,
+      "step": 3803
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2069249793899426e-05,
+      "loss": 5.2858,
+      "step": 3804
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.204863973619126e-05,
+      "loss": 4.7689,
+      "step": 3805
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.20280296784831e-05,
+      "loss": 5.0334,
+      "step": 3806
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.200741962077494e-05,
+      "loss": 5.1767,
+      "step": 3807
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.198680956306678e-05,
+      "loss": 4.8017,
+      "step": 3808
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1966199505358616e-05,
+      "loss": 4.9128,
+      "step": 3809
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.194558944765045e-05,
+      "loss": 4.7754,
+      "step": 3810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1924979389942296e-05,
+      "loss": 4.8946,
+      "step": 3811
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.190436933223413e-05,
+      "loss": 4.5352,
+      "step": 3812
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.188375927452597e-05,
+      "loss": 4.1154,
+      "step": 3813
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1863149216817806e-05,
+      "loss": 4.3339,
+      "step": 3814
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.184253915910965e-05,
+      "loss": 4.0952,
+      "step": 3815
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1821929101401486e-05,
+      "loss": 3.558,
+      "step": 3816
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.180131904369332e-05,
+      "loss": 3.3002,
+      "step": 3817
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1780708985985166e-05,
+      "loss": 3.7926,
+      "step": 3818
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1760098928276996e-05,
+      "loss": 6.0818,
+      "step": 3819
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.173948887056884e-05,
+      "loss": 5.8426,
+      "step": 3820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1718878812860676e-05,
+      "loss": 5.9946,
+      "step": 3821
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.169826875515252e-05,
+      "loss": 5.6831,
+      "step": 3822
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1677658697444356e-05,
+      "loss": 5.8113,
+      "step": 3823
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.165704863973619e-05,
+      "loss": 5.3997,
+      "step": 3824
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1636438582028036e-05,
+      "loss": 5.4732,
+      "step": 3825
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1615828524319865e-05,
+      "loss": 5.1668,
+      "step": 3826
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.159521846661171e-05,
+      "loss": 5.9104,
+      "step": 3827
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1574608408903545e-05,
+      "loss": 5.5768,
+      "step": 3828
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.155399835119538e-05,
+      "loss": 5.4835,
+      "step": 3829
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1533388293487225e-05,
+      "loss": 5.3707,
+      "step": 3830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.151277823577906e-05,
+      "loss": 5.7151,
+      "step": 3831
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1492168178070905e-05,
+      "loss": 5.2318,
+      "step": 3832
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1471558120362735e-05,
+      "loss": 5.1811,
+      "step": 3833
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.145094806265458e-05,
+      "loss": 5.1067,
+      "step": 3834
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1430338004946415e-05,
+      "loss": 5.2139,
+      "step": 3835
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.140972794723825e-05,
+      "loss": 5.3414,
+      "step": 3836
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.138911788953009e-05,
+      "loss": 5.558,
+      "step": 3837
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.136850783182193e-05,
+      "loss": 5.2758,
+      "step": 3838
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.134789777411377e-05,
+      "loss": 5.1165,
+      "step": 3839
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1327287716405605e-05,
+      "loss": 5.2383,
+      "step": 3840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.130667765869745e-05,
+      "loss": 5.2664,
+      "step": 3841
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1286067600989285e-05,
+      "loss": 5.0617,
+      "step": 3842
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.126545754328112e-05,
+      "loss": 5.102,
+      "step": 3843
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.124484748557296e-05,
+      "loss": 5.125,
+      "step": 3844
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.12242374278648e-05,
+      "loss": 4.9934,
+      "step": 3845
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.120362737015664e-05,
+      "loss": 5.1845,
+      "step": 3846
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1183017312448475e-05,
+      "loss": 5.3169,
+      "step": 3847
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.116240725474032e-05,
+      "loss": 5.3992,
+      "step": 3848
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1141797197032155e-05,
+      "loss": 5.0736,
+      "step": 3849
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.112118713932399e-05,
+      "loss": 4.9542,
+      "step": 3850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.110057708161583e-05,
+      "loss": 4.9805,
+      "step": 3851
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.107996702390767e-05,
+      "loss": 4.9067,
+      "step": 3852
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.10593569661995e-05,
+      "loss": 5.0144,
+      "step": 3853
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1038746908491344e-05,
+      "loss": 5.2052,
+      "step": 3854
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.101813685078319e-05,
+      "loss": 5.1837,
+      "step": 3855
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.0997526793075024e-05,
+      "loss": 4.9921,
+      "step": 3856
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.097691673536686e-05,
+      "loss": 5.4291,
+      "step": 3857
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.09563066776587e-05,
+      "loss": 4.9635,
+      "step": 3858
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.093569661995054e-05,
+      "loss": 5.2906,
+      "step": 3859
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.091508656224237e-05,
+      "loss": 4.7043,
+      "step": 3860
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.0894476504534214e-05,
+      "loss": 4.5993,
+      "step": 3861
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.087386644682606e-05,
+      "loss": 4.4296,
+      "step": 3862
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.085325638911789e-05,
+      "loss": 4.3902,
+      "step": 3863
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.083264633140973e-05,
+      "loss": 3.9293,
+      "step": 3864
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.081203627370157e-05,
+      "loss": 3.8899,
+      "step": 3865
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.079142621599341e-05,
+      "loss": 3.5067,
+      "step": 3866
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.077081615828524e-05,
+      "loss": 4.6393,
+      "step": 3867
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.0750206100577084e-05,
+      "loss": 3.4679,
+      "step": 3868
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.072959604286892e-05,
+      "loss": 6.391,
+      "step": 3869
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.070898598516076e-05,
+      "loss": 6.5247,
+      "step": 3870
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.06883759274526e-05,
+      "loss": 6.3184,
+      "step": 3871
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.066776586974444e-05,
+      "loss": 6.328,
+      "step": 3872
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.0647155812036274e-05,
+      "loss": 6.0418,
+      "step": 3873
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.062654575432811e-05,
+      "loss": 6.0158,
+      "step": 3874
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.0605935696619954e-05,
+      "loss": 6.0142,
+      "step": 3875
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.058532563891179e-05,
+      "loss": 5.4299,
+      "step": 3876
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.056471558120363e-05,
+      "loss": 5.3817,
+      "step": 3877
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.054410552349547e-05,
+      "loss": 5.5534,
+      "step": 3878
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.052349546578731e-05,
+      "loss": 5.7637,
+      "step": 3879
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.0502885408079147e-05,
+      "loss": 5.6737,
+      "step": 3880
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.048227535037098e-05,
+      "loss": 5.5298,
+      "step": 3881
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0461665292662824e-05,
+      "loss": 5.5723,
+      "step": 3882
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0441055234954657e-05,
+      "loss": 5.9206,
+      "step": 3883
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0420445177246497e-05,
+      "loss": 5.5486,
+      "step": 3884
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0399835119538333e-05,
+      "loss": 5.2267,
+      "step": 3885
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0379225061830173e-05,
+      "loss": 5.2051,
+      "step": 3886
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0358615004122017e-05,
+      "loss": 5.2276,
+      "step": 3887
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.033800494641385e-05,
+      "loss": 5.3668,
+      "step": 3888
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.031739488870569e-05,
+      "loss": 5.4339,
+      "step": 3889
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0296784830997527e-05,
+      "loss": 5.4779,
+      "step": 3890
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0276174773289367e-05,
+      "loss": 5.591,
+      "step": 3891
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0255564715581203e-05,
+      "loss": 5.379,
+      "step": 3892
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0234954657873043e-05,
+      "loss": 5.2998,
+      "step": 3893
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0214344600164883e-05,
+      "loss": 5.2557,
+      "step": 3894
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.019373454245672e-05,
+      "loss": 5.3758,
+      "step": 3895
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.017312448474856e-05,
+      "loss": 5.1167,
+      "step": 3896
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0152514427040396e-05,
+      "loss": 5.1509,
+      "step": 3897
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.0131904369332236e-05,
+      "loss": 5.2189,
+      "step": 3898
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0111294311624073e-05,
+      "loss": 4.8377,
+      "step": 3899
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0090684253915913e-05,
+      "loss": 5.3997,
+      "step": 3900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.007007419620775e-05,
+      "loss": 5.3032,
+      "step": 3901
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.004946413849959e-05,
+      "loss": 5.3173,
+      "step": 3902
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.002885408079143e-05,
+      "loss": 5.6327,
+      "step": 3903
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.0008244023083266e-05,
+      "loss": 5.2228,
+      "step": 3904
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.9987633965375106e-05,
+      "loss": 5.0053,
+      "step": 3905
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.9967023907666943e-05,
+      "loss": 4.9627,
+      "step": 3906
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.9946413849958783e-05,
+      "loss": 4.9666,
+      "step": 3907
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.9925803792250616e-05,
+      "loss": 5.0445,
+      "step": 3908
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.990519373454246e-05,
+      "loss": 5.1057,
+      "step": 3909
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.98845836768343e-05,
+      "loss": 5.0148,
+      "step": 3910
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.9863973619126136e-05,
+      "loss": 4.9458,
+      "step": 3911
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.9843363561417976e-05,
+      "loss": 4.4856,
+      "step": 3912
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.982275350370981e-05,
+      "loss": 4.4098,
+      "step": 3913
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.9802143446001652e-05,
+      "loss": 4.0592,
+      "step": 3914
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.9781533388293486e-05,
+      "loss": 4.4005,
+      "step": 3915
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.976092333058533e-05,
+      "loss": 3.8026,
+      "step": 3916
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9740313272877162e-05,
+      "loss": 3.9945,
+      "step": 3917
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9719703215169002e-05,
+      "loss": 3.5634,
+      "step": 3918
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9699093157460846e-05,
+      "loss": 6.1392,
+      "step": 3919
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.967848309975268e-05,
+      "loss": 6.0442,
+      "step": 3920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.965787304204452e-05,
+      "loss": 5.8057,
+      "step": 3921
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9637262984336355e-05,
+      "loss": 5.7901,
+      "step": 3922
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9616652926628195e-05,
+      "loss": 6.1277,
+      "step": 3923
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9596042868920032e-05,
+      "loss": 5.7292,
+      "step": 3924
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9575432811211872e-05,
+      "loss": 5.4492,
+      "step": 3925
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9554822753503712e-05,
+      "loss": 5.381,
+      "step": 3926
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.953421269579555e-05,
+      "loss": 5.226,
+      "step": 3927
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.951360263808739e-05,
+      "loss": 5.2521,
+      "step": 3928
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9492992580379225e-05,
+      "loss": 5.4909,
+      "step": 3929
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9472382522671065e-05,
+      "loss": 5.2135,
+      "step": 3930
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9451772464962902e-05,
+      "loss": 5.6963,
+      "step": 3931
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9431162407254742e-05,
+      "loss": 5.4955,
+      "step": 3932
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.9410552349546582e-05,
+      "loss": 5.7687,
+      "step": 3933
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.938994229183842e-05,
+      "loss": 5.4701,
+      "step": 3934
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.936933223413026e-05,
+      "loss": 5.2873,
+      "step": 3935
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9348722176422095e-05,
+      "loss": 5.054,
+      "step": 3936
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9328112118713935e-05,
+      "loss": 5.5048,
+      "step": 3937
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.930750206100577e-05,
+      "loss": 5.3579,
+      "step": 3938
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.928689200329761e-05,
+      "loss": 5.4576,
+      "step": 3939
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9266281945589448e-05,
+      "loss": 5.1547,
+      "step": 3940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9245671887881288e-05,
+      "loss": 5.2438,
+      "step": 3941
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9225061830173128e-05,
+      "loss": 5.0162,
+      "step": 3942
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9204451772464965e-05,
+      "loss": 5.2216,
+      "step": 3943
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9183841714756805e-05,
+      "loss": 5.2949,
+      "step": 3944
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9163231657048638e-05,
+      "loss": 5.0976,
+      "step": 3945
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.914262159934048e-05,
+      "loss": 5.1048,
+      "step": 3946
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9122011541632315e-05,
+      "loss": 5.1052,
+      "step": 3947
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9101401483924158e-05,
+      "loss": 5.2757,
+      "step": 3948
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9080791426215998e-05,
+      "loss": 5.2104,
+      "step": 3949
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.906018136850783e-05,
+      "loss": 4.9563,
+      "step": 3950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.9039571310799675e-05,
+      "loss": 4.9768,
+      "step": 3951
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.9018961253091508e-05,
+      "loss": 5.4829,
+      "step": 3952
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.899835119538335e-05,
+      "loss": 4.9419,
+      "step": 3953
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8977741137675184e-05,
+      "loss": 4.9943,
+      "step": 3954
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8957131079967024e-05,
+      "loss": 4.8308,
+      "step": 3955
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.893652102225886e-05,
+      "loss": 4.7126,
+      "step": 3956
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.89159109645507e-05,
+      "loss": 4.7118,
+      "step": 3957
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8895300906842544e-05,
+      "loss": 4.7946,
+      "step": 3958
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8874690849134378e-05,
+      "loss": 4.7423,
+      "step": 3959
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8854080791426218e-05,
+      "loss": 4.6509,
+      "step": 3960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8833470733718054e-05,
+      "loss": 4.3876,
+      "step": 3961
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8812860676009894e-05,
+      "loss": 4.3694,
+      "step": 3962
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.879225061830173e-05,
+      "loss": 4.2582,
+      "step": 3963
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.877164056059357e-05,
+      "loss": 4.3259,
+      "step": 3964
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.875103050288541e-05,
+      "loss": 4.1985,
+      "step": 3965
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8730420445177247e-05,
+      "loss": 4.2072,
+      "step": 3966
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8709810387469087e-05,
+      "loss": 3.7347,
+      "step": 3967
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8689200329760924e-05,
+      "loss": 3.2268,
+      "step": 3968
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.8668590272052764e-05,
+      "loss": 6.2438,
+      "step": 3969
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.86479802143446e-05,
+      "loss": 6.1856,
+      "step": 3970
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.862737015663644e-05,
+      "loss": 6.2211,
+      "step": 3971
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8606760098928277e-05,
+      "loss": 5.7582,
+      "step": 3972
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8586150041220117e-05,
+      "loss": 6.2252,
+      "step": 3973
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8565539983511957e-05,
+      "loss": 5.6915,
+      "step": 3974
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8544929925803794e-05,
+      "loss": 5.6016,
+      "step": 3975
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8524319868095634e-05,
+      "loss": 5.4556,
+      "step": 3976
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.850370981038747e-05,
+      "loss": 5.2926,
+      "step": 3977
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.848309975267931e-05,
+      "loss": 5.2015,
+      "step": 3978
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8462489694971144e-05,
+      "loss": 5.8511,
+      "step": 3979
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8441879637262987e-05,
+      "loss": 5.5824,
+      "step": 3980
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8421269579554827e-05,
+      "loss": 5.2944,
+      "step": 3981
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8400659521846663e-05,
+      "loss": 5.3682,
+      "step": 3982
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8380049464138503e-05,
+      "loss": 5.5538,
+      "step": 3983
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8359439406430337e-05,
+      "loss": 5.6107,
+      "step": 3984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.833882934872218e-05,
+      "loss": 5.4116,
+      "step": 3985
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8318219291014013e-05,
+      "loss": 5.1445,
+      "step": 3986
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.8297609233305857e-05,
+      "loss": 5.2883,
+      "step": 3987
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.827699917559769e-05,
+      "loss": 5.2516,
+      "step": 3988
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.825638911788953e-05,
+      "loss": 5.1569,
+      "step": 3989
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8235779060181373e-05,
+      "loss": 5.5986,
+      "step": 3990
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8215169002473206e-05,
+      "loss": 5.3054,
+      "step": 3991
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.819455894476505e-05,
+      "loss": 5.4124,
+      "step": 3992
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8173948887056883e-05,
+      "loss": 5.3458,
+      "step": 3993
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8153338829348723e-05,
+      "loss": 4.9954,
+      "step": 3994
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.813272877164056e-05,
+      "loss": 5.3204,
+      "step": 3995
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.81121187139324e-05,
+      "loss": 5.0079,
+      "step": 3996
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.809150865622424e-05,
+      "loss": 5.1027,
+      "step": 3997
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8070898598516076e-05,
+      "loss": 5.2928,
+      "step": 3998
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8050288540807916e-05,
+      "loss": 5.0121,
+      "step": 3999
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8029678483099753e-05,
+      "loss": 4.9793,
+      "step": 4000
+    },
+    {
+      "epoch": 2.24,
+      "eval_loss": 18.32020378112793,
+      "eval_runtime": 1340.7601,
+      "eval_samples_per_second": 1.971,
+      "eval_steps_per_second": 0.247,
+      "eval_wer": 1.000654502181674,
+      "step": 4000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.8009068425391593e-05,
+      "loss": 5.1098,
+      "step": 4001
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.798845836768343e-05,
+      "loss": 5.0272,
+      "step": 4002
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.796784830997527e-05,
+      "loss": 5.0944,
+      "step": 4003
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7947238252267106e-05,
+      "loss": 4.9297,
+      "step": 4004
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7926628194558946e-05,
+      "loss": 5.3979,
+      "step": 4005
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7906018136850786e-05,
+      "loss": 5.1523,
+      "step": 4006
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7885408079142623e-05,
+      "loss": 4.8398,
+      "step": 4007
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7864798021434463e-05,
+      "loss": 4.6662,
+      "step": 4008
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.78441879637263e-05,
+      "loss": 4.6269,
+      "step": 4009
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.782357790601814e-05,
+      "loss": 4.6741,
+      "step": 4010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7802967848309976e-05,
+      "loss": 4.5262,
+      "step": 4011
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7782357790601816e-05,
+      "loss": 4.5681,
+      "step": 4012
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7761747732893656e-05,
+      "loss": 4.0361,
+      "step": 4013
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7741137675185492e-05,
+      "loss": 3.8852,
+      "step": 4014
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7720527617477332e-05,
+      "loss": 3.6479,
+      "step": 4015
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.769991755976917e-05,
+      "loss": 4.1979,
+      "step": 4016
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.767930750206101e-05,
+      "loss": 3.4106,
+      "step": 4017
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7658697444352842e-05,
+      "loss": 3.6541,
+      "step": 4018
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7638087386644686e-05,
+      "loss": 5.8831,
+      "step": 4019
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.761747732893652e-05,
+      "loss": 5.681,
+      "step": 4020
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.759686727122836e-05,
+      "loss": 5.4914,
+      "step": 4021
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7576257213520202e-05,
+      "loss": 5.7874,
+      "step": 4022
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7555647155812035e-05,
+      "loss": 5.2653,
+      "step": 4023
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.753503709810388e-05,
+      "loss": 5.5166,
+      "step": 4024
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7514427040395712e-05,
+      "loss": 5.4618,
+      "step": 4025
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7493816982687552e-05,
+      "loss": 5.3301,
+      "step": 4026
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.747320692497939e-05,
+      "loss": 5.4028,
+      "step": 4027
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.745259686727123e-05,
+      "loss": 5.4027,
+      "step": 4028
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7431986809563072e-05,
+      "loss": 5.3485,
+      "step": 4029
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7411376751854905e-05,
+      "loss": 5.2613,
+      "step": 4030
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7390766694146745e-05,
+      "loss": 5.4898,
+      "step": 4031
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7370156636438582e-05,
+      "loss": 5.0799,
+      "step": 4032
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7349546578730422e-05,
+      "loss": 5.4537,
+      "step": 4033
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.732893652102226e-05,
+      "loss": 5.0737,
+      "step": 4034
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.73083264633141e-05,
+      "loss": 5.1496,
+      "step": 4035
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7287716405605935e-05,
+      "loss": 4.9672,
+      "step": 4036
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7267106347897775e-05,
+      "loss": 5.3236,
+      "step": 4037
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7246496290189615e-05,
+      "loss": 5.1814,
+      "step": 4038
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.722588623248145e-05,
+      "loss": 4.9978,
+      "step": 4039
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.720527617477329e-05,
+      "loss": 5.0351,
+      "step": 4040
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7184666117065128e-05,
+      "loss": 5.1337,
+      "step": 4041
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7164056059356968e-05,
+      "loss": 4.9615,
+      "step": 4042
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7143446001648805e-05,
+      "loss": 5.0481,
+      "step": 4043
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7122835943940645e-05,
+      "loss": 5.0776,
+      "step": 4044
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7102225886232485e-05,
+      "loss": 5.2648,
+      "step": 4045
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.708161582852432e-05,
+      "loss": 5.2063,
+      "step": 4046
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.706100577081616e-05,
+      "loss": 4.9996,
+      "step": 4047
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7040395713107998e-05,
+      "loss": 5.0965,
+      "step": 4048
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7019785655399838e-05,
+      "loss": 5.0997,
+      "step": 4049
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.699917559769167e-05,
+      "loss": 4.9985,
+      "step": 4050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.6978565539983514e-05,
+      "loss": 4.8428,
+      "step": 4051
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.6957955482275354e-05,
+      "loss": 5.2227,
+      "step": 4052
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.693734542456719e-05,
+      "loss": 5.2121,
+      "step": 4053
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.691673536685903e-05,
+      "loss": 4.8507,
+      "step": 4054
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.6896125309150864e-05,
+      "loss": 4.8432,
+      "step": 4055
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.6875515251442708e-05,
+      "loss": 4.9215,
+      "step": 4056
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.685490519373454e-05,
+      "loss": 4.9624,
+      "step": 4057
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.6834295136026384e-05,
+      "loss": 5.0619,
+      "step": 4058
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6813685078318217e-05,
+      "loss": 4.8239,
+      "step": 4059
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6793075020610057e-05,
+      "loss": 4.8018,
+      "step": 4060
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.67724649629019e-05,
+      "loss": 4.5594,
+      "step": 4061
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6751854905193734e-05,
+      "loss": 4.222,
+      "step": 4062
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6731244847485577e-05,
+      "loss": 4.845,
+      "step": 4063
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.671063478977741e-05,
+      "loss": 5.0746,
+      "step": 4064
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.669002473206925e-05,
+      "loss": 3.9455,
+      "step": 4065
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6669414674361087e-05,
+      "loss": 3.6605,
+      "step": 4066
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6648804616652927e-05,
+      "loss": 3.3944,
+      "step": 4067
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.662819455894477e-05,
+      "loss": 3.5655,
+      "step": 4068
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6607584501236604e-05,
+      "loss": 5.851,
+      "step": 4069
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6586974443528444e-05,
+      "loss": 5.7427,
+      "step": 4070
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.656636438582028e-05,
+      "loss": 5.6979,
+      "step": 4071
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.654575432811212e-05,
+      "loss": 5.7124,
+      "step": 4072
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6525144270403957e-05,
+      "loss": 5.4221,
+      "step": 4073
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6504534212695797e-05,
+      "loss": 5.367,
+      "step": 4074
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6483924154987634e-05,
+      "loss": 5.5586,
+      "step": 4075
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6463314097279474e-05,
+      "loss": 5.3854,
+      "step": 4076
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6442704039571314e-05,
+      "loss": 5.284,
+      "step": 4077
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.642209398186315e-05,
+      "loss": 5.303,
+      "step": 4078
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.640148392415499e-05,
+      "loss": 5.4532,
+      "step": 4079
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6380873866446827e-05,
+      "loss": 5.1557,
+      "step": 4080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6360263808738667e-05,
+      "loss": 5.1029,
+      "step": 4081
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6339653751030503e-05,
+      "loss": 5.1088,
+      "step": 4082
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6319043693322343e-05,
+      "loss": 5.4688,
+      "step": 4083
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6298433635614183e-05,
+      "loss": 5.4663,
+      "step": 4084
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.627782357790602e-05,
+      "loss": 5.2992,
+      "step": 4085
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.625721352019786e-05,
+      "loss": 5.1202,
+      "step": 4086
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6236603462489697e-05,
+      "loss": 5.2048,
+      "step": 4087
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6215993404781537e-05,
+      "loss": 5.1285,
+      "step": 4088
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.619538334707337e-05,
+      "loss": 5.47,
+      "step": 4089
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6174773289365213e-05,
+      "loss": 5.2477,
+      "step": 4090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6154163231657046e-05,
+      "loss": 5.4747,
+      "step": 4091
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.613355317394889e-05,
+      "loss": 5.3384,
+      "step": 4092
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.611294311624073e-05,
+      "loss": 5.5069,
+      "step": 4093
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6092333058532563e-05,
+      "loss": 5.2542,
+      "step": 4094
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6071723000824406e-05,
+      "loss": 5.2138,
+      "step": 4095
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.605111294311624e-05,
+      "loss": 5.2769,
+      "step": 4096
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.603050288540808e-05,
+      "loss": 5.181,
+      "step": 4097
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6009892827699916e-05,
+      "loss": 5.2214,
+      "step": 4098
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5989282769991756e-05,
+      "loss": 4.9507,
+      "step": 4099
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.59686727122836e-05,
+      "loss": 5.2979,
+      "step": 4100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5948062654575433e-05,
+      "loss": 4.8309,
+      "step": 4101
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5927452596867273e-05,
+      "loss": 5.0383,
+      "step": 4102
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.590684253915911e-05,
+      "loss": 4.7602,
+      "step": 4103
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.588623248145095e-05,
+      "loss": 5.1394,
+      "step": 4104
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5865622423742786e-05,
+      "loss": 4.8453,
+      "step": 4105
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5845012366034626e-05,
+      "loss": 4.861,
+      "step": 4106
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5824402308326463e-05,
+      "loss": 4.8054,
+      "step": 4107
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5803792250618303e-05,
+      "loss": 4.9264,
+      "step": 4108
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5783182192910143e-05,
+      "loss": 4.7961,
+      "step": 4109
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.576257213520198e-05,
+      "loss": 4.7316,
+      "step": 4110
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.574196207749382e-05,
+      "loss": 4.6066,
+      "step": 4111
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5721352019785656e-05,
+      "loss": 4.35,
+      "step": 4112
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5700741962077496e-05,
+      "loss": 3.8463,
+      "step": 4113
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5680131904369332e-05,
+      "loss": 4.1212,
+      "step": 4114
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5659521846661172e-05,
+      "loss": 3.798,
+      "step": 4115
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5638911788953012e-05,
+      "loss": 3.6899,
+      "step": 4116
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.561830173124485e-05,
+      "loss": 3.0788,
+      "step": 4117
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.559769167353669e-05,
+      "loss": 2.9089,
+      "step": 4118
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5577081615828525e-05,
+      "loss": 5.7189,
+      "step": 4119
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5556471558120365e-05,
+      "loss": 6.112,
+      "step": 4120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.55358615004122e-05,
+      "loss": 5.7605,
+      "step": 4121
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5515251442704042e-05,
+      "loss": 5.728,
+      "step": 4122
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5494641384995875e-05,
+      "loss": 5.6969,
+      "step": 4123
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.547403132728772e-05,
+      "loss": 5.5737,
+      "step": 4124
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.545342126957956e-05,
+      "loss": 5.3179,
+      "step": 4125
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5432811211871392e-05,
+      "loss": 5.1955,
+      "step": 4126
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5412201154163235e-05,
+      "loss": 5.4512,
+      "step": 4127
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.539159109645507e-05,
+      "loss": 5.2849,
+      "step": 4128
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5370981038746912e-05,
+      "loss": 5.2183,
+      "step": 4129
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5350370981038745e-05,
+      "loss": 5.307,
+      "step": 4130
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5329760923330585e-05,
+      "loss": 5.477,
+      "step": 4131
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.530915086562243e-05,
+      "loss": 5.2423,
+      "step": 4132
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.528854080791426e-05,
+      "loss": 5.6013,
+      "step": 4133
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5267930750206105e-05,
+      "loss": 5.2593,
+      "step": 4134
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5247320692497938e-05,
+      "loss": 5.2904,
+      "step": 4135
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5226710634789778e-05,
+      "loss": 5.2257,
+      "step": 4136
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5206100577081615e-05,
+      "loss": 5.4024,
+      "step": 4137
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5185490519373455e-05,
+      "loss": 5.3906,
+      "step": 4138
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.516488046166529e-05,
+      "loss": 5.3712,
+      "step": 4139
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.514427040395713e-05,
+      "loss": 5.1657,
+      "step": 4140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.512366034624897e-05,
+      "loss": 5.0435,
+      "step": 4141
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5103050288540808e-05,
+      "loss": 5.0295,
+      "step": 4142
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5082440230832648e-05,
+      "loss": 4.9462,
+      "step": 4143
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5061830173124485e-05,
+      "loss": 5.2205,
+      "step": 4144
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5041220115416325e-05,
+      "loss": 5.2407,
+      "step": 4145
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.502061005770816e-05,
+      "loss": 5.6528,
+      "step": 4146
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5e-05,
+      "loss": 5.2118,
+      "step": 4147
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4979389942291838e-05,
+      "loss": 5.0411,
+      "step": 4148
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4958779884583678e-05,
+      "loss": 4.8284,
+      "step": 4149
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4938169826875514e-05,
+      "loss": 5.0396,
+      "step": 4150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4917559769167358e-05,
+      "loss": 4.9942,
+      "step": 4151
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4896949711459194e-05,
+      "loss": 5.0745,
+      "step": 4152
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.487633965375103e-05,
+      "loss": 4.9857,
+      "step": 4153
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.485572959604287e-05,
+      "loss": 5.118,
+      "step": 4154
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4835119538334708e-05,
+      "loss": 4.5677,
+      "step": 4155
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4814509480626548e-05,
+      "loss": 4.9855,
+      "step": 4156
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4793899422918384e-05,
+      "loss": 4.7544,
+      "step": 4157
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4773289365210224e-05,
+      "loss": 4.8542,
+      "step": 4158
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4752679307502064e-05,
+      "loss": 4.8212,
+      "step": 4159
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.47320692497939e-05,
+      "loss": 4.5307,
+      "step": 4160
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.471145919208574e-05,
+      "loss": 4.6089,
+      "step": 4161
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4690849134377577e-05,
+      "loss": 4.4883,
+      "step": 4162
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4670239076669417e-05,
+      "loss": 3.8925,
+      "step": 4163
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4649629018961254e-05,
+      "loss": 4.4617,
+      "step": 4164
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.462901896125309e-05,
+      "loss": 3.8613,
+      "step": 4165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.460840890354493e-05,
+      "loss": 3.6588,
+      "step": 4166
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.458779884583677e-05,
+      "loss": 3.2388,
+      "step": 4167
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.456718878812861e-05,
+      "loss": 3.1461,
+      "step": 4168
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4546578730420447e-05,
+      "loss": 5.7497,
+      "step": 4169
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4525968672712284e-05,
+      "loss": 5.5748,
+      "step": 4170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4505358615004124e-05,
+      "loss": 5.5658,
+      "step": 4171
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.448474855729596e-05,
+      "loss": 5.3672,
+      "step": 4172
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.44641384995878e-05,
+      "loss": 5.5171,
+      "step": 4173
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4443528441879637e-05,
+      "loss": 5.368,
+      "step": 4174
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4422918384171477e-05,
+      "loss": 5.5314,
+      "step": 4175
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4402308326463317e-05,
+      "loss": 5.4965,
+      "step": 4176
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4381698268755154e-05,
+      "loss": 5.3976,
+      "step": 4177
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4361088211046994e-05,
+      "loss": 5.3817,
+      "step": 4178
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.434047815333883e-05,
+      "loss": 5.154,
+      "step": 4179
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.431986809563067e-05,
+      "loss": 5.4107,
+      "step": 4180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4299258037922507e-05,
+      "loss": 5.4311,
+      "step": 4181
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4278647980214343e-05,
+      "loss": 5.3246,
+      "step": 4182
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4258037922506187e-05,
+      "loss": 5.3993,
+      "step": 4183
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4237427864798023e-05,
+      "loss": 5.1767,
+      "step": 4184
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4216817807089863e-05,
+      "loss": 5.0747,
+      "step": 4185
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.41962077493817e-05,
+      "loss": 5.3067,
+      "step": 4186
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4175597691673536e-05,
+      "loss": 5.2692,
+      "step": 4187
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4154987633965376e-05,
+      "loss": 5.0494,
+      "step": 4188
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4134377576257213e-05,
+      "loss": 5.3409,
+      "step": 4189
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4113767518549053e-05,
+      "loss": 5.1526,
+      "step": 4190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4093157460840893e-05,
+      "loss": 4.9631,
+      "step": 4191
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.407254740313273e-05,
+      "loss": 4.8898,
+      "step": 4192
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.405193734542457e-05,
+      "loss": 5.0209,
+      "step": 4193
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4031327287716406e-05,
+      "loss": 5.1938,
+      "step": 4194
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.4010717230008246e-05,
+      "loss": 5.0939,
+      "step": 4195
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3990107172300083e-05,
+      "loss": 4.7577,
+      "step": 4196
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3969497114591923e-05,
+      "loss": 4.9789,
+      "step": 4197
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.394888705688376e-05,
+      "loss": 5.164,
+      "step": 4198
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.39282769991756e-05,
+      "loss": 5.0084,
+      "step": 4199
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.390766694146744e-05,
+      "loss": 4.8951,
+      "step": 4200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3887056883759276e-05,
+      "loss": 5.219,
+      "step": 4201
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3866446826051113e-05,
+      "loss": 5.1004,
+      "step": 4202
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3845836768342953e-05,
+      "loss": 5.1595,
+      "step": 4203
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.382522671063479e-05,
+      "loss": 4.8579,
+      "step": 4204
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.380461665292663e-05,
+      "loss": 4.9814,
+      "step": 4205
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3784006595218466e-05,
+      "loss": 4.8938,
+      "step": 4206
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3763396537510306e-05,
+      "loss": 4.8044,
+      "step": 4207
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3742786479802146e-05,
+      "loss": 4.662,
+      "step": 4208
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3722176422093982e-05,
+      "loss": 4.7572,
+      "step": 4209
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3701566364385822e-05,
+      "loss": 4.615,
+      "step": 4210
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.368095630667766e-05,
+      "loss": 4.562,
+      "step": 4211
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.36603462489695e-05,
+      "loss": 4.4754,
+      "step": 4212
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3639736191261336e-05,
+      "loss": 4.115,
+      "step": 4213
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3619126133553172e-05,
+      "loss": 3.7926,
+      "step": 4214
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3598516075845016e-05,
+      "loss": 4.1234,
+      "step": 4215
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3577906018136852e-05,
+      "loss": 3.874,
+      "step": 4216
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3557295960428692e-05,
+      "loss": 3.6244,
+      "step": 4217
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.353668590272053e-05,
+      "loss": 3.0319,
+      "step": 4218
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3516075845012365e-05,
+      "loss": 5.9924,
+      "step": 4219
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3495465787304205e-05,
+      "loss": 6.1216,
+      "step": 4220
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3474855729596042e-05,
+      "loss": 6.1324,
+      "step": 4221
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3454245671887882e-05,
+      "loss": 5.8872,
+      "step": 4222
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3433635614179722e-05,
+      "loss": 5.5875,
+      "step": 4223
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.341302555647156e-05,
+      "loss": 5.4947,
+      "step": 4224
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.33924154987634e-05,
+      "loss": 5.1355,
+      "step": 4225
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3371805441055235e-05,
+      "loss": 5.1328,
+      "step": 4226
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3351195383347075e-05,
+      "loss": 5.4122,
+      "step": 4227
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3330585325638912e-05,
+      "loss": 5.6085,
+      "step": 4228
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3309975267930752e-05,
+      "loss": 5.6609,
+      "step": 4229
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.328936521022259e-05,
+      "loss": 5.7647,
+      "step": 4230
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.326875515251443e-05,
+      "loss": 5.7039,
+      "step": 4231
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.324814509480627e-05,
+      "loss": 5.562,
+      "step": 4232
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3227535037098105e-05,
+      "loss": 5.1901,
+      "step": 4233
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3206924979389945e-05,
+      "loss": 5.3084,
+      "step": 4234
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.318631492168178e-05,
+      "loss": 4.8797,
+      "step": 4235
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.3165704863973618e-05,
+      "loss": 5.2335,
+      "step": 4236
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3145094806265458e-05,
+      "loss": 5.2223,
+      "step": 4237
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3124484748557295e-05,
+      "loss": 5.0863,
+      "step": 4238
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3103874690849138e-05,
+      "loss": 5.2302,
+      "step": 4239
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3083264633140975e-05,
+      "loss": 5.1929,
+      "step": 4240
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.306265457543281e-05,
+      "loss": 5.39,
+      "step": 4241
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.304204451772465e-05,
+      "loss": 5.1904,
+      "step": 4242
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3021434460016488e-05,
+      "loss": 4.8571,
+      "step": 4243
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.3000824402308328e-05,
+      "loss": 5.0474,
+      "step": 4244
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2980214344600165e-05,
+      "loss": 4.9549,
+      "step": 4245
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2959604286892005e-05,
+      "loss": 5.1377,
+      "step": 4246
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2938994229183844e-05,
+      "loss": 5.5412,
+      "step": 4247
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.291838417147568e-05,
+      "loss": 4.9601,
+      "step": 4248
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.289777411376752e-05,
+      "loss": 5.182,
+      "step": 4249
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2877164056059358e-05,
+      "loss": 4.6843,
+      "step": 4250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2856553998351198e-05,
+      "loss": 4.9547,
+      "step": 4251
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2835943940643034e-05,
+      "loss": 4.5669,
+      "step": 4252
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.281533388293487e-05,
+      "loss": 5.126,
+      "step": 4253
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.279472382522671e-05,
+      "loss": 4.9312,
+      "step": 4254
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.277411376751855e-05,
+      "loss": 5.0565,
+      "step": 4255
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.275350370981039e-05,
+      "loss": 4.7092,
+      "step": 4256
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2732893652102227e-05,
+      "loss": 4.7441,
+      "step": 4257
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2712283594394064e-05,
+      "loss": 4.7731,
+      "step": 4258
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2691673536685904e-05,
+      "loss": 4.9253,
+      "step": 4259
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.267106347897774e-05,
+      "loss": 4.7753,
+      "step": 4260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.265045342126958e-05,
+      "loss": 4.4365,
+      "step": 4261
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.262984336356142e-05,
+      "loss": 4.7487,
+      "step": 4262
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2609233305853257e-05,
+      "loss": 4.4326,
+      "step": 4263
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2588623248145097e-05,
+      "loss": 4.5492,
+      "step": 4264
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2568013190436934e-05,
+      "loss": 3.5039,
+      "step": 4265
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2547403132728774e-05,
+      "loss": 3.3502,
+      "step": 4266
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.252679307502061e-05,
+      "loss": 3.4128,
+      "step": 4267
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.250618301731245e-05,
+      "loss": 3.1935,
+      "step": 4268
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2485572959604287e-05,
+      "loss": 6.2947,
+      "step": 4269
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2464962901896127e-05,
+      "loss": 6.1014,
+      "step": 4270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2444352844187967e-05,
+      "loss": 6.1289,
+      "step": 4271
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2423742786479804e-05,
+      "loss": 5.7206,
+      "step": 4272
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2403132728771644e-05,
+      "loss": 5.8594,
+      "step": 4273
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.238252267106348e-05,
+      "loss": 5.7024,
+      "step": 4274
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2361912613355317e-05,
+      "loss": 5.5052,
+      "step": 4275
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2341302555647157e-05,
+      "loss": 5.643,
+      "step": 4276
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2320692497938993e-05,
+      "loss": 5.4368,
+      "step": 4277
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2300082440230833e-05,
+      "loss": 5.1624,
+      "step": 4278
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2279472382522673e-05,
+      "loss": 5.2466,
+      "step": 4279
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.225886232481451e-05,
+      "loss": 5.6408,
+      "step": 4280
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.223825226710635e-05,
+      "loss": 5.522,
+      "step": 4281
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2217642209398187e-05,
+      "loss": 5.0769,
+      "step": 4282
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2197032151690027e-05,
+      "loss": 5.366,
+      "step": 4283
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2176422093981863e-05,
+      "loss": 5.6432,
+      "step": 4284
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2155812036273703e-05,
+      "loss": 5.2993,
+      "step": 4285
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2135201978565543e-05,
+      "loss": 5.1372,
+      "step": 4286
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.211459192085738e-05,
+      "loss": 5.1979,
+      "step": 4287
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.209398186314922e-05,
+      "loss": 5.3735,
+      "step": 4288
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2073371805441056e-05,
+      "loss": 5.3382,
+      "step": 4289
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2052761747732893e-05,
+      "loss": 5.2439,
+      "step": 4290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2032151690024733e-05,
+      "loss": 5.2073,
+      "step": 4291
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.201154163231657e-05,
+      "loss": 5.212,
+      "step": 4292
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.199093157460841e-05,
+      "loss": 5.1937,
+      "step": 4293
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.197032151690025e-05,
+      "loss": 5.2774,
+      "step": 4294
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1949711459192086e-05,
+      "loss": 5.1807,
+      "step": 4295
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1929101401483926e-05,
+      "loss": 5.0505,
+      "step": 4296
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1908491343775763e-05,
+      "loss": 5.4344,
+      "step": 4297
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1887881286067603e-05,
+      "loss": 5.0043,
+      "step": 4298
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.186727122835944e-05,
+      "loss": 5.0872,
+      "step": 4299
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.184666117065128e-05,
+      "loss": 5.229,
+      "step": 4300
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1826051112943116e-05,
+      "loss": 5.0606,
+      "step": 4301
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1805441055234956e-05,
+      "loss": 4.9316,
+      "step": 4302
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1784830997526796e-05,
+      "loss": 5.107,
+      "step": 4303
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1764220939818633e-05,
+      "loss": 5.0591,
+      "step": 4304
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1743610882110473e-05,
+      "loss": 5.198,
+      "step": 4305
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.172300082440231e-05,
+      "loss": 4.8996,
+      "step": 4306
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1702390766694146e-05,
+      "loss": 5.0202,
+      "step": 4307
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.1681780708985986e-05,
+      "loss": 4.7587,
+      "step": 4308
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1661170651277822e-05,
+      "loss": 4.7843,
+      "step": 4309
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1640560593569666e-05,
+      "loss": 4.6884,
+      "step": 4310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1619950535861502e-05,
+      "loss": 4.6272,
+      "step": 4311
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.159934047815334e-05,
+      "loss": 4.7039,
+      "step": 4312
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.157873042044518e-05,
+      "loss": 4.5821,
+      "step": 4313
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1558120362737016e-05,
+      "loss": 4.286,
+      "step": 4314
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1537510305028855e-05,
+      "loss": 4.2771,
+      "step": 4315
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1516900247320692e-05,
+      "loss": 4.0421,
+      "step": 4316
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1496290189612532e-05,
+      "loss": 2.9828,
+      "step": 4317
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1475680131904372e-05,
+      "loss": 4.0162,
+      "step": 4318
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.145507007419621e-05,
+      "loss": 6.1461,
+      "step": 4319
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.143446001648805e-05,
+      "loss": 6.2737,
+      "step": 4320
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1413849958779885e-05,
+      "loss": 5.8952,
+      "step": 4321
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1393239901071725e-05,
+      "loss": 6.1974,
+      "step": 4322
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1372629843363562e-05,
+      "loss": 6.1323,
+      "step": 4323
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.13520197856554e-05,
+      "loss": 5.7696,
+      "step": 4324
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.133140972794724e-05,
+      "loss": 5.4925,
+      "step": 4325
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.131079967023908e-05,
+      "loss": 5.8638,
+      "step": 4326
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.129018961253092e-05,
+      "loss": 5.9042,
+      "step": 4327
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1269579554822755e-05,
+      "loss": 5.19,
+      "step": 4328
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.124896949711459e-05,
+      "loss": 5.2944,
+      "step": 4329
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.122835943940643e-05,
+      "loss": 4.9809,
+      "step": 4330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1207749381698268e-05,
+      "loss": 5.2796,
+      "step": 4331
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1187139323990108e-05,
+      "loss": 5.439,
+      "step": 4332
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1166529266281945e-05,
+      "loss": 5.4381,
+      "step": 4333
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1145919208573785e-05,
+      "loss": 5.2467,
+      "step": 4334
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1125309150865625e-05,
+      "loss": 5.1523,
+      "step": 4335
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.110469909315746e-05,
+      "loss": 5.4934,
+      "step": 4336
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.10840890354493e-05,
+      "loss": 5.4688,
+      "step": 4337
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1063478977741138e-05,
+      "loss": 5.4184,
+      "step": 4338
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1042868920032978e-05,
+      "loss": 5.2806,
+      "step": 4339
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1022258862324815e-05,
+      "loss": 5.0691,
+      "step": 4340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.100164880461665e-05,
+      "loss": 5.0744,
+      "step": 4341
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.0981038746908495e-05,
+      "loss": 5.0286,
+      "step": 4342
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.096042868920033e-05,
+      "loss": 5.4378,
+      "step": 4343
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.093981863149217e-05,
+      "loss": 5.2294,
+      "step": 4344
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0919208573784008e-05,
+      "loss": 5.3632,
+      "step": 4345
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0898598516075844e-05,
+      "loss": 4.7122,
+      "step": 4346
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0877988458367684e-05,
+      "loss": 5.2394,
+      "step": 4347
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.085737840065952e-05,
+      "loss": 4.9127,
+      "step": 4348
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.083676834295136e-05,
+      "loss": 5.3218,
+      "step": 4349
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.08161582852432e-05,
+      "loss": 4.9575,
+      "step": 4350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0795548227535038e-05,
+      "loss": 4.7738,
+      "step": 4351
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0774938169826878e-05,
+      "loss": 4.8735,
+      "step": 4352
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0754328112118714e-05,
+      "loss": 5.2518,
+      "step": 4353
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0733718054410554e-05,
+      "loss": 4.8533,
+      "step": 4354
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.071310799670239e-05,
+      "loss": 5.031,
+      "step": 4355
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.069249793899423e-05,
+      "loss": 5.0403,
+      "step": 4356
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0671887881286067e-05,
+      "loss": 5.1853,
+      "step": 4357
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0651277823577907e-05,
+      "loss": 4.63,
+      "step": 4358
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0630667765869747e-05,
+      "loss": 4.8955,
+      "step": 4359
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0610057708161584e-05,
+      "loss": 4.7766,
+      "step": 4360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0589447650453424e-05,
+      "loss": 4.8705,
+      "step": 4361
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.056883759274526e-05,
+      "loss": 4.6676,
+      "step": 4362
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0548227535037097e-05,
+      "loss": 4.4722,
+      "step": 4363
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0527617477328937e-05,
+      "loss": 4.5993,
+      "step": 4364
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0507007419620774e-05,
+      "loss": 4.172,
+      "step": 4365
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0486397361912614e-05,
+      "loss": 3.7304,
+      "step": 4366
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0465787304204454e-05,
+      "loss": 2.9871,
+      "step": 4367
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.044517724649629e-05,
+      "loss": 3.7096,
+      "step": 4368
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.042456718878813e-05,
+      "loss": 5.9517,
+      "step": 4369
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0403957131079967e-05,
+      "loss": 5.7611,
+      "step": 4370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0383347073371807e-05,
+      "loss": 5.768,
+      "step": 4371
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0362737015663644e-05,
+      "loss": 5.8462,
+      "step": 4372
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0342126957955484e-05,
+      "loss": 5.5798,
+      "step": 4373
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0321516900247324e-05,
+      "loss": 5.4721,
+      "step": 4374
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.030090684253916e-05,
+      "loss": 5.2936,
+      "step": 4375
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0280296784831e-05,
+      "loss": 5.1809,
+      "step": 4376
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0259686727122837e-05,
+      "loss": 5.3037,
+      "step": 4377
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0239076669414673e-05,
+      "loss": 5.2882,
+      "step": 4378
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0218466611706513e-05,
+      "loss": 5.394,
+      "step": 4379
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.019785655399835e-05,
+      "loss": 5.0617,
+      "step": 4380
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.0177246496290193e-05,
+      "loss": 4.9391,
+      "step": 4381
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.015663643858203e-05,
+      "loss": 5.3316,
+      "step": 4382
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.0136026380873867e-05,
+      "loss": 5.415,
+      "step": 4383
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.0115416323165706e-05,
+      "loss": 5.3648,
+      "step": 4384
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.0094806265457543e-05,
+      "loss": 5.0399,
+      "step": 4385
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.0074196207749383e-05,
+      "loss": 5.0292,
+      "step": 4386
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.005358615004122e-05,
+      "loss": 5.1416,
+      "step": 4387
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.003297609233306e-05,
+      "loss": 5.2037,
+      "step": 4388
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.00123660346249e-05,
+      "loss": 5.3079,
+      "step": 4389
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9991755976916736e-05,
+      "loss": 5.1063,
+      "step": 4390
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9971145919208576e-05,
+      "loss": 5.1406,
+      "step": 4391
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9950535861500413e-05,
+      "loss": 5.1629,
+      "step": 4392
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9929925803792253e-05,
+      "loss": 4.9435,
+      "step": 4393
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.990931574608409e-05,
+      "loss": 4.8435,
+      "step": 4394
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9888705688375926e-05,
+      "loss": 4.9584,
+      "step": 4395
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9868095630667766e-05,
+      "loss": 4.866,
+      "step": 4396
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9847485572959606e-05,
+      "loss": 5.293,
+      "step": 4397
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9826875515251446e-05,
+      "loss": 4.9339,
+      "step": 4398
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9806265457543283e-05,
+      "loss": 5.1405,
+      "step": 4399
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.978565539983512e-05,
+      "loss": 5.2538,
+      "step": 4400
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.976504534212696e-05,
+      "loss": 4.822,
+      "step": 4401
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9744435284418796e-05,
+      "loss": 4.7649,
+      "step": 4402
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9723825226710636e-05,
+      "loss": 4.807,
+      "step": 4403
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9703215169002472e-05,
+      "loss": 5.1051,
+      "step": 4404
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9682605111294312e-05,
+      "loss": 4.7506,
+      "step": 4405
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9661995053586152e-05,
+      "loss": 5.0901,
+      "step": 4406
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.964138499587799e-05,
+      "loss": 4.8788,
+      "step": 4407
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.962077493816983e-05,
+      "loss": 4.5705,
+      "step": 4408
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9600164880461666e-05,
+      "loss": 4.2101,
+      "step": 4409
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9579554822753506e-05,
+      "loss": 4.2901,
+      "step": 4410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9558944765045342e-05,
+      "loss": 4.3712,
+      "step": 4411
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.953833470733718e-05,
+      "loss": 4.4539,
+      "step": 4412
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9517724649629022e-05,
+      "loss": 3.8275,
+      "step": 4413
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.949711459192086e-05,
+      "loss": 3.8815,
+      "step": 4414
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.94765045342127e-05,
+      "loss": 4.0714,
+      "step": 4415
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9455894476504535e-05,
+      "loss": 3.5716,
+      "step": 4416
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9435284418796372e-05,
+      "loss": 3.5899,
+      "step": 4417
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9414674361088212e-05,
+      "loss": 3.9711,
+      "step": 4418
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.939406430338005e-05,
+      "loss": 5.5772,
+      "step": 4419
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.937345424567189e-05,
+      "loss": 5.4735,
+      "step": 4420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.935284418796373e-05,
+      "loss": 5.4202,
+      "step": 4421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9332234130255565e-05,
+      "loss": 5.4024,
+      "step": 4422
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9311624072547405e-05,
+      "loss": 5.5597,
+      "step": 4423
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9291014014839242e-05,
+      "loss": 5.1182,
+      "step": 4424
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9270403957131082e-05,
+      "loss": 5.2608,
+      "step": 4425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.924979389942292e-05,
+      "loss": 5.3052,
+      "step": 4426
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.922918384171476e-05,
+      "loss": 5.2307,
+      "step": 4427
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9208573784006595e-05,
+      "loss": 5.4074,
+      "step": 4428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9187963726298435e-05,
+      "loss": 5.328,
+      "step": 4429
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9167353668590275e-05,
+      "loss": 5.4193,
+      "step": 4430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.914674361088211e-05,
+      "loss": 5.2507,
+      "step": 4431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.912613355317395e-05,
+      "loss": 5.4038,
+      "step": 4432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9105523495465788e-05,
+      "loss": 5.1829,
+      "step": 4433
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9084913437757625e-05,
+      "loss": 4.9451,
+      "step": 4434
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9064303380049465e-05,
+      "loss": 5.4831,
+      "step": 4435
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.90436933223413e-05,
+      "loss": 5.2358,
+      "step": 4436
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9023083264633145e-05,
+      "loss": 4.9509,
+      "step": 4437
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.900247320692498e-05,
+      "loss": 5.3464,
+      "step": 4438
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8981863149216818e-05,
+      "loss": 5.2613,
+      "step": 4439
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8961253091508658e-05,
+      "loss": 5.2374,
+      "step": 4440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8940643033800495e-05,
+      "loss": 5.1689,
+      "step": 4441
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8920032976092335e-05,
+      "loss": 4.9535,
+      "step": 4442
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.889942291838417e-05,
+      "loss": 5.3084,
+      "step": 4443
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.887881286067601e-05,
+      "loss": 4.9553,
+      "step": 4444
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.885820280296785e-05,
+      "loss": 5.2971,
+      "step": 4445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8837592745259688e-05,
+      "loss": 5.3733,
+      "step": 4446
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8816982687551528e-05,
+      "loss": 5.0168,
+      "step": 4447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8796372629843364e-05,
+      "loss": 5.0924,
+      "step": 4448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8775762572135204e-05,
+      "loss": 5.0578,
+      "step": 4449
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.875515251442704e-05,
+      "loss": 5.0364,
+      "step": 4450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.8734542456718878e-05,
+      "loss": 5.1629,
+      "step": 4451
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8713932399010717e-05,
+      "loss": 4.8615,
+      "step": 4452
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8693322341302557e-05,
+      "loss": 4.9513,
+      "step": 4453
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8672712283594394e-05,
+      "loss": 4.8889,
+      "step": 4454
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8652102225886234e-05,
+      "loss": 4.9717,
+      "step": 4455
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.863149216817807e-05,
+      "loss": 4.8979,
+      "step": 4456
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.861088211046991e-05,
+      "loss": 5.5107,
+      "step": 4457
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8590272052761747e-05,
+      "loss": 4.4709,
+      "step": 4458
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8569661995053587e-05,
+      "loss": 4.7312,
+      "step": 4459
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8549051937345424e-05,
+      "loss": 4.7823,
+      "step": 4460
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8528441879637264e-05,
+      "loss": 4.097,
+      "step": 4461
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8507831821929104e-05,
+      "loss": 4.2814,
+      "step": 4462
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.848722176422094e-05,
+      "loss": 4.2304,
+      "step": 4463
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.846661170651278e-05,
+      "loss": 3.8894,
+      "step": 4464
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8446001648804617e-05,
+      "loss": 4.0412,
+      "step": 4465
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8425391591096454e-05,
+      "loss": 3.5995,
+      "step": 4466
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.8404781533388294e-05,
+      "loss": 3.9243,
+      "step": 4467
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.838417147568013e-05,
+      "loss": 3.2126,
+      "step": 4468
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8363561417971974e-05,
+      "loss": 5.942,
+      "step": 4469
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.834295136026381e-05,
+      "loss": 5.7833,
+      "step": 4470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8322341302555647e-05,
+      "loss": 5.8002,
+      "step": 4471
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8301731244847487e-05,
+      "loss": 5.8344,
+      "step": 4472
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8281121187139323e-05,
+      "loss": 5.348,
+      "step": 4473
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8260511129431163e-05,
+      "loss": 5.2916,
+      "step": 4474
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8239901071723e-05,
+      "loss": 5.2287,
+      "step": 4475
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.821929101401484e-05,
+      "loss": 5.2813,
+      "step": 4476
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.819868095630668e-05,
+      "loss": 5.2662,
+      "step": 4477
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8178070898598517e-05,
+      "loss": 5.2559,
+      "step": 4478
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8157460840890357e-05,
+      "loss": 5.0452,
+      "step": 4479
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8136850783182193e-05,
+      "loss": 5.2912,
+      "step": 4480
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8116240725474033e-05,
+      "loss": 5.4133,
+      "step": 4481
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.809563066776587e-05,
+      "loss": 5.4443,
+      "step": 4482
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8075020610057706e-05,
+      "loss": 5.0242,
+      "step": 4483
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.805441055234955e-05,
+      "loss": 4.9355,
+      "step": 4484
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8033800494641386e-05,
+      "loss": 5.0504,
+      "step": 4485
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.8013190436933226e-05,
+      "loss": 5.0545,
+      "step": 4486
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7992580379225063e-05,
+      "loss": 4.7357,
+      "step": 4487
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.79719703215169e-05,
+      "loss": 5.403,
+      "step": 4488
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.795136026380874e-05,
+      "loss": 5.231,
+      "step": 4489
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7930750206100576e-05,
+      "loss": 5.2498,
+      "step": 4490
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7910140148392416e-05,
+      "loss": 5.0503,
+      "step": 4491
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7889530090684256e-05,
+      "loss": 4.8945,
+      "step": 4492
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7868920032976093e-05,
+      "loss": 4.9551,
+      "step": 4493
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7848309975267933e-05,
+      "loss": 5.1155,
+      "step": 4494
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.782769991755977e-05,
+      "loss": 4.9611,
+      "step": 4495
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.780708985985161e-05,
+      "loss": 4.9293,
+      "step": 4496
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7786479802143446e-05,
+      "loss": 5.0014,
+      "step": 4497
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7765869744435286e-05,
+      "loss": 4.871,
+      "step": 4498
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7745259686727123e-05,
+      "loss": 4.9855,
+      "step": 4499
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7724649629018963e-05,
+      "loss": 5.2332,
+      "step": 4500
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 19.54161262512207,
+      "eval_runtime": 1336.7962,
+      "eval_samples_per_second": 1.976,
+      "eval_steps_per_second": 0.248,
+      "eval_wer": 1.0008131693772313,
+      "step": 4500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7704039571310803e-05,
+      "loss": 5.123,
+      "step": 4501
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.768342951360264e-05,
+      "loss": 5.1154,
+      "step": 4502
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.766281945589448e-05,
+      "loss": 5.2048,
+      "step": 4503
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7642209398186316e-05,
+      "loss": 5.1437,
+      "step": 4504
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7621599340478152e-05,
+      "loss": 4.7699,
+      "step": 4505
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7600989282769992e-05,
+      "loss": 4.7523,
+      "step": 4506
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.758037922506183e-05,
+      "loss": 4.9403,
+      "step": 4507
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7559769167353672e-05,
+      "loss": 4.5738,
+      "step": 4508
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7559769167353672e-05,
+      "loss": 5.0021,
+      "step": 4509
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.753915910964551e-05,
+      "loss": 4.6907,
+      "step": 4510
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7518549051937346e-05,
+      "loss": 4.6278,
+      "step": 4511
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7497938994229186e-05,
+      "loss": 4.6609,
+      "step": 4512
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7477328936521022e-05,
+      "loss": 4.7447,
+      "step": 4513
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7456718878812862e-05,
+      "loss": 4.403,
+      "step": 4514
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.74361088211047e-05,
+      "loss": 3.6336,
+      "step": 4515
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.741549876339654e-05,
+      "loss": 3.7642,
+      "step": 4516
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.739488870568838e-05,
+      "loss": 3.1187,
+      "step": 4517
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7374278647980215e-05,
+      "loss": 2.7915,
+      "step": 4518
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7353668590272055e-05,
+      "loss": 6.018,
+      "step": 4519
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7333058532563892e-05,
+      "loss": 5.9712,
+      "step": 4520
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7312448474855732e-05,
+      "loss": 6.0018,
+      "step": 4521
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.729183841714757e-05,
+      "loss": 5.9353,
+      "step": 4522
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7271228359439405e-05,
+      "loss": 5.6326,
+      "step": 4523
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7250618301731245e-05,
+      "loss": 5.6639,
+      "step": 4524
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7230008244023085e-05,
+      "loss": 5.629,
+      "step": 4525
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7209398186314925e-05,
+      "loss": 5.3567,
+      "step": 4526
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.718878812860676e-05,
+      "loss": 5.2605,
+      "step": 4527
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7168178070898598e-05,
+      "loss": 5.3579,
+      "step": 4528
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7147568013190438e-05,
+      "loss": 5.0081,
+      "step": 4529
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7126957955482275e-05,
+      "loss": 5.3893,
+      "step": 4530
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7106347897774115e-05,
+      "loss": 5.2728,
+      "step": 4531
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.708573784006595e-05,
+      "loss": 5.3676,
+      "step": 4532
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.706512778235779e-05,
+      "loss": 5.2521,
+      "step": 4533
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.704451772464963e-05,
+      "loss": 5.1503,
+      "step": 4534
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7023907666941468e-05,
+      "loss": 4.9875,
+      "step": 4535
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7003297609233308e-05,
+      "loss": 5.2658,
+      "step": 4536
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6982687551525145e-05,
+      "loss": 5.2065,
+      "step": 4537
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6962077493816985e-05,
+      "loss": 4.8733,
+      "step": 4538
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.694146743610882e-05,
+      "loss": 5.1496,
+      "step": 4539
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.6920857378400658e-05,
+      "loss": 5.1355,
+      "step": 4540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.69002473206925e-05,
+      "loss": 5.0809,
+      "step": 4541
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6879637262984338e-05,
+      "loss": 4.6728,
+      "step": 4542
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6859027205276174e-05,
+      "loss": 5.0352,
+      "step": 4543
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6838417147568014e-05,
+      "loss": 5.0726,
+      "step": 4544
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.681780708985985e-05,
+      "loss": 4.9712,
+      "step": 4545
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.679719703215169e-05,
+      "loss": 4.802,
+      "step": 4546
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6776586974443528e-05,
+      "loss": 4.98,
+      "step": 4547
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6755976916735368e-05,
+      "loss": 5.4892,
+      "step": 4548
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6735366859027208e-05,
+      "loss": 4.9091,
+      "step": 4549
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6714756801319044e-05,
+      "loss": 4.9348,
+      "step": 4550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6694146743610884e-05,
+      "loss": 4.8223,
+      "step": 4551
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.667353668590272e-05,
+      "loss": 5.0853,
+      "step": 4552
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.665292662819456e-05,
+      "loss": 4.5983,
+      "step": 4553
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6632316570486397e-05,
+      "loss": 5.2738,
+      "step": 4554
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6611706512778234e-05,
+      "loss": 4.7383,
+      "step": 4555
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6591096455070074e-05,
+      "loss": 4.6844,
+      "step": 4556
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6570486397361914e-05,
+      "loss": 4.8503,
+      "step": 4557
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.6549876339653754e-05,
+      "loss": 5.3885,
+      "step": 4558
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.652926628194559e-05,
+      "loss": 4.5072,
+      "step": 4559
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6508656224237427e-05,
+      "loss": 4.3443,
+      "step": 4560
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6488046166529267e-05,
+      "loss": 4.4435,
+      "step": 4561
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6467436108821104e-05,
+      "loss": 4.3293,
+      "step": 4562
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6446826051112944e-05,
+      "loss": 4.2233,
+      "step": 4563
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.642621599340478e-05,
+      "loss": 3.9418,
+      "step": 4564
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.640560593569662e-05,
+      "loss": 4.1555,
+      "step": 4565
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.638499587798846e-05,
+      "loss": 3.6353,
+      "step": 4566
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6364385820280297e-05,
+      "loss": 3.3844,
+      "step": 4567
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6343775762572137e-05,
+      "loss": 3.3375,
+      "step": 4568
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6323165704863974e-05,
+      "loss": 5.9592,
+      "step": 4569
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6302555647155814e-05,
+      "loss": 5.8454,
+      "step": 4570
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.628194558944765e-05,
+      "loss": 5.7155,
+      "step": 4571
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6261335531739487e-05,
+      "loss": 5.6593,
+      "step": 4572
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.624072547403133e-05,
+      "loss": 5.7745,
+      "step": 4573
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6220115416323167e-05,
+      "loss": 5.3609,
+      "step": 4574
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.6199505358615007e-05,
+      "loss": 5.0432,
+      "step": 4575
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6178895300906843e-05,
+      "loss": 4.9757,
+      "step": 4576
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.615828524319868e-05,
+      "loss": 5.0136,
+      "step": 4577
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.613767518549052e-05,
+      "loss": 5.5721,
+      "step": 4578
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6117065127782357e-05,
+      "loss": 5.3818,
+      "step": 4579
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6096455070074197e-05,
+      "loss": 5.3473,
+      "step": 4580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6075845012366037e-05,
+      "loss": 5.8505,
+      "step": 4581
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6055234954657873e-05,
+      "loss": 5.5361,
+      "step": 4582
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.6034624896949713e-05,
+      "loss": 5.4926,
+      "step": 4583
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.601401483924155e-05,
+      "loss": 5.2687,
+      "step": 4584
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.599340478153339e-05,
+      "loss": 5.5114,
+      "step": 4585
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5972794723825226e-05,
+      "loss": 4.9758,
+      "step": 4586
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5952184666117066e-05,
+      "loss": 5.1631,
+      "step": 4587
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5931574608408903e-05,
+      "loss": 4.846,
+      "step": 4588
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5910964550700743e-05,
+      "loss": 5.1495,
+      "step": 4589
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5890354492992583e-05,
+      "loss": 4.8411,
+      "step": 4590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.586974443528442e-05,
+      "loss": 5.2266,
+      "step": 4591
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.584913437757626e-05,
+      "loss": 5.239,
+      "step": 4592
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.5828524319868096e-05,
+      "loss": 5.1138,
+      "step": 4593
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5807914262159933e-05,
+      "loss": 4.7442,
+      "step": 4594
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5787304204451773e-05,
+      "loss": 4.8861,
+      "step": 4595
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5766694146743613e-05,
+      "loss": 4.7221,
+      "step": 4596
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5746084089035453e-05,
+      "loss": 5.0963,
+      "step": 4597
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.572547403132729e-05,
+      "loss": 5.0493,
+      "step": 4598
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5704863973619126e-05,
+      "loss": 4.9951,
+      "step": 4599
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5684253915910966e-05,
+      "loss": 5.0208,
+      "step": 4600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5663643858202802e-05,
+      "loss": 5.1159,
+      "step": 4601
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5643033800494642e-05,
+      "loss": 4.9198,
+      "step": 4602
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.562242374278648e-05,
+      "loss": 5.1606,
+      "step": 4603
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.560181368507832e-05,
+      "loss": 4.7225,
+      "step": 4604
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.558120362737016e-05,
+      "loss": 4.9843,
+      "step": 4605
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5560593569661996e-05,
+      "loss": 4.9399,
+      "step": 4606
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5539983511953836e-05,
+      "loss": 4.8815,
+      "step": 4607
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5519373454245672e-05,
+      "loss": 4.6983,
+      "step": 4608
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5498763396537512e-05,
+      "loss": 4.8812,
+      "step": 4609
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.547815333882935e-05,
+      "loss": 4.8366,
+      "step": 4610
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.5457543281121185e-05,
+      "loss": 4.6595,
+      "step": 4611
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.543693322341303e-05,
+      "loss": 4.3515,
+      "step": 4612
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5416323165704865e-05,
+      "loss": 4.4063,
+      "step": 4613
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5395713107996705e-05,
+      "loss": 4.4038,
+      "step": 4614
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5375103050288542e-05,
+      "loss": 3.2651,
+      "step": 4615
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.535449299258038e-05,
+      "loss": 3.3817,
+      "step": 4616
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.533388293487222e-05,
+      "loss": 3.2292,
+      "step": 4617
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5313272877164055e-05,
+      "loss": 2.5181,
+      "step": 4618
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5292662819455895e-05,
+      "loss": 6.1351,
+      "step": 4619
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5272052761747735e-05,
+      "loss": 6.0788,
+      "step": 4620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5251442704039573e-05,
+      "loss": 5.8742,
+      "step": 4621
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5230832646331412e-05,
+      "loss": 6.1194,
+      "step": 4622
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5210222588623248e-05,
+      "loss": 5.8599,
+      "step": 4623
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5189612530915087e-05,
+      "loss": 5.8313,
+      "step": 4624
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5169002473206925e-05,
+      "loss": 5.8715,
+      "step": 4625
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5148392415498763e-05,
+      "loss": 5.6002,
+      "step": 4626
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5127782357790602e-05,
+      "loss": 5.6315,
+      "step": 4627
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.5107172300082442e-05,
+      "loss": 5.2351,
+      "step": 4628
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.508656224237428e-05,
+      "loss": 5.2491,
+      "step": 4629
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5065952184666118e-05,
+      "loss": 5.543,
+      "step": 4630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5045342126957956e-05,
+      "loss": 5.0543,
+      "step": 4631
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5024732069249795e-05,
+      "loss": 5.4549,
+      "step": 4632
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.5004122011541633e-05,
+      "loss": 5.2722,
+      "step": 4633
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4983511953833471e-05,
+      "loss": 5.1119,
+      "step": 4634
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4962901896125308e-05,
+      "loss": 5.3516,
+      "step": 4635
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.494229183841715e-05,
+      "loss": 5.0637,
+      "step": 4636
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4921681780708988e-05,
+      "loss": 5.0638,
+      "step": 4637
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4901071723000826e-05,
+      "loss": 5.2384,
+      "step": 4638
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4880461665292665e-05,
+      "loss": 5.1097,
+      "step": 4639
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4859851607584501e-05,
+      "loss": 4.9293,
+      "step": 4640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.483924154987634e-05,
+      "loss": 5.1256,
+      "step": 4641
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4818631492168178e-05,
+      "loss": 5.0647,
+      "step": 4642
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4798021434460016e-05,
+      "loss": 5.2194,
+      "step": 4643
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4777411376751856e-05,
+      "loss": 5.2367,
+      "step": 4644
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4756801319043694e-05,
+      "loss": 4.9294,
+      "step": 4645
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4736191261335533e-05,
+      "loss": 5.2209,
+      "step": 4646
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.4715581203627371e-05,
+      "loss": 5.3206,
+      "step": 4647
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.469497114591921e-05,
+      "loss": 4.9015,
+      "step": 4648
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4674361088211048e-05,
+      "loss": 5.2153,
+      "step": 4649
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4653751030502886e-05,
+      "loss": 4.9614,
+      "step": 4650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4633140972794724e-05,
+      "loss": 5.0746,
+      "step": 4651
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4612530915086564e-05,
+      "loss": 4.8597,
+      "step": 4652
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4591920857378402e-05,
+      "loss": 5.2064,
+      "step": 4653
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.457131079967024e-05,
+      "loss": 5.0643,
+      "step": 4654
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4550700741962079e-05,
+      "loss": 5.009,
+      "step": 4655
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4530090684253916e-05,
+      "loss": 4.868,
+      "step": 4656
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4509480626545754e-05,
+      "loss": 4.8306,
+      "step": 4657
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4488870568837592e-05,
+      "loss": 4.8185,
+      "step": 4658
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.446826051112943e-05,
+      "loss": 4.6676,
+      "step": 4659
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4447650453421272e-05,
+      "loss": 4.3897,
+      "step": 4660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4427040395713109e-05,
+      "loss": 4.4645,
+      "step": 4661
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4406430338004947e-05,
+      "loss": 4.4046,
+      "step": 4662
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4385820280296785e-05,
+      "loss": 3.9317,
+      "step": 4663
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4365210222588624e-05,
+      "loss": 3.8195,
+      "step": 4664
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.4344600164880462e-05,
+      "loss": 3.969,
+      "step": 4665
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.43239901071723e-05,
+      "loss": 3.499,
+      "step": 4666
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4303380049464139e-05,
+      "loss": 3.862,
+      "step": 4667
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4282769991755979e-05,
+      "loss": 2.8942,
+      "step": 4668
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4262159934047817e-05,
+      "loss": 5.429,
+      "step": 4669
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4241549876339655e-05,
+      "loss": 5.6371,
+      "step": 4670
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4220939818631493e-05,
+      "loss": 5.3697,
+      "step": 4671
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4200329760923332e-05,
+      "loss": 5.306,
+      "step": 4672
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4179719703215168e-05,
+      "loss": 5.2006,
+      "step": 4673
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4159109645507007e-05,
+      "loss": 5.3761,
+      "step": 4674
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4138499587798845e-05,
+      "loss": 5.2137,
+      "step": 4675
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4117889530090687e-05,
+      "loss": 5.1829,
+      "step": 4676
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4097279472382525e-05,
+      "loss": 4.8796,
+      "step": 4677
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4076669414674362e-05,
+      "loss": 5.3649,
+      "step": 4678
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.40560593569662e-05,
+      "loss": 5.3002,
+      "step": 4679
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4035449299258038e-05,
+      "loss": 5.4689,
+      "step": 4680
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.4014839241549876e-05,
+      "loss": 5.1358,
+      "step": 4681
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.3994229183841715e-05,
+      "loss": 4.8624,
+      "step": 4682
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.3973619126133553e-05,
+      "loss": 5.409,
+      "step": 4683
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3953009068425393e-05,
+      "loss": 4.975,
+      "step": 4684
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3932399010717231e-05,
+      "loss": 5.449,
+      "step": 4685
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.391178895300907e-05,
+      "loss": 5.1602,
+      "step": 4686
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3891178895300908e-05,
+      "loss": 5.2282,
+      "step": 4687
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3870568837592746e-05,
+      "loss": 4.8521,
+      "step": 4688
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3849958779884584e-05,
+      "loss": 4.8571,
+      "step": 4689
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3829348722176421e-05,
+      "loss": 4.9942,
+      "step": 4690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.380873866446826e-05,
+      "loss": 5.1394,
+      "step": 4691
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3788128606760101e-05,
+      "loss": 5.019,
+      "step": 4692
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.376751854905194e-05,
+      "loss": 4.9152,
+      "step": 4693
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3746908491343776e-05,
+      "loss": 5.1694,
+      "step": 4694
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3726298433635614e-05,
+      "loss": 5.2256,
+      "step": 4695
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3705688375927453e-05,
+      "loss": 5.0988,
+      "step": 4696
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3685078318219291e-05,
+      "loss": 5.0182,
+      "step": 4697
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.366446826051113e-05,
+      "loss": 4.9901,
+      "step": 4698
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3643858202802967e-05,
+      "loss": 5.3336,
+      "step": 4699
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.3623248145094807e-05,
+      "loss": 4.7714,
+      "step": 4700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3602638087386646e-05,
+      "loss": 5.1735,
+      "step": 4701
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3582028029678484e-05,
+      "loss": 4.7007,
+      "step": 4702
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3561417971970322e-05,
+      "loss": 4.927,
+      "step": 4703
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.354080791426216e-05,
+      "loss": 4.9655,
+      "step": 4704
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3520197856553999e-05,
+      "loss": 4.7445,
+      "step": 4705
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3499587798845836e-05,
+      "loss": 4.9622,
+      "step": 4706
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3478977741137677e-05,
+      "loss": 4.7726,
+      "step": 4707
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3458367683429516e-05,
+      "loss": 4.4683,
+      "step": 4708
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3437757625721354e-05,
+      "loss": 4.6015,
+      "step": 4709
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3417147568013192e-05,
+      "loss": 4.7519,
+      "step": 4710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3396537510305029e-05,
+      "loss": 4.6081,
+      "step": 4711
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3375927452596867e-05,
+      "loss": 4.6676,
+      "step": 4712
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3355317394888705e-05,
+      "loss": 3.8955,
+      "step": 4713
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3334707337180544e-05,
+      "loss": 3.8866,
+      "step": 4714
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3314097279472385e-05,
+      "loss": 4.2565,
+      "step": 4715
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3293487221764222e-05,
+      "loss": 4.0587,
+      "step": 4716
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.327287716405606e-05,
+      "loss": 4.0142,
+      "step": 4717
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.3252267106347899e-05,
+      "loss": 3.4287,
+      "step": 4718
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3231657048639737e-05,
+      "loss": 5.7851,
+      "step": 4719
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3211046990931575e-05,
+      "loss": 5.5089,
+      "step": 4720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3190436933223413e-05,
+      "loss": 5.5559,
+      "step": 4721
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3169826875515252e-05,
+      "loss": 5.6857,
+      "step": 4722
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3149216817807092e-05,
+      "loss": 5.345,
+      "step": 4723
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.312860676009893e-05,
+      "loss": 5.5466,
+      "step": 4724
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3107996702390768e-05,
+      "loss": 5.2777,
+      "step": 4725
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3087386644682607e-05,
+      "loss": 5.1564,
+      "step": 4726
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3066776586974445e-05,
+      "loss": 5.2438,
+      "step": 4727
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3046166529266281e-05,
+      "loss": 4.8076,
+      "step": 4728
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.302555647155812e-05,
+      "loss": 5.2649,
+      "step": 4729
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.3004946413849958e-05,
+      "loss": 5.2452,
+      "step": 4730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.29843363561418e-05,
+      "loss": 5.2303,
+      "step": 4731
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2963726298433636e-05,
+      "loss": 5.5413,
+      "step": 4732
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2943116240725475e-05,
+      "loss": 5.0739,
+      "step": 4733
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2922506183017313e-05,
+      "loss": 5.1729,
+      "step": 4734
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.2901896125309151e-05,
+      "loss": 5.1729,
+      "step": 4735
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.288128606760099e-05,
+      "loss": 5.0421,
+      "step": 4736
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2860676009892828e-05,
+      "loss": 5.0838,
+      "step": 4737
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2840065952184666e-05,
+      "loss": 5.0516,
+      "step": 4738
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2819455894476506e-05,
+      "loss": 5.1414,
+      "step": 4739
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2798845836768344e-05,
+      "loss": 4.9941,
+      "step": 4740
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2778235779060183e-05,
+      "loss": 4.9446,
+      "step": 4741
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2757625721352021e-05,
+      "loss": 4.8422,
+      "step": 4742
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.273701566364386e-05,
+      "loss": 5.1293,
+      "step": 4743
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2716405605935696e-05,
+      "loss": 4.8858,
+      "step": 4744
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2695795548227534e-05,
+      "loss": 4.9495,
+      "step": 4745
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2675185490519373e-05,
+      "loss": 4.9746,
+      "step": 4746
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2654575432811214e-05,
+      "loss": 5.0871,
+      "step": 4747
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2633965375103053e-05,
+      "loss": 4.8242,
+      "step": 4748
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2613355317394889e-05,
+      "loss": 4.769,
+      "step": 4749
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2592745259686727e-05,
+      "loss": 5.2947,
+      "step": 4750
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2572135201978566e-05,
+      "loss": 5.0034,
+      "step": 4751
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2551525144270404e-05,
+      "loss": 4.7879,
+      "step": 4752
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.2530915086562242e-05,
+      "loss": 4.7251,
+      "step": 4753
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.251030502885408e-05,
+      "loss": 4.8182,
+      "step": 4754
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2489694971145919e-05,
+      "loss": 4.9844,
+      "step": 4755
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2469084913437757e-05,
+      "loss": 4.7295,
+      "step": 4756
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2448474855729597e-05,
+      "loss": 4.667,
+      "step": 4757
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2427864798021435e-05,
+      "loss": 4.9337,
+      "step": 4758
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2407254740313274e-05,
+      "loss": 4.8693,
+      "step": 4759
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2386644682605112e-05,
+      "loss": 4.8661,
+      "step": 4760
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.236603462489695e-05,
+      "loss": 4.5056,
+      "step": 4761
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2345424567188789e-05,
+      "loss": 4.0508,
+      "step": 4762
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2324814509480627e-05,
+      "loss": 4.1134,
+      "step": 4763
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2304204451772465e-05,
+      "loss": 4.1416,
+      "step": 4764
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2283594394064305e-05,
+      "loss": 3.9043,
+      "step": 4765
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2262984336356142e-05,
+      "loss": 3.8532,
+      "step": 4766
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.224237427864798e-05,
+      "loss": 3.521,
+      "step": 4767
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2221764220939818e-05,
+      "loss": 2.8608,
+      "step": 4768
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2201154163231658e-05,
+      "loss": 5.7447,
+      "step": 4769
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2180544105523497e-05,
+      "loss": 5.4452,
+      "step": 4770
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2159934047815335e-05,
+      "loss": 5.6244,
+      "step": 4771
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.2139323990107172e-05,
+      "loss": 5.4044,
+      "step": 4772
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2118713932399012e-05,
+      "loss": 5.3677,
+      "step": 4773
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.209810387469085e-05,
+      "loss": 5.3471,
+      "step": 4774
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2077493816982688e-05,
+      "loss": 5.1157,
+      "step": 4775
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2056883759274527e-05,
+      "loss": 5.0118,
+      "step": 4776
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2036273701566365e-05,
+      "loss": 5.0719,
+      "step": 4777
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.2015663643858203e-05,
+      "loss": 4.6052,
+      "step": 4778
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1995053586150041e-05,
+      "loss": 5.53,
+      "step": 4779
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.197444352844188e-05,
+      "loss": 5.1533,
+      "step": 4780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.195383347073372e-05,
+      "loss": 5.1219,
+      "step": 4781
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1933223413025556e-05,
+      "loss": 5.292,
+      "step": 4782
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1912613355317395e-05,
+      "loss": 4.8821,
+      "step": 4783
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1892003297609233e-05,
+      "loss": 4.7307,
+      "step": 4784
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1892003297609233e-05,
+      "loss": 5.5673,
+      "step": 4785
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1871393239901073e-05,
+      "loss": 4.9396,
+      "step": 4786
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1850783182192911e-05,
+      "loss": 5.1287,
+      "step": 4787
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.183017312448475e-05,
+      "loss": 5.0082,
+      "step": 4788
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1809563066776586e-05,
+      "loss": 4.826,
+      "step": 4789
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.1788953009068426e-05,
+      "loss": 5.0319,
+      "step": 4790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1768342951360264e-05,
+      "loss": 4.9444,
+      "step": 4791
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1747732893652103e-05,
+      "loss": 4.9291,
+      "step": 4792
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1727122835943941e-05,
+      "loss": 4.8865,
+      "step": 4793
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.170651277823578e-05,
+      "loss": 5.4163,
+      "step": 4794
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1685902720527618e-05,
+      "loss": 5.1865,
+      "step": 4795
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1665292662819456e-05,
+      "loss": 4.961,
+      "step": 4796
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1644682605111294e-05,
+      "loss": 4.9457,
+      "step": 4797
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1624072547403134e-05,
+      "loss": 5.0736,
+      "step": 4798
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1603462489694972e-05,
+      "loss": 4.8695,
+      "step": 4799
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1582852431986809e-05,
+      "loss": 5.1756,
+      "step": 4800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1562242374278647e-05,
+      "loss": 4.8373,
+      "step": 4801
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1541632316570487e-05,
+      "loss": 4.8387,
+      "step": 4802
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1521022258862326e-05,
+      "loss": 4.5614,
+      "step": 4803
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1500412201154164e-05,
+      "loss": 4.8019,
+      "step": 4804
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1479802143446002e-05,
+      "loss": 5.3571,
+      "step": 4805
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.145919208573784e-05,
+      "loss": 5.1733,
+      "step": 4806
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.1438582028029679e-05,
+      "loss": 4.7448,
+      "step": 4807
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1417971970321517e-05,
+      "loss": 4.3336,
+      "step": 4808
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1397361912613355e-05,
+      "loss": 4.5375,
+      "step": 4809
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1376751854905195e-05,
+      "loss": 4.7866,
+      "step": 4810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1356141797197032e-05,
+      "loss": 4.5435,
+      "step": 4811
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.133553173948887e-05,
+      "loss": 4.4659,
+      "step": 4812
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.131492168178071e-05,
+      "loss": 4.2504,
+      "step": 4813
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1294311624072549e-05,
+      "loss": 4.0033,
+      "step": 4814
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1273701566364387e-05,
+      "loss": 4.1132,
+      "step": 4815
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1253091508656225e-05,
+      "loss": 3.8681,
+      "step": 4816
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1232481450948064e-05,
+      "loss": 3.3735,
+      "step": 4817
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1211871393239902e-05,
+      "loss": 2.7238,
+      "step": 4818
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.119126133553174e-05,
+      "loss": 5.5376,
+      "step": 4819
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1170651277823578e-05,
+      "loss": 5.3701,
+      "step": 4820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1150041220115417e-05,
+      "loss": 5.4754,
+      "step": 4821
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1129431162407255e-05,
+      "loss": 5.3637,
+      "step": 4822
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1108821104699093e-05,
+      "loss": 5.2137,
+      "step": 4823
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1088211046990932e-05,
+      "loss": 5.0932,
+      "step": 4824
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.1067600989282772e-05,
+      "loss": 5.2597,
+      "step": 4825
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.104699093157461e-05,
+      "loss": 5.155,
+      "step": 4826
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1026380873866446e-05,
+      "loss": 5.5922,
+      "step": 4827
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.1005770816158285e-05,
+      "loss": 5.058,
+      "step": 4828
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0985160758450125e-05,
+      "loss": 4.8975,
+      "step": 4829
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0964550700741963e-05,
+      "loss": 5.2986,
+      "step": 4830
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0943940643033801e-05,
+      "loss": 4.9683,
+      "step": 4831
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.092333058532564e-05,
+      "loss": 5.1106,
+      "step": 4832
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0902720527617478e-05,
+      "loss": 4.9802,
+      "step": 4833
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0882110469909316e-05,
+      "loss": 5.067,
+      "step": 4834
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0861500412201155e-05,
+      "loss": 4.8896,
+      "step": 4835
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0840890354492993e-05,
+      "loss": 5.0764,
+      "step": 4836
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0820280296784833e-05,
+      "loss": 4.9161,
+      "step": 4837
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.079967023907667e-05,
+      "loss": 5.2063,
+      "step": 4838
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0779060181368508e-05,
+      "loss": 4.8832,
+      "step": 4839
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0758450123660346e-05,
+      "loss": 5.2586,
+      "step": 4840
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0737840065952186e-05,
+      "loss": 5.1788,
+      "step": 4841
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0717230008244024e-05,
+      "loss": 5.0235,
+      "step": 4842
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.0696619950535863e-05,
+      "loss": 4.8517,
+      "step": 4843
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.06760098928277e-05,
+      "loss": 4.7311,
+      "step": 4844
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.065539983511954e-05,
+      "loss": 5.1198,
+      "step": 4845
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0634789777411378e-05,
+      "loss": 4.7417,
+      "step": 4846
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0614179719703216e-05,
+      "loss": 4.691,
+      "step": 4847
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0593569661995054e-05,
+      "loss": 4.8982,
+      "step": 4848
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0572959604286892e-05,
+      "loss": 4.9373,
+      "step": 4849
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.055234954657873e-05,
+      "loss": 5.4365,
+      "step": 4850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0531739488870569e-05,
+      "loss": 4.7536,
+      "step": 4851
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0511129431162407e-05,
+      "loss": 4.9543,
+      "step": 4852
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0490519373454247e-05,
+      "loss": 4.8326,
+      "step": 4853
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0469909315746086e-05,
+      "loss": 5.4828,
+      "step": 4854
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0449299258037922e-05,
+      "loss": 5.0004,
+      "step": 4855
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.042868920032976e-05,
+      "loss": 4.886,
+      "step": 4856
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.04080791426216e-05,
+      "loss": 4.5349,
+      "step": 4857
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0387469084913439e-05,
+      "loss": 4.814,
+      "step": 4858
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0366859027205277e-05,
+      "loss": 4.6566,
+      "step": 4859
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0346248969497115e-05,
+      "loss": 4.8235,
+      "step": 4860
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.0325638911788954e-05,
+      "loss": 4.4494,
+      "step": 4861
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0305028854080792e-05,
+      "loss": 4.6863,
+      "step": 4862
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.028441879637263e-05,
+      "loss": 4.092,
+      "step": 4863
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0263808738664469e-05,
+      "loss": 4.0433,
+      "step": 4864
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0243198680956307e-05,
+      "loss": 3.8307,
+      "step": 4865
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0222588623248145e-05,
+      "loss": 3.6246,
+      "step": 4866
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0201978565539983e-05,
+      "loss": 2.9351,
+      "step": 4867
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0181368507831822e-05,
+      "loss": 2.486,
+      "step": 4868
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0160758450123662e-05,
+      "loss": 5.9463,
+      "step": 4869
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.01401483924155e-05,
+      "loss": 5.6454,
+      "step": 4870
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0119538334707337e-05,
+      "loss": 5.6587,
+      "step": 4871
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0098928276999175e-05,
+      "loss": 5.4793,
+      "step": 4872
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0078318219291015e-05,
+      "loss": 5.6331,
+      "step": 4873
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0057708161582853e-05,
+      "loss": 5.4195,
+      "step": 4874
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.0037098103874692e-05,
+      "loss": 5.4622,
+      "step": 4875
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.001648804616653e-05,
+      "loss": 5.2223,
+      "step": 4876
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.995877988458368e-06,
+      "loss": 5.295,
+      "step": 4877
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.975267930750206e-06,
+      "loss": 5.1182,
+      "step": 4878
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.954657873042045e-06,
+      "loss": 4.8489,
+      "step": 4879
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.934047815333883e-06,
+      "loss": 5.3366,
+      "step": 4880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.913437757625723e-06,
+      "loss": 4.8296,
+      "step": 4881
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.89282769991756e-06,
+      "loss": 5.1767,
+      "step": 4882
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.872217642209398e-06,
+      "loss": 5.2692,
+      "step": 4883
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.851607584501236e-06,
+      "loss": 4.968,
+      "step": 4884
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.830997526793076e-06,
+      "loss": 5.5031,
+      "step": 4885
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.810387469084915e-06,
+      "loss": 5.0568,
+      "step": 4886
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.789777411376753e-06,
+      "loss": 5.263,
+      "step": 4887
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.76916735366859e-06,
+      "loss": 4.8673,
+      "step": 4888
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.74855729596043e-06,
+      "loss": 5.1399,
+      "step": 4889
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.727947238252268e-06,
+      "loss": 4.9183,
+      "step": 4890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.707337180544106e-06,
+      "loss": 5.1612,
+      "step": 4891
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.686727122835944e-06,
+      "loss": 4.9598,
+      "step": 4892
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.666117065127783e-06,
+      "loss": 5.0708,
+      "step": 4893
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.645507007419621e-06,
+      "loss": 4.799,
+      "step": 4894
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.62489694971146e-06,
+      "loss": 5.0378,
+      "step": 4895
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.604286892003297e-06,
+      "loss": 4.8376,
+      "step": 4896
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.583676834295137e-06,
+      "loss": 4.9034,
+      "step": 4897
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.563066776586976e-06,
+      "loss": 5.1397,
+      "step": 4898
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.542456718878812e-06,
+      "loss": 4.8653,
+      "step": 4899
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.52184666117065e-06,
+      "loss": 4.7822,
+      "step": 4900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.50123660346249e-06,
+      "loss": 4.7661,
+      "step": 4901
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.480626545754329e-06,
+      "loss": 5.0049,
+      "step": 4902
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.460016488046167e-06,
+      "loss": 4.8534,
+      "step": 4903
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.439406430338006e-06,
+      "loss": 4.6969,
+      "step": 4904
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.418796372629844e-06,
+      "loss": 4.6595,
+      "step": 4905
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.398186314921682e-06,
+      "loss": 4.5619,
+      "step": 4906
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.37757625721352e-06,
+      "loss": 4.5803,
+      "step": 4907
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.356966199505359e-06,
+      "loss": 4.9269,
+      "step": 4908
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.336356141797197e-06,
+      "loss": 4.8006,
+      "step": 4909
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.315746084089035e-06,
+      "loss": 4.691,
+      "step": 4910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.295136026380874e-06,
+      "loss": 4.4799,
+      "step": 4911
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.274525968672712e-06,
+      "loss": 4.2392,
+      "step": 4912
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.253915910964552e-06,
+      "loss": 4.1713,
+      "step": 4913
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.23330585325639e-06,
+      "loss": 3.8492,
+      "step": 4914
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.212695795548227e-06,
+      "loss": 3.9258,
+      "step": 4915
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.192085737840065e-06,
+      "loss": 3.633,
+      "step": 4916
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.171475680131905e-06,
+      "loss": 3.5011,
+      "step": 4917
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.150865622423743e-06,
+      "loss": 2.9365,
+      "step": 4918
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.130255564715582e-06,
+      "loss": 5.7961,
+      "step": 4919
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.10964550700742e-06,
+      "loss": 5.6237,
+      "step": 4920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.089035449299258e-06,
+      "loss": 5.6298,
+      "step": 4921
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.068425391591097e-06,
+      "loss": 5.9162,
+      "step": 4922
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.047815333882935e-06,
+      "loss": 5.534,
+      "step": 4923
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.027205276174775e-06,
+      "loss": 5.5484,
+      "step": 4924
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.006595218466613e-06,
+      "loss": 5.6853,
+      "step": 4925
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.98598516075845e-06,
+      "loss": 5.5569,
+      "step": 4926
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.965375103050288e-06,
+      "loss": 5.4454,
+      "step": 4927
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.944765045342128e-06,
+      "loss": 5.4268,
+      "step": 4928
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.924154987633966e-06,
+      "loss": 5.0813,
+      "step": 4929
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.903544929925805e-06,
+      "loss": 4.9812,
+      "step": 4930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.882934872217643e-06,
+      "loss": 5.0007,
+      "step": 4931
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.862324814509481e-06,
+      "loss": 5.3056,
+      "step": 4932
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.84171475680132e-06,
+      "loss": 5.4851,
+      "step": 4933
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.821104699093158e-06,
+      "loss": 5.1525,
+      "step": 4934
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.800494641384996e-06,
+      "loss": 5.0617,
+      "step": 4935
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.779884583676836e-06,
+      "loss": 5.216,
+      "step": 4936
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.759274525968673e-06,
+      "loss": 5.4057,
+      "step": 4937
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.738664468260511e-06,
+      "loss": 5.1924,
+      "step": 4938
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.71805441055235e-06,
+      "loss": 5.105,
+      "step": 4939
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.69744435284419e-06,
+      "loss": 5.0465,
+      "step": 4940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.676834295136028e-06,
+      "loss": 4.8918,
+      "step": 4941
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.656224237427866e-06,
+      "loss": 4.911,
+      "step": 4942
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.635614179719703e-06,
+      "loss": 4.9606,
+      "step": 4943
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.615004122011543e-06,
+      "loss": 5.4342,
+      "step": 4944
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.59439406430338e-06,
+      "loss": 4.8832,
+      "step": 4945
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.573784006595219e-06,
+      "loss": 4.9707,
+      "step": 4946
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.553173948887057e-06,
+      "loss": 4.9893,
+      "step": 4947
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.532563891178896e-06,
+      "loss": 5.3547,
+      "step": 4948
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.511953833470734e-06,
+      "loss": 5.0437,
+      "step": 4949
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.491343775762572e-06,
+      "loss": 4.8788,
+      "step": 4950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.47073371805441e-06,
+      "loss": 5.1736,
+      "step": 4951
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.45012366034625e-06,
+      "loss": 4.881,
+      "step": 4952
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.429513602638087e-06,
+      "loss": 5.0827,
+      "step": 4953
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.408903544929926e-06,
+      "loss": 4.5825,
+      "step": 4954
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.388293487221764e-06,
+      "loss": 5.0815,
+      "step": 4955
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.367683429513604e-06,
+      "loss": 4.7131,
+      "step": 4956
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.347073371805442e-06,
+      "loss": 4.7564,
+      "step": 4957
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.32646331409728e-06,
+      "loss": 4.764,
+      "step": 4958
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.305853256389117e-06,
+      "loss": 4.3501,
+      "step": 4959
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.285243198680957e-06,
+      "loss": 4.7096,
+      "step": 4960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.264633140972795e-06,
+      "loss": 4.1267,
+      "step": 4961
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.244023083264634e-06,
+      "loss": 4.2533,
+      "step": 4962
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.223413025556472e-06,
+      "loss": 3.9581,
+      "step": 4963
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.20280296784831e-06,
+      "loss": 3.9697,
+      "step": 4964
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.182192910140148e-06,
+      "loss": 3.5526,
+      "step": 4965
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.161582852431987e-06,
+      "loss": 3.5424,
+      "step": 4966
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.140972794723825e-06,
+      "loss": 3.0776,
+      "step": 4967
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.120362737015665e-06,
+      "loss": 2.9366,
+      "step": 4968
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.099752679307503e-06,
+      "loss": 5.6801,
+      "step": 4969
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.07914262159934e-06,
+      "loss": 5.5831,
+      "step": 4970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.058532563891178e-06,
+      "loss": 5.7967,
+      "step": 4971
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.037922506183018e-06,
+      "loss": 5.5855,
+      "step": 4972
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.017312448474857e-06,
+      "loss": 5.5869,
+      "step": 4973
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.996702390766695e-06,
+      "loss": 5.3305,
+      "step": 4974
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.976092333058533e-06,
+      "loss": 5.3512,
+      "step": 4975
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.955482275350371e-06,
+      "loss": 5.6472,
+      "step": 4976
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.93487221764221e-06,
+      "loss": 5.3351,
+      "step": 4977
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.914262159934048e-06,
+      "loss": 5.2073,
+      "step": 4978
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.893652102225886e-06,
+      "loss": 5.0253,
+      "step": 4979
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.873042044517726e-06,
+      "loss": 5.0729,
+      "step": 4980
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.852431986809563e-06,
+      "loss": 5.3109,
+      "step": 4981
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.831821929101401e-06,
+      "loss": 4.8816,
+      "step": 4982
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.81121187139324e-06,
+      "loss": 5.2607,
+      "step": 4983
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.79060181368508e-06,
+      "loss": 4.9707,
+      "step": 4984
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.769991755976918e-06,
+      "loss": 5.2019,
+      "step": 4985
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.749381698268756e-06,
+      "loss": 5.3264,
+      "step": 4986
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.728771640560593e-06,
+      "loss": 4.9444,
+      "step": 4987
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.708161582852433e-06,
+      "loss": 5.1415,
+      "step": 4988
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.687551525144271e-06,
+      "loss": 4.9631,
+      "step": 4989
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.66694146743611e-06,
+      "loss": 5.1895,
+      "step": 4990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.646331409727948e-06,
+      "loss": 5.0302,
+      "step": 4991
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.625721352019787e-06,
+      "loss": 4.9718,
+      "step": 4992
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.605111294311624e-06,
+      "loss": 5.0668,
+      "step": 4993
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.5845012366034625e-06,
+      "loss": 4.7996,
+      "step": 4994
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.563891178895301e-06,
+      "loss": 5.2613,
+      "step": 4995
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.54328112118714e-06,
+      "loss": 4.7361,
+      "step": 4996
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.522671063478978e-06,
+      "loss": 4.8266,
+      "step": 4997
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.5020610057708165e-06,
+      "loss": 4.9903,
+      "step": 4998
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.481450948062654e-06,
+      "loss": 4.9355,
+      "step": 4999
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.460840890354494e-06,
+      "loss": 4.9792,
+      "step": 5000
+    },
+    {
+      "epoch": 2.8,
+      "eval_loss": 20.59589195251465,
+      "eval_runtime": 1341.2706,
+      "eval_samples_per_second": 1.97,
+      "eval_steps_per_second": 0.247,
+      "eval_wer": 1.0008131693772313,
+      "step": 5000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.440230832646332e-06,
+      "loss": 4.9942,
+      "step": 5001
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.41962077493817e-06,
+      "loss": 5.224,
+      "step": 5002
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.399010717230008e-06,
+      "loss": 4.8965,
+      "step": 5003
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 7.378400659521847e-06,
+      "loss": 4.5096,
+      "step": 5004
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.3577906018136855e-06,
+      "loss": 4.8244,
+      "step": 5005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.337180544105524e-06,
+      "loss": 4.8861,
+      "step": 5006
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.316570486397362e-06,
+      "loss": 4.8351,
+      "step": 5007
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.295960428689201e-06,
+      "loss": 4.6445,
+      "step": 5008
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.2753503709810395e-06,
+      "loss": 4.5059,
+      "step": 5009
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.254740313272877e-06,
+      "loss": 4.3818,
+      "step": 5010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.234130255564715e-06,
+      "loss": 4.6193,
+      "step": 5011
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.213520197856554e-06,
+      "loss": 4.316,
+      "step": 5012
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.192910140148393e-06,
+      "loss": 4.5101,
+      "step": 5013
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.172300082440231e-06,
+      "loss": 4.2116,
+      "step": 5014
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.151690024732069e-06,
+      "loss": 3.8023,
+      "step": 5015
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.1310799670239084e-06,
+      "loss": 4.0554,
+      "step": 5016
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.110469909315747e-06,
+      "loss": 3.2494,
+      "step": 5017
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.089859851607584e-06,
+      "loss": 2.8905,
+      "step": 5018
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.0692497938994225e-06,
+      "loss": 5.5251,
+      "step": 5019
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.0486397361912625e-06,
+      "loss": 5.463,
+      "step": 5020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 7.0280296784831e-06,
+      "loss": 5.5152,
+      "step": 5021
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 7.007419620774938e-06,
+      "loss": 5.6887,
+      "step": 5022
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.9868095630667765e-06,
+      "loss": 5.6961,
+      "step": 5023
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.966199505358616e-06,
+      "loss": 5.2732,
+      "step": 5024
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.945589447650454e-06,
+      "loss": 5.328,
+      "step": 5025
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.924979389942292e-06,
+      "loss": 5.3101,
+      "step": 5026
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.90436933223413e-06,
+      "loss": 5.0332,
+      "step": 5027
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.88375927452597e-06,
+      "loss": 5.2356,
+      "step": 5028
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.863149216817807e-06,
+      "loss": 5.2781,
+      "step": 5029
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.8425391591096454e-06,
+      "loss": 5.0614,
+      "step": 5030
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.821929101401484e-06,
+      "loss": 5.044,
+      "step": 5031
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.801319043693323e-06,
+      "loss": 4.7606,
+      "step": 5032
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.780708985985161e-06,
+      "loss": 4.9385,
+      "step": 5033
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.7600989282769995e-06,
+      "loss": 5.0088,
+      "step": 5034
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.739488870568839e-06,
+      "loss": 4.9419,
+      "step": 5035
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.718878812860677e-06,
+      "loss": 5.0568,
+      "step": 5036
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.698268755152514e-06,
+      "loss": 4.9848,
+      "step": 5037
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.677658697444353e-06,
+      "loss": 4.9024,
+      "step": 5038
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.657048639736193e-06,
+      "loss": 5.2979,
+      "step": 5039
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.63643858202803e-06,
+      "loss": 5.1915,
+      "step": 5040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.615828524319868e-06,
+      "loss": 4.9671,
+      "step": 5041
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.595218466611707e-06,
+      "loss": 5.0166,
+      "step": 5042
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.574608408903546e-06,
+      "loss": 4.8823,
+      "step": 5043
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.553998351195384e-06,
+      "loss": 5.4211,
+      "step": 5044
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.5333882934872224e-06,
+      "loss": 4.6098,
+      "step": 5045
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.51277823577906e-06,
+      "loss": 5.0251,
+      "step": 5046
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.4921681780709e-06,
+      "loss": 5.0019,
+      "step": 5047
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.471558120362737e-06,
+      "loss": 4.9634,
+      "step": 5048
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.450948062654576e-06,
+      "loss": 4.7749,
+      "step": 5049
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.430338004946414e-06,
+      "loss": 4.8876,
+      "step": 5050
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.409727947238253e-06,
+      "loss": 4.7889,
+      "step": 5051
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.389117889530091e-06,
+      "loss": 4.6369,
+      "step": 5052
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.36850783182193e-06,
+      "loss": 4.9409,
+      "step": 5053
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.347897774113767e-06,
+      "loss": 4.6601,
+      "step": 5054
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.327287716405607e-06,
+      "loss": 5.024,
+      "step": 5055
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.3066776586974446e-06,
+      "loss": 4.8488,
+      "step": 5056
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.286067600989283e-06,
+      "loss": 4.787,
+      "step": 5057
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.265457543281121e-06,
+      "loss": 4.6067,
+      "step": 5058
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.2448474855729595e-06,
+      "loss": 4.86,
+      "step": 5059
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.224237427864799e-06,
+      "loss": 4.6003,
+      "step": 5060
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.203627370156637e-06,
+      "loss": 4.5757,
+      "step": 5061
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.183017312448475e-06,
+      "loss": 4.4151,
+      "step": 5062
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.1624072547403135e-06,
+      "loss": 4.0246,
+      "step": 5063
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.141797197032153e-06,
+      "loss": 4.3471,
+      "step": 5064
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.12118713932399e-06,
+      "loss": 3.9279,
+      "step": 5065
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.100577081615829e-06,
+      "loss": 3.3636,
+      "step": 5066
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.0799670239076675e-06,
+      "loss": 3.729,
+      "step": 5067
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.059356966199506e-06,
+      "loss": 3.9029,
+      "step": 5068
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.038746908491344e-06,
+      "loss": 5.5911,
+      "step": 5069
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.018136850783182e-06,
+      "loss": 5.9112,
+      "step": 5070
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.997526793075021e-06,
+      "loss": 5.7745,
+      "step": 5071
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.97691673536686e-06,
+      "loss": 5.5739,
+      "step": 5072
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.956306677658697e-06,
+      "loss": 5.736,
+      "step": 5073
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.9356966199505365e-06,
+      "loss": 5.7831,
+      "step": 5074
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.915086562242375e-06,
+      "loss": 5.6374,
+      "step": 5075
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.894476504534213e-06,
+      "loss": 5.4493,
+      "step": 5076
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.873866446826051e-06,
+      "loss": 5.7552,
+      "step": 5077
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.85325638911789e-06,
+      "loss": 5.4905,
+      "step": 5078
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.832646331409728e-06,
+      "loss": 5.3928,
+      "step": 5079
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.812036273701567e-06,
+      "loss": 5.3172,
+      "step": 5080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.7914262159934045e-06,
+      "loss": 5.0844,
+      "step": 5081
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.770816158285244e-06,
+      "loss": 4.7566,
+      "step": 5082
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.750206100577082e-06,
+      "loss": 5.2953,
+      "step": 5083
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.72959604286892e-06,
+      "loss": 4.8955,
+      "step": 5084
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.7089859851607586e-06,
+      "loss": 5.1817,
+      "step": 5085
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.688375927452598e-06,
+      "loss": 4.8461,
+      "step": 5086
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.667765869744435e-06,
+      "loss": 5.0782,
+      "step": 5087
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.647155812036274e-06,
+      "loss": 4.8095,
+      "step": 5088
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.626545754328113e-06,
+      "loss": 4.6509,
+      "step": 5089
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.605935696619951e-06,
+      "loss": 4.9819,
+      "step": 5090
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.585325638911789e-06,
+      "loss": 5.0318,
+      "step": 5091
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.5647155812036275e-06,
+      "loss": 5.074,
+      "step": 5092
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.544105523495466e-06,
+      "loss": 4.9083,
+      "step": 5093
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.523495465787305e-06,
+      "loss": 4.975,
+      "step": 5094
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.502885408079142e-06,
+      "loss": 4.7963,
+      "step": 5095
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.4822753503709815e-06,
+      "loss": 4.9607,
+      "step": 5096
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.46166529266282e-06,
+      "loss": 4.5474,
+      "step": 5097
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.441055234954658e-06,
+      "loss": 4.8999,
+      "step": 5098
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.4204451772464964e-06,
+      "loss": 4.9331,
+      "step": 5099
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.399835119538335e-06,
+      "loss": 4.5656,
+      "step": 5100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.379225061830173e-06,
+      "loss": 4.4848,
+      "step": 5101
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.358615004122012e-06,
+      "loss": 4.706,
+      "step": 5102
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.33800494641385e-06,
+      "loss": 4.8986,
+      "step": 5103
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.317394888705689e-06,
+      "loss": 4.7054,
+      "step": 5104
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.296784830997527e-06,
+      "loss": 5.2068,
+      "step": 5105
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.276174773289365e-06,
+      "loss": 4.733,
+      "step": 5106
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.255564715581204e-06,
+      "loss": 4.512,
+      "step": 5107
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.234954657873043e-06,
+      "loss": 4.4166,
+      "step": 5108
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.21434460016488e-06,
+      "loss": 4.4602,
+      "step": 5109
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.193734542456719e-06,
+      "loss": 4.5878,
+      "step": 5110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.173124484748558e-06,
+      "loss": 4.1263,
+      "step": 5111
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.152514427040396e-06,
+      "loss": 4.0768,
+      "step": 5112
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.131904369332234e-06,
+      "loss": 3.9025,
+      "step": 5113
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.111294311624073e-06,
+      "loss": 3.6999,
+      "step": 5114
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.090684253915911e-06,
+      "loss": 3.8803,
+      "step": 5115
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.07007419620775e-06,
+      "loss": 3.4543,
+      "step": 5116
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.0494641384995875e-06,
+      "loss": 3.8122,
+      "step": 5117
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.028854080791427e-06,
+      "loss": 2.4936,
+      "step": 5118
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.008244023083265e-06,
+      "loss": 5.6857,
+      "step": 5119
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.987633965375103e-06,
+      "loss": 5.5729,
+      "step": 5120
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.9670239076669415e-06,
+      "loss": 5.3622,
+      "step": 5121
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.94641384995878e-06,
+      "loss": 5.6823,
+      "step": 5122
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.925803792250618e-06,
+      "loss": 5.2408,
+      "step": 5123
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.905193734542457e-06,
+      "loss": 5.38,
+      "step": 5124
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.884583676834295e-06,
+      "loss": 5.6711,
+      "step": 5125
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.863973619126134e-06,
+      "loss": 5.5224,
+      "step": 5126
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.843363561417972e-06,
+      "loss": 5.2924,
+      "step": 5127
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.8227535037098104e-06,
+      "loss": 5.4853,
+      "step": 5128
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.802143446001649e-06,
+      "loss": 5.2362,
+      "step": 5129
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.781533388293488e-06,
+      "loss": 4.9919,
+      "step": 5130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.760923330585325e-06,
+      "loss": 5.4204,
+      "step": 5131
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.7403132728771645e-06,
+      "loss": 4.9496,
+      "step": 5132
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.719703215169003e-06,
+      "loss": 5.1621,
+      "step": 5133
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.699093157460841e-06,
+      "loss": 4.8392,
+      "step": 5134
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.678483099752679e-06,
+      "loss": 5.1011,
+      "step": 5135
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.657873042044518e-06,
+      "loss": 5.0214,
+      "step": 5136
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.637262984336356e-06,
+      "loss": 5.0409,
+      "step": 5137
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.616652926628195e-06,
+      "loss": 4.7765,
+      "step": 5138
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5960428689200326e-06,
+      "loss": 4.8654,
+      "step": 5139
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.575432811211872e-06,
+      "loss": 5.0282,
+      "step": 5140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.55482275350371e-06,
+      "loss": 5.0863,
+      "step": 5141
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.534212695795548e-06,
+      "loss": 5.0416,
+      "step": 5142
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.5136026380873874e-06,
+      "loss": 4.7986,
+      "step": 5143
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.492992580379225e-06,
+      "loss": 4.7688,
+      "step": 5144
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.472382522671064e-06,
+      "loss": 4.9015,
+      "step": 5145
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.451772464962902e-06,
+      "loss": 4.7375,
+      "step": 5146
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.431162407254741e-06,
+      "loss": 5.1159,
+      "step": 5147
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.410552349546579e-06,
+      "loss": 4.7153,
+      "step": 5148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.389942291838418e-06,
+      "loss": 4.8147,
+      "step": 5149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.3693322341302555e-06,
+      "loss": 4.8218,
+      "step": 5150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.348722176422095e-06,
+      "loss": 4.6215,
+      "step": 5151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.328112118713933e-06,
+      "loss": 4.7447,
+      "step": 5152
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.307502061005771e-06,
+      "loss": 5.1067,
+      "step": 5153
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.2868920032976096e-06,
+      "loss": 4.74,
+      "step": 5154
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.266281945589448e-06,
+      "loss": 4.7101,
+      "step": 5155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.245671887881286e-06,
+      "loss": 4.5644,
+      "step": 5156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.225061830173125e-06,
+      "loss": 4.5044,
+      "step": 5157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.204451772464963e-06,
+      "loss": 4.9838,
+      "step": 5158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.183841714756802e-06,
+      "loss": 4.4263,
+      "step": 5159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.16323165704864e-06,
+      "loss": 4.4137,
+      "step": 5160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.1426215993404785e-06,
+      "loss": 4.5501,
+      "step": 5161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.122011541632317e-06,
+      "loss": 4.3387,
+      "step": 5162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.101401483924155e-06,
+      "loss": 4.1609,
+      "step": 5163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.080791426215993e-06,
+      "loss": 4.2749,
+      "step": 5164
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.0601813685078325e-06,
+      "loss": 3.6598,
+      "step": 5165
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.03957131079967e-06,
+      "loss": 4.0748,
+      "step": 5166
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.018961253091509e-06,
+      "loss": 3.1914,
+      "step": 5167
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.9983511953833474e-06,
+      "loss": 2.6383,
+      "step": 5168
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.977741137675186e-06,
+      "loss": 5.6315,
+      "step": 5169
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.957131079967024e-06,
+      "loss": 5.3108,
+      "step": 5170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.936521022258863e-06,
+      "loss": 5.2475,
+      "step": 5171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.915910964550701e-06,
+      "loss": 5.3597,
+      "step": 5172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.89530090684254e-06,
+      "loss": 5.0077,
+      "step": 5173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.874690849134378e-06,
+      "loss": 5.4304,
+      "step": 5174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.854080791426216e-06,
+      "loss": 5.2624,
+      "step": 5175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.833470733718055e-06,
+      "loss": 5.0721,
+      "step": 5176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.8128606760098934e-06,
+      "loss": 5.0394,
+      "step": 5177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.7922506183017312e-06,
+      "loss": 5.0931,
+      "step": 5178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.77164056059357e-06,
+      "loss": 4.9715,
+      "step": 5179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.7510305028854083e-06,
+      "loss": 5.3158,
+      "step": 5180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.730420445177247e-06,
+      "loss": 5.1167,
+      "step": 5181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.709810387469085e-06,
+      "loss": 4.8948,
+      "step": 5182
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6892003297609236e-06,
+      "loss": 4.8175,
+      "step": 5183
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.668590272052762e-06,
+      "loss": 5.2443,
+      "step": 5184
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6479802143446006e-06,
+      "loss": 5.1076,
+      "step": 5185
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6273701566364385e-06,
+      "loss": 4.9619,
+      "step": 5186
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.606760098928277e-06,
+      "loss": 5.2482,
+      "step": 5187
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5861500412201155e-06,
+      "loss": 4.9039,
+      "step": 5188
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5655399835119542e-06,
+      "loss": 4.6636,
+      "step": 5189
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.544929925803792e-06,
+      "loss": 4.8549,
+      "step": 5190
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5243198680956312e-06,
+      "loss": 4.7344,
+      "step": 5191
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.503709810387469e-06,
+      "loss": 4.7425,
+      "step": 5192
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.483099752679308e-06,
+      "loss": 4.6267,
+      "step": 5193
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.462489694971146e-06,
+      "loss": 4.9347,
+      "step": 5194
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.441879637262985e-06,
+      "loss": 4.9913,
+      "step": 5195
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.4212695795548227e-06,
+      "loss": 4.6617,
+      "step": 5196
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.4006595218466614e-06,
+      "loss": 4.8447,
+      "step": 5197
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3800494641384997e-06,
+      "loss": 4.8013,
+      "step": 5198
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3594394064303385e-06,
+      "loss": 5.3862,
+      "step": 5199
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3388293487221763e-06,
+      "loss": 4.8574,
+      "step": 5200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.318219291014015e-06,
+      "loss": 4.7946,
+      "step": 5201
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2976092333058533e-06,
+      "loss": 4.7291,
+      "step": 5202
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.276999175597692e-06,
+      "loss": 4.7675,
+      "step": 5203
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.25638911788953e-06,
+      "loss": 4.991,
+      "step": 5204
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2357790601813687e-06,
+      "loss": 4.4183,
+      "step": 5205
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.215169002473207e-06,
+      "loss": 4.5025,
+      "step": 5206
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1945589447650457e-06,
+      "loss": 4.6671,
+      "step": 5207
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1739488870568836e-06,
+      "loss": 5.0126,
+      "step": 5208
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1533388293487223e-06,
+      "loss": 4.6605,
+      "step": 5209
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1327287716405606e-06,
+      "loss": 4.4703,
+      "step": 5210
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1121187139323993e-06,
+      "loss": 4.4237,
+      "step": 5211
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.0915086562242376e-06,
+      "loss": 4.5546,
+      "step": 5212
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.0708985985160763e-06,
+      "loss": 3.8409,
+      "step": 5213
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.0502885408079146e-06,
+      "loss": 4.0227,
+      "step": 5214
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.029678483099753e-06,
+      "loss": 4.22,
+      "step": 5215
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.009068425391591e-06,
+      "loss": 3.6168,
+      "step": 5216
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.98845836768343e-06,
+      "loss": 3.33,
+      "step": 5217
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.9678483099752682e-06,
+      "loss": 3.0703,
+      "step": 5218
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9472382522671065e-06,
+      "loss": 5.4585,
+      "step": 5219
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.926628194558945e-06,
+      "loss": 5.4281,
+      "step": 5220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9060181368507835e-06,
+      "loss": 5.4324,
+      "step": 5221
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.885408079142622e-06,
+      "loss": 5.2781,
+      "step": 5222
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.86479802143446e-06,
+      "loss": 5.3337,
+      "step": 5223
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.844187963726299e-06,
+      "loss": 5.3403,
+      "step": 5224
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.823577906018137e-06,
+      "loss": 5.4458,
+      "step": 5225
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8029678483099755e-06,
+      "loss": 5.2583,
+      "step": 5226
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7823577906018138e-06,
+      "loss": 5.2006,
+      "step": 5227
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7617477328936525e-06,
+      "loss": 5.088,
+      "step": 5228
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7411376751854908e-06,
+      "loss": 4.9515,
+      "step": 5229
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.720527617477329e-06,
+      "loss": 5.2933,
+      "step": 5230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6999175597691674e-06,
+      "loss": 4.9053,
+      "step": 5231
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.679307502061006e-06,
+      "loss": 5.1863,
+      "step": 5232
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6586974443528444e-06,
+      "loss": 4.8394,
+      "step": 5233
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6380873866446827e-06,
+      "loss": 4.9371,
+      "step": 5234
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6174773289365214e-06,
+      "loss": 5.3673,
+      "step": 5235
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5968672712283597e-06,
+      "loss": 5.008,
+      "step": 5236
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.576257213520198e-06,
+      "loss": 4.7417,
+      "step": 5237
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5556471558120363e-06,
+      "loss": 5.1261,
+      "step": 5238
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.535037098103875e-06,
+      "loss": 5.1043,
+      "step": 5239
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5144270403957133e-06,
+      "loss": 4.8423,
+      "step": 5240
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4938169826875516e-06,
+      "loss": 4.8992,
+      "step": 5241
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.47320692497939e-06,
+      "loss": 4.7645,
+      "step": 5242
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4525968672712286e-06,
+      "loss": 4.9741,
+      "step": 5243
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.431986809563067e-06,
+      "loss": 4.8511,
+      "step": 5244
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4113767518549052e-06,
+      "loss": 4.9497,
+      "step": 5245
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.390766694146744e-06,
+      "loss": 4.5519,
+      "step": 5246
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3701566364385822e-06,
+      "loss": 5.0908,
+      "step": 5247
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3495465787304205e-06,
+      "loss": 4.6203,
+      "step": 5248
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.328936521022259e-06,
+      "loss": 4.7718,
+      "step": 5249
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3083264633140976e-06,
+      "loss": 4.851,
+      "step": 5250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.287716405605936e-06,
+      "loss": 4.6061,
+      "step": 5251
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.267106347897774e-06,
+      "loss": 4.7297,
+      "step": 5252
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2464962901896124e-06,
+      "loss": 4.6439,
+      "step": 5253
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.225886232481451e-06,
+      "loss": 4.8564,
+      "step": 5254
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.2052761747732895e-06,
+      "loss": 4.5379,
+      "step": 5255
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1846661170651278e-06,
+      "loss": 4.6354,
+      "step": 5256
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1640560593569665e-06,
+      "loss": 4.819,
+      "step": 5257
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1434460016488048e-06,
+      "loss": 4.3441,
+      "step": 5258
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.122835943940643e-06,
+      "loss": 4.3509,
+      "step": 5259
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1022258862324814e-06,
+      "loss": 4.4016,
+      "step": 5260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.08161582852432e-06,
+      "loss": 4.2381,
+      "step": 5261
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0610057708161584e-06,
+      "loss": 4.1698,
+      "step": 5262
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0403957131079967e-06,
+      "loss": 4.3645,
+      "step": 5263
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.019785655399835e-06,
+      "loss": 3.8847,
+      "step": 5264
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9991755976916737e-06,
+      "loss": 3.3239,
+      "step": 5265
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.978565539983512e-06,
+      "loss": 3.5787,
+      "step": 5266
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9579554822753503e-06,
+      "loss": 3.1489,
+      "step": 5267
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.937345424567189e-06,
+      "loss": 3.192,
+      "step": 5268
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9167353668590273e-06,
+      "loss": 5.6025,
+      "step": 5269
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8961253091508656e-06,
+      "loss": 5.0864,
+      "step": 5270
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8755152514427041e-06,
+      "loss": 5.3036,
+      "step": 5271
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8549051937345424e-06,
+      "loss": 5.5223,
+      "step": 5272
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.834295136026381e-06,
+      "loss": 5.2175,
+      "step": 5273
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8136850783182192e-06,
+      "loss": 4.9827,
+      "step": 5274
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7930750206100577e-06,
+      "loss": 5.0062,
+      "step": 5275
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.772464962901896e-06,
+      "loss": 5.0195,
+      "step": 5276
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7518549051937346e-06,
+      "loss": 5.2456,
+      "step": 5277
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.731244847485573e-06,
+      "loss": 5.1905,
+      "step": 5278
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7106347897774114e-06,
+      "loss": 5.0046,
+      "step": 5279
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6900247320692499e-06,
+      "loss": 5.1373,
+      "step": 5280
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6694146743610882e-06,
+      "loss": 4.76,
+      "step": 5281
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6488046166529267e-06,
+      "loss": 5.1101,
+      "step": 5282
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.628194558944765e-06,
+      "loss": 4.9121,
+      "step": 5283
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6075845012366035e-06,
+      "loss": 5.1893,
+      "step": 5284
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5869744435284418e-06,
+      "loss": 4.9847,
+      "step": 5285
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5663643858202803e-06,
+      "loss": 4.9223,
+      "step": 5286
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5457543281121188e-06,
+      "loss": 5.0814,
+      "step": 5287
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5251442704039573e-06,
+      "loss": 4.8673,
+      "step": 5288
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5045342126957956e-06,
+      "loss": 4.7769,
+      "step": 5289
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4839241549876341e-06,
+      "loss": 4.7936,
+      "step": 5290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4633140972794724e-06,
+      "loss": 4.8707,
+      "step": 5291
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.442704039571311e-06,
+      "loss": 4.6411,
+      "step": 5292
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4220939818631494e-06,
+      "loss": 4.8262,
+      "step": 5293
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4014839241549877e-06,
+      "loss": 4.7487,
+      "step": 5294
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3808738664468262e-06,
+      "loss": 4.869,
+      "step": 5295
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3602638087386645e-06,
+      "loss": 5.1843,
+      "step": 5296
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.339653751030503e-06,
+      "loss": 4.8043,
+      "step": 5297
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.3190436933223413e-06,
+      "loss": 4.7771,
+      "step": 5298
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2984336356141798e-06,
+      "loss": 4.6642,
+      "step": 5299
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2778235779060181e-06,
+      "loss": 4.9434,
+      "step": 5300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2572135201978567e-06,
+      "loss": 5.1796,
+      "step": 5301
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.236603462489695e-06,
+      "loss": 4.8209,
+      "step": 5302
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.2159934047815335e-06,
+      "loss": 5.1376,
+      "step": 5303
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.195383347073372e-06,
+      "loss": 4.6316,
+      "step": 5304
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1747732893652103e-06,
+      "loss": 4.2954,
+      "step": 5305
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1541632316570488e-06,
+      "loss": 4.8585,
+      "step": 5306
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.133553173948887e-06,
+      "loss": 4.6769,
+      "step": 5307
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1129431162407256e-06,
+      "loss": 4.6862,
+      "step": 5308
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0923330585325639e-06,
+      "loss": 4.7266,
+      "step": 5309
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0717230008244024e-06,
+      "loss": 4.699,
+      "step": 5310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0511129431162407e-06,
+      "loss": 4.1216,
+      "step": 5311
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0305028854080792e-06,
+      "loss": 4.9831,
+      "step": 5312
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0098928276999175e-06,
+      "loss": 4.4008,
+      "step": 5313
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.89282769991756e-07,
+      "loss": 3.9894,
+      "step": 5314
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.686727122835945e-07,
+      "loss": 3.9423,
+      "step": 5315
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.480626545754328e-07,
+      "loss": 3.214,
+      "step": 5316
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.274525968672712e-07,
+      "loss": 3.733,
+      "step": 5317
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.068425391591096e-07,
+      "loss": 3.2678,
+      "step": 5318
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.86232481450948e-07,
+      "loss": 5.383,
+      "step": 5319
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.656224237427865e-07,
+      "loss": 5.4783,
+      "step": 5320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.450123660346249e-07,
+      "loss": 5.1129,
+      "step": 5321
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.244023083264633e-07,
+      "loss": 5.1164,
+      "step": 5322
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.037922506183017e-07,
+      "loss": 5.0721,
+      "step": 5323
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.831821929101401e-07,
+      "loss": 5.1312,
+      "step": 5324
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.625721352019787e-07,
+      "loss": 5.1583,
+      "step": 5325
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.419620774938171e-07,
+      "loss": 4.9809,
+      "step": 5326
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.213520197856555e-07,
+      "loss": 5.3964,
+      "step": 5327
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.007419620774939e-07,
+      "loss": 5.2341,
+      "step": 5328
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.801319043693323e-07,
+      "loss": 5.3491,
+      "step": 5329
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.595218466611707e-07,
+      "loss": 4.9862,
+      "step": 5330
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.389117889530091e-07,
+      "loss": 5.1888,
+      "step": 5331
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.183017312448475e-07,
+      "loss": 4.9862,
+      "step": 5332
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.97691673536686e-07,
+      "loss": 5.0949,
+      "step": 5333
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.770816158285244e-07,
+      "loss": 5.1073,
+      "step": 5334
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.564715581203628e-07,
+      "loss": 5.2934,
+      "step": 5335
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.358615004122012e-07,
+      "loss": 4.9402,
+      "step": 5336
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.152514427040396e-07,
+      "loss": 5.1293,
+      "step": 5337
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.94641384995878e-07,
+      "loss": 4.7629,
+      "step": 5338
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.740313272877164e-07,
+      "loss": 4.8445,
+      "step": 5339
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.534212695795548e-07,
+      "loss": 4.9831,
+      "step": 5340
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3281121187139327e-07,
+      "loss": 5.0559,
+      "step": 5341
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1220115416323167e-07,
+      "loss": 4.7817,
+      "step": 5342
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.9159109645507007e-07,
+      "loss": 4.418,
+      "step": 5343
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.7098103874690853e-07,
+      "loss": 4.5948,
+      "step": 5344
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5037098103874693e-07,
+      "loss": 4.681,
+      "step": 5345
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2976092333058533e-07,
+      "loss": 4.5751,
+      "step": 5346
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0915086562242374e-07,
+      "loss": 4.5757,
+      "step": 5347
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.885408079142622e-07,
+      "loss": 4.5069,
+      "step": 5348
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.679307502061006e-07,
+      "loss": 4.2085,
+      "step": 5349
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.47320692497939e-07,
+      "loss": 4.1064,
+      "step": 5350
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.267106347897774e-07,
+      "loss": 3.474,
+      "step": 5351
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0610057708161583e-07,
+      "loss": 2.5302,
+      "step": 5352
+    },
+    {
+      "epoch": 3.0,
+      "step": 5352,
+      "total_flos": 0.0,
+      "train_loss": 5.8440248690913075,
+      "train_runtime": 31164.8267,
+      "train_samples_per_second": 2.747,
+      "train_steps_per_second": 0.172
+    }
+  ],
+  "max_steps": 5352,
+  "num_train_epochs": 3,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}