diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,43884 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 7309,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 10.8205,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 10.7294,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 10.2463,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 9.3174,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 8.7481,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 9.0121,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1818181818181814e-05,
+      "loss": 8.5476,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 7.9042,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.090909090909091e-05,
+      "loss": 7.2121,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 6.6885,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 6.2664,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 5.9377,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.909090909090909e-05,
+      "loss": 5.9705,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.363636363636363e-05,
+      "loss": 5.7789,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.818181818181818e-05,
+      "loss": 5.5084,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 5.4386,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.727272727272727e-05,
+      "loss": 5.6003,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.181818181818182e-05,
+      "loss": 5.2592,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.636363636363636e-05,
+      "loss": 5.3642,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 5.0485,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.545454545454546e-05,
+      "loss": 5.1937,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 5.0947,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010454545454545454,
+      "loss": 4.8366,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010909090909090909,
+      "loss": 4.7965,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011363636363636364,
+      "loss": 4.8484,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011818181818181818,
+      "loss": 4.9717,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012272727272727272,
+      "loss": 4.837,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012727272727272725,
+      "loss": 4.6255,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001318181818181818,
+      "loss": 4.6368,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013636363636363637,
+      "loss": 4.5921,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001409090909090909,
+      "loss": 4.6467,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014545454545454546,
+      "loss": 4.5508,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015,
+      "loss": 4.5712,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015454545454545454,
+      "loss": 4.2924,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001590909090909091,
+      "loss": 4.4814,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016363636363636363,
+      "loss": 4.2961,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001681818181818182,
+      "loss": 4.28,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017272727272727272,
+      "loss": 4.3724,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017727272727272728,
+      "loss": 4.3964,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 4.1129,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018636363636363636,
+      "loss": 4.0768,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019090909090909092,
+      "loss": 4.1871,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019545454545454548,
+      "loss": 4.161,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 4.3697,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020454545454545457,
+      "loss": 4.0646,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020909090909090907,
+      "loss": 3.9815,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021363636363636363,
+      "loss": 4.0339,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021818181818181818,
+      "loss": 4.032,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022272727272727272,
+      "loss": 3.993,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022727272727272727,
+      "loss": 4.1206,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023181818181818183,
+      "loss": 3.8883,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023636363636363636,
+      "loss": 3.7391,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024090909090909092,
+      "loss": 3.8651,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024545454545454545,
+      "loss": 3.7375,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025,
+      "loss": 3.5616,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002545454545454545,
+      "loss": 3.543,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002590909090909091,
+      "loss": 3.7165,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002636363636363636,
+      "loss": 3.4769,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002681818181818182,
+      "loss": 3.641,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027272727272727274,
+      "loss": 3.5868,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002772727272727273,
+      "loss": 3.47,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002818181818181818,
+      "loss": 3.5154,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028636363636363636,
+      "loss": 3.6115,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002909090909090909,
+      "loss": 3.6548,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029545454545454547,
+      "loss": 3.6292,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003,
+      "loss": 3.567,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003045454545454546,
+      "loss": 3.4438,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003090909090909091,
+      "loss": 3.5862,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031363636363636365,
+      "loss": 3.4379,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003181818181818182,
+      "loss": 3.5734,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032272727272727276,
+      "loss": 3.4958,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032727272727272726,
+      "loss": 3.5345,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003318181818181819,
+      "loss": 3.4996,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003363636363636364,
+      "loss": 3.6252,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003409090909090909,
+      "loss": 3.5475,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034545454545454544,
+      "loss": 3.4351,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035,
+      "loss": 3.6096,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035454545454545455,
+      "loss": 3.324,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035909090909090906,
+      "loss": 3.4085,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036363636363636367,
+      "loss": 3.3894,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036818181818181817,
+      "loss": 3.3863,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037272727272727273,
+      "loss": 3.4806,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003772727272727273,
+      "loss": 3.4038,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038181818181818184,
+      "loss": 3.3773,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038636363636363635,
+      "loss": 3.4236,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039090909090909096,
+      "loss": 3.2357,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039545454545454546,
+      "loss": 3.4037,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004,
+      "loss": 3.3351,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004045454545454546,
+      "loss": 3.34,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040909090909090913,
+      "loss": 3.4174,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041363636363636364,
+      "loss": 3.3644,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041818181818181814,
+      "loss": 3.4968,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042272727272727275,
+      "loss": 3.3806,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042727272727272726,
+      "loss": 3.3374,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004318181818181818,
+      "loss": 3.2532,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043636363636363637,
+      "loss": 3.4145,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044090909090909093,
+      "loss": 3.3584,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044545454545454543,
+      "loss": 3.352,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 3.2563,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045454545454545455,
+      "loss": 3.4545,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004590909090909091,
+      "loss": 3.3864,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046363636363636366,
+      "loss": 3.3335,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004681818181818182,
+      "loss": 3.1003,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004727272727272727,
+      "loss": 3.3679,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004772727272727273,
+      "loss": 3.3695,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00048181818181818184,
+      "loss": 3.3797,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004863636363636364,
+      "loss": 3.2521,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004909090909090909,
+      "loss": 3.4547,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004954545454545455,
+      "loss": 3.1907,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005,
+      "loss": 3.2594,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005045454545454546,
+      "loss": 3.1924,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000509090909090909,
+      "loss": 3.4055,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005136363636363637,
+      "loss": 3.2156,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005181818181818182,
+      "loss": 3.6265,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005227272727272727,
+      "loss": 3.2202,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005272727272727272,
+      "loss": 3.2062,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005318181818181819,
+      "loss": 3.2424,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005363636363636364,
+      "loss": 3.1729,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005409090909090909,
+      "loss": 3.2404,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005454545454545455,
+      "loss": 3.0505,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00055,
+      "loss": 3.2245,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005545454545454546,
+      "loss": 3.2894,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005590909090909091,
+      "loss": 3.0334,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005636363636363636,
+      "loss": 3.2942,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005681818181818183,
+      "loss": 3.1683,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005727272727272727,
+      "loss": 3.128,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005772727272727273,
+      "loss": 3.296,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005818181818181818,
+      "loss": 3.0149,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005863636363636363,
+      "loss": 3.127,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005909090909090909,
+      "loss": 2.976,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005954545454545455,
+      "loss": 3.1792,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006,
+      "loss": 3.1626,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006045454545454545,
+      "loss": 2.9979,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006090909090909092,
+      "loss": 3.1639,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006136363636363636,
+      "loss": 3.1585,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006181818181818182,
+      "loss": 3.1086,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006227272727272727,
+      "loss": 3.1021,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006272727272727273,
+      "loss": 3.0745,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006318181818181818,
+      "loss": 3.04,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006363636363636364,
+      "loss": 2.9901,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006409090909090909,
+      "loss": 2.9622,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006454545454545455,
+      "loss": 3.1155,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006500000000000001,
+      "loss": 3.2157,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006545454545454545,
+      "loss": 3.2052,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006590909090909091,
+      "loss": 3.0264,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006636363636363638,
+      "loss": 3.0874,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006681818181818182,
+      "loss": 2.9431,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006727272727272728,
+      "loss": 2.8745,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006772727272727273,
+      "loss": 2.9539,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006818181818181818,
+      "loss": 2.955,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006863636363636364,
+      "loss": 3.038,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006909090909090909,
+      "loss": 3.027,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006954545454545454,
+      "loss": 2.8695,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007,
+      "loss": 3.1004,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007045454545454546,
+      "loss": 2.8385,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007090909090909091,
+      "loss": 2.967,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007136363636363637,
+      "loss": 2.9112,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007181818181818181,
+      "loss": 2.9583,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007227272727272728,
+      "loss": 2.8338,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007272727272727273,
+      "loss": 3.0555,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007318181818181818,
+      "loss": 2.9622,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007363636363636363,
+      "loss": 2.9459,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000740909090909091,
+      "loss": 2.962,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007454545454545455,
+      "loss": 2.9549,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00075,
+      "loss": 2.9166,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007545454545454546,
+      "loss": 2.8638,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000759090909090909,
+      "loss": 3.0436,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007636363636363637,
+      "loss": 2.8727,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007681818181818182,
+      "loss": 3.0167,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007727272727272727,
+      "loss": 2.8207,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007772727272727273,
+      "loss": 2.8446,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007818181818181819,
+      "loss": 3.0391,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007863636363636364,
+      "loss": 2.9902,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007909090909090909,
+      "loss": 2.8875,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007954545454545455,
+      "loss": 2.8679,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008,
+      "loss": 2.9319,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008045454545454546,
+      "loss": 2.9933,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008090909090909092,
+      "loss": 3.0451,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008136363636363636,
+      "loss": 2.8871,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008181818181818183,
+      "loss": 2.7976,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008227272727272727,
+      "loss": 2.8725,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008272727272727273,
+      "loss": 3.0641,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008318181818181818,
+      "loss": 2.8082,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008363636363636363,
+      "loss": 3.0134,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000840909090909091,
+      "loss": 2.7583,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008454545454545455,
+      "loss": 2.7133,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00085,
+      "loss": 3.036,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008545454545454545,
+      "loss": 2.8152,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008590909090909092,
+      "loss": 2.8079,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008636363636363636,
+      "loss": 3.0115,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008681818181818182,
+      "loss": 2.8557,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008727272727272727,
+      "loss": 2.9172,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008772727272727273,
+      "loss": 2.8949,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008818181818181819,
+      "loss": 2.7417,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008863636363636364,
+      "loss": 2.8346,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008909090909090909,
+      "loss": 2.7531,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008954545454545455,
+      "loss": 2.8142,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009000000000000001,
+      "loss": 2.8586,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009045454545454545,
+      "loss": 2.8641,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009090909090909091,
+      "loss": 2.7462,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009136363636363638,
+      "loss": 2.7543,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009181818181818182,
+      "loss": 2.7598,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009227272727272728,
+      "loss": 2.6929,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009272727272727273,
+      "loss": 2.7076,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009318181818181818,
+      "loss": 2.6266,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009363636363636364,
+      "loss": 2.6822,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009409090909090909,
+      "loss": 2.8912,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009454545454545454,
+      "loss": 2.7226,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00095,
+      "loss": 2.917,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009545454545454546,
+      "loss": 2.7955,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009590909090909091,
+      "loss": 2.7627,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009636363636363637,
+      "loss": 2.5904,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009681818181818181,
+      "loss": 2.7939,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009727272727272728,
+      "loss": 2.9654,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009772727272727272,
+      "loss": 2.9098,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009818181818181818,
+      "loss": 2.7547,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009863636363636363,
+      "loss": 2.9632,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000990909090909091,
+      "loss": 2.9979,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009954545454545455,
+      "loss": 3.0575,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 3.3979,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999999509013224,
+      "loss": 3.1153,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999998036052998,
+      "loss": 2.9291,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999995581119606,
+      "loss": 3.1291,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999999214421353,
+      "loss": 3.2198,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999998772533545,
+      "loss": 3.3452,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999982324486228,
+      "loss": 4.9216,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999997594166693,
+      "loss": 5.1045,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999968576878803,
+      "loss": 3.8569,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00099999602301233,
+      "loss": 3.9035,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999950901402061,
+      "loss": 3.7915,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999940590716913,
+      "loss": 3.995,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999992929806988,
+      "loss": 3.9122,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999917023463185,
+      "loss": 3.9515,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999903766899237,
+      "loss": 3.7743,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999889528380637,
+      "loss": 3.6741,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999874307910185,
+      "loss": 3.5462,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999858105490867,
+      "loss": 3.4325,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999840921125866,
+      "loss": 3.2928,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999822754818558,
+      "loss": 3.2609,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999803606572512,
+      "loss": 3.3829,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999783476391485,
+      "loss": 3.326,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999976236427943,
+      "loss": 3.3889,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00099997402702405,
+      "loss": 3.3585,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999717194279026,
+      "loss": 3.2799,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999693136399546,
+      "loss": 3.2862,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999668096606783,
+      "loss": 3.2278,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999642074905652,
+      "loss": 3.4077,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999615071301267,
+      "loss": 3.1562,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999587085798931,
+      "loss": 3.2337,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999558118404138,
+      "loss": 3.1048,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999528169122578,
+      "loss": 3.1428,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999497237960136,
+      "loss": 3.1748,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999465324922881,
+      "loss": 3.2321,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999432430017083,
+      "loss": 3.1789,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999398553249205,
+      "loss": 3.1482,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999363694625897,
+      "loss": 3.1449,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999327854154007,
+      "loss": 2.9967,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999291031840573,
+      "loss": 3.2882,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999253227692827,
+      "loss": 3.1363,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999921444171819,
+      "loss": 3.296,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999174673924287,
+      "loss": 3.0782,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999133924318924,
+      "loss": 3.2433,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999092192910103,
+      "loss": 3.1429,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999049479706018,
+      "loss": 3.094,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999005784715063,
+      "loss": 3.2449,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998961107945815,
+      "loss": 3.2728,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998915449407051,
+      "loss": 3.047,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998868809107738,
+      "loss": 3.1537,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998821187057034,
+      "loss": 3.4405,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998772583264292,
+      "loss": 3.1822,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999872299773906,
+      "loss": 3.1438,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998672430491073,
+      "loss": 3.0593,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998620881530266,
+      "loss": 3.0432,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999856835086676,
+      "loss": 3.1011,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999851483851087,
+      "loss": 3.1027,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999846034447311,
+      "loss": 3.133,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999840486876418,
+      "loss": 3.3376,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998348411394977,
+      "loss": 3.1525,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998290972376585,
+      "loss": 3.1156,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998232551720289,
+      "loss": 3.2853,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999817314943756,
+      "loss": 3.1627,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998112765540064,
+      "loss": 3.1936,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998051400039662,
+      "loss": 3.3981,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997989052948405,
+      "loss": 3.2889,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997925724278538,
+      "loss": 3.1066,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997861414042498,
+      "loss": 3.4403,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997796122252914,
+      "loss": 3.3754,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999772984892261,
+      "loss": 3.2848,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997662594064603,
+      "loss": 3.4274,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00099975943576921,
+      "loss": 2.9668,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997525139818502,
+      "loss": 3.2376,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997454940457403,
+      "loss": 3.1088,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997383759622592,
+      "loss": 3.2045,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997311597328045,
+      "loss": 3.5338,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997238453587937,
+      "loss": 3.2816,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997164328416634,
+      "loss": 3.2957,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999708922182869,
+      "loss": 3.5241,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997013133838859,
+      "loss": 3.2935,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996936064462082,
+      "loss": 3.1871,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996858013713496,
+      "loss": 3.2818,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999677898160843,
+      "loss": 3.1905,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996698968162405,
+      "loss": 3.2109,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996617973391133,
+      "loss": 3.118,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996535997310527,
+      "loss": 3.4992,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999645303993668,
+      "loss": 3.2182,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999636910128589,
+      "loss": 3.3357,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999628418137464,
+      "loss": 3.0179,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996198280219604,
+      "loss": 3.1481,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996111397837658,
+      "loss": 3.3103,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999602353424586,
+      "loss": 3.1584,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995934689461471,
+      "loss": 3.1612,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999584486350194,
+      "loss": 3.1135,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995754056384904,
+      "loss": 3.2291,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995662268128197,
+      "loss": 3.4126,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995569498749853,
+      "loss": 3.3521,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999547574826808,
+      "loss": 3.2728,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995381016701301,
+      "loss": 3.113,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995285304068113,
+      "loss": 3.5013,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995188610387319,
+      "loss": 3.2544,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995090935677907,
+      "loss": 3.321,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994992279959056,
+      "loss": 3.1994,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994892643250147,
+      "loss": 3.4079,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994792025570746,
+      "loss": 3.3971,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994690426940614,
+      "loss": 3.5278,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994587847379703,
+      "loss": 3.1195,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994484286908161,
+      "loss": 3.2929,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994379745546327,
+      "loss": 3.3441,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999427422331473,
+      "loss": 3.2995,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994167720234095,
+      "loss": 3.4812,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999406023632534,
+      "loss": 3.4045,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993951771609573,
+      "loss": 3.1097,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993842326108096,
+      "loss": 3.2454,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993731899842406,
+      "loss": 3.0564,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993620492834187,
+      "loss": 3.0995,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993508105105317,
+      "loss": 3.2412,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993394736677872,
+      "loss": 3.0226,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993280387574118,
+      "loss": 3.2141,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993165057816508,
+      "loss": 3.0272,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993048747427694,
+      "loss": 3.2075,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992931456430524,
+      "loss": 3.3617,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992813184848024,
+      "loss": 3.2186,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992693932703428,
+      "loss": 3.2713,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992573700020157,
+      "loss": 3.181,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992452486821822,
+      "loss": 3.2901,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992330293132226,
+      "loss": 3.0983,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992207118975371,
+      "loss": 3.1792,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992082964375448,
+      "loss": 3.2029,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999195782935684,
+      "loss": 3.1011,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999183171394412,
+      "loss": 3.1476,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999170461816206,
+      "loss": 3.3017,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991576542035616,
+      "loss": 3.1404,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991447485589947,
+      "loss": 3.0323,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991317448850396,
+      "loss": 3.16,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991186431842502,
+      "loss": 3.2319,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991054434591998,
+      "loss": 3.1105,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990921457124805,
+      "loss": 3.0105,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999078749946704,
+      "loss": 3.0864,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990652561645012,
+      "loss": 2.9718,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990516643685222,
+      "loss": 3.2003,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999037974561436,
+      "loss": 3.2211,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990241867459317,
+      "loss": 3.0078,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999010300924717,
+      "loss": 3.0533,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989963171005193,
+      "loss": 3.0636,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989822352760842,
+      "loss": 3.0096,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989680554541777,
+      "loss": 3.1583,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998953777637585,
+      "loss": 3.1681,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989394018291096,
+      "loss": 3.0294,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989249280315752,
+      "loss": 3.0218,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989103562478242,
+      "loss": 2.9156,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988956864807185,
+      "loss": 3.0824,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988809187331392,
+      "loss": 3.0298,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988660530079864,
+      "loss": 3.1511,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988510893081798,
+      "loss": 3.1341,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988360276366585,
+      "loss": 2.8812,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988208679963802,
+      "loss": 2.961,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998805610390322,
+      "loss": 3.0121,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987902548214806,
+      "loss": 3.1044,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987748012928717,
+      "loss": 2.9193,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987592498075307,
+      "loss": 3.093,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987436003685113,
+      "loss": 3.0393,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998727852978887,
+      "loss": 2.8089,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987120076417509,
+      "loss": 3.0743,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986960643602145,
+      "loss": 3.0953,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986800231374094,
+      "loss": 2.9464,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986638839764857,
+      "loss": 2.7562,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998647646880613,
+      "loss": 2.9317,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986313118529803,
+      "loss": 2.8908,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998614878896796,
+      "loss": 3.091,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998598348015287,
+      "loss": 2.9681,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985817192117,
+      "loss": 3.0202,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998564992489301,
+      "loss": 3.0657,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985481678513749,
+      "loss": 3.0384,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998531245301226,
+      "loss": 2.9546,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985142248421775,
+      "loss": 2.9828,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984971064775727,
+      "loss": 2.8052,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998479890210773,
+      "loss": 2.9294,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099846257604516,
+      "loss": 2.8164,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998445163984134,
+      "loss": 2.873,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984276540311143,
+      "loss": 2.8174,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984100461895404,
+      "loss": 2.8371,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009983923404628698,
+      "loss": 2.9555,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983745368545801,
+      "loss": 2.8616,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998356635368168,
+      "loss": 2.9616,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983386360071487,
+      "loss": 2.9255,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983205387750576,
+      "loss": 2.8244,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983023436754487,
+      "loss": 2.8018,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998284050711896,
+      "loss": 2.8535,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982656598879912,
+      "loss": 2.9172,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998247171207347,
+      "loss": 2.9148,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982285846735937,
+      "loss": 2.8728,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982099002903823,
+      "loss": 2.7338,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981911180613821,
+      "loss": 2.8387,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981722379902816,
+      "loss": 2.9344,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998153260080789,
+      "loss": 2.8903,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981341843366312,
+      "loss": 3.019,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981150107615548,
+      "loss": 2.9517,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980957393593256,
+      "loss": 2.9988,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980763701337278,
+      "loss": 2.9295,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980569030885657,
+      "loss": 2.9707,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998037338227663,
+      "loss": 2.9289,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980176755548613,
+      "loss": 3.0496,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979979150740229,
+      "loss": 2.8074,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979780567890282,
+      "loss": 2.8995,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979581007037775,
+      "loss": 2.9067,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099793804682219,
+      "loss": 2.8455,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979178951482043,
+      "loss": 2.8618,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978976456857783,
+      "loss": 2.9595,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978772984388883,
+      "loss": 2.7414,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978568534115306,
+      "loss": 3.0303,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978363106077205,
+      "loss": 2.9246,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978156700314927,
+      "loss": 2.916,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977949316869008,
+      "loss": 2.8719,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977740955780176,
+      "loss": 2.9844,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977531617089352,
+      "loss": 3.0645,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977321300837652,
+      "loss": 2.9834,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977110007066377,
+      "loss": 2.8488,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976897735817026,
+      "loss": 2.8592,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976684487131289,
+      "loss": 2.8629,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976470261051043,
+      "loss": 2.9175,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976255057618365,
+      "loss": 2.877,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976038876875516,
+      "loss": 2.9054,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975821718864958,
+      "loss": 2.8338,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975603583629335,
+      "loss": 2.9015,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997538447121149,
+      "loss": 3.0678,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975164381654455,
+      "loss": 2.9625,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974943315001455,
+      "loss": 2.8954,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974721271295904,
+      "loss": 2.9305,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974498250581412,
+      "loss": 2.9253,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974274252901778,
+      "loss": 2.8557,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974049278300996,
+      "loss": 2.7892,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973823326823248,
+      "loss": 2.7983,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973596398512913,
+      "loss": 3.0201,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973368493414552,
+      "loss": 2.8807,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997313961157293,
+      "loss": 2.9416,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972909753032996,
+      "loss": 2.74,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972678917839894,
+      "loss": 2.9776,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972447106038956,
+      "loss": 2.8082,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972214317675713,
+      "loss": 2.8254,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971980552795881,
+      "loss": 2.8342,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997174581144537,
+      "loss": 2.89,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971510093670281,
+      "loss": 2.9454,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971273399516911,
+      "loss": 3.0026,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971035729031742,
+      "loss": 2.8681,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970797082261454,
+      "loss": 2.705,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970557459252913,
+      "loss": 2.7532,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970316860053183,
+      "loss": 2.9341,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970075284709515,
+      "loss": 2.8529,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969832733269352,
+      "loss": 2.869,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969589205780331,
+      "loss": 2.8113,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996934470229028,
+      "loss": 2.8222,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969099222847217,
+      "loss": 2.875,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968852767499353,
+      "loss": 2.9714,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996860533629509,
+      "loss": 2.7762,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968356929283026,
+      "loss": 2.9108,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968107546511943,
+      "loss": 3.0201,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967857188030818,
+      "loss": 2.7191,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996760585388882,
+      "loss": 2.822,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967353544135314,
+      "loss": 3.0122,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967100258819848,
+      "loss": 2.8845,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966845997992167,
+      "loss": 2.8714,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966590761702207,
+      "loss": 2.9597,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966334550000095,
+      "loss": 2.9673,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996607736293615,
+      "loss": 2.9124,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996581920056088,
+      "loss": 2.8937,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996556006292499,
+      "loss": 2.7614,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965299950079373,
+      "loss": 2.8007,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996503886207511,
+      "loss": 2.83,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996477679896348,
+      "loss": 2.8919,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996451376079595,
+      "loss": 2.8531,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964249747624178,
+      "loss": 2.8707,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963984759500022,
+      "loss": 2.9464,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963718796475514,
+      "loss": 2.8051,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963451858602894,
+      "loss": 2.9094,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963183945934588,
+      "loss": 3.0153,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962915058523208,
+      "loss": 2.8264,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962645196421565,
+      "loss": 2.8448,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996237435968266,
+      "loss": 2.8701,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962102548359678,
+      "loss": 2.8328,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961829762506007,
+      "loss": 2.8768,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996155600217522,
+      "loss": 2.938,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961281267421081,
+      "loss": 2.8068,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961005558297544,
+      "loss": 2.7023,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960728874858762,
+      "loss": 2.7803,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996045121715907,
+      "loss": 2.9175,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960172585253,
+      "loss": 2.7208,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959892979195272,
+      "loss": 2.7717,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959612399040804,
+      "loss": 3.0029,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959330844844696,
+      "loss": 2.9425,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959048316662245,
+      "loss": 2.9049,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958764814548939,
+      "loss": 2.6122,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958480338560455,
+      "loss": 2.8502,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958194888752662,
+      "loss": 2.8749,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957908465181623,
+      "loss": 2.832,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957621067903588,
+      "loss": 2.7619,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957332696975002,
+      "loss": 2.7804,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957043352452498,
+      "loss": 2.8949,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956753034392903,
+      "loss": 3.0272,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956461742853234,
+      "loss": 2.8014,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956169477890698,
+      "loss": 2.8439,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955876239562695,
+      "loss": 2.9642,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955582027926814,
+      "loss": 2.8117,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995528684304084,
+      "loss": 2.6605,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954990684962741,
+      "loss": 2.7939,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954693553750687,
+      "loss": 2.862,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954395449463028,
+      "loss": 2.8626,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995409637215831,
+      "loss": 2.7584,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953796321895273,
+      "loss": 2.7942,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953495298732843,
+      "loss": 2.7996,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953193302730144,
+      "loss": 2.9177,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995289033394648,
+      "loss": 2.6818,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952586392441357,
+      "loss": 2.8035,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952281478274464,
+      "loss": 2.918,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951975591505687,
+      "loss": 2.7548,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00099516687321951,
+      "loss": 2.7599,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995136090040297,
+      "loss": 2.6644,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995105209618975,
+      "loss": 2.8659,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995074231961609,
+      "loss": 2.9687,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950431570742828,
+      "loss": 2.9515,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950119849630994,
+      "loss": 2.8449,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949807156341807,
+      "loss": 2.9559,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994949349093668,
+      "loss": 2.9629,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949178853477215,
+      "loss": 2.8395,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948863244025202,
+      "loss": 2.9484,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994854666264263,
+      "loss": 2.6951,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994822910939167,
+      "loss": 2.7635,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947910584334689,
+      "loss": 2.889,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947591087534245,
+      "loss": 2.8117,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947270619053083,
+      "loss": 2.7553,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946949178954144,
+      "loss": 3.0119,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946626767300556,
+      "loss": 2.9951,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946303384155636,
+      "loss": 2.8049,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099459790295829,
+      "loss": 3.1122,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945653703646046,
+      "loss": 2.7796,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994532740640897,
+      "loss": 2.9381,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994500013793575,
+      "loss": 2.7606,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994467189829066,
+      "loss": 2.8107,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994434268753817,
+      "loss": 2.8986,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944012505742929,
+      "loss": 2.8117,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994368135296979,
+      "loss": 2.8826,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943349229283782,
+      "loss": 2.8043,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943016134750136,
+      "loss": 2.7561,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942682069434272,
+      "loss": 2.8645,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942347033401794,
+      "loss": 2.9337,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942011026718505,
+      "loss": 2.6578,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941674049450393,
+      "loss": 2.9626,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994133610166364,
+      "loss": 2.8972,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940997183424617,
+      "loss": 2.9167,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940657294799885,
+      "loss": 2.8221,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940316435856195,
+      "loss": 2.7721,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939974606660493,
+      "loss": 2.8602,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939631807279912,
+      "loss": 2.827,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939288037781773,
+      "loss": 2.8409,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938943298233593,
+      "loss": 2.7989,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938597588703075,
+      "loss": 2.7707,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938250909258118,
+      "loss": 2.8433,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937903259966804,
+      "loss": 2.7588,
+      "step": 576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937554640897413,
+      "loss": 2.7779,
+      "step": 577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993720505211841,
+      "loss": 2.7164,
+      "step": 578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936854493698454,
+      "loss": 2.6121,
+      "step": 579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993650296570639,
+      "loss": 2.9118,
+      "step": 580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936150468211255,
+      "loss": 2.8647,
+      "step": 581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935797001282285,
+      "loss": 2.7732,
+      "step": 582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935442564988893,
+      "loss": 2.7803,
+      "step": 583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935087159400689,
+      "loss": 2.62,
+      "step": 584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934730784587475,
+      "loss": 2.9463,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934373440619237,
+      "loss": 2.8687,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934015127566161,
+      "loss": 2.7684,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933655845498614,
+      "loss": 2.7964,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993329559448716,
+      "loss": 2.6234,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932934374602548,
+      "loss": 2.8432,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932572185915717,
+      "loss": 2.6546,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932209028497806,
+      "loss": 2.8982,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993184490242013,
+      "loss": 2.7541,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931479807754207,
+      "loss": 2.6009,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993111374457174,
+      "loss": 2.791,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930746712944615,
+      "loss": 3.0007,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993037871294492,
+      "loss": 2.7818,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930009744644928,
+      "loss": 2.9116,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929639808117102,
+      "loss": 2.7656,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929268903434096,
+      "loss": 2.8878,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928897030668754,
+      "loss": 2.8864,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928524189894109,
+      "loss": 2.9916,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928150381183383,
+      "loss": 2.8284,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927775604609995,
+      "loss": 2.7708,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927399860247545,
+      "loss": 2.8586,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927023148169829,
+      "loss": 2.7665,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992664546845083,
+      "loss": 2.7202,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926266821164724,
+      "loss": 2.751,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925887206385874,
+      "loss": 2.7996,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925506624188837,
+      "loss": 2.7967,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925125074648352,
+      "loss": 2.6855,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924742557839356,
+      "loss": 2.9056,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924359073836975,
+      "loss": 2.8114,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923974622716523,
+      "loss": 2.7669,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923589204553502,
+      "loss": 2.796,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923202819423606,
+      "loss": 2.758,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922815467402722,
+      "loss": 2.7371,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922427148566922,
+      "loss": 2.5971,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992203786299247,
+      "loss": 2.9231,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992164761075582,
+      "loss": 3.0265,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921256391933614,
+      "loss": 2.74,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920864206602684,
+      "loss": 2.8527,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920471054840057,
+      "loss": 2.8499,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920076936722946,
+      "loss": 2.7957,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991968185232875,
+      "loss": 2.8211,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919285801735062,
+      "loss": 2.8245,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991888878501967,
+      "loss": 2.706,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918490802260538,
+      "loss": 2.8619,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918091853535834,
+      "loss": 2.9703,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917691938923904,
+      "loss": 2.8998,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917291058503296,
+      "loss": 3.0656,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916889212352733,
+      "loss": 2.7204,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991648640055114,
+      "loss": 2.7919,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916082623177627,
+      "loss": 2.8476,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915677880311493,
+      "loss": 2.8439,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915272172032227,
+      "loss": 2.8083,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991486549841951,
+      "loss": 2.7382,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914457859553207,
+      "loss": 2.8828,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914049255513378,
+      "loss": 2.7557,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991363968638027,
+      "loss": 2.7026,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913229152234324,
+      "loss": 2.752,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991281765315616,
+      "loss": 2.7188,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099124051892266,
+      "loss": 2.7634,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911991760526648,
+      "loss": 2.7231,
+      "step": 644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911577367137498,
+      "loss": 2.6923,
+      "step": 645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911162009140536,
+      "loss": 2.7852,
+      "step": 646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910745686617336,
+      "loss": 2.7881,
+      "step": 647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991032839964966,
+      "loss": 2.8368,
+      "step": 648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990991014831946,
+      "loss": 2.6689,
+      "step": 649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909490932708885,
+      "loss": 2.7725,
+      "step": 650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990907075290026,
+      "loss": 2.7084,
+      "step": 651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908649608976108,
+      "loss": 2.8167,
+      "step": 652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990822750101914,
+      "loss": 2.779,
+      "step": 653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907804429112256,
+      "loss": 3.0487,
+      "step": 654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907380393338544,
+      "loss": 2.7972,
+      "step": 655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906955393781283,
+      "loss": 2.7495,
+      "step": 656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990652943052394,
+      "loss": 2.7169,
+      "step": 657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906102503650174,
+      "loss": 2.7907,
+      "step": 658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990567461324383,
+      "loss": 2.9171,
+      "step": 659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990524575938894,
+      "loss": 2.8581,
+      "step": 660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904815942169737,
+      "loss": 2.7891,
+      "step": 661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904385161670624,
+      "loss": 2.8196,
+      "step": 662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903953417976213,
+      "loss": 2.6944,
+      "step": 663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903520711171292,
+      "loss": 2.9116,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903087041340842,
+      "loss": 2.7876,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902652408570036,
+      "loss": 2.802,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902216812944231,
+      "loss": 2.7182,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901780254548977,
+      "loss": 2.7382,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901342733470014,
+      "loss": 2.8358,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900904249793264,
+      "loss": 2.7785,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900464803604845,
+      "loss": 2.7359,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900024394991062,
+      "loss": 2.8051,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989958302403841,
+      "loss": 2.8752,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899140690833572,
+      "loss": 2.7479,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989869739546342,
+      "loss": 2.8558,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989825313801501,
+      "loss": 2.7665,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897807918575596,
+      "loss": 2.7789,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897361737232619,
+      "loss": 2.798,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896914594073702,
+      "loss": 2.7344,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896466489186665,
+      "loss": 2.8625,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989601742265951,
+      "loss": 2.8027,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895567394580436,
+      "loss": 2.8264,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895116405037822,
+      "loss": 2.6315,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894664454120241,
+      "loss": 2.6777,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894211541916454,
+      "loss": 2.7527,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893757668515413,
+      "loss": 2.7688,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989330283400625,
+      "loss": 2.7612,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892847038478298,
+      "loss": 2.8565,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892390282021074,
+      "loss": 2.6305,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891932564724275,
+      "loss": 2.693,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891473886677802,
+      "loss": 2.8319,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891014247971732,
+      "loss": 2.6554,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890553648696336,
+      "loss": 2.688,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890092088942078,
+      "loss": 2.7255,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098896295687996,
+      "loss": 2.8188,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988916608835974,
+      "loss": 2.6895,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888701647713528,
+      "loss": 2.7679,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888236246952172,
+      "loss": 2.6331,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887769886167074,
+      "loss": 2.7784,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988730256544983,
+      "loss": 2.6677,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886834284892213,
+      "loss": 2.7335,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886365044586196,
+      "loss": 2.9033,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885894844623932,
+      "loss": 2.7929,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885423685097768,
+      "loss": 2.6297,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884951566100236,
+      "loss": 2.8098,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884478487724058,
+      "loss": 2.6235,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884004450062143,
+      "loss": 2.7175,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883529453207592,
+      "loss": 2.6654,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883053497253692,
+      "loss": 2.6789,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882576582293913,
+      "loss": 2.7834,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882098708421925,
+      "loss": 2.6819,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881619875731576,
+      "loss": 2.7745,
+      "step": 712
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881140084316907,
+      "loss": 2.712,
+      "step": 713
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880659334272146,
+      "loss": 2.8058,
+      "step": 714
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880177625691713,
+      "loss": 2.6098,
+      "step": 715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879694958670208,
+      "loss": 2.7254,
+      "step": 716
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987921133330243,
+      "loss": 2.7222,
+      "step": 717
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878726749683354,
+      "loss": 2.6778,
+      "step": 718
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878241207908157,
+      "loss": 2.6095,
+      "step": 719
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987775470807219,
+      "loss": 2.9887,
+      "step": 720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877267250271,
+      "loss": 2.6981,
+      "step": 721
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876778834600325,
+      "loss": 2.8867,
+      "step": 722
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876289461156088,
+      "loss": 2.7736,
+      "step": 723
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875799130034393,
+      "loss": 2.7942,
+      "step": 724
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875307841331542,
+      "loss": 2.7159,
+      "step": 725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874815595144022,
+      "loss": 2.7598,
+      "step": 726
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987432239156851,
+      "loss": 2.8929,
+      "step": 727
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873828230701862,
+      "loss": 2.7379,
+      "step": 728
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873333112641132,
+      "loss": 2.617,
+      "step": 729
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987283703748356,
+      "loss": 2.8189,
+      "step": 730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872340005326573,
+      "loss": 2.6843,
+      "step": 731
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987184201626778,
+      "loss": 2.8218,
+      "step": 732
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987134307040499,
+      "loss": 2.8174,
+      "step": 733
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987084316783619,
+      "loss": 2.8401,
+      "step": 734
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870342308659558,
+      "loss": 2.6268,
+      "step": 735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869840492973461,
+      "loss": 2.7504,
+      "step": 736
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869337720876455,
+      "loss": 2.7281,
+      "step": 737
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868833992467277,
+      "loss": 2.8575,
+      "step": 738
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986832930784486,
+      "loss": 2.7218,
+      "step": 739
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986782366710832,
+      "loss": 2.7523,
+      "step": 740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867317070356963,
+      "loss": 2.6892,
+      "step": 741
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866809517690284,
+      "loss": 2.7342,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866301009207957,
+      "loss": 2.7333,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865791545009856,
+      "loss": 2.7306,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865281125196036,
+      "loss": 2.7259,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864769749866743,
+      "loss": 2.7712,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864257419122404,
+      "loss": 2.7284,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986374413306364,
+      "loss": 2.7563,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863229891791257,
+      "loss": 2.7385,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862714695406253,
+      "loss": 2.7341,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862198544009803,
+      "loss": 2.7876,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986168143770328,
+      "loss": 2.7,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861163376588243,
+      "loss": 2.6251,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860644360766434,
+      "loss": 2.6694,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860124390339785,
+      "loss": 2.8806,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859603465410417,
+      "loss": 2.7944,
+      "step": 756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859081586080635,
+      "loss": 2.6908,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858558752452935,
+      "loss": 2.8373,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858034964629998,
+      "loss": 2.7603,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857510222714692,
+      "loss": 2.5494,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856984526810076,
+      "loss": 2.7603,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985645787701939,
+      "loss": 2.7986,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985593027344607,
+      "loss": 2.7752,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855401716193733,
+      "loss": 2.7721,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854872205366182,
+      "loss": 2.7202,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854341741067415,
+      "loss": 2.6336,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853810323401607,
+      "loss": 2.8809,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985327795247313,
+      "loss": 2.6347,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852744628386534,
+      "loss": 2.7036,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852210351246568,
+      "loss": 2.7744,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851675121158156,
+      "loss": 2.788,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851138938226417,
+      "loss": 2.7638,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985060180255665,
+      "loss": 2.6946,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850063714254352,
+      "loss": 2.6215,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849524673425196,
+      "loss": 2.6645,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848984680175049,
+      "loss": 2.5119,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984844373460996,
+      "loss": 2.7582,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984790183683617,
+      "loss": 2.777,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847358986960109,
+      "loss": 2.657,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846815185088384,
+      "loss": 2.8744,
+      "step": 780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846270431327793,
+      "loss": 2.791,
+      "step": 781
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984572472578533,
+      "loss": 2.8641,
+      "step": 782
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845178068568162,
+      "loss": 2.7462,
+      "step": 783
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844630459783655,
+      "loss": 2.7338,
+      "step": 784
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844081899539352,
+      "loss": 2.9679,
+      "step": 785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984353238794299,
+      "loss": 2.7302,
+      "step": 786
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984298192510249,
+      "loss": 2.6201,
+      "step": 787
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984243051112596,
+      "loss": 2.7364,
+      "step": 788
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841878146121693,
+      "loss": 2.7855,
+      "step": 789
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841324830198173,
+      "loss": 2.6904,
+      "step": 790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840770563464066,
+      "loss": 2.6109,
+      "step": 791
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840215346028228,
+      "loss": 2.6918,
+      "step": 792
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098396591779997,
+      "loss": 2.5541,
+      "step": 793
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839102059487714,
+      "loss": 2.6811,
+      "step": 794
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983854399060168,
+      "loss": 2.7387,
+      "step": 795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837984971451203,
+      "loss": 2.9578,
+      "step": 796
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837425002146073,
+      "loss": 2.8018,
+      "step": 797
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983686408279626,
+      "loss": 2.6281,
+      "step": 798
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983630221351193,
+      "loss": 2.6242,
+      "step": 799
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835739394403427,
+      "loss": 2.6081,
+      "step": 800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835175625581291,
+      "loss": 2.748,
+      "step": 801
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834610907156237,
+      "loss": 2.6294,
+      "step": 802
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983404523923918,
+      "loss": 2.8108,
+      "step": 803
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983347862194121,
+      "loss": 2.7053,
+      "step": 804
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832911055373606,
+      "loss": 2.6711,
+      "step": 805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832342539647838,
+      "loss": 2.5881,
+      "step": 806
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831773074875556,
+      "loss": 2.6376,
+      "step": 807
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831202661168606,
+      "loss": 2.6434,
+      "step": 808
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830631298639008,
+      "loss": 2.6791,
+      "step": 809
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830058987398977,
+      "loss": 2.7307,
+      "step": 810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829485727560915,
+      "loss": 2.8056,
+      "step": 811
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828911519237401,
+      "loss": 2.7675,
+      "step": 812
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982833636254121,
+      "loss": 2.7416,
+      "step": 813
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827760257585302,
+      "loss": 2.6687,
+      "step": 814
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827183204482817,
+      "loss": 2.8378,
+      "step": 815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826605203347085,
+      "loss": 2.7817,
+      "step": 816
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826026254291626,
+      "loss": 2.7038,
+      "step": 817
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982544635743014,
+      "loss": 2.7732,
+      "step": 818
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824865512876515,
+      "loss": 2.8923,
+      "step": 819
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824283720744828,
+      "loss": 2.4652,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982370098114934,
+      "loss": 2.7251,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009823117294204496,
+      "loss": 2.7499,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822532660024928,
+      "loss": 2.6879,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982194707872546,
+      "loss": 2.8146,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821360550421093,
+      "loss": 2.6892,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820773075227016,
+      "loss": 2.7544,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820184653258613,
+      "loss": 2.653,
+      "step": 827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819595284631443,
+      "loss": 2.723,
+      "step": 828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981900496946125,
+      "loss": 2.6164,
+      "step": 829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981841370786398,
+      "loss": 2.7081,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817821499955744,
+      "loss": 2.5643,
+      "step": 831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817228345852852,
+      "loss": 2.8546,
+      "step": 832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816634245671796,
+      "loss": 2.6067,
+      "step": 833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816039199529256,
+      "loss": 2.7391,
+      "step": 834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815443207542092,
+      "loss": 2.7385,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814846269827358,
+      "loss": 2.7026,
+      "step": 836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814248386502285,
+      "loss": 2.8108,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813649557684299,
+      "loss": 2.6514,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813049783491002,
+      "loss": 2.7022,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981244906404019,
+      "loss": 2.61,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811847399449837,
+      "loss": 2.7505,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811244789838115,
+      "loss": 2.6773,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810641235323363,
+      "loss": 2.6978,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810036736024124,
+      "loss": 2.7371,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009809431292059113,
+      "loss": 2.7226,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980882490354724,
+      "loss": 2.6999,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808217570607593,
+      "loss": 2.6874,
+      "step": 847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807609293359452,
+      "loss": 2.4661,
+      "step": 848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807000071922279,
+      "loss": 2.737,
+      "step": 849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980638990641572,
+      "loss": 2.6802,
+      "step": 850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805778796959609,
+      "loss": 2.7264,
+      "step": 851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805166743673966,
+      "loss": 2.6227,
+      "step": 852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009804553746678992,
+      "loss": 2.8043,
+      "step": 853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980393980609508,
+      "loss": 2.8163,
+      "step": 854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803324922042804,
+      "loss": 2.6841,
+      "step": 855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802709094642922,
+      "loss": 2.7734,
+      "step": 856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980209232401638,
+      "loss": 2.8419,
+      "step": 857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980147461028431,
+      "loss": 2.7986,
+      "step": 858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800855953568024,
+      "loss": 2.7682,
+      "step": 859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800236353989028,
+      "loss": 2.781,
+      "step": 860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009799615811669003,
+      "loss": 2.7495,
+      "step": 861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798994326729825,
+      "loss": 2.9475,
+      "step": 862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798371899293547,
+      "loss": 2.7745,
+      "step": 863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797748529482412,
+      "loss": 2.863,
+      "step": 864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797124217418847,
+      "loss": 2.8293,
+      "step": 865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796498963225459,
+      "loss": 2.7853,
+      "step": 866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795872767025052,
+      "loss": 2.7194,
+      "step": 867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795245628940602,
+      "loss": 2.6158,
+      "step": 868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794617549095277,
+      "loss": 2.6654,
+      "step": 869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793988527612432,
+      "loss": 2.784,
+      "step": 870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793358564615597,
+      "loss": 2.7278,
+      "step": 871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00097927276602285,
+      "loss": 2.7479,
+      "step": 872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979209581457504,
+      "loss": 2.7635,
+      "step": 873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791463027779318,
+      "loss": 2.6617,
+      "step": 874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00097908292999656,
+      "loss": 2.6033,
+      "step": 875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009790194631258352,
+      "loss": 2.6939,
+      "step": 876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978955902178222,
+      "loss": 2.6085,
+      "step": 877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788922471662032,
+      "loss": 2.6142,
+      "step": 878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788284981022805,
+      "loss": 2.7056,
+      "step": 879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787646549989737,
+      "loss": 2.618,
+      "step": 880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787007178688213,
+      "loss": 2.6256,
+      "step": 881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786366867243802,
+      "loss": 2.7132,
+      "step": 882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785725615782262,
+      "loss": 2.5542,
+      "step": 883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785083424429524,
+      "loss": 2.6687,
+      "step": 884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784440293311716,
+      "loss": 2.7282,
+      "step": 885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783796222555144,
+      "loss": 2.7853,
+      "step": 886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00097831512122863,
+      "loss": 2.7427,
+      "step": 887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782505262631862,
+      "loss": 2.6232,
+      "step": 888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978185837371869,
+      "loss": 2.7762,
+      "step": 889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978121054567383,
+      "loss": 2.6983,
+      "step": 890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780561778624508,
+      "loss": 2.5982,
+      "step": 891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779912072698147,
+      "loss": 2.6945,
+      "step": 892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779261428022336,
+      "loss": 2.6329,
+      "step": 893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778609844724867,
+      "loss": 2.821,
+      "step": 894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00097779573229337,
+      "loss": 2.7312,
+      "step": 895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777303862776993,
+      "loss": 2.7998,
+      "step": 896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977664946438308,
+      "loss": 2.7343,
+      "step": 897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775994127880479,
+      "loss": 2.6572,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775337853397896,
+      "loss": 2.7292,
+      "step": 899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009774680641064222,
+      "loss": 2.6033,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009774022491008528,
+      "loss": 2.6175,
+      "step": 901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977336340336007,
+      "loss": 2.6591,
+      "step": 902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772703378248292,
+      "loss": 2.9249,
+      "step": 903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772042415802818,
+      "loss": 2.8807,
+      "step": 904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977138051615346,
+      "loss": 2.6575,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770717679430205,
+      "loss": 2.7774,
+      "step": 906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770053905763239,
+      "loss": 2.6516,
+      "step": 907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769389195282915,
+      "loss": 2.9628,
+      "step": 908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768723548119788,
+      "loss": 2.6432,
+      "step": 909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768056964404581,
+      "loss": 2.6433,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009767389444268208,
+      "loss": 2.7655,
+      "step": 911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976672098784177,
+      "loss": 2.7505,
+      "step": 912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766051595256546,
+      "loss": 2.718,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765381266644,
+      "loss": 2.6595,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764710002135784,
+      "loss": 2.5328,
+      "step": 915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764037801863727,
+      "loss": 2.816,
+      "step": 916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763364665959849,
+      "loss": 2.6107,
+      "step": 917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762690594556347,
+      "loss": 2.734,
+      "step": 918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762015587785609,
+      "loss": 2.7671,
+      "step": 919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097613396457802,
+      "loss": 2.7174,
+      "step": 920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760662768672872,
+      "loss": 2.6318,
+      "step": 921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009759984956596562,
+      "loss": 2.7471,
+      "step": 922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009759306209684384,
+      "loss": 2.7115,
+      "step": 923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758626528069646,
+      "loss": 2.7578,
+      "step": 924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975794591188583,
+      "loss": 2.7224,
+      "step": 925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757264361266607,
+      "loss": 2.6638,
+      "step": 926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756581876345829,
+      "loss": 2.7052,
+      "step": 927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755898457257532,
+      "loss": 2.6091,
+      "step": 928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755214104135939,
+      "loss": 2.7879,
+      "step": 929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754528817115452,
+      "loss": 2.6028,
+      "step": 930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753842596330653,
+      "loss": 2.6033,
+      "step": 931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753155441916319,
+      "loss": 2.8349,
+      "step": 932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752467354007399,
+      "loss": 2.6471,
+      "step": 933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751778332739033,
+      "loss": 2.8591,
+      "step": 934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751088378246539,
+      "loss": 2.6536,
+      "step": 935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975039749066542,
+      "loss": 2.7523,
+      "step": 936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749705670131363,
+      "loss": 2.6325,
+      "step": 937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749012916780241,
+      "loss": 2.667,
+      "step": 938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748319230748104,
+      "loss": 2.7902,
+      "step": 939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747624612171188,
+      "loss": 2.7208,
+      "step": 940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746929061185912,
+      "loss": 2.7544,
+      "step": 941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746232577928881,
+      "loss": 2.7189,
+      "step": 942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745535162536881,
+      "loss": 2.6996,
+      "step": 943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744836815146877,
+      "loss": 2.5719,
+      "step": 944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744137535896022,
+      "loss": 2.6251,
+      "step": 945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743437324921655,
+      "loss": 2.6032,
+      "step": 946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742736182361289,
+      "loss": 2.8309,
+      "step": 947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742034108352624,
+      "loss": 2.6571,
+      "step": 948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741331103033547,
+      "loss": 2.6091,
+      "step": 949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740627166542125,
+      "loss": 2.7121,
+      "step": 950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739922299016601,
+      "loss": 2.7393,
+      "step": 951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739216500595416,
+      "loss": 2.7044,
+      "step": 952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738509771417179,
+      "loss": 2.5732,
+      "step": 953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737802111620691,
+      "loss": 2.868,
+      "step": 954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737093521344931,
+      "loss": 2.7834,
+      "step": 955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736384000729062,
+      "loss": 2.5783,
+      "step": 956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000973567354991243,
+      "loss": 2.6701,
+      "step": 957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734962169034567,
+      "loss": 2.6024,
+      "step": 958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734249858235181,
+      "loss": 2.7184,
+      "step": 959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733536617654166,
+      "loss": 2.7449,
+      "step": 960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732822447431602,
+      "loss": 2.699,
+      "step": 961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732107347707745,
+      "loss": 2.6222,
+      "step": 962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731391318623038,
+      "loss": 2.7406,
+      "step": 963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730674360318106,
+      "loss": 2.6728,
+      "step": 964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729956472933754,
+      "loss": 2.6216,
+      "step": 965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729237656610972,
+      "loss": 2.613,
+      "step": 966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728517911490932,
+      "loss": 2.7182,
+      "step": 967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972779723771499,
+      "loss": 2.6606,
+      "step": 968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727075635424681,
+      "loss": 2.7901,
+      "step": 969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972635310476172,
+      "loss": 2.6808,
+      "step": 970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725629645868014,
+      "loss": 2.7874,
+      "step": 971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724905258885646,
+      "loss": 2.6497,
+      "step": 972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724179943956878,
+      "loss": 2.8652,
+      "step": 973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723453701224161,
+      "loss": 2.5505,
+      "step": 974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722726530830123,
+      "loss": 2.6208,
+      "step": 975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009721998432917578,
+      "loss": 2.7838,
+      "step": 976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972126940762952,
+      "loss": 2.5574,
+      "step": 977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720539455109128,
+      "loss": 2.7849,
+      "step": 978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719808575499758,
+      "loss": 2.7261,
+      "step": 979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971907676894495,
+      "loss": 2.7798,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718344035588431,
+      "loss": 2.61,
+      "step": 981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717610375574102,
+      "loss": 2.8083,
+      "step": 982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716875789046053,
+      "loss": 2.6793,
+      "step": 983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716140276148549,
+      "loss": 2.618,
+      "step": 984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715403837026045,
+      "loss": 2.682,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009714666471823172,
+      "loss": 2.8072,
+      "step": 986
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713928180684746,
+      "loss": 2.6973,
+      "step": 987
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000971318896375576,
+      "loss": 2.6326,
+      "step": 988
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712448821181394,
+      "loss": 2.7197,
+      "step": 989
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009711707753107011,
+      "loss": 2.594,
+      "step": 990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710965759678149,
+      "loss": 2.7043,
+      "step": 991
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710222841040532,
+      "loss": 2.8493,
+      "step": 992
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709478997340068,
+      "loss": 2.6878,
+      "step": 993
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708734228722841,
+      "loss": 2.6572,
+      "step": 994
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707988535335121,
+      "loss": 2.6788,
+      "step": 995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707241917323358,
+      "loss": 2.6209,
+      "step": 996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706494374834185,
+      "loss": 2.5304,
+      "step": 997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705745908014414,
+      "loss": 2.7306,
+      "step": 998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704996517011039,
+      "loss": 2.9155,
+      "step": 999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704246201971238,
+      "loss": 2.6373,
+      "step": 1000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000970349496304237,
+      "loss": 2.6731,
+      "step": 1001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702742800371971,
+      "loss": 2.6439,
+      "step": 1002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701989714107764,
+      "loss": 3.0199,
+      "step": 1003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701235704397652,
+      "loss": 2.8486,
+      "step": 1004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700480771389717,
+      "loss": 2.6364,
+      "step": 1005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699724915232224,
+      "loss": 2.8553,
+      "step": 1006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698968136073619,
+      "loss": 2.7737,
+      "step": 1007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698210434062529,
+      "loss": 2.8553,
+      "step": 1008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697451809347764,
+      "loss": 2.881,
+      "step": 1009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696692262078315,
+      "loss": 2.7999,
+      "step": 1010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969593179240335,
+      "loss": 2.8114,
+      "step": 1011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009695170400472222,
+      "loss": 2.7979,
+      "step": 1012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009694408086434465,
+      "loss": 2.7991,
+      "step": 1013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693644850439796,
+      "loss": 2.6841,
+      "step": 1014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692880692638104,
+      "loss": 2.7801,
+      "step": 1015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692115613179473,
+      "loss": 2.7174,
+      "step": 1016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009691349612214156,
+      "loss": 2.71,
+      "step": 1017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690582689892591,
+      "loss": 2.7014,
+      "step": 1018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689814846365399,
+      "loss": 2.6669,
+      "step": 1019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689046081783381,
+      "loss": 2.8821,
+      "step": 1020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968827639629752,
+      "loss": 2.6768,
+      "step": 1021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687505790058973,
+      "loss": 2.7116,
+      "step": 1022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686734263219087,
+      "loss": 2.5558,
+      "step": 1023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685961815929384,
+      "loss": 2.7073,
+      "step": 1024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685188448341571,
+      "loss": 2.7237,
+      "step": 1025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968441416060753,
+      "loss": 2.651,
+      "step": 1026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009683638952879328,
+      "loss": 2.5866,
+      "step": 1027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682862825309214,
+      "loss": 2.7044,
+      "step": 1028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682085778049613,
+      "loss": 2.9246,
+      "step": 1029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681307811253133,
+      "loss": 2.8066,
+      "step": 1030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680528925072563,
+      "loss": 2.8306,
+      "step": 1031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679749119660873,
+      "loss": 2.7944,
+      "step": 1032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678968395171213,
+      "loss": 2.7713,
+      "step": 1033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967818675175691,
+      "loss": 2.6559,
+      "step": 1034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677404189571478,
+      "loss": 2.7324,
+      "step": 1035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676620708768607,
+      "loss": 2.7322,
+      "step": 1036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675836309502166,
+      "loss": 2.6505,
+      "step": 1037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967505099192621,
+      "loss": 2.7586,
+      "step": 1038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674264756194973,
+      "loss": 2.6895,
+      "step": 1039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673477602462861,
+      "loss": 2.7067,
+      "step": 1040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009672689530884473,
+      "loss": 2.7258,
+      "step": 1041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967190054161458,
+      "loss": 2.8083,
+      "step": 1042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671110634808135,
+      "loss": 2.7842,
+      "step": 1043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009670319810620271,
+      "loss": 2.6322,
+      "step": 1044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669528069206302,
+      "loss": 2.7927,
+      "step": 1045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668735410721722,
+      "loss": 2.7345,
+      "step": 1046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667941835322207,
+      "loss": 2.7653,
+      "step": 1047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667147343163608,
+      "loss": 2.5679,
+      "step": 1048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966635193440196,
+      "loss": 2.781,
+      "step": 1049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665555609193477,
+      "loss": 2.7378,
+      "step": 1050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009664758367694554,
+      "loss": 2.7201,
+      "step": 1051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663960210061764,
+      "loss": 2.7256,
+      "step": 1052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663161136451861,
+      "loss": 2.5816,
+      "step": 1053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662361147021779,
+      "loss": 2.6003,
+      "step": 1054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661560241928632,
+      "loss": 2.6675,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009660758421329714,
+      "loss": 2.7553,
+      "step": 1056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009659955685382497,
+      "loss": 2.9413,
+      "step": 1057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009659152034244634,
+      "loss": 2.6819,
+      "step": 1058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658347468073959,
+      "loss": 2.7033,
+      "step": 1059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009657541987028484,
+      "loss": 2.6522,
+      "step": 1060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656735591266401,
+      "loss": 2.7269,
+      "step": 1061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655928280946083,
+      "loss": 2.8156,
+      "step": 1062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655120056226081,
+      "loss": 2.6485,
+      "step": 1063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009654310917265124,
+      "loss": 2.6358,
+      "step": 1064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653500864222127,
+      "loss": 2.6159,
+      "step": 1065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652689897256175,
+      "loss": 2.6695,
+      "step": 1066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000965187801652654,
+      "loss": 2.6306,
+      "step": 1067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009651065222192672,
+      "loss": 2.854,
+      "step": 1068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650251514414198,
+      "loss": 2.8573,
+      "step": 1069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009649436893350928,
+      "loss": 2.6682,
+      "step": 1070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648621359162847,
+      "loss": 2.7215,
+      "step": 1071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647804912010123,
+      "loss": 2.662,
+      "step": 1072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009646987552053099,
+      "loss": 2.7018,
+      "step": 1073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009646169279452305,
+      "loss": 2.6295,
+      "step": 1074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645350094368441,
+      "loss": 2.7635,
+      "step": 1075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644529996962395,
+      "loss": 2.5787,
+      "step": 1076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643708987395226,
+      "loss": 2.8403,
+      "step": 1077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642887065828176,
+      "loss": 2.5664,
+      "step": 1078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964206423242267,
+      "loss": 2.5636,
+      "step": 1079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641240487340303,
+      "loss": 2.8759,
+      "step": 1080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640415830742857,
+      "loss": 2.7837,
+      "step": 1081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963959026279229,
+      "loss": 2.6878,
+      "step": 1082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638763783650739,
+      "loss": 2.4771,
+      "step": 1083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963793639348052,
+      "loss": 2.7892,
+      "step": 1084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637108092444129,
+      "loss": 2.7498,
+      "step": 1085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009636278880704239,
+      "loss": 2.7257,
+      "step": 1086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009635448758423702,
+      "loss": 2.8946,
+      "step": 1087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634617725765551,
+      "loss": 2.6435,
+      "step": 1088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633785782892997,
+      "loss": 2.7145,
+      "step": 1089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632952929969425,
+      "loss": 2.5622,
+      "step": 1090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632119167158409,
+      "loss": 2.6657,
+      "step": 1091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631284494623691,
+      "loss": 2.4791,
+      "step": 1092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009630448912529199,
+      "loss": 2.9066,
+      "step": 1093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629612421039036,
+      "loss": 2.7727,
+      "step": 1094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628775020317483,
+      "loss": 2.7523,
+      "step": 1095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627936710529003,
+      "loss": 2.5034,
+      "step": 1096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627097491838234,
+      "loss": 2.619,
+      "step": 1097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626257364409997,
+      "loss": 2.7025,
+      "step": 1098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625416328409286,
+      "loss": 2.6571,
+      "step": 1099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009624574384001276,
+      "loss": 2.6978,
+      "step": 1100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623731531351322,
+      "loss": 2.6141,
+      "step": 1101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622887770624953,
+      "loss": 2.8868,
+      "step": 1102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622043101987885,
+      "loss": 2.8167,
+      "step": 1103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621197525605999,
+      "loss": 2.6598,
+      "step": 1104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009620351041645366,
+      "loss": 2.7155,
+      "step": 1105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619503650272231,
+      "loss": 2.7927,
+      "step": 1106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618655351653016,
+      "loss": 2.6242,
+      "step": 1107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617806145954322,
+      "loss": 2.7573,
+      "step": 1108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000961695603334293,
+      "loss": 2.7283,
+      "step": 1109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616105013985798,
+      "loss": 2.6826,
+      "step": 1110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615253088050058,
+      "loss": 2.4381,
+      "step": 1111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009614400255703029,
+      "loss": 2.5973,
+      "step": 1112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613546517112199,
+      "loss": 2.6334,
+      "step": 1113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009612691872445239,
+      "loss": 2.6711,
+      "step": 1114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611836321869996,
+      "loss": 2.6977,
+      "step": 1115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009610979865554497,
+      "loss": 2.6782,
+      "step": 1116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009610122503666944,
+      "loss": 2.677,
+      "step": 1117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609264236375719,
+      "loss": 2.59,
+      "step": 1118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009608405063849381,
+      "loss": 2.6094,
+      "step": 1119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607544986256667,
+      "loss": 2.5851,
+      "step": 1120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009606684003766493,
+      "loss": 2.6789,
+      "step": 1121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605822116547949,
+      "loss": 2.5148,
+      "step": 1122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009604959324770306,
+      "loss": 2.5889,
+      "step": 1123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009604095628603013,
+      "loss": 2.7381,
+      "step": 1124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603231028215693,
+      "loss": 2.6713,
+      "step": 1125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602365523778152,
+      "loss": 2.6614,
+      "step": 1126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009601499115460368,
+      "loss": 2.6893,
+      "step": 1127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009600631803432499,
+      "loss": 2.5197,
+      "step": 1128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009599763587864882,
+      "loss": 2.5372,
+      "step": 1129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000959889446892803,
+      "loss": 2.7565,
+      "step": 1130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598024446792632,
+      "loss": 2.7159,
+      "step": 1131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009597153521629556,
+      "loss": 2.5786,
+      "step": 1132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009596281693609848,
+      "loss": 2.7303,
+      "step": 1133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009595408962904728,
+      "loss": 2.6355,
+      "step": 1134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00095945353296856,
+      "loss": 2.7378,
+      "step": 1135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593660794124038,
+      "loss": 2.6375,
+      "step": 1136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009592785356391795,
+      "loss": 2.4884,
+      "step": 1137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591909016660806,
+      "loss": 2.6537,
+      "step": 1138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591031775103176,
+      "loss": 2.6473,
+      "step": 1139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009590153631891193,
+      "loss": 2.8148,
+      "step": 1140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589274587197318,
+      "loss": 2.6422,
+      "step": 1141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588394641194193,
+      "loss": 2.6583,
+      "step": 1142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009587513794054632,
+      "loss": 2.8454,
+      "step": 1143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958663204595163,
+      "loss": 2.6365,
+      "step": 1144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585749397058357,
+      "loss": 2.6585,
+      "step": 1145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584865847548162,
+      "loss": 2.7124,
+      "step": 1146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583981397594569,
+      "loss": 2.7068,
+      "step": 1147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583096047371278,
+      "loss": 2.7912,
+      "step": 1148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582209797052168,
+      "loss": 2.6725,
+      "step": 1149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009581322646811294,
+      "loss": 2.7083,
+      "step": 1150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009580434596822889,
+      "loss": 2.5889,
+      "step": 1151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009579545647261359,
+      "loss": 2.6992,
+      "step": 1152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578655798301289,
+      "loss": 2.66,
+      "step": 1153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009577765050117441,
+      "loss": 2.5729,
+      "step": 1154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576873402884756,
+      "loss": 2.6134,
+      "step": 1155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009575980856778344,
+      "loss": 2.5795,
+      "step": 1156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009575087411973501,
+      "loss": 2.7248,
+      "step": 1157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009574193068645692,
+      "loss": 2.6625,
+      "step": 1158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573297826970562,
+      "loss": 2.7067,
+      "step": 1159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009572401687123931,
+      "loss": 2.7871,
+      "step": 1160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009571504649281795,
+      "loss": 2.621,
+      "step": 1161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009570606713620333,
+      "loss": 2.6579,
+      "step": 1162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569707880315889,
+      "loss": 2.6155,
+      "step": 1163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009568808149544992,
+      "loss": 2.7266,
+      "step": 1164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567907521484343,
+      "loss": 2.6247,
+      "step": 1165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000956700599631082,
+      "loss": 2.6834,
+      "step": 1166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009566103574201482,
+      "loss": 2.556,
+      "step": 1167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565200255333555,
+      "loss": 2.6808,
+      "step": 1168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564296039884447,
+      "loss": 2.5459,
+      "step": 1169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563390928031743,
+      "loss": 2.6348,
+      "step": 1170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009562484919953202,
+      "loss": 2.5148,
+      "step": 1171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561578015826757,
+      "loss": 2.6011,
+      "step": 1172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009560670215830522,
+      "loss": 2.7482,
+      "step": 1173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009559761520142781,
+      "loss": 2.8027,
+      "step": 1174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009558851928942,
+      "loss": 2.6451,
+      "step": 1175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557941442406817,
+      "loss": 2.6731,
+      "step": 1176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557030060716045,
+      "loss": 2.7459,
+      "step": 1177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009556117784048677,
+      "loss": 2.6517,
+      "step": 1178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009555204612583878,
+      "loss": 2.6891,
+      "step": 1179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554290546500989,
+      "loss": 2.6196,
+      "step": 1180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009553375585979532,
+      "loss": 2.5863,
+      "step": 1181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552459731199195,
+      "loss": 2.6454,
+      "step": 1182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009551542982339849,
+      "loss": 2.6235,
+      "step": 1183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000955062533958154,
+      "loss": 2.4909,
+      "step": 1184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549706803104488,
+      "loss": 2.6358,
+      "step": 1185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009548787373089087,
+      "loss": 2.6116,
+      "step": 1186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009547867049715908,
+      "loss": 2.7389,
+      "step": 1187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546945833165701,
+      "loss": 2.6505,
+      "step": 1188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546023723619386,
+      "loss": 2.6142,
+      "step": 1189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009545100721258059,
+      "loss": 2.614,
+      "step": 1190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009544176826262994,
+      "loss": 2.5221,
+      "step": 1191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954325203881564,
+      "loss": 2.6014,
+      "step": 1192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009542326359097619,
+      "loss": 2.6427,
+      "step": 1193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954139978729073,
+      "loss": 2.6675,
+      "step": 1194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009540472323576948,
+      "loss": 2.6528,
+      "step": 1195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539543968138419,
+      "loss": 2.495,
+      "step": 1196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538614721157471,
+      "loss": 2.6264,
+      "step": 1197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537684582816599,
+      "loss": 2.6483,
+      "step": 1198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009536753553298481,
+      "loss": 2.6299,
+      "step": 1199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009535821632785964,
+      "loss": 2.5555,
+      "step": 1200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009534888821462074,
+      "loss": 2.58,
+      "step": 1201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533955119510008,
+      "loss": 2.5903,
+      "step": 1202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533020527113142,
+      "loss": 2.5047,
+      "step": 1203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009532085044455023,
+      "loss": 2.5663,
+      "step": 1204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009531148671719378,
+      "loss": 2.5585,
+      "step": 1205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009530211409090103,
+      "loss": 2.7651,
+      "step": 1206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529273256751272,
+      "loss": 2.6383,
+      "step": 1207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528334214887133,
+      "loss": 2.5677,
+      "step": 1208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009527394283682111,
+      "loss": 2.607,
+      "step": 1209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526453463320801,
+      "loss": 2.683,
+      "step": 1210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009525511753987974,
+      "loss": 2.61,
+      "step": 1211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524569155868581,
+      "loss": 2.6352,
+      "step": 1212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009523625669147739,
+      "loss": 2.6461,
+      "step": 1213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009522681294010747,
+      "loss": 2.5844,
+      "step": 1214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009521736030643074,
+      "loss": 2.4651,
+      "step": 1215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520789879230365,
+      "loss": 2.5766,
+      "step": 1216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009519842839958438,
+      "loss": 2.5423,
+      "step": 1217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009518894913013288,
+      "loss": 2.5529,
+      "step": 1218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517946098581081,
+      "loss": 2.5841,
+      "step": 1219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000951699639684816,
+      "loss": 2.5814,
+      "step": 1220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009516045808001044,
+      "loss": 2.7076,
+      "step": 1221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515094332226419,
+      "loss": 2.5705,
+      "step": 1222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009514141969711155,
+      "loss": 2.6465,
+      "step": 1223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009513188720642285,
+      "loss": 2.516,
+      "step": 1224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512234585207027,
+      "loss": 2.5712,
+      "step": 1225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009511279563592767,
+      "loss": 2.5652,
+      "step": 1226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510323655987063,
+      "loss": 2.5077,
+      "step": 1227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009509366862577656,
+      "loss": 2.4909,
+      "step": 1228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508409183552449,
+      "loss": 2.3999,
+      "step": 1229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009507450619099529,
+      "loss": 2.547,
+      "step": 1230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506491169407152,
+      "loss": 2.7065,
+      "step": 1231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009505530834663748,
+      "loss": 2.5344,
+      "step": 1232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009504569615057923,
+      "loss": 2.5674,
+      "step": 1233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009503607510778456,
+      "loss": 2.713,
+      "step": 1234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009502644522014297,
+      "loss": 2.7668,
+      "step": 1235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501680648954573,
+      "loss": 2.6352,
+      "step": 1236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009500715891788584,
+      "loss": 2.5926,
+      "step": 1237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499750250705803,
+      "loss": 2.5452,
+      "step": 1238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498783725895877,
+      "loss": 2.4862,
+      "step": 1239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009497816317548624,
+      "loss": 2.5893,
+      "step": 1240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009496848025854042,
+      "loss": 2.6225,
+      "step": 1241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009495878851002294,
+      "loss": 2.7667,
+      "step": 1242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009494908793183724,
+      "loss": 2.7403,
+      "step": 1243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009493937852588845,
+      "loss": 2.6676,
+      "step": 1244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009492966029408345,
+      "loss": 2.4896,
+      "step": 1245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009491993323833086,
+      "loss": 2.5579,
+      "step": 1246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00094910197360541,
+      "loss": 2.4753,
+      "step": 1247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490045266262595,
+      "loss": 2.6143,
+      "step": 1248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009489069914649953,
+      "loss": 2.8279,
+      "step": 1249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488093681407725,
+      "loss": 2.6495,
+      "step": 1250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009487116566727643,
+      "loss": 2.6199,
+      "step": 1251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009486138570801604,
+      "loss": 2.6864,
+      "step": 1252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000948515969382168,
+      "loss": 2.6887,
+      "step": 1253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009484179935980119,
+      "loss": 2.5861,
+      "step": 1254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000948319929746934,
+      "loss": 2.6544,
+      "step": 1255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482217778481936,
+      "loss": 2.6354,
+      "step": 1256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009481235379210671,
+      "loss": 2.5664,
+      "step": 1257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480252099848484,
+      "loss": 2.5557,
+      "step": 1258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009479267940588485,
+      "loss": 2.5722,
+      "step": 1259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478282901623958,
+      "loss": 2.5365,
+      "step": 1260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009477296983148358,
+      "loss": 2.5721,
+      "step": 1261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476310185355317,
+      "loss": 2.6401,
+      "step": 1262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475322508438635,
+      "loss": 2.567,
+      "step": 1263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009474333952592287,
+      "loss": 2.5328,
+      "step": 1264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009473344518010421,
+      "loss": 2.6091,
+      "step": 1265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009472354204887355,
+      "loss": 2.5024,
+      "step": 1266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009471363013417581,
+      "loss": 2.5085,
+      "step": 1267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009470370943795765,
+      "loss": 2.5579,
+      "step": 1268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009469377996216746,
+      "loss": 2.6886,
+      "step": 1269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009468384170875531,
+      "loss": 2.6581,
+      "step": 1270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009467389467967301,
+      "loss": 2.6025,
+      "step": 1271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466393887687414,
+      "loss": 2.5383,
+      "step": 1272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009465397430231393,
+      "loss": 2.6753,
+      "step": 1273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009464400095794941,
+      "loss": 2.4244,
+      "step": 1274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009463401884573925,
+      "loss": 2.587,
+      "step": 1275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462402796764391,
+      "loss": 2.5549,
+      "step": 1276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009461402832562554,
+      "loss": 2.6183,
+      "step": 1277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009460401992164802,
+      "loss": 2.5696,
+      "step": 1278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009459400275767696,
+      "loss": 2.6912,
+      "step": 1279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009458397683567963,
+      "loss": 2.7235,
+      "step": 1280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009457394215762513,
+      "loss": 2.5961,
+      "step": 1281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009456389872548416,
+      "loss": 2.5666,
+      "step": 1282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009455384654122925,
+      "loss": 2.4844,
+      "step": 1283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009454378560683456,
+      "loss": 2.6179,
+      "step": 1284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009453371592427601,
+      "loss": 2.5425,
+      "step": 1285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452363749553124,
+      "loss": 2.4307,
+      "step": 1286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945135503225796,
+      "loss": 2.6022,
+      "step": 1287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009450345440740216,
+      "loss": 2.583,
+      "step": 1288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009449334975198168,
+      "loss": 2.5139,
+      "step": 1289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448323635830269,
+      "loss": 2.5182,
+      "step": 1290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000944731142283514,
+      "loss": 2.5817,
+      "step": 1291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009446298336411574,
+      "loss": 2.5491,
+      "step": 1292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445284376758536,
+      "loss": 2.58,
+      "step": 1293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009444269544075162,
+      "loss": 2.6736,
+      "step": 1294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009443253838560759,
+      "loss": 2.5894,
+      "step": 1295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009442237260414807,
+      "loss": 2.572,
+      "step": 1296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009441219809836959,
+      "loss": 2.7336,
+      "step": 1297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009440201487027031,
+      "loss": 2.5702,
+      "step": 1298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439182292185022,
+      "loss": 2.6086,
+      "step": 1299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438162225511096,
+      "loss": 2.6984,
+      "step": 1300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009437141287205584,
+      "loss": 2.6087,
+      "step": 1301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436119477468996,
+      "loss": 2.551,
+      "step": 1302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000943509679650201,
+      "loss": 2.493,
+      "step": 1303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009434073244505476,
+      "loss": 2.7052,
+      "step": 1304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433048821680413,
+      "loss": 2.6433,
+      "step": 1305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009432023528228011,
+      "loss": 2.5696,
+      "step": 1306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430997364349634,
+      "loss": 2.5937,
+      "step": 1307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009429970330246816,
+      "loss": 2.7415,
+      "step": 1308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009428942426121258,
+      "loss": 2.6257,
+      "step": 1309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000942791365217484,
+      "loss": 2.5128,
+      "step": 1310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009426884008609602,
+      "loss": 2.5887,
+      "step": 1311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009425853495627763,
+      "loss": 2.5708,
+      "step": 1312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009424822113431712,
+      "loss": 2.7295,
+      "step": 1313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423789862224004,
+      "loss": 2.6315,
+      "step": 1314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009422756742207368,
+      "loss": 2.6129,
+      "step": 1315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421722753584706,
+      "loss": 2.5638,
+      "step": 1316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009420687896559086,
+      "loss": 2.5305,
+      "step": 1317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009419652171333748,
+      "loss": 2.52,
+      "step": 1318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418615578112104,
+      "loss": 2.8141,
+      "step": 1319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009417578117097736,
+      "loss": 2.6438,
+      "step": 1320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416539788494394,
+      "loss": 2.696,
+      "step": 1321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009415500592506001,
+      "loss": 2.6233,
+      "step": 1322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000941446052933665,
+      "loss": 2.5698,
+      "step": 1323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413419599190604,
+      "loss": 2.4925,
+      "step": 1324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009412377802272297,
+      "loss": 2.4825,
+      "step": 1325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000941133513878633,
+      "loss": 2.7592,
+      "step": 1326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009410291608937478,
+      "loss": 2.6297,
+      "step": 1327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009409247212930683,
+      "loss": 2.7053,
+      "step": 1328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009408201950971063,
+      "loss": 2.5441,
+      "step": 1329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009407155823263898,
+      "loss": 2.6082,
+      "step": 1330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009406108830014644,
+      "loss": 2.4319,
+      "step": 1331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009405060971428923,
+      "loss": 2.5761,
+      "step": 1332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940401224771253,
+      "loss": 2.4344,
+      "step": 1333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009402962659071431,
+      "loss": 2.4801,
+      "step": 1334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009401912205711755,
+      "loss": 2.5717,
+      "step": 1335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009400860887839811,
+      "loss": 2.4775,
+      "step": 1336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399808705662068,
+      "loss": 2.7347,
+      "step": 1337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009398755659385168,
+      "loss": 2.679,
+      "step": 1338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009397701749215929,
+      "loss": 2.6516,
+      "step": 1339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009396646975361329,
+      "loss": 2.7721,
+      "step": 1340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009395591338028522,
+      "loss": 2.5705,
+      "step": 1341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000939453483742483,
+      "loss": 2.6498,
+      "step": 1342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393477473757741,
+      "loss": 2.6924,
+      "step": 1343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009392419247234921,
+      "loss": 2.6029,
+      "step": 1344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009391360158064192,
+      "loss": 2.6713,
+      "step": 1345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009390300206453563,
+      "loss": 2.823,
+      "step": 1346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009389239392611197,
+      "loss": 2.4499,
+      "step": 1347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009388177716745432,
+      "loss": 2.438,
+      "step": 1348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009387115179064777,
+      "loss": 2.5244,
+      "step": 1349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000938605177977791,
+      "loss": 2.506,
+      "step": 1350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009384987519093677,
+      "loss": 2.5148,
+      "step": 1351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009383922397221089,
+      "loss": 2.668,
+      "step": 1352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009382856414369336,
+      "loss": 2.6473,
+      "step": 1353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009381789570747768,
+      "loss": 2.5637,
+      "step": 1354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380721866565908,
+      "loss": 2.6201,
+      "step": 1355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009379653302033446,
+      "loss": 2.5333,
+      "step": 1356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009378583877360246,
+      "loss": 2.6075,
+      "step": 1357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009377513592756335,
+      "loss": 2.624,
+      "step": 1358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000937644244843191,
+      "loss": 2.5498,
+      "step": 1359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009375370444597342,
+      "loss": 2.5372,
+      "step": 1360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374297581463163,
+      "loss": 2.4945,
+      "step": 1361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009373223859240079,
+      "loss": 2.6343,
+      "step": 1362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009372149278138964,
+      "loss": 2.5363,
+      "step": 1363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000937107383837086,
+      "loss": 2.5319,
+      "step": 1364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009369997540146977,
+      "loss": 2.6916,
+      "step": 1365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368920383678694,
+      "loss": 2.6667,
+      "step": 1366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009367842369177559,
+      "loss": 2.596,
+      "step": 1367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000936676349685529,
+      "loss": 2.5091,
+      "step": 1368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009365683766923771,
+      "loss": 2.6912,
+      "step": 1369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364603179595053,
+      "loss": 2.6461,
+      "step": 1370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000936352173508136,
+      "loss": 2.5132,
+      "step": 1371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009362439433595083,
+      "loss": 2.5436,
+      "step": 1372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361356275348776,
+      "loss": 2.4362,
+      "step": 1373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009360272260555169,
+      "loss": 2.5181,
+      "step": 1374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009359187389427157,
+      "loss": 2.5394,
+      "step": 1375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358101662177802,
+      "loss": 2.7203,
+      "step": 1376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009357015079020334,
+      "loss": 2.5988,
+      "step": 1377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009355927640168154,
+      "loss": 2.6141,
+      "step": 1378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000935483934583483,
+      "loss": 2.5374,
+      "step": 1379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353750196234092,
+      "loss": 2.4861,
+      "step": 1380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009352660191579849,
+      "loss": 2.5837,
+      "step": 1381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009351569332086171,
+      "loss": 2.5572,
+      "step": 1382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350477617967295,
+      "loss": 2.5613,
+      "step": 1383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009349385049437629,
+      "loss": 2.5505,
+      "step": 1384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009348291626711747,
+      "loss": 2.7324,
+      "step": 1385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009347197350004393,
+      "loss": 2.6618,
+      "step": 1386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009346102219530476,
+      "loss": 2.7436,
+      "step": 1387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009345006235505076,
+      "loss": 2.6648,
+      "step": 1388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009343909398143433,
+      "loss": 2.564,
+      "step": 1389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009342811707660966,
+      "loss": 2.6511,
+      "step": 1390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009341713164273253,
+      "loss": 2.5274,
+      "step": 1391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009340613768196042,
+      "loss": 2.4894,
+      "step": 1392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009339513519645248,
+      "loss": 2.5909,
+      "step": 1393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009338412418836957,
+      "loss": 2.6448,
+      "step": 1394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009337310465987416,
+      "loss": 2.5533,
+      "step": 1395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009336207661313045,
+      "loss": 2.6237,
+      "step": 1396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009335104005030425,
+      "loss": 2.4425,
+      "step": 1397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009333999497356314,
+      "loss": 2.5438,
+      "step": 1398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009332894138507628,
+      "loss": 2.4882,
+      "step": 1399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009331787928701455,
+      "loss": 2.5342,
+      "step": 1400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009330680868155048,
+      "loss": 2.5165,
+      "step": 1401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009329572957085827,
+      "loss": 2.5841,
+      "step": 1402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009328464195711381,
+      "loss": 2.5997,
+      "step": 1403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327354584249465,
+      "loss": 2.5247,
+      "step": 1404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326244122918,
+      "loss": 2.4094,
+      "step": 1405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009325132811935075,
+      "loss": 2.5467,
+      "step": 1406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009324020651518948,
+      "loss": 2.5283,
+      "step": 1407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322907641888037,
+      "loss": 2.616,
+      "step": 1408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009321793783260934,
+      "loss": 2.4909,
+      "step": 1409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009320679075856395,
+      "loss": 2.5091,
+      "step": 1410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009319563519893341,
+      "loss": 2.5515,
+      "step": 1411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009318447115590864,
+      "loss": 2.4538,
+      "step": 1412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317329863168217,
+      "loss": 2.6048,
+      "step": 1413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009316211762844824,
+      "loss": 2.5744,
+      "step": 1414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009315092814840273,
+      "loss": 2.5636,
+      "step": 1415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931397301937432,
+      "loss": 2.5198,
+      "step": 1416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931285237666689,
+      "loss": 2.4129,
+      "step": 1417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009311730886938064,
+      "loss": 2.4155,
+      "step": 1418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009310608550408104,
+      "loss": 2.463,
+      "step": 1419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009309485367297427,
+      "loss": 2.6226,
+      "step": 1420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000930836133782662,
+      "loss": 2.4944,
+      "step": 1421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000930723646221644,
+      "loss": 2.6588,
+      "step": 1422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009306110740687803,
+      "loss": 2.5804,
+      "step": 1423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009304984173461796,
+      "loss": 2.5517,
+      "step": 1424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009303856760759671,
+      "loss": 2.6828,
+      "step": 1425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009302728502802845,
+      "loss": 2.6047,
+      "step": 1426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301599399812904,
+      "loss": 2.4898,
+      "step": 1427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009300469452011595,
+      "loss": 2.4888,
+      "step": 1428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009299338659620836,
+      "loss": 2.6169,
+      "step": 1429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009298207022862708,
+      "loss": 2.5717,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297074541959458,
+      "loss": 2.5192,
+      "step": 1431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00092959412171335,
+      "loss": 2.4818,
+      "step": 1432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009294807048607412,
+      "loss": 2.6192,
+      "step": 1433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000929367203660394,
+      "loss": 2.6659,
+      "step": 1434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009292536181345994,
+      "loss": 2.7339,
+      "step": 1435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009291399483056648,
+      "loss": 2.5439,
+      "step": 1436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009290261941959146,
+      "loss": 2.6092,
+      "step": 1437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009289123558276895,
+      "loss": 2.6237,
+      "step": 1438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287984332233466,
+      "loss": 2.5943,
+      "step": 1439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009286844264052599,
+      "loss": 2.5187,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009285703353958195,
+      "loss": 2.4741,
+      "step": 1441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009284561602174324,
+      "loss": 2.3556,
+      "step": 1442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009283419008925218,
+      "loss": 2.469,
+      "step": 1443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009282275574435281,
+      "loss": 2.4239,
+      "step": 1444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009281131298929074,
+      "loss": 2.5685,
+      "step": 1445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009279986182631325,
+      "loss": 2.5347,
+      "step": 1446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009278840225766932,
+      "loss": 2.6405,
+      "step": 1447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009277693428560953,
+      "loss": 2.5043,
+      "step": 1448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009276545791238618,
+      "loss": 2.5845,
+      "step": 1449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009275397314025309,
+      "loss": 2.5087,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009274247997146585,
+      "loss": 2.5416,
+      "step": 1451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009273097840828165,
+      "loss": 2.5337,
+      "step": 1452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009271946845295934,
+      "loss": 2.6427,
+      "step": 1453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270795010775943,
+      "loss": 2.7211,
+      "step": 1454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009269642337494401,
+      "loss": 2.4151,
+      "step": 1455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009268488825677695,
+      "loss": 2.4926,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926733447555236,
+      "loss": 2.4339,
+      "step": 1457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009266179287345111,
+      "loss": 2.6819,
+      "step": 1458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009265023261282817,
+      "loss": 2.5789,
+      "step": 1459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009263866397592517,
+      "loss": 2.7057,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009262708696501412,
+      "loss": 2.525,
+      "step": 1461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926155015823687,
+      "loss": 2.6703,
+      "step": 1462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926039078302642,
+      "loss": 2.4888,
+      "step": 1463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009259230571097757,
+      "loss": 2.3839,
+      "step": 1464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258069522678743,
+      "loss": 2.5614,
+      "step": 1465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009256907637997398,
+      "loss": 2.6886,
+      "step": 1466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009255744917281914,
+      "loss": 2.5061,
+      "step": 1467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925458136076064,
+      "loss": 2.6564,
+      "step": 1468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009253416968662094,
+      "loss": 2.5107,
+      "step": 1469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009252251741214957,
+      "loss": 2.5866,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009251085678648072,
+      "loss": 2.5304,
+      "step": 1471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009249918781190447,
+      "loss": 2.5951,
+      "step": 1472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009248751049071255,
+      "loss": 2.4193,
+      "step": 1473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247582482519837,
+      "loss": 2.4874,
+      "step": 1474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009246413081765686,
+      "loss": 2.5104,
+      "step": 1475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009245242847038469,
+      "loss": 2.5632,
+      "step": 1476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009244071778568015,
+      "loss": 2.6109,
+      "step": 1477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009242899876584316,
+      "loss": 2.6666,
+      "step": 1478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009241727141317527,
+      "loss": 2.727,
+      "step": 1479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009240553572997963,
+      "loss": 2.663,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009239379171856113,
+      "loss": 2.5527,
+      "step": 1481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009238203938122618,
+      "loss": 2.7052,
+      "step": 1482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009237027872028292,
+      "loss": 2.621,
+      "step": 1483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009235850973804104,
+      "loss": 2.5609,
+      "step": 1484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009234673243681194,
+      "loss": 2.5976,
+      "step": 1485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000923349468189086,
+      "loss": 2.5384,
+      "step": 1486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232315288664564,
+      "loss": 2.5502,
+      "step": 1487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009231135064233935,
+      "loss": 2.5247,
+      "step": 1488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009229954008830764,
+      "loss": 2.4872,
+      "step": 1489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009228772122687001,
+      "loss": 2.5593,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009227589406034764,
+      "loss": 2.4817,
+      "step": 1491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009226405859106332,
+      "loss": 2.668,
+      "step": 1492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009225221482134147,
+      "loss": 2.5292,
+      "step": 1493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009224036275350813,
+      "loss": 2.5389,
+      "step": 1494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009222850238989102,
+      "loss": 2.5152,
+      "step": 1495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009221663373281942,
+      "loss": 2.5022,
+      "step": 1496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009220475678462427,
+      "loss": 2.64,
+      "step": 1497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009219287154763818,
+      "loss": 2.4247,
+      "step": 1498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921809780241953,
+      "loss": 2.6964,
+      "step": 1499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009216907621663147,
+      "loss": 2.5631,
+      "step": 1500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009215716612728416,
+      "loss": 2.4999,
+      "step": 1501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009214524775849243,
+      "loss": 2.5807,
+      "step": 1502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00092133321112597,
+      "loss": 2.468,
+      "step": 1503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009212138619194017,
+      "loss": 2.6831,
+      "step": 1504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009210944299886594,
+      "loss": 2.4534,
+      "step": 1505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009209749153571984,
+      "loss": 2.6345,
+      "step": 1506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009208553180484911,
+      "loss": 2.5088,
+      "step": 1507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009207356380860256,
+      "loss": 2.5291,
+      "step": 1508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009206158754933067,
+      "loss": 2.5034,
+      "step": 1509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009204960302938548,
+      "loss": 2.462,
+      "step": 1510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000920376102511207,
+      "loss": 2.5952,
+      "step": 1511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009202560921689165,
+      "loss": 2.5315,
+      "step": 1512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009201359992905525,
+      "loss": 2.5756,
+      "step": 1513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000920015823899701,
+      "loss": 2.4727,
+      "step": 1514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198955660199634,
+      "loss": 2.5658,
+      "step": 1515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919775225674958,
+      "loss": 2.5241,
+      "step": 1516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009196548028883189,
+      "loss": 2.5641,
+      "step": 1517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009195342976836964,
+      "loss": 2.5196,
+      "step": 1518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009194137100847573,
+      "loss": 2.5898,
+      "step": 1519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009192930401151842,
+      "loss": 2.5706,
+      "step": 1520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191722877986762,
+      "loss": 2.5217,
+      "step": 1521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919051453158948,
+      "loss": 2.5269,
+      "step": 1522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009189305362197316,
+      "loss": 2.4255,
+      "step": 1523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009188095370047738,
+      "loss": 2.6643,
+      "step": 1524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009186884555378384,
+      "loss": 2.455,
+      "step": 1525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009185672918427053,
+      "loss": 2.5979,
+      "step": 1526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009184460459431702,
+      "loss": 2.71,
+      "step": 1527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009183247178630455,
+      "loss": 2.3507,
+      "step": 1528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009182033076261591,
+      "loss": 2.5615,
+      "step": 1529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009180818152563552,
+      "loss": 2.5508,
+      "step": 1530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009179602407774947,
+      "loss": 2.4866,
+      "step": 1531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009178385842134538,
+      "loss": 2.651,
+      "step": 1532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009177168455881254,
+      "loss": 2.3857,
+      "step": 1533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175950249254183,
+      "loss": 2.4479,
+      "step": 1534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009174731222492573,
+      "loss": 2.5381,
+      "step": 1535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009173511375835835,
+      "loss": 2.6982,
+      "step": 1536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009172290709523542,
+      "loss": 2.5032,
+      "step": 1537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009171069223795424,
+      "loss": 2.627,
+      "step": 1538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009169846918891376,
+      "loss": 2.7959,
+      "step": 1539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009168623795051451,
+      "loss": 2.6547,
+      "step": 1540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009167399852515868,
+      "loss": 2.6253,
+      "step": 1541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009166175091524997,
+      "loss": 2.4522,
+      "step": 1542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009164949512319377,
+      "loss": 2.5448,
+      "step": 1543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009163723115139706,
+      "loss": 2.4917,
+      "step": 1544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009162495900226842,
+      "loss": 2.4697,
+      "step": 1545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009161267867821802,
+      "loss": 2.3172,
+      "step": 1546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009160039018165767,
+      "loss": 2.6264,
+      "step": 1547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009158809351500074,
+      "loss": 2.5316,
+      "step": 1548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009157578868066225,
+      "loss": 2.6721,
+      "step": 1549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009156347568105881,
+      "loss": 2.5188,
+      "step": 1550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009155115451860863,
+      "loss": 2.5508,
+      "step": 1551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009153882519573149,
+      "loss": 2.5271,
+      "step": 1552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009152648771484881,
+      "loss": 2.4284,
+      "step": 1553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009151414207838364,
+      "loss": 2.5928,
+      "step": 1554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009150178828876058,
+      "loss": 2.5141,
+      "step": 1555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009148942634840584,
+      "loss": 2.5621,
+      "step": 1556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009147705625974725,
+      "loss": 2.4195,
+      "step": 1557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009146467802521421,
+      "loss": 2.5692,
+      "step": 1558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009145229164723778,
+      "loss": 2.5869,
+      "step": 1559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009143989712825053,
+      "loss": 2.5966,
+      "step": 1560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009142749447068671,
+      "loss": 2.4702,
+      "step": 1561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009141508367698213,
+      "loss": 2.5779,
+      "step": 1562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914026647495742,
+      "loss": 2.6277,
+      "step": 1563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009139023769090195,
+      "loss": 2.5906,
+      "step": 1564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009137780250340595,
+      "loss": 2.547,
+      "step": 1565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009136535918952844,
+      "loss": 2.8705,
+      "step": 1566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009135290775171321,
+      "loss": 2.4814,
+      "step": 1567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009134044819240566,
+      "loss": 2.4925,
+      "step": 1568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009132798051405277,
+      "loss": 2.6059,
+      "step": 1569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009131550471910314,
+      "loss": 2.6538,
+      "step": 1570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009130302081000694,
+      "loss": 2.6473,
+      "step": 1571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009129052878921595,
+      "loss": 2.5408,
+      "step": 1572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009127802865918353,
+      "loss": 2.3946,
+      "step": 1573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009126552042236464,
+      "loss": 2.7446,
+      "step": 1574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009125300408121584,
+      "loss": 2.5895,
+      "step": 1575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009124047963819526,
+      "loss": 2.5662,
+      "step": 1576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009122794709576265,
+      "loss": 2.6426,
+      "step": 1577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009121540645637933,
+      "loss": 2.7835,
+      "step": 1578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009120285772250821,
+      "loss": 2.646,
+      "step": 1579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000911903008966138,
+      "loss": 2.684,
+      "step": 1580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009117773598116219,
+      "loss": 2.6761,
+      "step": 1581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009116516297862107,
+      "loss": 2.5583,
+      "step": 1582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000911525818914597,
+      "loss": 2.5359,
+      "step": 1583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113999272214895,
+      "loss": 2.5142,
+      "step": 1584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009112739547316124,
+      "loss": 2.6148,
+      "step": 1585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009111479014697065,
+      "loss": 2.4566,
+      "step": 1586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110217674605277,
+      "loss": 2.6042,
+      "step": 1587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009108955527288479,
+      "loss": 2.5221,
+      "step": 1588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009107692572994554,
+      "loss": 2.5347,
+      "step": 1589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009106428811971537,
+      "loss": 2.5483,
+      "step": 1590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009105164244467623,
+      "loss": 2.5773,
+      "step": 1591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000910389887073117,
+      "loss": 2.4872,
+      "step": 1592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009102632691010688,
+      "loss": 2.4701,
+      "step": 1593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009101365705554847,
+      "loss": 2.6173,
+      "step": 1594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009100097914612479,
+      "loss": 2.7601,
+      "step": 1595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009098829318432571,
+      "loss": 2.6777,
+      "step": 1596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009097559917264266,
+      "loss": 2.4623,
+      "step": 1597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009096289711356871,
+      "loss": 2.6251,
+      "step": 1598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009095018700959845,
+      "loss": 2.6745,
+      "step": 1599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000909374688632281,
+      "loss": 2.5492,
+      "step": 1600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009092474267695543,
+      "loss": 2.6277,
+      "step": 1601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009091200845327978,
+      "loss": 2.6148,
+      "step": 1602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908992661947021,
+      "loss": 2.416,
+      "step": 1603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009088651590372489,
+      "loss": 2.4851,
+      "step": 1604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009087375758285225,
+      "loss": 2.6175,
+      "step": 1605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009086099123458985,
+      "loss": 2.6871,
+      "step": 1606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009084821686144493,
+      "loss": 2.5346,
+      "step": 1607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908354344659263,
+      "loss": 2.6242,
+      "step": 1608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009082264405054437,
+      "loss": 2.6429,
+      "step": 1609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009080984561781109,
+      "loss": 2.5819,
+      "step": 1610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009079703917024002,
+      "loss": 2.5628,
+      "step": 1611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009078422471034627,
+      "loss": 2.55,
+      "step": 1612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009077140224064655,
+      "loss": 2.5859,
+      "step": 1613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000907585717636591,
+      "loss": 2.5239,
+      "step": 1614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009074573328190377,
+      "loss": 2.6013,
+      "step": 1615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009073288679790197,
+      "loss": 2.6266,
+      "step": 1616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000907200323141767,
+      "loss": 2.6105,
+      "step": 1617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009070716983325247,
+      "loss": 2.6889,
+      "step": 1618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009069429935765542,
+      "loss": 2.4946,
+      "step": 1619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009068142088991329,
+      "loss": 2.5472,
+      "step": 1620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009066853443255527,
+      "loss": 2.5108,
+      "step": 1621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065563998811223,
+      "loss": 2.5931,
+      "step": 1622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009064273755911658,
+      "loss": 2.3652,
+      "step": 1623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009062982714810226,
+      "loss": 2.7756,
+      "step": 1624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009061690875760483,
+      "loss": 2.5467,
+      "step": 1625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009060398239016139,
+      "loss": 2.7498,
+      "step": 1626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009059104804831059,
+      "loss": 2.6172,
+      "step": 1627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009057810573459269,
+      "loss": 2.5026,
+      "step": 1628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009056515545154948,
+      "loss": 2.5205,
+      "step": 1629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009055219720172432,
+      "loss": 2.6418,
+      "step": 1630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009053923098766217,
+      "loss": 2.5311,
+      "step": 1631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052625681190951,
+      "loss": 2.5784,
+      "step": 1632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009051327467701438,
+      "loss": 2.5663,
+      "step": 1633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009050028458552642,
+      "loss": 2.6848,
+      "step": 1634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009048728653999683,
+      "loss": 2.6014,
+      "step": 1635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009047428054297834,
+      "loss": 2.5502,
+      "step": 1636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046126659702525,
+      "loss": 2.4932,
+      "step": 1637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009044824470469345,
+      "loss": 2.4455,
+      "step": 1638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009043521486854036,
+      "loss": 2.6012,
+      "step": 1639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009042217709112496,
+      "loss": 2.6341,
+      "step": 1640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040913137500784,
+      "loss": 2.5505,
+      "step": 1641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009039607772275107,
+      "loss": 2.5443,
+      "step": 1642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009038301613691835,
+      "loss": 2.6387,
+      "step": 1643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009036994662007487,
+      "loss": 2.5698,
+      "step": 1644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009035686917478743,
+      "loss": 2.5874,
+      "step": 1645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009034378380362438,
+      "loss": 2.381,
+      "step": 1646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009033069050915562,
+      "loss": 2.4305,
+      "step": 1647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009031758929395257,
+      "loss": 2.5653,
+      "step": 1648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009030448016058828,
+      "loss": 2.5055,
+      "step": 1649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009029136311163729,
+      "loss": 2.6391,
+      "step": 1650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009027823814967573,
+      "loss": 2.5421,
+      "step": 1651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009026510527728128,
+      "loss": 2.5732,
+      "step": 1652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009025196449703314,
+      "loss": 2.5068,
+      "step": 1653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009023881581151212,
+      "loss": 2.6481,
+      "step": 1654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009022565922330055,
+      "loss": 2.429,
+      "step": 1655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009021249473498227,
+      "loss": 2.4222,
+      "step": 1656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009019932234914279,
+      "loss": 2.5379,
+      "step": 1657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009018614206836904,
+      "loss": 2.5736,
+      "step": 1658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009017295389524956,
+      "loss": 2.6894,
+      "step": 1659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009015975783237447,
+      "loss": 2.6925,
+      "step": 1660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009014655388233537,
+      "loss": 2.5209,
+      "step": 1661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009013334204772547,
+      "loss": 2.5623,
+      "step": 1662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009012012233113951,
+      "loss": 2.4941,
+      "step": 1663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009010689473517375,
+      "loss": 2.6103,
+      "step": 1664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009009365926242603,
+      "loss": 2.4182,
+      "step": 1665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009008041591549574,
+      "loss": 2.6239,
+      "step": 1666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009006716469698378,
+      "loss": 2.6262,
+      "step": 1667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009005390560949262,
+      "loss": 2.5492,
+      "step": 1668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009004063865562631,
+      "loss": 2.5256,
+      "step": 1669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009002736383799036,
+      "loss": 2.6144,
+      "step": 1670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009001408115919192,
+      "loss": 2.5433,
+      "step": 1671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009000079062183959,
+      "loss": 2.5466,
+      "step": 1672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008998749222854361,
+      "loss": 2.5913,
+      "step": 1673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008997418598191569,
+      "loss": 2.4818,
+      "step": 1674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899608718845691,
+      "loss": 2.6857,
+      "step": 1675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008994754993911867,
+      "loss": 2.6969,
+      "step": 1676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008993422014818075,
+      "loss": 2.628,
+      "step": 1677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008992088251437325,
+      "loss": 2.4814,
+      "step": 1678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899075370403156,
+      "loss": 2.5075,
+      "step": 1679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000898941837286288,
+      "loss": 2.5392,
+      "step": 1680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008988082258193535,
+      "loss": 2.3371,
+      "step": 1681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008986745360285932,
+      "loss": 2.5012,
+      "step": 1682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008985407679402631,
+      "loss": 2.509,
+      "step": 1683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008984069215806343,
+      "loss": 2.5904,
+      "step": 1684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008982729969759937,
+      "loss": 2.4125,
+      "step": 1685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008981389941526435,
+      "loss": 2.4171,
+      "step": 1686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008980049131369008,
+      "loss": 2.5834,
+      "step": 1687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008978707539550989,
+      "loss": 2.4248,
+      "step": 1688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008977365166335854,
+      "loss": 2.4481,
+      "step": 1689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008976022011987244,
+      "loss": 2.5533,
+      "step": 1690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008974678076768941,
+      "loss": 2.4718,
+      "step": 1691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008973333360944891,
+      "loss": 2.3477,
+      "step": 1692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971987864779189,
+      "loss": 2.5197,
+      "step": 1693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897064158853608,
+      "loss": 2.4321,
+      "step": 1694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000896929453247997,
+      "loss": 2.4774,
+      "step": 1695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008967946696875411,
+      "loss": 2.5492,
+      "step": 1696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008966598081987109,
+      "loss": 2.5208,
+      "step": 1697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000896524868807993,
+      "loss": 2.6394,
+      "step": 1698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008963898515418885,
+      "loss": 2.4572,
+      "step": 1699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008962547564269139,
+      "loss": 2.5262,
+      "step": 1700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008961195834896014,
+      "loss": 2.5047,
+      "step": 1701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008959843327564983,
+      "loss": 2.5192,
+      "step": 1702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008958490042541669,
+      "loss": 2.6291,
+      "step": 1703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000895713598009185,
+      "loss": 2.6031,
+      "step": 1704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000895578114048146,
+      "loss": 2.6939,
+      "step": 1705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000895442552397658,
+      "loss": 2.4759,
+      "step": 1706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008953069130843446,
+      "loss": 2.4833,
+      "step": 1707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008951711961348445,
+      "loss": 2.5586,
+      "step": 1708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008950354015758122,
+      "loss": 2.4943,
+      "step": 1709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008948995294339166,
+      "loss": 2.3784,
+      "step": 1710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008947635797358425,
+      "loss": 2.6008,
+      "step": 1711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008946275525082896,
+      "loss": 2.4608,
+      "step": 1712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008944914477779731,
+      "loss": 2.6124,
+      "step": 1713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000894355265571623,
+      "loss": 2.6033,
+      "step": 1714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000894219005915985,
+      "loss": 2.4977,
+      "step": 1715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008940826688378196,
+      "loss": 2.6175,
+      "step": 1716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008939462543639027,
+      "loss": 2.5801,
+      "step": 1717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008938097625210256,
+      "loss": 2.5698,
+      "step": 1718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008936731933359942,
+      "loss": 2.526,
+      "step": 1719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008935365468356304,
+      "loss": 2.4538,
+      "step": 1720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933998230467704,
+      "loss": 2.6776,
+      "step": 1721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008932630219962664,
+      "loss": 2.5834,
+      "step": 1722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008931261437109852,
+      "loss": 2.5565,
+      "step": 1723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000892989188217809,
+      "loss": 2.5333,
+      "step": 1724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008928521555436353,
+      "loss": 2.3316,
+      "step": 1725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008927150457153763,
+      "loss": 2.647,
+      "step": 1726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008925778587599598,
+      "loss": 2.5207,
+      "step": 1727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008924405947043286,
+      "loss": 2.3682,
+      "step": 1728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923032535754407,
+      "loss": 2.4226,
+      "step": 1729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000892165835400269,
+      "loss": 2.5926,
+      "step": 1730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008920283402058019,
+      "loss": 2.4293,
+      "step": 1731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008918907680190426,
+      "loss": 2.5167,
+      "step": 1732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008917531188670095,
+      "loss": 2.3711,
+      "step": 1733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008916153927767363,
+      "loss": 2.5143,
+      "step": 1734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008914775897752716,
+      "loss": 2.3578,
+      "step": 1735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913397098896792,
+      "loss": 2.5966,
+      "step": 1736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000891201753147038,
+      "loss": 2.348,
+      "step": 1737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000891063719574442,
+      "loss": 2.5814,
+      "step": 1738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008909256091990001,
+      "loss": 2.5079,
+      "step": 1739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008907874220478365,
+      "loss": 2.4858,
+      "step": 1740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008906491581480906,
+      "loss": 2.4115,
+      "step": 1741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008905108175269167,
+      "loss": 2.6508,
+      "step": 1742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008903724002114839,
+      "loss": 2.5534,
+      "step": 1743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008902339062289768,
+      "loss": 2.424,
+      "step": 1744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008900953356065949,
+      "loss": 2.3881,
+      "step": 1745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008899566883715526,
+      "loss": 2.4524,
+      "step": 1746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008898179645510797,
+      "loss": 2.441,
+      "step": 1747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008896791641724207,
+      "loss": 2.435,
+      "step": 1748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008895402872628352,
+      "loss": 2.3852,
+      "step": 1749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008894013338495981,
+      "loss": 2.4965,
+      "step": 1750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008892623039599988,
+      "loss": 2.5289,
+      "step": 1751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008891231976213423,
+      "loss": 2.5419,
+      "step": 1752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008889840148609483,
+      "loss": 2.4945,
+      "step": 1753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008888447557061515,
+      "loss": 2.4236,
+      "step": 1754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008887054201843017,
+      "loss": 2.462,
+      "step": 1755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008885660083227636,
+      "loss": 2.4714,
+      "step": 1756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000888426520148917,
+      "loss": 2.5187,
+      "step": 1757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008882869556901566,
+      "loss": 2.5631,
+      "step": 1758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008881473149738921,
+      "loss": 2.4551,
+      "step": 1759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008880075980275484,
+      "loss": 2.5032,
+      "step": 1760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000887867804878565,
+      "loss": 2.5178,
+      "step": 1761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008877279355543966,
+      "loss": 2.5567,
+      "step": 1762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008875879900825125,
+      "loss": 2.4227,
+      "step": 1763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008874479684903977,
+      "loss": 2.4831,
+      "step": 1764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008873078708055515,
+      "loss": 2.1996,
+      "step": 1765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008871676970554883,
+      "loss": 2.604,
+      "step": 1766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008870274472677376,
+      "loss": 2.5114,
+      "step": 1767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008868871214698436,
+      "loss": 2.6251,
+      "step": 1768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008867467196893655,
+      "loss": 2.5463,
+      "step": 1769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008866062419538776,
+      "loss": 2.5561,
+      "step": 1770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008864656882909689,
+      "loss": 2.5847,
+      "step": 1771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008863250587282434,
+      "loss": 2.3572,
+      "step": 1772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008861843532933202,
+      "loss": 2.5142,
+      "step": 1773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008860435720138329,
+      "loss": 2.4837,
+      "step": 1774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008859027149174302,
+      "loss": 2.2456,
+      "step": 1775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008857617820317757,
+      "loss": 2.4758,
+      "step": 1776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008856207733845481,
+      "loss": 2.6075,
+      "step": 1777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008854796890034402,
+      "loss": 2.6712,
+      "step": 1778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008853385289161608,
+      "loss": 2.5337,
+      "step": 1779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008851972931504329,
+      "loss": 2.5951,
+      "step": 1780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008850559817339941,
+      "loss": 2.5115,
+      "step": 1781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008849145946945976,
+      "loss": 2.4495,
+      "step": 1782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008847731320600108,
+      "loss": 2.4596,
+      "step": 1783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008846315938580162,
+      "loss": 2.4369,
+      "step": 1784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844899801164116,
+      "loss": 2.5148,
+      "step": 1785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008843482908630087,
+      "loss": 2.56,
+      "step": 1786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008842065261256347,
+      "loss": 2.4684,
+      "step": 1787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008840646859321315,
+      "loss": 2.5301,
+      "step": 1788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008839227703103555,
+      "loss": 2.4397,
+      "step": 1789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008837807792881785,
+      "loss": 2.4783,
+      "step": 1790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008836387128934866,
+      "loss": 2.488,
+      "step": 1791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008834965711541811,
+      "loss": 2.5294,
+      "step": 1792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008833543540981777,
+      "loss": 2.4528,
+      "step": 1793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000883212061753407,
+      "loss": 2.5178,
+      "step": 1794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008830696941478146,
+      "loss": 2.5143,
+      "step": 1795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882927251309361,
+      "loss": 2.6777,
+      "step": 1796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008827847332660206,
+      "loss": 2.4602,
+      "step": 1797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008826421400457836,
+      "loss": 2.6114,
+      "step": 1798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008824994716766545,
+      "loss": 2.6677,
+      "step": 1799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008823567281866526,
+      "loss": 2.4178,
+      "step": 1800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008822139096038119,
+      "loss": 2.5885,
+      "step": 1801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008820710159561814,
+      "loss": 2.6608,
+      "step": 1802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008819280472718245,
+      "loss": 2.6237,
+      "step": 1803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817850035788195,
+      "loss": 2.4061,
+      "step": 1804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008816418849052594,
+      "loss": 2.4512,
+      "step": 1805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008814986912792521,
+      "loss": 2.4324,
+      "step": 1806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00088135542272892,
+      "loss": 2.6788,
+      "step": 1807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008812120792824001,
+      "loss": 2.4346,
+      "step": 1808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008810686609678446,
+      "loss": 2.5085,
+      "step": 1809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809251678134199,
+      "loss": 2.558,
+      "step": 1810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008807815998473074,
+      "loss": 2.6676,
+      "step": 1811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000880637957097703,
+      "loss": 2.6124,
+      "step": 1812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008804942395928173,
+      "loss": 2.3013,
+      "step": 1813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008803504473608759,
+      "loss": 2.4991,
+      "step": 1814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008802065804301187,
+      "loss": 2.6263,
+      "step": 1815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800626388288003,
+      "loss": 2.5225,
+      "step": 1816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008799186225851903,
+      "loss": 2.6643,
+      "step": 1817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008797745317275726,
+      "loss": 2.6496,
+      "step": 1818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008796303662842457,
+      "loss": 2.521,
+      "step": 1819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008794861262835234,
+      "loss": 2.524,
+      "step": 1820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008793418117537333,
+      "loss": 2.5801,
+      "step": 1821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000879197422723218,
+      "loss": 2.6525,
+      "step": 1822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000879052959220335,
+      "loss": 2.5394,
+      "step": 1823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000878908421273456,
+      "loss": 2.5214,
+      "step": 1824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008787638089109673,
+      "loss": 2.4773,
+      "step": 1825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008786191221612702,
+      "loss": 2.4845,
+      "step": 1826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008784743610527806,
+      "loss": 2.4906,
+      "step": 1827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008783295256139285,
+      "loss": 2.4355,
+      "step": 1828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008781846158731592,
+      "loss": 2.5113,
+      "step": 1829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008780396318589317,
+      "loss": 2.5426,
+      "step": 1830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008778945735997204,
+      "loss": 2.558,
+      "step": 1831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008777494411240139,
+      "loss": 2.5806,
+      "step": 1832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008776042344603155,
+      "loss": 2.5492,
+      "step": 1833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008774589536371429,
+      "loss": 2.5179,
+      "step": 1834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008773135986830287,
+      "loss": 2.6773,
+      "step": 1835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008771681696265196,
+      "loss": 2.5029,
+      "step": 1836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008770226664961773,
+      "loss": 2.6339,
+      "step": 1837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008768770893205779,
+      "loss": 2.4997,
+      "step": 1838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008767314381283118,
+      "loss": 2.4398,
+      "step": 1839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000876585712947984,
+      "loss": 2.5465,
+      "step": 1840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008764399138082144,
+      "loss": 2.5293,
+      "step": 1841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762940407376372,
+      "loss": 2.4119,
+      "step": 1842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008761480937649009,
+      "loss": 2.3839,
+      "step": 1843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008760020729186689,
+      "loss": 2.4882,
+      "step": 1844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008758559782276187,
+      "loss": 2.3858,
+      "step": 1845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008757098097204428,
+      "loss": 2.4849,
+      "step": 1846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008755635674258476,
+      "loss": 2.5071,
+      "step": 1847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008754172513725547,
+      "loss": 2.4995,
+      "step": 1848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008752708615892996,
+      "loss": 2.4809,
+      "step": 1849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008751243981048323,
+      "loss": 2.555,
+      "step": 1850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008749778609479178,
+      "loss": 2.5345,
+      "step": 1851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000874831250147335,
+      "loss": 2.5037,
+      "step": 1852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008746845657318776,
+      "loss": 2.45,
+      "step": 1853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008745378077303535,
+      "loss": 2.3962,
+      "step": 1854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008743909761715854,
+      "loss": 2.386,
+      "step": 1855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00087424407108441,
+      "loss": 2.719,
+      "step": 1856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008740970924976788,
+      "loss": 2.5998,
+      "step": 1857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008739500404402577,
+      "loss": 2.5549,
+      "step": 1858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008738029149410267,
+      "loss": 2.6825,
+      "step": 1859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008736557160288807,
+      "loss": 2.5916,
+      "step": 1860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735084437327288,
+      "loss": 2.492,
+      "step": 1861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008733610980814944,
+      "loss": 2.4697,
+      "step": 1862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008732136791041155,
+      "loss": 2.6003,
+      "step": 1863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008730661868295442,
+      "loss": 2.3711,
+      "step": 1864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008729186212867475,
+      "loss": 2.5136,
+      "step": 1865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008727709825047061,
+      "loss": 2.6008,
+      "step": 1866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008726232705124158,
+      "loss": 2.4536,
+      "step": 1867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008724754853388865,
+      "loss": 2.5207,
+      "step": 1868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008723276270131421,
+      "loss": 2.3743,
+      "step": 1869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008721796955642215,
+      "loss": 2.3625,
+      "step": 1870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008720316910211776,
+      "loss": 2.4297,
+      "step": 1871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008718836134130774,
+      "loss": 2.5797,
+      "step": 1872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008717354627690029,
+      "loss": 2.32,
+      "step": 1873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008715872391180502,
+      "loss": 2.3695,
+      "step": 1874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008714389424893292,
+      "loss": 2.5682,
+      "step": 1875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008712905729119649,
+      "loss": 2.5044,
+      "step": 1876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008711421304150962,
+      "loss": 2.4911,
+      "step": 1877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008709936150278765,
+      "loss": 2.3395,
+      "step": 1878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008708450267794734,
+      "loss": 2.3825,
+      "step": 1879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008706963656990687,
+      "loss": 2.3493,
+      "step": 1880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008705476318158588,
+      "loss": 2.5973,
+      "step": 1881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008703988251590543,
+      "loss": 2.5117,
+      "step": 1882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00087024994575788,
+      "loss": 2.4853,
+      "step": 1883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008701009936415749,
+      "loss": 2.661,
+      "step": 1884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008699519688393924,
+      "loss": 2.3602,
+      "step": 1885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008698028713806004,
+      "loss": 2.5624,
+      "step": 1886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008696537012944808,
+      "loss": 2.4597,
+      "step": 1887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008695044586103295,
+      "loss": 2.5487,
+      "step": 1888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008693551433574575,
+      "loss": 2.4405,
+      "step": 1889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008692057555651891,
+      "loss": 2.5174,
+      "step": 1890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008690562952628633,
+      "loss": 2.4458,
+      "step": 1891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008689067624798337,
+      "loss": 2.5367,
+      "step": 1892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008687571572454674,
+      "loss": 2.6302,
+      "step": 1893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000868607479589146,
+      "loss": 2.4314,
+      "step": 1894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008684577295402657,
+      "loss": 2.439,
+      "step": 1895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008683079071282365,
+      "loss": 2.4273,
+      "step": 1896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008681580123824825,
+      "loss": 2.4939,
+      "step": 1897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008680080453324426,
+      "loss": 2.4672,
+      "step": 1898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008678580060075693,
+      "loss": 2.6047,
+      "step": 1899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008677078944373295,
+      "loss": 2.6015,
+      "step": 1900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008675577106512047,
+      "loss": 2.4862,
+      "step": 1901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008674074546786897,
+      "loss": 2.4054,
+      "step": 1902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008672571265492943,
+      "loss": 2.4736,
+      "step": 1903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000867106726292542,
+      "loss": 2.3328,
+      "step": 1904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008669562539379707,
+      "loss": 2.3896,
+      "step": 1905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008668057095151323,
+      "loss": 2.4964,
+      "step": 1906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008666550930535931,
+      "loss": 2.6187,
+      "step": 1907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008665044045829331,
+      "loss": 2.5515,
+      "step": 1908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008663536441327469,
+      "loss": 2.521,
+      "step": 1909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000866202811732643,
+      "loss": 2.4461,
+      "step": 1910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008660519074122439,
+      "loss": 2.3575,
+      "step": 1911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000865900931201187,
+      "loss": 2.4053,
+      "step": 1912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008657498831291225,
+      "loss": 2.6542,
+      "step": 1913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008655987632257161,
+      "loss": 2.3922,
+      "step": 1914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008654475715206463,
+      "loss": 2.4628,
+      "step": 1915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008652963080436069,
+      "loss": 2.4556,
+      "step": 1916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008651449728243049,
+      "loss": 2.3317,
+      "step": 1917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008649935658924619,
+      "loss": 2.6442,
+      "step": 1918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008648420872778136,
+      "loss": 2.4617,
+      "step": 1919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000864690537010109,
+      "loss": 2.5134,
+      "step": 1920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008645389151191125,
+      "loss": 2.3719,
+      "step": 1921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008643872216346014,
+      "loss": 2.4253,
+      "step": 1922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008642354565863674,
+      "loss": 2.4452,
+      "step": 1923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008640836200042168,
+      "loss": 2.3999,
+      "step": 1924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008639317119179691,
+      "loss": 2.6056,
+      "step": 1925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008637797323574585,
+      "loss": 2.493,
+      "step": 1926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008636276813525328,
+      "loss": 2.4346,
+      "step": 1927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000863475558933054,
+      "loss": 2.4129,
+      "step": 1928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008633233651288983,
+      "loss": 2.4307,
+      "step": 1929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008631710999699556,
+      "loss": 2.3873,
+      "step": 1930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008630187634861302,
+      "loss": 2.4487,
+      "step": 1931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00086286635570734,
+      "loss": 2.3842,
+      "step": 1932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862713876663517,
+      "loss": 2.4447,
+      "step": 1933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008625613263846076,
+      "loss": 2.4842,
+      "step": 1934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008624087049005716,
+      "loss": 2.6333,
+      "step": 1935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862256012241383,
+      "loss": 2.6403,
+      "step": 1936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008621032484370299,
+      "loss": 2.2833,
+      "step": 1937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008619504135175145,
+      "loss": 2.4725,
+      "step": 1938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008617975075128526,
+      "loss": 2.4192,
+      "step": 1939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000861644530453074,
+      "loss": 2.3764,
+      "step": 1940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008614914823682229,
+      "loss": 2.5312,
+      "step": 1941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008613383632883568,
+      "loss": 2.5017,
+      "step": 1942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008611851732435478,
+      "loss": 2.5654,
+      "step": 1943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008610319122638814,
+      "loss": 2.4891,
+      "step": 1944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008608785803794572,
+      "loss": 2.4536,
+      "step": 1945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860725177620389,
+      "loss": 2.4888,
+      "step": 1946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860571704016804,
+      "loss": 2.4521,
+      "step": 1947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860418159598844,
+      "loss": 2.5552,
+      "step": 1948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008602645443966639,
+      "loss": 2.4249,
+      "step": 1949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008601108584404331,
+      "loss": 2.5224,
+      "step": 1950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008599571017603348,
+      "loss": 2.3766,
+      "step": 1951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008598032743865659,
+      "loss": 2.4483,
+      "step": 1952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008596493763493372,
+      "loss": 2.4611,
+      "step": 1953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008594954076788736,
+      "loss": 2.4674,
+      "step": 1954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008593413684054136,
+      "loss": 2.4169,
+      "step": 1955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008591872585592099,
+      "loss": 2.4981,
+      "step": 1956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008590330781705287,
+      "loss": 2.6809,
+      "step": 1957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588788272696501,
+      "loss": 2.5194,
+      "step": 1958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008587245058868685,
+      "loss": 2.5283,
+      "step": 1959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008585701140524915,
+      "loss": 2.3617,
+      "step": 1960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008584156517968409,
+      "loss": 2.3891,
+      "step": 1961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008582611191502523,
+      "loss": 2.5321,
+      "step": 1962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008581065161430752,
+      "loss": 2.5021,
+      "step": 1963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008579518428056727,
+      "loss": 2.5185,
+      "step": 1964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008577970991684218,
+      "loss": 2.526,
+      "step": 1965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008576422852617136,
+      "loss": 2.536,
+      "step": 1966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008574874011159524,
+      "loss": 2.6002,
+      "step": 1967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008573324467615566,
+      "loss": 2.4784,
+      "step": 1968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008571774222289587,
+      "loss": 2.44,
+      "step": 1969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008570223275486044,
+      "loss": 2.4005,
+      "step": 1970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008568671627509539,
+      "loss": 2.4618,
+      "step": 1971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008567119278664803,
+      "loss": 2.6387,
+      "step": 1972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008565566229256711,
+      "loss": 2.5326,
+      "step": 1973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008564012479590273,
+      "loss": 2.4432,
+      "step": 1974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008562458029970639,
+      "loss": 2.6029,
+      "step": 1975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008560902880703092,
+      "loss": 2.3948,
+      "step": 1976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008559347032093057,
+      "loss": 2.4283,
+      "step": 1977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008557790484446094,
+      "loss": 2.5173,
+      "step": 1978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008556233238067901,
+      "loss": 2.4806,
+      "step": 1979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008554675293264313,
+      "loss": 2.5405,
+      "step": 1980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008553116650341301,
+      "loss": 2.4067,
+      "step": 1981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008551557309604974,
+      "loss": 2.5054,
+      "step": 1982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000854999727136158,
+      "loss": 2.4685,
+      "step": 1983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008548436535917503,
+      "loss": 2.3449,
+      "step": 1984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000854687510357926,
+      "loss": 2.3079,
+      "step": 1985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008545312974653512,
+      "loss": 2.4781,
+      "step": 1986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008543750149447049,
+      "loss": 2.5123,
+      "step": 1987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008542186628266802,
+      "loss": 2.3133,
+      "step": 1988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000854062241141984,
+      "loss": 2.6818,
+      "step": 1989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008539057499213367,
+      "loss": 2.4391,
+      "step": 1990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008537491891954724,
+      "loss": 2.6327,
+      "step": 1991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008535925589951385,
+      "loss": 2.624,
+      "step": 1992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008534358593510967,
+      "loss": 2.5428,
+      "step": 1993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008532790902941217,
+      "loss": 2.56,
+      "step": 1994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008531222518550021,
+      "loss": 2.5674,
+      "step": 1995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008529653440645405,
+      "loss": 2.5332,
+      "step": 1996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008528083669535524,
+      "loss": 2.5604,
+      "step": 1997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008526513205528674,
+      "loss": 2.5849,
+      "step": 1998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008524942048933286,
+      "loss": 2.5483,
+      "step": 1999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008523370200057927,
+      "loss": 2.7671,
+      "step": 2000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008521797659211299,
+      "loss": 2.7001,
+      "step": 2001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008520224426702242,
+      "loss": 2.6102,
+      "step": 2002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008518650502839729,
+      "loss": 2.5767,
+      "step": 2003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008517075887932871,
+      "loss": 2.6571,
+      "step": 2004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008515500582290913,
+      "loss": 2.5858,
+      "step": 2005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008513924586223239,
+      "loss": 2.3939,
+      "step": 2006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008512347900039366,
+      "loss": 2.6353,
+      "step": 2007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008510770524048944,
+      "loss": 2.4402,
+      "step": 2008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008509192458561765,
+      "loss": 2.6307,
+      "step": 2009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008507613703887749,
+      "loss": 2.514,
+      "step": 2010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008506034260336959,
+      "loss": 2.5891,
+      "step": 2011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008504454128219585,
+      "loss": 2.5545,
+      "step": 2012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008502873307845959,
+      "loss": 2.3066,
+      "step": 2013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008501291799526547,
+      "loss": 2.4763,
+      "step": 2014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008499709603571947,
+      "loss": 2.3882,
+      "step": 2015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008498126720292896,
+      "loss": 2.4575,
+      "step": 2016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849654315000026,
+      "loss": 2.4399,
+      "step": 2017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008494958893005047,
+      "loss": 2.6046,
+      "step": 2018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008493373949618396,
+      "loss": 2.7568,
+      "step": 2019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008491788320151581,
+      "loss": 2.5044,
+      "step": 2020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008490202004916012,
+      "loss": 2.6164,
+      "step": 2021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008488615004223233,
+      "loss": 2.6106,
+      "step": 2022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000848702731838492,
+      "loss": 2.7857,
+      "step": 2023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008485438947712889,
+      "loss": 2.6081,
+      "step": 2024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008483849892519088,
+      "loss": 2.6412,
+      "step": 2025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008482260153115597,
+      "loss": 2.5169,
+      "step": 2026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008480669729814634,
+      "loss": 2.5741,
+      "step": 2027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008479078622928547,
+      "loss": 2.6401,
+      "step": 2028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008477486832769825,
+      "loss": 2.6356,
+      "step": 2029
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008475894359651084,
+      "loss": 2.6836,
+      "step": 2030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008474301203885078,
+      "loss": 2.7497,
+      "step": 2031
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008472707365784696,
+      "loss": 2.7831,
+      "step": 2032
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008471112845662956,
+      "loss": 2.6549,
+      "step": 2033
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008469517643833017,
+      "loss": 2.7274,
+      "step": 2034
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008467921760608166,
+      "loss": 2.7368,
+      "step": 2035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008466325196301827,
+      "loss": 2.8133,
+      "step": 2036
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008464727951227557,
+      "loss": 2.8613,
+      "step": 2037
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008463130025699046,
+      "loss": 2.5908,
+      "step": 2038
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008461531420030117,
+      "loss": 2.7933,
+      "step": 2039
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000845993213453473,
+      "loss": 2.7542,
+      "step": 2040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008458332169526973,
+      "loss": 2.7387,
+      "step": 2041
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008456731525321073,
+      "loss": 2.8597,
+      "step": 2042
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008455130202231389,
+      "loss": 2.9117,
+      "step": 2043
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008453528200572409,
+      "loss": 2.7967,
+      "step": 2044
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008451925520658761,
+      "loss": 2.8,
+      "step": 2045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008450322162805202,
+      "loss": 2.7385,
+      "step": 2046
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008448718127326621,
+      "loss": 2.6728,
+      "step": 2047
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008447113414538045,
+      "loss": 2.6967,
+      "step": 2048
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000844550802475463,
+      "loss": 2.9346,
+      "step": 2049
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008443901958291665,
+      "loss": 2.8396,
+      "step": 2050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008442295215464573,
+      "loss": 2.5017,
+      "step": 2051
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008440687796588912,
+      "loss": 2.9092,
+      "step": 2052
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008439079701980368,
+      "loss": 2.79,
+      "step": 2053
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008437470931954765,
+      "loss": 2.7641,
+      "step": 2054
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008435861486828053,
+      "loss": 2.69,
+      "step": 2055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008434251366916323,
+      "loss": 2.7803,
+      "step": 2056
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000843264057253579,
+      "loss": 2.6877,
+      "step": 2057
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008431029104002807,
+      "loss": 2.8668,
+      "step": 2058
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000842941696163386,
+      "loss": 2.7213,
+      "step": 2059
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008427804145745563,
+      "loss": 2.8281,
+      "step": 2060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008426190656654664,
+      "loss": 2.8045,
+      "step": 2061
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008424576494678046,
+      "loss": 2.5083,
+      "step": 2062
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000842296166013272,
+      "loss": 2.6146,
+      "step": 2063
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008421346153335832,
+      "loss": 2.7615,
+      "step": 2064
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008419729974604659,
+      "loss": 2.68,
+      "step": 2065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008418113124256608,
+      "loss": 2.7641,
+      "step": 2066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008416495602609222,
+      "loss": 2.6628,
+      "step": 2067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008414877409980173,
+      "loss": 2.7654,
+      "step": 2068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008413258546687265,
+      "loss": 2.8425,
+      "step": 2069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008411639013048435,
+      "loss": 2.6215,
+      "step": 2070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000841001880938175,
+      "loss": 2.7608,
+      "step": 2071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008408397936005411,
+      "loss": 2.6669,
+      "step": 2072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008406776393237748,
+      "loss": 2.702,
+      "step": 2073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008405154181397221,
+      "loss": 2.9023,
+      "step": 2074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000840353130080243,
+      "loss": 2.63,
+      "step": 2075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008401907751772093,
+      "loss": 2.6868,
+      "step": 2076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000840028353462507,
+      "loss": 2.6798,
+      "step": 2077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008398658649680351,
+      "loss": 2.5034,
+      "step": 2078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000839703309725705,
+      "loss": 2.6224,
+      "step": 2079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000839540687767442,
+      "loss": 2.6357,
+      "step": 2080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008393779991251839,
+      "loss": 2.664,
+      "step": 2081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008392152438308824,
+      "loss": 2.9644,
+      "step": 2082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008390524219165014,
+      "loss": 2.7387,
+      "step": 2083
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008388895334140181,
+      "loss": 2.778,
+      "step": 2084
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008387265783554235,
+      "loss": 2.7435,
+      "step": 2085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008385635567727206,
+      "loss": 2.6841,
+      "step": 2086
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008384004686979261,
+      "loss": 2.7919,
+      "step": 2087
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008382373141630698,
+      "loss": 2.8561,
+      "step": 2088
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008380740932001943,
+      "loss": 2.601,
+      "step": 2089
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008379108058413552,
+      "loss": 2.6273,
+      "step": 2090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008377474521186215,
+      "loss": 2.733,
+      "step": 2091
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837584032064075,
+      "loss": 2.7281,
+      "step": 2092
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008374205457098103,
+      "loss": 2.7586,
+      "step": 2093
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008372569930879353,
+      "loss": 2.6411,
+      "step": 2094
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837093374230571,
+      "loss": 2.6083,
+      "step": 2095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008369296891698513,
+      "loss": 2.6676,
+      "step": 2096
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008367659379379231,
+      "loss": 2.6983,
+      "step": 2097
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008366021205669462,
+      "loss": 2.5166,
+      "step": 2098
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008364382370890933,
+      "loss": 2.5932,
+      "step": 2099
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008362742875365505,
+      "loss": 2.734,
+      "step": 2100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008361102719415163,
+      "loss": 2.5892,
+      "step": 2101
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000835946190336203,
+      "loss": 2.7354,
+      "step": 2102
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008357820427528351,
+      "loss": 2.7269,
+      "step": 2103
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00083561782922365,
+      "loss": 2.755,
+      "step": 2104
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008354535497808991,
+      "loss": 2.5308,
+      "step": 2105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008352892044568454,
+      "loss": 2.635,
+      "step": 2106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008351247932837654,
+      "loss": 2.7426,
+      "step": 2107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008349603162939491,
+      "loss": 2.72,
+      "step": 2108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008347957735196985,
+      "loss": 2.6445,
+      "step": 2109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008346311649933292,
+      "loss": 2.8041,
+      "step": 2110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008344664907471691,
+      "loss": 2.5835,
+      "step": 2111
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008343017508135596,
+      "loss": 2.7083,
+      "step": 2112
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008341369452248548,
+      "loss": 2.6338,
+      "step": 2113
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008339720740134216,
+      "loss": 2.6038,
+      "step": 2114
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008338071372116396,
+      "loss": 2.6911,
+      "step": 2115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008336421348519019,
+      "loss": 2.6323,
+      "step": 2116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008334770669666138,
+      "loss": 2.7067,
+      "step": 2117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008333119335881938,
+      "loss": 2.6039,
+      "step": 2118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008331467347490733,
+      "loss": 2.4865,
+      "step": 2119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008329814704816965,
+      "loss": 2.6622,
+      "step": 2120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008328161408185204,
+      "loss": 2.7792,
+      "step": 2121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008326507457920147,
+      "loss": 2.633,
+      "step": 2122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008324852854346624,
+      "loss": 2.6841,
+      "step": 2123
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008323197597789589,
+      "loss": 2.8055,
+      "step": 2124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008321541688574125,
+      "loss": 2.7518,
+      "step": 2125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008319885127025445,
+      "loss": 2.5645,
+      "step": 2126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008318227913468888,
+      "loss": 2.6877,
+      "step": 2127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008316570048229922,
+      "loss": 2.5301,
+      "step": 2128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008314911531634144,
+      "loss": 2.7097,
+      "step": 2129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008313252364007277,
+      "loss": 2.791,
+      "step": 2130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008311592545675173,
+      "loss": 2.4743,
+      "step": 2131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008309932076963811,
+      "loss": 2.7581,
+      "step": 2132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00083082709581993,
+      "loss": 2.7558,
+      "step": 2133
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008306609189707873,
+      "loss": 2.6119,
+      "step": 2134
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008304946771815891,
+      "loss": 2.6346,
+      "step": 2135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008303283704849849,
+      "loss": 2.6422,
+      "step": 2136
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008301619989136361,
+      "loss": 2.7192,
+      "step": 2137
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008299955625002173,
+      "loss": 2.6798,
+      "step": 2138
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008298290612774156,
+      "loss": 2.6012,
+      "step": 2139
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008296624952779313,
+      "loss": 2.4614,
+      "step": 2140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008294958645344765,
+      "loss": 2.529,
+      "step": 2141
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008293291690797773,
+      "loss": 2.6144,
+      "step": 2142
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008291624089465712,
+      "loss": 2.6281,
+      "step": 2143
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008289955841676096,
+      "loss": 2.6824,
+      "step": 2144
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008288286947756554,
+      "loss": 2.4802,
+      "step": 2145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008286617408034851,
+      "loss": 2.6013,
+      "step": 2146
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008284947222838875,
+      "loss": 2.64,
+      "step": 2147
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008283276392496642,
+      "loss": 2.626,
+      "step": 2148
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008281604917336294,
+      "loss": 2.5215,
+      "step": 2149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00082799327976861,
+      "loss": 2.5882,
+      "step": 2150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008278260033874456,
+      "loss": 2.6048,
+      "step": 2151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008276586626229885,
+      "loss": 2.6089,
+      "step": 2152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008274912575081031,
+      "loss": 2.6759,
+      "step": 2153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008273237880756675,
+      "loss": 2.4387,
+      "step": 2154
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008271562543585711,
+      "loss": 2.6659,
+      "step": 2155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008269886563897172,
+      "loss": 2.7716,
+      "step": 2156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008268209942020211,
+      "loss": 2.5355,
+      "step": 2157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008266532678284103,
+      "loss": 2.7096,
+      "step": 2158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826485477301826,
+      "loss": 2.7055,
+      "step": 2159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008263176226552209,
+      "loss": 2.5535,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826149703921561,
+      "loss": 2.6893,
+      "step": 2161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008259817211338247,
+      "loss": 2.6616,
+      "step": 2162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008258136743250026,
+      "loss": 2.5922,
+      "step": 2163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008256455635280986,
+      "loss": 2.5818,
+      "step": 2164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008254773887761285,
+      "loss": 2.7003,
+      "step": 2165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000825309150102121,
+      "loss": 2.6441,
+      "step": 2166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008251408475391176,
+      "loss": 2.6466,
+      "step": 2167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008249724811201715,
+      "loss": 2.688,
+      "step": 2168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008248040508783492,
+      "loss": 2.6775,
+      "step": 2169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008246355568467296,
+      "loss": 2.6098,
+      "step": 2170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000824466999058404,
+      "loss": 2.5227,
+      "step": 2171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008242983775464761,
+      "loss": 2.597,
+      "step": 2172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008241296923440624,
+      "loss": 2.5543,
+      "step": 2173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000823960943484292,
+      "loss": 2.6652,
+      "step": 2174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008237921310003058,
+      "loss": 2.4666,
+      "step": 2175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008236232549252582,
+      "loss": 2.6362,
+      "step": 2176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008234543152923151,
+      "loss": 2.6113,
+      "step": 2177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008232853121346558,
+      "loss": 2.5322,
+      "step": 2178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008231162454854711,
+      "loss": 2.5869,
+      "step": 2179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008229471153779652,
+      "loss": 2.5244,
+      "step": 2180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008227779218453543,
+      "loss": 2.7641,
+      "step": 2181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008226086649208669,
+      "loss": 2.722,
+      "step": 2182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008224393446377445,
+      "loss": 2.6775,
+      "step": 2183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008222699610292405,
+      "loss": 2.5679,
+      "step": 2184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008221005141286208,
+      "loss": 2.6692,
+      "step": 2185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008219310039691642,
+      "loss": 2.5299,
+      "step": 2186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008217614305841614,
+      "loss": 2.5761,
+      "step": 2187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215917940069158,
+      "loss": 2.4366,
+      "step": 2188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000821422094270743,
+      "loss": 2.8101,
+      "step": 2189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008212523314089714,
+      "loss": 2.6305,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008210825054549411,
+      "loss": 2.6547,
+      "step": 2191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008209126164420055,
+      "loss": 2.6574,
+      "step": 2192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008207426644035295,
+      "loss": 2.5838,
+      "step": 2193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205726493728911,
+      "loss": 2.5636,
+      "step": 2194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008204025713834802,
+      "loss": 2.5107,
+      "step": 2195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000820232430468699,
+      "loss": 2.4607,
+      "step": 2196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008200622266619626,
+      "loss": 2.7235,
+      "step": 2197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008198919599966981,
+      "loss": 2.7609,
+      "step": 2198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008197216305063449,
+      "loss": 2.6194,
+      "step": 2199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008195512382243547,
+      "loss": 2.4446,
+      "step": 2200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008193807831841918,
+      "loss": 2.6234,
+      "step": 2201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008192102654193327,
+      "loss": 2.5519,
+      "step": 2202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008190396849632659,
+      "loss": 2.5638,
+      "step": 2203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008188690418494928,
+      "loss": 2.6987,
+      "step": 2204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008186983361115268,
+      "loss": 2.6308,
+      "step": 2205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008185275677828935,
+      "loss": 2.654,
+      "step": 2206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008183567368971308,
+      "loss": 2.4698,
+      "step": 2207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008181858434877891,
+      "loss": 2.7385,
+      "step": 2208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000818014887588431,
+      "loss": 2.5237,
+      "step": 2209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008178438692326314,
+      "loss": 2.4838,
+      "step": 2210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008176727884539772,
+      "loss": 2.6195,
+      "step": 2211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008175016452860677,
+      "loss": 2.5686,
+      "step": 2212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008173304397625147,
+      "loss": 2.5917,
+      "step": 2213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008171591719169422,
+      "loss": 2.5448,
+      "step": 2214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000816987841782986,
+      "loss": 2.6558,
+      "step": 2215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008168164493942947,
+      "loss": 2.5386,
+      "step": 2216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008166449947845285,
+      "loss": 2.4368,
+      "step": 2217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008164734779873604,
+      "loss": 2.651,
+      "step": 2218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008163018990364754,
+      "loss": 2.691,
+      "step": 2219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008161302579655709,
+      "loss": 2.5602,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008159585548083558,
+      "loss": 2.6108,
+      "step": 2221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000815786789598552,
+      "loss": 2.4564,
+      "step": 2222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008156149623698933,
+      "loss": 2.6878,
+      "step": 2223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008154430731561257,
+      "loss": 2.5952,
+      "step": 2224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008152711219910071,
+      "loss": 2.5984,
+      "step": 2225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000815099108908308,
+      "loss": 2.7077,
+      "step": 2226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008149270339418107,
+      "loss": 2.4802,
+      "step": 2227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00081475489712531,
+      "loss": 2.5913,
+      "step": 2228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008145826984926126,
+      "loss": 2.5466,
+      "step": 2229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008144104380775372,
+      "loss": 2.5064,
+      "step": 2230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008142381159139152,
+      "loss": 2.5323,
+      "step": 2231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008140657320355896,
+      "loss": 2.6333,
+      "step": 2232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008138932864764155,
+      "loss": 2.4979,
+      "step": 2233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008137207792702604,
+      "loss": 2.5891,
+      "step": 2234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008135482104510039,
+      "loss": 2.5779,
+      "step": 2235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008133755800525375,
+      "loss": 2.502,
+      "step": 2236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008132028881087651,
+      "loss": 2.6698,
+      "step": 2237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008130301346536022,
+      "loss": 2.46,
+      "step": 2238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008128573197209767,
+      "loss": 2.4975,
+      "step": 2239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126844433448289,
+      "loss": 2.7823,
+      "step": 2240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008125115055591103,
+      "loss": 2.6283,
+      "step": 2241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008123385063977853,
+      "loss": 2.693,
+      "step": 2242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00081216544589483,
+      "loss": 2.5765,
+      "step": 2243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008119923240842324,
+      "loss": 2.5764,
+      "step": 2244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008118191409999929,
+      "loss": 2.6275,
+      "step": 2245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008116458966761235,
+      "loss": 2.6042,
+      "step": 2246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008114725911466487,
+      "loss": 2.4196,
+      "step": 2247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008112992244456046,
+      "loss": 2.4472,
+      "step": 2248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008111257966070398,
+      "loss": 2.6297,
+      "step": 2249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008109523076650141,
+      "loss": 2.5167,
+      "step": 2250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008107787576536003,
+      "loss": 2.6081,
+      "step": 2251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008106051466068827,
+      "loss": 2.5315,
+      "step": 2252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008104314745589571,
+      "loss": 2.6549,
+      "step": 2253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008102577415439323,
+      "loss": 2.6542,
+      "step": 2254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008100839475959283,
+      "loss": 2.6921,
+      "step": 2255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008099100927490772,
+      "loss": 2.6073,
+      "step": 2256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008097361770375236,
+      "loss": 2.6335,
+      "step": 2257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008095622004954231,
+      "loss": 2.5856,
+      "step": 2258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008093881631569441,
+      "loss": 2.7557,
+      "step": 2259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008092140650562665,
+      "loss": 2.5244,
+      "step": 2260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008090399062275822,
+      "loss": 2.5271,
+      "step": 2261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008088656867050953,
+      "loss": 2.555,
+      "step": 2262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008086914065230215,
+      "loss": 2.6978,
+      "step": 2263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008085170657155883,
+      "loss": 2.4915,
+      "step": 2264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008083426643170355,
+      "loss": 2.4939,
+      "step": 2265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008081682023616146,
+      "loss": 2.6147,
+      "step": 2266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008079936798835888,
+      "loss": 2.5647,
+      "step": 2267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008078190969172338,
+      "loss": 2.5683,
+      "step": 2268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008076444534968364,
+      "loss": 2.6089,
+      "step": 2269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008074697496566959,
+      "loss": 2.5345,
+      "step": 2270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008072949854311229,
+      "loss": 2.6561,
+      "step": 2271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008071201608544405,
+      "loss": 2.5435,
+      "step": 2272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008069452759609832,
+      "loss": 2.419,
+      "step": 2273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008067703307850973,
+      "loss": 2.6533,
+      "step": 2274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008065953253611415,
+      "loss": 2.6249,
+      "step": 2275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008064202597234855,
+      "loss": 2.5421,
+      "step": 2276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008062451339065116,
+      "loss": 2.597,
+      "step": 2277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008060699479446133,
+      "loss": 2.4583,
+      "step": 2278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008058947018721965,
+      "loss": 2.5077,
+      "step": 2279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008057193957236782,
+      "loss": 2.4561,
+      "step": 2280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008055440295334881,
+      "loss": 2.7011,
+      "step": 2281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000805368603336067,
+      "loss": 2.4203,
+      "step": 2282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008051931171658673,
+      "loss": 2.5068,
+      "step": 2283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008050175710573541,
+      "loss": 2.4736,
+      "step": 2284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008048419650450034,
+      "loss": 2.5083,
+      "step": 2285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008046662991633034,
+      "loss": 2.6198,
+      "step": 2286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000804490573446754,
+      "loss": 2.4987,
+      "step": 2287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008043147879298666,
+      "loss": 2.6743,
+      "step": 2288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008041389426471648,
+      "loss": 2.5697,
+      "step": 2289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008039630376331837,
+      "loss": 2.5586,
+      "step": 2290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008037870729224697,
+      "loss": 2.8225,
+      "step": 2291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008036110485495818,
+      "loss": 2.7875,
+      "step": 2292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008034349645490901,
+      "loss": 2.6292,
+      "step": 2293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008032588209555765,
+      "loss": 2.5978,
+      "step": 2294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008030826178036345,
+      "loss": 2.7381,
+      "step": 2295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008029063551278699,
+      "loss": 2.8343,
+      "step": 2296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008027300329628996,
+      "loss": 2.5174,
+      "step": 2297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008025536513433522,
+      "loss": 2.7701,
+      "step": 2298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008023772103038682,
+      "loss": 2.5995,
+      "step": 2299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008022007098790995,
+      "loss": 2.3499,
+      "step": 2300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008020241501037103,
+      "loss": 2.6189,
+      "step": 2301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008018475310123756,
+      "loss": 2.5506,
+      "step": 2302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008016708526397827,
+      "loss": 2.5499,
+      "step": 2303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008014941150206301,
+      "loss": 2.5239,
+      "step": 2304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008013173181896282,
+      "loss": 2.4739,
+      "step": 2305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000801140462181499,
+      "loss": 2.5338,
+      "step": 2306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008009635470309762,
+      "loss": 2.5398,
+      "step": 2307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008007865727728047,
+      "loss": 2.5599,
+      "step": 2308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008006095394417417,
+      "loss": 2.536,
+      "step": 2309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000800432447072555,
+      "loss": 2.5543,
+      "step": 2310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008002552957000253,
+      "loss": 2.6228,
+      "step": 2311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008000780853589437,
+      "loss": 2.4121,
+      "step": 2312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007999008160841137,
+      "loss": 2.5487,
+      "step": 2313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007997234879103498,
+      "loss": 2.5865,
+      "step": 2314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007995461008724785,
+      "loss": 2.6284,
+      "step": 2315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007993686550053376,
+      "loss": 2.5818,
+      "step": 2316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007991911503437767,
+      "loss": 2.6037,
+      "step": 2317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007990135869226562,
+      "loss": 2.7262,
+      "step": 2318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007988359647768494,
+      "loss": 2.5115,
+      "step": 2319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007986582839412397,
+      "loss": 2.6302,
+      "step": 2320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007984805444507231,
+      "loss": 2.5301,
+      "step": 2321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007983027463402065,
+      "loss": 2.4779,
+      "step": 2322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007981248896446085,
+      "loss": 2.4846,
+      "step": 2323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007979469743988593,
+      "loss": 2.5721,
+      "step": 2324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007977690006379006,
+      "loss": 2.4347,
+      "step": 2325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007975909683966851,
+      "loss": 2.4489,
+      "step": 2326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007974128777101779,
+      "loss": 2.4648,
+      "step": 2327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007972347286133547,
+      "loss": 2.4995,
+      "step": 2328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007970565211412033,
+      "loss": 2.4368,
+      "step": 2329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007968782553287226,
+      "loss": 2.5169,
+      "step": 2330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007966999312109231,
+      "loss": 2.579,
+      "step": 2331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007965215488228265,
+      "loss": 2.5878,
+      "step": 2332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007963431081994663,
+      "loss": 2.5603,
+      "step": 2333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007961646093758875,
+      "loss": 2.6781,
+      "step": 2334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000795986052387146,
+      "loss": 2.5296,
+      "step": 2335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007958074372683095,
+      "loss": 2.5505,
+      "step": 2336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007956287640544572,
+      "loss": 2.5076,
+      "step": 2337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007954500327806797,
+      "loss": 2.5213,
+      "step": 2338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007952712434820786,
+      "loss": 2.4125,
+      "step": 2339
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007950923961937673,
+      "loss": 2.7134,
+      "step": 2340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007949134909508704,
+      "loss": 2.4999,
+      "step": 2341
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007947345277885241,
+      "loss": 2.749,
+      "step": 2342
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007945555067418755,
+      "loss": 2.6555,
+      "step": 2343
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007943764278460836,
+      "loss": 2.4876,
+      "step": 2344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007941972911363187,
+      "loss": 2.4641,
+      "step": 2345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007940180966477619,
+      "loss": 2.5302,
+      "step": 2346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007938388444156066,
+      "loss": 2.7041,
+      "step": 2347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007936595344750564,
+      "loss": 2.5699,
+      "step": 2348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007934801668613272,
+      "loss": 2.498,
+      "step": 2349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007933007416096457,
+      "loss": 2.5022,
+      "step": 2350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007931212587552501,
+      "loss": 2.5171,
+      "step": 2351
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00079294171833339,
+      "loss": 2.522,
+      "step": 2352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000792762120379326,
+      "loss": 2.7184,
+      "step": 2353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007925824649283304,
+      "loss": 2.654,
+      "step": 2354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007924027520156865,
+      "loss": 2.5825,
+      "step": 2355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007922229816766886,
+      "loss": 2.4986,
+      "step": 2356
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007920431539466433,
+      "loss": 2.6102,
+      "step": 2357
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007918632688608673,
+      "loss": 2.6244,
+      "step": 2358
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007916833264546892,
+      "loss": 2.562,
+      "step": 2359
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000791503326763449,
+      "loss": 2.5861,
+      "step": 2360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007913232698224975,
+      "loss": 2.4881,
+      "step": 2361
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007911431556671967,
+      "loss": 2.5152,
+      "step": 2362
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007909629843329205,
+      "loss": 2.4865,
+      "step": 2363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007907827558550532,
+      "loss": 2.5257,
+      "step": 2364
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007906024702689909,
+      "loss": 2.5202,
+      "step": 2365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000790422127610141,
+      "loss": 2.4407,
+      "step": 2366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007902417279139213,
+      "loss": 2.5644,
+      "step": 2367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007900612712157618,
+      "loss": 2.4804,
+      "step": 2368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007898807575511029,
+      "loss": 2.4769,
+      "step": 2369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007897001869553968,
+      "loss": 2.5184,
+      "step": 2370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007895195594641065,
+      "loss": 2.6921,
+      "step": 2371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007893388751127061,
+      "loss": 2.756,
+      "step": 2372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007891581339366816,
+      "loss": 2.561,
+      "step": 2373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007889773359715291,
+      "loss": 2.602,
+      "step": 2374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007887964812527564,
+      "loss": 2.3355,
+      "step": 2375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007886155698158826,
+      "loss": 2.4786,
+      "step": 2376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007884346016964378,
+      "loss": 2.6163,
+      "step": 2377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007882535769299628,
+      "loss": 2.6521,
+      "step": 2378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007880724955520105,
+      "loss": 2.5902,
+      "step": 2379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007878913575981437,
+      "loss": 2.6286,
+      "step": 2380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007877101631039372,
+      "loss": 2.6032,
+      "step": 2381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000787528912104977,
+      "loss": 2.5286,
+      "step": 2382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000787347604636859,
+      "loss": 2.7495,
+      "step": 2383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007871662407351918,
+      "loss": 2.697,
+      "step": 2384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000786984820435594,
+      "loss": 2.6404,
+      "step": 2385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007868033437736957,
+      "loss": 2.6287,
+      "step": 2386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000786621810785138,
+      "loss": 2.5768,
+      "step": 2387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007864402215055726,
+      "loss": 2.5497,
+      "step": 2388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007862585759706631,
+      "loss": 2.5106,
+      "step": 2389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007860768742160837,
+      "loss": 2.5064,
+      "step": 2390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007858951162775194,
+      "loss": 2.4631,
+      "step": 2391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007857133021906668,
+      "loss": 2.7128,
+      "step": 2392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000785531431991233,
+      "loss": 2.4814,
+      "step": 2393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007853495057149364,
+      "loss": 2.5571,
+      "step": 2394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007851675233975065,
+      "loss": 2.433,
+      "step": 2395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007849854850746834,
+      "loss": 2.5798,
+      "step": 2396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007848033907822186,
+      "loss": 2.4522,
+      "step": 2397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007846212405558746,
+      "loss": 2.4694,
+      "step": 2398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007844390344314247,
+      "loss": 2.6356,
+      "step": 2399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007842567724446528,
+      "loss": 2.567,
+      "step": 2400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007840744546313549,
+      "loss": 2.5428,
+      "step": 2401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007838920810273365,
+      "loss": 2.4802,
+      "step": 2402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007837096516684155,
+      "loss": 2.5474,
+      "step": 2403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007835271665904195,
+      "loss": 2.5457,
+      "step": 2404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007833446258291881,
+      "loss": 2.599,
+      "step": 2405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007831620294205709,
+      "loss": 2.7285,
+      "step": 2406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007829793774004291,
+      "loss": 2.4826,
+      "step": 2407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007827966698046346,
+      "loss": 2.4303,
+      "step": 2408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00078261390666907,
+      "loss": 2.6835,
+      "step": 2409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007824310880296294,
+      "loss": 2.5364,
+      "step": 2410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007822482139222171,
+      "loss": 2.5949,
+      "step": 2411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007820652843827486,
+      "loss": 2.5729,
+      "step": 2412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007818822994471504,
+      "loss": 2.5029,
+      "step": 2413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007816992591513598,
+      "loss": 2.6292,
+      "step": 2414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007815161635313249,
+      "loss": 2.3869,
+      "step": 2415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007813330126230047,
+      "loss": 2.5478,
+      "step": 2416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007811498064623691,
+      "loss": 2.5958,
+      "step": 2417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007809665450853988,
+      "loss": 2.634,
+      "step": 2418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007807832285280854,
+      "loss": 2.5645,
+      "step": 2419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007805998568264312,
+      "loss": 2.5404,
+      "step": 2420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007804164300164493,
+      "loss": 2.5037,
+      "step": 2421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007802329481341643,
+      "loss": 2.5414,
+      "step": 2422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007800494112156107,
+      "loss": 2.5492,
+      "step": 2423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000779865819296834,
+      "loss": 2.4287,
+      "step": 2424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007796821724138911,
+      "loss": 2.6811,
+      "step": 2425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000779498470602849,
+      "loss": 2.5252,
+      "step": 2426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007793147138997857,
+      "loss": 2.5728,
+      "step": 2427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007791309023407904,
+      "loss": 2.3709,
+      "step": 2428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007789470359619621,
+      "loss": 2.646,
+      "step": 2429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000778763114799412,
+      "loss": 2.5048,
+      "step": 2430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007785791388892607,
+      "loss": 2.5449,
+      "step": 2431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007783951082676402,
+      "loss": 2.5307,
+      "step": 2432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007782110229706932,
+      "loss": 2.6681,
+      "step": 2433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007780268830345728,
+      "loss": 2.5541,
+      "step": 2434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007778426884954434,
+      "loss": 2.6642,
+      "step": 2435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00077765843938948,
+      "loss": 2.5042,
+      "step": 2436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007774741357528678,
+      "loss": 2.6675,
+      "step": 2437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007772897776218032,
+      "loss": 2.5213,
+      "step": 2438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007771053650324931,
+      "loss": 2.5978,
+      "step": 2439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007769208980211553,
+      "loss": 2.5223,
+      "step": 2440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000776736376624018,
+      "loss": 2.7209,
+      "step": 2441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007765518008773203,
+      "loss": 2.5828,
+      "step": 2442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007763671708173117,
+      "loss": 2.6029,
+      "step": 2443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007761824864802529,
+      "loss": 2.4956,
+      "step": 2444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007759977479024148,
+      "loss": 2.4936,
+      "step": 2445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000775812955120079,
+      "loss": 2.6728,
+      "step": 2446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007756281081695379,
+      "loss": 2.6014,
+      "step": 2447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007754432070870943,
+      "loss": 2.4386,
+      "step": 2448
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007752582519090621,
+      "loss": 2.6583,
+      "step": 2449
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007750732426717653,
+      "loss": 2.6511,
+      "step": 2450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007748881794115387,
+      "loss": 2.4792,
+      "step": 2451
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000774703062164728,
+      "loss": 2.4169,
+      "step": 2452
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007745178909676891,
+      "loss": 2.5026,
+      "step": 2453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007743326658567884,
+      "loss": 2.4591,
+      "step": 2454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007741473868684035,
+      "loss": 2.417,
+      "step": 2455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000773962054038922,
+      "loss": 2.3697,
+      "step": 2456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007737766674047424,
+      "loss": 2.397,
+      "step": 2457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007735912270022735,
+      "loss": 2.5794,
+      "step": 2458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007734057328679352,
+      "loss": 2.6012,
+      "step": 2459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000773220185038157,
+      "loss": 2.6369,
+      "step": 2460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00077303458354938,
+      "loss": 2.5061,
+      "step": 2461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007728489284380549,
+      "loss": 2.4048,
+      "step": 2462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007726632197406439,
+      "loss": 2.6058,
+      "step": 2463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007724774574936189,
+      "loss": 2.3667,
+      "step": 2464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007722916417334625,
+      "loss": 2.6322,
+      "step": 2465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007721057724966682,
+      "loss": 2.4905,
+      "step": 2466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007719198498197397,
+      "loss": 2.4863,
+      "step": 2467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007717338737391911,
+      "loss": 2.5972,
+      "step": 2468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007715478442915471,
+      "loss": 2.4453,
+      "step": 2469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007713617615133431,
+      "loss": 2.8122,
+      "step": 2470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007711756254411246,
+      "loss": 2.4532,
+      "step": 2471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007709894361114478,
+      "loss": 2.6741,
+      "step": 2472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007708031935608793,
+      "loss": 2.6015,
+      "step": 2473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000770616897825996,
+      "loss": 2.4985,
+      "step": 2474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007704305489433855,
+      "loss": 2.5301,
+      "step": 2475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007702441469496461,
+      "loss": 2.5332,
+      "step": 2476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007700576918813857,
+      "loss": 2.3709,
+      "step": 2477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000769871183775223,
+      "loss": 2.585,
+      "step": 2478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007696846226677879,
+      "loss": 2.536,
+      "step": 2479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007694980085957193,
+      "loss": 2.5361,
+      "step": 2480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007693113415956674,
+      "loss": 2.5576,
+      "step": 2481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007691246217042927,
+      "loss": 2.4877,
+      "step": 2482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768937848958266,
+      "loss": 2.5821,
+      "step": 2483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007687510233942684,
+      "loss": 2.6114,
+      "step": 2484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007685641450489915,
+      "loss": 2.6243,
+      "step": 2485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007683772139591373,
+      "loss": 2.5264,
+      "step": 2486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007681902301614181,
+      "loss": 2.4862,
+      "step": 2487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007680031936925562,
+      "loss": 2.3482,
+      "step": 2488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007678161045892849,
+      "loss": 2.5553,
+      "step": 2489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007676289628883475,
+      "loss": 2.6731,
+      "step": 2490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007674417686264974,
+      "loss": 2.6029,
+      "step": 2491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007672545218404989,
+      "loss": 2.4346,
+      "step": 2492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000767067222567126,
+      "loss": 2.5607,
+      "step": 2493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007668798708431634,
+      "loss": 2.5713,
+      "step": 2494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007666924667054058,
+      "loss": 2.5699,
+      "step": 2495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007665050101906588,
+      "loss": 2.5429,
+      "step": 2496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007663175013357375,
+      "loss": 2.259,
+      "step": 2497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007661299401774677,
+      "loss": 2.5024,
+      "step": 2498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007659423267526857,
+      "loss": 2.5284,
+      "step": 2499
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007657546610982375,
+      "loss": 2.5,
+      "step": 2500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007655669432509796,
+      "loss": 2.2774,
+      "step": 2501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007653791732477789,
+      "loss": 2.5101,
+      "step": 2502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007651913511255125,
+      "loss": 2.4871,
+      "step": 2503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007650034769210676,
+      "loss": 2.5605,
+      "step": 2504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007648155506713417,
+      "loss": 2.3992,
+      "step": 2505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007646275724132425,
+      "loss": 2.5511,
+      "step": 2506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007644395421836881,
+      "loss": 2.6099,
+      "step": 2507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007642514600196064,
+      "loss": 2.715,
+      "step": 2508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007640633259579359,
+      "loss": 2.5477,
+      "step": 2509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000763875140035625,
+      "loss": 2.4964,
+      "step": 2510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007636869022896326,
+      "loss": 2.6145,
+      "step": 2511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007634986127569275,
+      "loss": 2.5737,
+      "step": 2512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007633102714744886,
+      "loss": 2.432,
+      "step": 2513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007631218784793054,
+      "loss": 2.5763,
+      "step": 2514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007629334338083773,
+      "loss": 2.5576,
+      "step": 2515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007627449374987136,
+      "loss": 2.6029,
+      "step": 2516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007625563895873342,
+      "loss": 2.5068,
+      "step": 2517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007623677901112689,
+      "loss": 2.4398,
+      "step": 2518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007621791391075573,
+      "loss": 2.4865,
+      "step": 2519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007619904366132499,
+      "loss": 2.4673,
+      "step": 2520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007618016826654066,
+      "loss": 2.4017,
+      "step": 2521
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007616128773010977,
+      "loss": 2.4666,
+      "step": 2522
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007614240205574039,
+      "loss": 2.4357,
+      "step": 2523
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007612351124714152,
+      "loss": 2.6189,
+      "step": 2524
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007610461530802324,
+      "loss": 2.4342,
+      "step": 2525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007608571424209661,
+      "loss": 2.5494,
+      "step": 2526
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007606680805307369,
+      "loss": 2.6195,
+      "step": 2527
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007604789674466757,
+      "loss": 2.7033,
+      "step": 2528
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007602898032059231,
+      "loss": 2.5355,
+      "step": 2529
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007601005878456302,
+      "loss": 2.6852,
+      "step": 2530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007599113214029578,
+      "loss": 2.4278,
+      "step": 2531
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007597220039150767,
+      "loss": 2.4616,
+      "step": 2532
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007595326354191682,
+      "loss": 2.5393,
+      "step": 2533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007593432159524229,
+      "loss": 2.492,
+      "step": 2534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007591537455520419,
+      "loss": 2.5167,
+      "step": 2535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007589642242552361,
+      "loss": 2.4334,
+      "step": 2536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007587746520992269,
+      "loss": 2.5436,
+      "step": 2537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007585850291212447,
+      "loss": 2.4882,
+      "step": 2538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007583953553585307,
+      "loss": 2.4043,
+      "step": 2539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000758205630848336,
+      "loss": 2.538,
+      "step": 2540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007580158556279212,
+      "loss": 2.4868,
+      "step": 2541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007578260297345572,
+      "loss": 2.6456,
+      "step": 2542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007576361532055249,
+      "loss": 2.5619,
+      "step": 2543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007574462260781152,
+      "loss": 2.5689,
+      "step": 2544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007572562483896284,
+      "loss": 2.5465,
+      "step": 2545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007570662201773754,
+      "loss": 2.4494,
+      "step": 2546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007568761414786767,
+      "loss": 2.5832,
+      "step": 2547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007566860123308626,
+      "loss": 2.3414,
+      "step": 2548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007564958327712734,
+      "loss": 2.6914,
+      "step": 2549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007563056028372598,
+      "loss": 2.4483,
+      "step": 2550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007561153225661816,
+      "loss": 2.4019,
+      "step": 2551
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007559249919954087,
+      "loss": 2.5146,
+      "step": 2552
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007557346111623216,
+      "loss": 2.4365,
+      "step": 2553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007555441801043093,
+      "loss": 2.4307,
+      "step": 2554
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007553536988587721,
+      "loss": 2.3704,
+      "step": 2555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007551631674631192,
+      "loss": 2.4537,
+      "step": 2556
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007549725859547701,
+      "loss": 2.4372,
+      "step": 2557
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007547819543711538,
+      "loss": 2.439,
+      "step": 2558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007545912727497097,
+      "loss": 2.5593,
+      "step": 2559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007544005411278862,
+      "loss": 2.4461,
+      "step": 2560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007542097595431424,
+      "loss": 2.5029,
+      "step": 2561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007540189280329464,
+      "loss": 2.504,
+      "step": 2562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007538280466347769,
+      "loss": 2.617,
+      "step": 2563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007536371153861217,
+      "loss": 2.52,
+      "step": 2564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007534461343244787,
+      "loss": 2.4561,
+      "step": 2565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007532551034873558,
+      "loss": 2.534,
+      "step": 2566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007530640229122703,
+      "loss": 2.4866,
+      "step": 2567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007528728926367493,
+      "loss": 2.4817,
+      "step": 2568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00075268171269833,
+      "loss": 2.4278,
+      "step": 2569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007524904831345589,
+      "loss": 2.5157,
+      "step": 2570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007522992039829928,
+      "loss": 2.5049,
+      "step": 2571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007521078752811974,
+      "loss": 2.4215,
+      "step": 2572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007519164970667491,
+      "loss": 2.5319,
+      "step": 2573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007517250693772336,
+      "loss": 2.5364,
+      "step": 2574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000751533592250246,
+      "loss": 2.5585,
+      "step": 2575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007513420657233914,
+      "loss": 2.536,
+      "step": 2576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007511504898342848,
+      "loss": 2.5595,
+      "step": 2577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007509588646205505,
+      "loss": 2.5739,
+      "step": 2578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007507671901198227,
+      "loss": 2.5658,
+      "step": 2579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007505754663697456,
+      "loss": 2.3956,
+      "step": 2580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007503836934079723,
+      "loss": 2.6116,
+      "step": 2581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007501918712721662,
+      "loss": 2.536,
+      "step": 2582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00075,
+      "loss": 2.4109,
+      "step": 2583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007498080796291564,
+      "loss": 2.5162,
+      "step": 2584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007496161101973273,
+      "loss": 2.5722,
+      "step": 2585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007494240917422149,
+      "loss": 2.5947,
+      "step": 2586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007492320243015303,
+      "loss": 2.4036,
+      "step": 2587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007490399079129944,
+      "loss": 2.4514,
+      "step": 2588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000748847742614338,
+      "loss": 2.4648,
+      "step": 2589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007486555284433016,
+      "loss": 2.4128,
+      "step": 2590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007484632654376346,
+      "loss": 2.4651,
+      "step": 2591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007482709536350967,
+      "loss": 2.5629,
+      "step": 2592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007480785930734568,
+      "loss": 2.4352,
+      "step": 2593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007478861837904938,
+      "loss": 2.4944,
+      "step": 2594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007476937258239953,
+      "loss": 2.5269,
+      "step": 2595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007475012192117597,
+      "loss": 2.4443,
+      "step": 2596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007473086639915938,
+      "loss": 2.4577,
+      "step": 2597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007471160602013145,
+      "loss": 2.4503,
+      "step": 2598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007469234078787482,
+      "loss": 2.5609,
+      "step": 2599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007467307070617308,
+      "loss": 2.3763,
+      "step": 2600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000746537957788108,
+      "loss": 2.5023,
+      "step": 2601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007463451600957342,
+      "loss": 2.443,
+      "step": 2602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007461523140224742,
+      "loss": 2.498,
+      "step": 2603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000745959419606202,
+      "loss": 2.6459,
+      "step": 2604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007457664768848008,
+      "loss": 2.3521,
+      "step": 2605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007455734858961637,
+      "loss": 2.5046,
+      "step": 2606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000745380446678193,
+      "loss": 2.5839,
+      "step": 2607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007451873592688008,
+      "loss": 2.6044,
+      "step": 2608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007449942237059082,
+      "loss": 2.4008,
+      "step": 2609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007448010400274461,
+      "loss": 2.4615,
+      "step": 2610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007446078082713547,
+      "loss": 2.4846,
+      "step": 2611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007444145284755837,
+      "loss": 2.5564,
+      "step": 2612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007442212006780924,
+      "loss": 2.4388,
+      "step": 2613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007440278249168491,
+      "loss": 2.471,
+      "step": 2614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007438344012298319,
+      "loss": 2.6386,
+      "step": 2615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007436409296550283,
+      "loss": 2.5886,
+      "step": 2616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000743447410230435,
+      "loss": 2.5399,
+      "step": 2617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007432538429940581,
+      "loss": 2.4808,
+      "step": 2618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007430602279839131,
+      "loss": 2.5844,
+      "step": 2619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007428665652380255,
+      "loss": 2.7935,
+      "step": 2620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000742672854794429,
+      "loss": 2.3645,
+      "step": 2621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007424790966911675,
+      "loss": 2.381,
+      "step": 2622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007422852909662942,
+      "loss": 2.4632,
+      "step": 2623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007420914376578717,
+      "loss": 2.4755,
+      "step": 2624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007418975368039711,
+      "loss": 2.2774,
+      "step": 2625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007417035884426743,
+      "loss": 2.4028,
+      "step": 2626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007415095926120711,
+      "loss": 2.4376,
+      "step": 2627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007413155493502616,
+      "loss": 2.5337,
+      "step": 2628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007411214586953548,
+      "loss": 2.4299,
+      "step": 2629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740927320685469,
+      "loss": 2.4836,
+      "step": 2630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007407331353587319,
+      "loss": 2.582,
+      "step": 2631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007405389027532806,
+      "loss": 2.5921,
+      "step": 2632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740344622907261,
+      "loss": 2.5228,
+      "step": 2633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007401502958588292,
+      "loss": 2.4363,
+      "step": 2634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007399559216461495,
+      "loss": 2.5449,
+      "step": 2635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007397615003073963,
+      "loss": 2.4271,
+      "step": 2636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007395670318807528,
+      "loss": 2.5016,
+      "step": 2637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007393725164044114,
+      "loss": 2.5262,
+      "step": 2638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007391779539165741,
+      "loss": 2.5885,
+      "step": 2639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007389833444554521,
+      "loss": 2.6966,
+      "step": 2640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007387886880592654,
+      "loss": 2.6122,
+      "step": 2641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007385939847662434,
+      "loss": 2.5699,
+      "step": 2642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007383992346146251,
+      "loss": 2.5225,
+      "step": 2643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007382044376426582,
+      "loss": 2.3909,
+      "step": 2644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007380095938886,
+      "loss": 2.3627,
+      "step": 2645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007378147033907163,
+      "loss": 2.4571,
+      "step": 2646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007376197661872832,
+      "loss": 2.5015,
+      "step": 2647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007374247823165851,
+      "loss": 2.4626,
+      "step": 2648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007372297518169154,
+      "loss": 2.5985,
+      "step": 2649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007370346747265777,
+      "loss": 2.5201,
+      "step": 2650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007368395510838838,
+      "loss": 2.5487,
+      "step": 2651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007366443809271547,
+      "loss": 2.5174,
+      "step": 2652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007364491642947213,
+      "loss": 2.4468,
+      "step": 2653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007362539012249228,
+      "loss": 2.4,
+      "step": 2654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007360585917561078,
+      "loss": 2.4487,
+      "step": 2655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007358632359266341,
+      "loss": 2.3165,
+      "step": 2656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007356678337748687,
+      "loss": 2.4317,
+      "step": 2657
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007354723853391875,
+      "loss": 2.4321,
+      "step": 2658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007352768906579753,
+      "loss": 2.4894,
+      "step": 2659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007350813497696265,
+      "loss": 2.5573,
+      "step": 2660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007348857627125442,
+      "loss": 2.5833,
+      "step": 2661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007346901295251406,
+      "loss": 2.4243,
+      "step": 2662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007344944502458371,
+      "loss": 2.544,
+      "step": 2663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007342987249130641,
+      "loss": 2.6683,
+      "step": 2664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007341029535652609,
+      "loss": 2.5758,
+      "step": 2665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000733907136240876,
+      "loss": 2.4003,
+      "step": 2666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007337112729783671,
+      "loss": 2.5539,
+      "step": 2667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007335153638162004,
+      "loss": 2.5031,
+      "step": 2668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007333194087928515,
+      "loss": 2.357,
+      "step": 2669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000733123407946805,
+      "loss": 2.3477,
+      "step": 2670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007329273613165546,
+      "loss": 2.3633,
+      "step": 2671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007327312689406023,
+      "loss": 2.5739,
+      "step": 2672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007325351308574602,
+      "loss": 2.3746,
+      "step": 2673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007323389471056485,
+      "loss": 2.3664,
+      "step": 2674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007321427177236965,
+      "loss": 2.5956,
+      "step": 2675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007319464427501429,
+      "loss": 2.509,
+      "step": 2676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000731750122223535,
+      "loss": 2.4716,
+      "step": 2677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000731553756182429,
+      "loss": 2.5404,
+      "step": 2678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007313573446653902,
+      "loss": 2.4351,
+      "step": 2679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007311608877109928,
+      "loss": 2.5639,
+      "step": 2680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00073096438535782,
+      "loss": 2.415,
+      "step": 2681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007307678376444636,
+      "loss": 2.6072,
+      "step": 2682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007305712446095248,
+      "loss": 2.7389,
+      "step": 2683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000730374606291613,
+      "loss": 2.4825,
+      "step": 2684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007301779227293475,
+      "loss": 2.4225,
+      "step": 2685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007299811939613554,
+      "loss": 2.4817,
+      "step": 2686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007297844200262736,
+      "loss": 2.42,
+      "step": 2687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007295876009627472,
+      "loss": 2.6397,
+      "step": 2688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007293907368094304,
+      "loss": 2.4682,
+      "step": 2689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007291938276049865,
+      "loss": 2.4266,
+      "step": 2690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007289968733880874,
+      "loss": 2.4994,
+      "step": 2691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007287998741974136,
+      "loss": 2.5522,
+      "step": 2692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007286028300716548,
+      "loss": 2.4823,
+      "step": 2693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007284057410495097,
+      "loss": 2.5147,
+      "step": 2694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007282086071696851,
+      "loss": 2.5329,
+      "step": 2695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007280114284708975,
+      "loss": 2.5093,
+      "step": 2696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007278142049918715,
+      "loss": 2.425,
+      "step": 2697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007276169367713407,
+      "loss": 2.3889,
+      "step": 2698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007274196238480476,
+      "loss": 2.3415,
+      "step": 2699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007272222662607436,
+      "loss": 2.6197,
+      "step": 2700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007270248640481883,
+      "loss": 2.4327,
+      "step": 2701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007268274172491508,
+      "loss": 2.5309,
+      "step": 2702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007266299259024084,
+      "loss": 2.5162,
+      "step": 2703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007264323900467476,
+      "loss": 2.5433,
+      "step": 2704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726234809720963,
+      "loss": 2.5905,
+      "step": 2705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007260371849638587,
+      "loss": 2.4827,
+      "step": 2706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000725839515814247,
+      "loss": 2.5542,
+      "step": 2707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007256418023109492,
+      "loss": 2.5132,
+      "step": 2708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007254440444927948,
+      "loss": 2.5938,
+      "step": 2709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007252462423986229,
+      "loss": 2.5681,
+      "step": 2710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007250483960672805,
+      "loss": 2.5361,
+      "step": 2711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007248505055376236,
+      "loss": 2.6314,
+      "step": 2712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007246525708485169,
+      "loss": 2.4204,
+      "step": 2713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007244545920388337,
+      "loss": 2.417,
+      "step": 2714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000724256569147456,
+      "loss": 2.3653,
+      "step": 2715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007240585022132744,
+      "loss": 2.6029,
+      "step": 2716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007238603912751882,
+      "loss": 2.4267,
+      "step": 2717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007236622363721055,
+      "loss": 2.4477,
+      "step": 2718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007234640375429427,
+      "loss": 2.4984,
+      "step": 2719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000723265794826625,
+      "loss": 2.3979,
+      "step": 2720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007230675082620865,
+      "loss": 2.606,
+      "step": 2721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007228691778882692,
+      "loss": 2.6016,
+      "step": 2722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007226708037441243,
+      "loss": 2.4229,
+      "step": 2723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007224723858686117,
+      "loss": 2.4551,
+      "step": 2724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007222739243006993,
+      "loss": 2.5417,
+      "step": 2725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000722075419079364,
+      "loss": 2.6248,
+      "step": 2726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007218768702435911,
+      "loss": 2.299,
+      "step": 2727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007216782778323747,
+      "loss": 2.476,
+      "step": 2728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007214796418847172,
+      "loss": 2.4404,
+      "step": 2729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007212809624396295,
+      "loss": 2.6341,
+      "step": 2730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007210822395361318,
+      "loss": 2.516,
+      "step": 2731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007208834732132514,
+      "loss": 2.4243,
+      "step": 2732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007206846635100254,
+      "loss": 2.4647,
+      "step": 2733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007204858104654991,
+      "loss": 2.6793,
+      "step": 2734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007202869141187259,
+      "loss": 2.8023,
+      "step": 2735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007200879745087681,
+      "loss": 2.3661,
+      "step": 2736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007198889916746964,
+      "loss": 2.5741,
+      "step": 2737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007196899656555899,
+      "loss": 2.511,
+      "step": 2738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007194908964905363,
+      "loss": 2.3751,
+      "step": 2739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007192917842186317,
+      "loss": 2.3528,
+      "step": 2740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007190926288789807,
+      "loss": 2.3186,
+      "step": 2741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007188934305106967,
+      "loss": 2.4455,
+      "step": 2742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007186941891529007,
+      "loss": 2.5572,
+      "step": 2743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007184949048447227,
+      "loss": 2.2856,
+      "step": 2744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007182955776253015,
+      "loss": 2.4761,
+      "step": 2745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007180962075337834,
+      "loss": 2.3895,
+      "step": 2746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000717896794609324,
+      "loss": 2.3752,
+      "step": 2747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000717697338891087,
+      "loss": 2.3275,
+      "step": 2748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000717497840418244,
+      "loss": 2.3981,
+      "step": 2749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007172982992299758,
+      "loss": 2.4174,
+      "step": 2750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007170987153654712,
+      "loss": 2.4551,
+      "step": 2751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007168990888639273,
+      "loss": 2.6231,
+      "step": 2752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007166994197645496,
+      "loss": 2.6611,
+      "step": 2753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007164997081065526,
+      "loss": 2.4453,
+      "step": 2754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007162999539291579,
+      "loss": 2.4955,
+      "step": 2755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007161001572715965,
+      "loss": 2.5788,
+      "step": 2756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007159003181731075,
+      "loss": 2.4839,
+      "step": 2757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000715700436672938,
+      "loss": 2.5323,
+      "step": 2758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000715500512810344,
+      "loss": 2.4739,
+      "step": 2759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007153005466245891,
+      "loss": 2.5181,
+      "step": 2760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007151005381549459,
+      "loss": 2.4299,
+      "step": 2761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007149004874406948,
+      "loss": 2.4806,
+      "step": 2762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007147003945211249,
+      "loss": 2.4557,
+      "step": 2763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007145002594355332,
+      "loss": 2.6029,
+      "step": 2764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007143000822232254,
+      "loss": 2.4967,
+      "step": 2765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007140998629235149,
+      "loss": 2.4244,
+      "step": 2766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007138996015757241,
+      "loss": 2.4693,
+      "step": 2767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000713699298219183,
+      "loss": 2.4494,
+      "step": 2768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007134989528932302,
+      "loss": 2.5574,
+      "step": 2769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007132985656372125,
+      "loss": 2.4102,
+      "step": 2770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007130981364904848,
+      "loss": 2.5994,
+      "step": 2771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007128976654924105,
+      "loss": 2.4056,
+      "step": 2772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007126971526823609,
+      "loss": 2.5196,
+      "step": 2773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007124965980997156,
+      "loss": 2.624,
+      "step": 2774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007122960017838626,
+      "loss": 2.3026,
+      "step": 2775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007120953637741978,
+      "loss": 2.5639,
+      "step": 2776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007118946841101258,
+      "loss": 2.4492,
+      "step": 2777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007116939628310587,
+      "loss": 2.3885,
+      "step": 2778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000711493199976417,
+      "loss": 2.5101,
+      "step": 2779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007112923955856298,
+      "loss": 2.3936,
+      "step": 2780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007110915496981339,
+      "loss": 2.3815,
+      "step": 2781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007108906623533741,
+      "loss": 2.3287,
+      "step": 2782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007106897335908038,
+      "loss": 2.3309,
+      "step": 2783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007104887634498847,
+      "loss": 2.3973,
+      "step": 2784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007102877519700857,
+      "loss": 2.447,
+      "step": 2785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007100866991908845,
+      "loss": 2.5434,
+      "step": 2786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007098856051517673,
+      "loss": 2.3332,
+      "step": 2787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007096844698922274,
+      "loss": 2.3072,
+      "step": 2788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007094832934517668,
+      "loss": 2.5049,
+      "step": 2789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007092820758698954,
+      "loss": 2.3347,
+      "step": 2790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007090808171861319,
+      "loss": 2.4075,
+      "step": 2791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007088795174400015,
+      "loss": 2.6006,
+      "step": 2792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007086781766710389,
+      "loss": 2.4185,
+      "step": 2793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007084767949187865,
+      "loss": 2.4524,
+      "step": 2794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007082753722227943,
+      "loss": 2.4996,
+      "step": 2795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007080739086226208,
+      "loss": 2.4595,
+      "step": 2796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007078724041578325,
+      "loss": 2.502,
+      "step": 2797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007076708588680034,
+      "loss": 2.4444,
+      "step": 2798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007074692727927163,
+      "loss": 2.5514,
+      "step": 2799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007072676459715618,
+      "loss": 2.6171,
+      "step": 2800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007070659784441378,
+      "loss": 2.5864,
+      "step": 2801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007068642702500512,
+      "loss": 2.4203,
+      "step": 2802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000706662521428916,
+      "loss": 2.525,
+      "step": 2803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007064607320203551,
+      "loss": 2.2703,
+      "step": 2804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007062589020639986,
+      "loss": 2.5117,
+      "step": 2805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007060570315994845,
+      "loss": 2.5935,
+      "step": 2806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007058551206664598,
+      "loss": 2.5288,
+      "step": 2807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007056531693045784,
+      "loss": 2.3933,
+      "step": 2808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007054511775535022,
+      "loss": 2.4713,
+      "step": 2809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007052491454529017,
+      "loss": 2.5866,
+      "step": 2810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000705047073042455,
+      "loss": 2.5232,
+      "step": 2811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007048449603618475,
+      "loss": 2.3716,
+      "step": 2812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007046428074507736,
+      "loss": 2.5076,
+      "step": 2813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007044406143489348,
+      "loss": 2.4464,
+      "step": 2814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000704238381096041,
+      "loss": 2.4501,
+      "step": 2815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007040361077318095,
+      "loss": 2.457,
+      "step": 2816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007038337942959659,
+      "loss": 2.3826,
+      "step": 2817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007036314408282433,
+      "loss": 2.3387,
+      "step": 2818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007034290473683829,
+      "loss": 2.3581,
+      "step": 2819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007032266139561336,
+      "loss": 2.4035,
+      "step": 2820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007030241406312527,
+      "loss": 2.633,
+      "step": 2821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007028216274335044,
+      "loss": 2.5026,
+      "step": 2822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007026190744026615,
+      "loss": 2.5144,
+      "step": 2823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007024164815785042,
+      "loss": 2.4084,
+      "step": 2824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007022138490008206,
+      "loss": 2.5083,
+      "step": 2825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000702011176709407,
+      "loss": 2.3229,
+      "step": 2826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007018084647440667,
+      "loss": 2.6349,
+      "step": 2827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007016057131446116,
+      "loss": 2.4732,
+      "step": 2828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000701402921950861,
+      "loss": 2.4987,
+      "step": 2829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007012000912026419,
+      "loss": 2.6202,
+      "step": 2830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007009972209397893,
+      "loss": 2.5137,
+      "step": 2831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007007943112021459,
+      "loss": 2.3941,
+      "step": 2832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007005913620295617,
+      "loss": 2.626,
+      "step": 2833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007003883734618952,
+      "loss": 2.4566,
+      "step": 2834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007001853455390123,
+      "loss": 2.4393,
+      "step": 2835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006999822783007865,
+      "loss": 2.4365,
+      "step": 2836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006997791717870992,
+      "loss": 2.4979,
+      "step": 2837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006995760260378395,
+      "loss": 2.5551,
+      "step": 2838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006993728410929037,
+      "loss": 2.6249,
+      "step": 2839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006991696169921968,
+      "loss": 2.3341,
+      "step": 2840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006989663537756309,
+      "loss": 2.6287,
+      "step": 2841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006987630514831255,
+      "loss": 2.3586,
+      "step": 2842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006985597101546082,
+      "loss": 2.4915,
+      "step": 2843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006983563298300142,
+      "loss": 2.4021,
+      "step": 2844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006981529105492864,
+      "loss": 2.5247,
+      "step": 2845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006979494523523752,
+      "loss": 2.3997,
+      "step": 2846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006977459552792387,
+      "loss": 2.5655,
+      "step": 2847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006975424193698428,
+      "loss": 2.4909,
+      "step": 2848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006973388446641605,
+      "loss": 2.3564,
+      "step": 2849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006971352312021732,
+      "loss": 2.5191,
+      "step": 2850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006969315790238692,
+      "loss": 2.4621,
+      "step": 2851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000696727888169245,
+      "loss": 2.472,
+      "step": 2852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006965241586783042,
+      "loss": 2.4381,
+      "step": 2853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006963203905910584,
+      "loss": 2.429,
+      "step": 2854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006961165839475262,
+      "loss": 2.6136,
+      "step": 2855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006959127387877345,
+      "loss": 2.4602,
+      "step": 2856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006957088551517173,
+      "loss": 2.3897,
+      "step": 2857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006955049330795162,
+      "loss": 2.4916,
+      "step": 2858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006953009726111805,
+      "loss": 2.4632,
+      "step": 2859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006950969737867669,
+      "loss": 2.4556,
+      "step": 2860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006948929366463397,
+      "loss": 2.4655,
+      "step": 2861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006946888612299709,
+      "loss": 2.6236,
+      "step": 2862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006944847475777394,
+      "loss": 2.5438,
+      "step": 2863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006942805957297325,
+      "loss": 2.5321,
+      "step": 2864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006940764057260443,
+      "loss": 2.4724,
+      "step": 2865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006938721776067767,
+      "loss": 2.4559,
+      "step": 2866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006936679114120389,
+      "loss": 2.6654,
+      "step": 2867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006934636071819479,
+      "loss": 2.4499,
+      "step": 2868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006932592649566278,
+      "loss": 2.4233,
+      "step": 2869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006930548847762104,
+      "loss": 2.409,
+      "step": 2870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006928504666808351,
+      "loss": 2.3846,
+      "step": 2871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006926460107106483,
+      "loss": 2.5107,
+      "step": 2872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006924415169058038,
+      "loss": 2.3987,
+      "step": 2873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006922369853064636,
+      "loss": 2.5159,
+      "step": 2874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006920324159527965,
+      "loss": 2.5277,
+      "step": 2875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006918278088849786,
+      "loss": 2.5539,
+      "step": 2876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000691623164143194,
+      "loss": 2.5707,
+      "step": 2877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006914184817676336,
+      "loss": 2.3486,
+      "step": 2878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006912137617984959,
+      "loss": 2.5354,
+      "step": 2879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006910090042759869,
+      "loss": 2.6177,
+      "step": 2880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006908042092403201,
+      "loss": 2.4778,
+      "step": 2881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006905993767317157,
+      "loss": 2.4623,
+      "step": 2882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006903945067904021,
+      "loss": 2.5354,
+      "step": 2883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006901895994566147,
+      "loss": 2.4578,
+      "step": 2884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000689984654770596,
+      "loss": 2.4888,
+      "step": 2885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006897796727725959,
+      "loss": 2.5442,
+      "step": 2886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006895746535028723,
+      "loss": 2.3752,
+      "step": 2887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006893695970016894,
+      "loss": 2.4779,
+      "step": 2888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006891645033093196,
+      "loss": 2.5573,
+      "step": 2889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006889593724660418,
+      "loss": 2.3252,
+      "step": 2890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006887542045121431,
+      "loss": 2.5169,
+      "step": 2891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000688548999487917,
+      "loss": 2.4919,
+      "step": 2892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006883437574336648,
+      "loss": 2.4369,
+      "step": 2893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006881384783896951,
+      "loss": 2.4194,
+      "step": 2894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006879331623963237,
+      "loss": 2.3575,
+      "step": 2895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000687727809493873,
+      "loss": 2.2123,
+      "step": 2896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006875224197226737,
+      "loss": 2.3883,
+      "step": 2897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006873169931230633,
+      "loss": 2.458,
+      "step": 2898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006871115297353862,
+      "loss": 2.4507,
+      "step": 2899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006869060295999945,
+      "loss": 2.4854,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006867004927572474,
+      "loss": 2.315,
+      "step": 2901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006864949192475112,
+      "loss": 2.4777,
+      "step": 2902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006862893091111594,
+      "loss": 2.3959,
+      "step": 2903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006860836623885727,
+      "loss": 2.4605,
+      "step": 2904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006858779791201391,
+      "loss": 2.4711,
+      "step": 2905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006856722593462538,
+      "loss": 2.2359,
+      "step": 2906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006854665031073189,
+      "loss": 2.3525,
+      "step": 2907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000685260710443744,
+      "loss": 2.3597,
+      "step": 2908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006850548813959455,
+      "loss": 2.451,
+      "step": 2909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006848490160043472,
+      "loss": 2.4287,
+      "step": 2910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006846431143093802,
+      "loss": 2.4462,
+      "step": 2911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006844371763514822,
+      "loss": 2.5817,
+      "step": 2912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006842312021710983,
+      "loss": 2.3136,
+      "step": 2913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006840251918086812,
+      "loss": 2.4357,
+      "step": 2914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006838191453046898,
+      "loss": 2.431,
+      "step": 2915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006836130626995905,
+      "loss": 2.3947,
+      "step": 2916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006834069440338573,
+      "loss": 2.437,
+      "step": 2917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006832007893479703,
+      "loss": 2.4479,
+      "step": 2918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006829945986824176,
+      "loss": 2.4504,
+      "step": 2919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006827883720776933,
+      "loss": 2.5699,
+      "step": 2920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006825821095743002,
+      "loss": 2.4355,
+      "step": 2921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006823758112127465,
+      "loss": 2.3785,
+      "step": 2922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006821694770335481,
+      "loss": 2.3743,
+      "step": 2923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006819631070772283,
+      "loss": 2.3381,
+      "step": 2924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006817567013843169,
+      "loss": 2.4623,
+      "step": 2925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006815502599953506,
+      "loss": 2.5346,
+      "step": 2926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006813437829508738,
+      "loss": 2.4159,
+      "step": 2927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006811372702914371,
+      "loss": 2.4266,
+      "step": 2928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006809307220575991,
+      "loss": 2.5515,
+      "step": 2929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006807241382899242,
+      "loss": 2.4409,
+      "step": 2930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006805175190289847,
+      "loss": 2.3989,
+      "step": 2931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006803108643153592,
+      "loss": 2.3964,
+      "step": 2932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000680104174189634,
+      "loss": 2.3746,
+      "step": 2933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006798974486924016,
+      "loss": 2.569,
+      "step": 2934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006796906878642621,
+      "loss": 2.4191,
+      "step": 2935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006794838917458221,
+      "loss": 2.3798,
+      "step": 2936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000679277060377695,
+      "loss": 2.421,
+      "step": 2937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006790701938005018,
+      "loss": 2.4491,
+      "step": 2938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006788632920548698,
+      "loss": 2.5782,
+      "step": 2939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006786563551814333,
+      "loss": 2.4627,
+      "step": 2940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000678449383220834,
+      "loss": 2.4647,
+      "step": 2941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006782423762137197,
+      "loss": 2.3358,
+      "step": 2942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006780353342007454,
+      "loss": 2.4703,
+      "step": 2943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006778282572225736,
+      "loss": 2.528,
+      "step": 2944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006776211453198728,
+      "loss": 2.4455,
+      "step": 2945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006774139985333185,
+      "loss": 2.3012,
+      "step": 2946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006772068169035935,
+      "loss": 2.504,
+      "step": 2947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000676999600471387,
+      "loss": 2.3791,
+      "step": 2948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006767923492773955,
+      "loss": 2.4742,
+      "step": 2949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006765850633623215,
+      "loss": 2.5111,
+      "step": 2950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006763777427668755,
+      "loss": 2.5671,
+      "step": 2951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006761703875317738,
+      "loss": 2.6391,
+      "step": 2952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00067596299769774,
+      "loss": 2.3263,
+      "step": 2953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006757555733055041,
+      "loss": 2.3524,
+      "step": 2954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006755481143958035,
+      "loss": 2.4998,
+      "step": 2955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006753406210093818,
+      "loss": 2.4035,
+      "step": 2956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006751330931869897,
+      "loss": 2.531,
+      "step": 2957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006749255309693846,
+      "loss": 2.4697,
+      "step": 2958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006747179343973307,
+      "loss": 2.4906,
+      "step": 2959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006745103035115985,
+      "loss": 2.5877,
+      "step": 2960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000674302638352966,
+      "loss": 2.3788,
+      "step": 2961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006740949389622174,
+      "loss": 2.5087,
+      "step": 2962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006738872053801437,
+      "loss": 2.4322,
+      "step": 2963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006736794376475428,
+      "loss": 2.3794,
+      "step": 2964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000673471635805219,
+      "loss": 2.3476,
+      "step": 2965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006732637998939836,
+      "loss": 2.3565,
+      "step": 2966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006730559299546546,
+      "loss": 2.5144,
+      "step": 2967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006728480260280565,
+      "loss": 2.5082,
+      "step": 2968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006726400881550203,
+      "loss": 2.5215,
+      "step": 2969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000672432116376384,
+      "loss": 2.5277,
+      "step": 2970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006722241107329925,
+      "loss": 2.4647,
+      "step": 2971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006720160712656965,
+      "loss": 2.6049,
+      "step": 2972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000671807998015354,
+      "loss": 2.4816,
+      "step": 2973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006715998910228296,
+      "loss": 2.3413,
+      "step": 2974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006713917503289946,
+      "loss": 2.4611,
+      "step": 2975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006711835759747264,
+      "loss": 2.4094,
+      "step": 2976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006709753680009092,
+      "loss": 2.5509,
+      "step": 2977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006707671264484345,
+      "loss": 2.2821,
+      "step": 2978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006705588513581994,
+      "loss": 2.5482,
+      "step": 2979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006703505427711081,
+      "loss": 2.451,
+      "step": 2980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006701422007280714,
+      "loss": 2.4133,
+      "step": 2981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006699338252700067,
+      "loss": 2.4385,
+      "step": 2982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006697254164378375,
+      "loss": 2.4113,
+      "step": 2983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006695169742724943,
+      "loss": 2.4546,
+      "step": 2984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006693084988149141,
+      "loss": 2.5974,
+      "step": 2985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006690999901060406,
+      "loss": 2.4397,
+      "step": 2986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006688914481868235,
+      "loss": 2.3717,
+      "step": 2987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006686828730982192,
+      "loss": 2.523,
+      "step": 2988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006684742648811913,
+      "loss": 2.4265,
+      "step": 2989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006682656235767089,
+      "loss": 2.5085,
+      "step": 2990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000668056949225748,
+      "loss": 2.3695,
+      "step": 2991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006678482418692915,
+      "loss": 2.4572,
+      "step": 2992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006676395015483281,
+      "loss": 2.3743,
+      "step": 2993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006674307283038534,
+      "loss": 2.5069,
+      "step": 2994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006672219221768694,
+      "loss": 2.5424,
+      "step": 2995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006670130832083845,
+      "loss": 2.3543,
+      "step": 2996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006668042114394136,
+      "loss": 2.5298,
+      "step": 2997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000666595306910978,
+      "loss": 2.4818,
+      "step": 2998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006663863696641052,
+      "loss": 2.379,
+      "step": 2999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006661773997398298,
+      "loss": 2.3236,
+      "step": 3000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006659683971791922,
+      "loss": 2.4166,
+      "step": 3001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006657593620232393,
+      "loss": 2.403,
+      "step": 3002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006655502943130245,
+      "loss": 2.4258,
+      "step": 3003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006653411940896077,
+      "loss": 2.3701,
+      "step": 3004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006651320613940552,
+      "loss": 2.3635,
+      "step": 3005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006649228962674393,
+      "loss": 2.3936,
+      "step": 3006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006647136987508389,
+      "loss": 2.4549,
+      "step": 3007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006645044688853395,
+      "loss": 2.2911,
+      "step": 3008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006642952067120327,
+      "loss": 2.3534,
+      "step": 3009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006640859122720163,
+      "loss": 2.4273,
+      "step": 3010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006638765856063948,
+      "loss": 2.4381,
+      "step": 3011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006636672267562788,
+      "loss": 2.5561,
+      "step": 3012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006634578357627854,
+      "loss": 2.321,
+      "step": 3013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006632484126670377,
+      "loss": 2.3776,
+      "step": 3014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006630389575101651,
+      "loss": 2.3945,
+      "step": 3015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000662829470333304,
+      "loss": 2.3208,
+      "step": 3016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006626199511775962,
+      "loss": 2.3286,
+      "step": 3017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006624104000841902,
+      "loss": 2.4152,
+      "step": 3018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000662200817094241,
+      "loss": 2.6144,
+      "step": 3019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006619912022489091,
+      "loss": 2.3458,
+      "step": 3020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006617815555893621,
+      "loss": 2.4071,
+      "step": 3021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006615718771567736,
+      "loss": 2.5471,
+      "step": 3022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006613621669923228,
+      "loss": 2.503,
+      "step": 3023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006611524251371963,
+      "loss": 2.3721,
+      "step": 3024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000660942651632586,
+      "loss": 2.4212,
+      "step": 3025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006607328465196901,
+      "loss": 2.5351,
+      "step": 3026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006605230098397135,
+      "loss": 2.5044,
+      "step": 3027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000660313141633867,
+      "loss": 2.3282,
+      "step": 3028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006601032419433676,
+      "loss": 2.3679,
+      "step": 3029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006598933108094382,
+      "loss": 2.4347,
+      "step": 3030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006596833482733086,
+      "loss": 2.396,
+      "step": 3031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006594733543762142,
+      "loss": 2.4088,
+      "step": 3032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006592633291593965,
+      "loss": 2.2939,
+      "step": 3033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006590532726641035,
+      "loss": 2.5035,
+      "step": 3034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000658843184931589,
+      "loss": 2.4825,
+      "step": 3035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006586330660031135,
+      "loss": 2.5233,
+      "step": 3036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006584229159199427,
+      "loss": 2.3621,
+      "step": 3037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006582127347233497,
+      "loss": 2.4291,
+      "step": 3038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006580025224546122,
+      "loss": 2.459,
+      "step": 3039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006577922791550151,
+      "loss": 2.4302,
+      "step": 3040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006575820048658495,
+      "loss": 2.584,
+      "step": 3041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006573716996284113,
+      "loss": 2.3658,
+      "step": 3042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006571613634840041,
+      "loss": 2.3322,
+      "step": 3043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006569509964739365,
+      "loss": 2.312,
+      "step": 3044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006567405986395232,
+      "loss": 2.4391,
+      "step": 3045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006565301700220858,
+      "loss": 2.5507,
+      "step": 3046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006563197106629508,
+      "loss": 2.294,
+      "step": 3047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006561092206034517,
+      "loss": 2.4185,
+      "step": 3048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006558986998849275,
+      "loss": 2.382,
+      "step": 3049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006556881485487233,
+      "loss": 2.4828,
+      "step": 3050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006554775666361902,
+      "loss": 2.4762,
+      "step": 3051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006552669541886858,
+      "loss": 2.4783,
+      "step": 3052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006550563112475729,
+      "loss": 2.4182,
+      "step": 3053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006548456378542204,
+      "loss": 2.4167,
+      "step": 3054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006546349340500042,
+      "loss": 2.4616,
+      "step": 3055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006544241998763047,
+      "loss": 2.3475,
+      "step": 3056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006542134353745093,
+      "loss": 2.4558,
+      "step": 3057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000654002640586011,
+      "loss": 2.467,
+      "step": 3058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006537918155522088,
+      "loss": 2.356,
+      "step": 3059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006535809603145076,
+      "loss": 2.4729,
+      "step": 3060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006533700749143181,
+      "loss": 2.5895,
+      "step": 3061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006531591593930574,
+      "loss": 2.4423,
+      "step": 3062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006529482137921479,
+      "loss": 2.3111,
+      "step": 3063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006527372381530183,
+      "loss": 2.3273,
+      "step": 3064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006525262325171032,
+      "loss": 2.4305,
+      "step": 3065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006523151969258429,
+      "loss": 2.3917,
+      "step": 3066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006521041314206835,
+      "loss": 2.4991,
+      "step": 3067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006518930360430774,
+      "loss": 2.4223,
+      "step": 3068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006516819108344827,
+      "loss": 2.321,
+      "step": 3069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000651470755836363,
+      "loss": 2.5479,
+      "step": 3070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006512595710901882,
+      "loss": 2.3125,
+      "step": 3071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006510483566374338,
+      "loss": 2.5453,
+      "step": 3072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006508371125195812,
+      "loss": 2.3968,
+      "step": 3073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006506258387781175,
+      "loss": 2.39,
+      "step": 3074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006504145354545359,
+      "loss": 2.528,
+      "step": 3075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006502032025903355,
+      "loss": 2.4427,
+      "step": 3076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006499918402270204,
+      "loss": 2.4703,
+      "step": 3077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006497804484061014,
+      "loss": 2.4258,
+      "step": 3078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006495690271690948,
+      "loss": 2.4582,
+      "step": 3079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006493575765575225,
+      "loss": 2.5534,
+      "step": 3080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006491460966129121,
+      "loss": 2.4435,
+      "step": 3081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006489345873767975,
+      "loss": 2.4804,
+      "step": 3082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006487230488907177,
+      "loss": 2.3938,
+      "step": 3083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006485114811962177,
+      "loss": 2.4937,
+      "step": 3084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006482998843348487,
+      "loss": 2.3682,
+      "step": 3085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006480882583481668,
+      "loss": 2.4574,
+      "step": 3086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006478766032777343,
+      "loss": 2.4015,
+      "step": 3087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006476649191651192,
+      "loss": 2.4467,
+      "step": 3088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006474532060518951,
+      "loss": 2.3626,
+      "step": 3089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006472414639796416,
+      "loss": 2.355,
+      "step": 3090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006470296929899432,
+      "loss": 2.3091,
+      "step": 3091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006468178931243911,
+      "loss": 2.3634,
+      "step": 3092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006466060644245814,
+      "loss": 2.321,
+      "step": 3093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006463942069321161,
+      "loss": 2.3431,
+      "step": 3094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006461823206886031,
+      "loss": 2.3683,
+      "step": 3095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006459704057356557,
+      "loss": 2.4896,
+      "step": 3096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006457584621148925,
+      "loss": 2.4322,
+      "step": 3097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006455464898679386,
+      "loss": 2.4786,
+      "step": 3098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006453344890364241,
+      "loss": 2.3971,
+      "step": 3099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006451224596619847,
+      "loss": 2.3119,
+      "step": 3100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006449104017862619,
+      "loss": 2.2989,
+      "step": 3101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006446983154509028,
+      "loss": 2.3305,
+      "step": 3102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00064448620069756,
+      "loss": 2.5452,
+      "step": 3103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006442740575678918,
+      "loss": 2.4929,
+      "step": 3104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006440618861035618,
+      "loss": 2.5116,
+      "step": 3105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006438496863462395,
+      "loss": 2.4951,
+      "step": 3106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006436374583375997,
+      "loss": 2.3569,
+      "step": 3107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000643425202119323,
+      "loss": 2.4417,
+      "step": 3108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006432129177330955,
+      "loss": 2.4202,
+      "step": 3109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006430006052206082,
+      "loss": 2.4754,
+      "step": 3110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006427882646235587,
+      "loss": 2.5373,
+      "step": 3111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006425758959836494,
+      "loss": 2.3623,
+      "step": 3112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006423634993425883,
+      "loss": 2.2881,
+      "step": 3113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006421510747420891,
+      "loss": 2.4565,
+      "step": 3114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006419386222238706,
+      "loss": 2.4957,
+      "step": 3115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641726141829658,
+      "loss": 2.4813,
+      "step": 3116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006415136336011804,
+      "loss": 2.5379,
+      "step": 3117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641301097580174,
+      "loss": 2.4344,
+      "step": 3118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006410885338083796,
+      "loss": 2.3208,
+      "step": 3119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006408759423275434,
+      "loss": 2.5249,
+      "step": 3120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006406633231794173,
+      "loss": 2.5037,
+      "step": 3121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006404506764057589,
+      "loss": 2.4752,
+      "step": 3122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006402380020483305,
+      "loss": 2.3459,
+      "step": 3123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006400253001489002,
+      "loss": 2.4284,
+      "step": 3124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006398125707492419,
+      "loss": 2.296,
+      "step": 3125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006395998138911342,
+      "loss": 2.5649,
+      "step": 3126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006393870296163615,
+      "loss": 2.5236,
+      "step": 3127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006391742179667137,
+      "loss": 2.3263,
+      "step": 3128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006389613789839856,
+      "loss": 2.2864,
+      "step": 3129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006387485127099781,
+      "loss": 2.2788,
+      "step": 3130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006385356191864963,
+      "loss": 2.4138,
+      "step": 3131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006383226984553518,
+      "loss": 2.4432,
+      "step": 3132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006381097505583612,
+      "loss": 2.3551,
+      "step": 3133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006378967755373459,
+      "loss": 2.3716,
+      "step": 3134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006376837734341338,
+      "loss": 2.4802,
+      "step": 3135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006374707442905567,
+      "loss": 2.295,
+      "step": 3136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006372576881484526,
+      "loss": 2.4555,
+      "step": 3137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006370446050496648,
+      "loss": 2.3826,
+      "step": 3138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006368314950360416,
+      "loss": 2.3908,
+      "step": 3139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006366183581494364,
+      "loss": 2.4262,
+      "step": 3140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006364051944317084,
+      "loss": 2.2684,
+      "step": 3141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006361920039247219,
+      "loss": 2.3118,
+      "step": 3142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006359787866703462,
+      "loss": 2.2428,
+      "step": 3143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006357655427104562,
+      "loss": 2.4999,
+      "step": 3144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006355522720869318,
+      "loss": 2.5078,
+      "step": 3145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006353389748416583,
+      "loss": 2.5049,
+      "step": 3146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000635125651016526,
+      "loss": 2.2865,
+      "step": 3147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006349123006534305,
+      "loss": 2.4711,
+      "step": 3148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006346989237942731,
+      "loss": 2.5677,
+      "step": 3149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006344855204809596,
+      "loss": 2.3015,
+      "step": 3150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006342720907554012,
+      "loss": 2.4314,
+      "step": 3151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006340586346595146,
+      "loss": 2.4456,
+      "step": 3152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006338451522352212,
+      "loss": 2.3899,
+      "step": 3153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000633631643524448,
+      "loss": 2.4231,
+      "step": 3154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006334181085691271,
+      "loss": 2.4015,
+      "step": 3155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006332045474111951,
+      "loss": 2.4646,
+      "step": 3156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006329909600925949,
+      "loss": 2.388,
+      "step": 3157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006327773466552737,
+      "loss": 2.4913,
+      "step": 3158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006325637071411839,
+      "loss": 2.4078,
+      "step": 3159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006323500415922834,
+      "loss": 2.3362,
+      "step": 3160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006321363500505347,
+      "loss": 2.5312,
+      "step": 3161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006319226325579058,
+      "loss": 2.3987,
+      "step": 3162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006317088891563701,
+      "loss": 2.3788,
+      "step": 3163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000631495119887905,
+      "loss": 2.2755,
+      "step": 3164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000631281324794494,
+      "loss": 2.4512,
+      "step": 3165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006310675039181253,
+      "loss": 2.3454,
+      "step": 3166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006308536573007923,
+      "loss": 2.5006,
+      "step": 3167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006306397849844931,
+      "loss": 2.3676,
+      "step": 3168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006304258870112314,
+      "loss": 2.5191,
+      "step": 3169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006302119634230152,
+      "loss": 2.3319,
+      "step": 3170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006299980142618583,
+      "loss": 2.4516,
+      "step": 3171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006297840395697792,
+      "loss": 2.5644,
+      "step": 3172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006295700393888012,
+      "loss": 2.4082,
+      "step": 3173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006293560137609528,
+      "loss": 2.5475,
+      "step": 3174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006291419627282677,
+      "loss": 2.4987,
+      "step": 3175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006289278863327844,
+      "loss": 2.4221,
+      "step": 3176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000628713784616546,
+      "loss": 2.5011,
+      "step": 3177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006284996576216013,
+      "loss": 2.4288,
+      "step": 3178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006282855053900038,
+      "loss": 2.286,
+      "step": 3179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006280713279638116,
+      "loss": 2.3549,
+      "step": 3180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006278571253850878,
+      "loss": 2.281,
+      "step": 3181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006276428976959013,
+      "loss": 2.2292,
+      "step": 3182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006274286449383247,
+      "loss": 2.3855,
+      "step": 3183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006272143671544365,
+      "loss": 2.4438,
+      "step": 3184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006270000643863195,
+      "loss": 2.6213,
+      "step": 3185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006267857366760617,
+      "loss": 2.444,
+      "step": 3186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006265713840657559,
+      "loss": 2.4781,
+      "step": 3187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006263570065974998,
+      "loss": 2.4108,
+      "step": 3188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000626142604313396,
+      "loss": 2.5994,
+      "step": 3189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006259281772555521,
+      "loss": 2.3402,
+      "step": 3190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006257137254660803,
+      "loss": 2.5545,
+      "step": 3191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006254992489870978,
+      "loss": 2.4053,
+      "step": 3192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006252847478607269,
+      "loss": 2.3708,
+      "step": 3193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006250702221290941,
+      "loss": 2.5484,
+      "step": 3194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006248556718343313,
+      "loss": 2.4531,
+      "step": 3195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006246410970185751,
+      "loss": 2.5369,
+      "step": 3196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006244264977239668,
+      "loss": 2.3813,
+      "step": 3197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006242118739926525,
+      "loss": 2.3335,
+      "step": 3198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006239972258667834,
+      "loss": 2.3571,
+      "step": 3199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006237825533885149,
+      "loss": 2.549,
+      "step": 3200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006235678566000078,
+      "loss": 2.4349,
+      "step": 3201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006233531355434274,
+      "loss": 2.3868,
+      "step": 3202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006231383902609438,
+      "loss": 2.3341,
+      "step": 3203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006229236207947315,
+      "loss": 2.3774,
+      "step": 3204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006227088271869704,
+      "loss": 2.3596,
+      "step": 3205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000622494009479845,
+      "loss": 2.5148,
+      "step": 3206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006222791677155439,
+      "loss": 2.4398,
+      "step": 3207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006220643019362612,
+      "loss": 2.3898,
+      "step": 3208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006218494121841954,
+      "loss": 2.2962,
+      "step": 3209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006216344985015495,
+      "loss": 2.3836,
+      "step": 3210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006214195609305315,
+      "loss": 2.3235,
+      "step": 3211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006212045995133543,
+      "loss": 2.3043,
+      "step": 3212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006209896142922348,
+      "loss": 2.3837,
+      "step": 3213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000620774605309395,
+      "loss": 2.4707,
+      "step": 3214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006205595726070617,
+      "loss": 2.426,
+      "step": 3215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006203445162274662,
+      "loss": 2.3912,
+      "step": 3216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006201294362128442,
+      "loss": 2.393,
+      "step": 3217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006199143326054365,
+      "loss": 2.4388,
+      "step": 3218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006196992054474882,
+      "loss": 2.3663,
+      "step": 3219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006194840547812491,
+      "loss": 2.5334,
+      "step": 3220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006192688806489736,
+      "loss": 2.4475,
+      "step": 3221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006190536830929213,
+      "loss": 2.4479,
+      "step": 3222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006188384621553551,
+      "loss": 2.5346,
+      "step": 3223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006186232178785436,
+      "loss": 2.4983,
+      "step": 3224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006184079503047598,
+      "loss": 2.4004,
+      "step": 3225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006181926594762808,
+      "loss": 2.3765,
+      "step": 3226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006179773454353888,
+      "loss": 2.5023,
+      "step": 3227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006177620082243701,
+      "loss": 2.3383,
+      "step": 3228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000617546647885516,
+      "loss": 2.3818,
+      "step": 3229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006173312644611223,
+      "loss": 2.3897,
+      "step": 3230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006171158579934888,
+      "loss": 2.5206,
+      "step": 3231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006169004285249203,
+      "loss": 2.3441,
+      "step": 3232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006166849760977261,
+      "loss": 2.3803,
+      "step": 3233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006164695007542198,
+      "loss": 2.2653,
+      "step": 3234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006162540025367197,
+      "loss": 2.319,
+      "step": 3235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006160384814875486,
+      "loss": 2.4632,
+      "step": 3236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006158229376490334,
+      "loss": 2.3401,
+      "step": 3237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006156073710635061,
+      "loss": 2.373,
+      "step": 3238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006153917817733025,
+      "loss": 2.5397,
+      "step": 3239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006151761698207636,
+      "loss": 2.4396,
+      "step": 3240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006149605352482342,
+      "loss": 2.4432,
+      "step": 3241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006147448780980637,
+      "loss": 2.4588,
+      "step": 3242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006145291984126063,
+      "loss": 2.364,
+      "step": 3243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006143134962342201,
+      "loss": 2.3101,
+      "step": 3244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000614097771605268,
+      "loss": 2.3603,
+      "step": 3245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000613882024568117,
+      "loss": 2.4843,
+      "step": 3246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000613666255165139,
+      "loss": 2.3658,
+      "step": 3247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006134504634387097,
+      "loss": 2.3256,
+      "step": 3248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006132346494312096,
+      "loss": 2.341,
+      "step": 3249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006130188131850234,
+      "loss": 2.3473,
+      "step": 3250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006128029547425399,
+      "loss": 2.2921,
+      "step": 3251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006125870741461531,
+      "loss": 2.5426,
+      "step": 3252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006123711714382605,
+      "loss": 2.413,
+      "step": 3253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006121552466612641,
+      "loss": 2.2718,
+      "step": 3254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006119392998575707,
+      "loss": 2.4868,
+      "step": 3255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000611723331069591,
+      "loss": 2.4174,
+      "step": 3256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00061150734033974,
+      "loss": 2.3217,
+      "step": 3257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006112913277104372,
+      "loss": 2.4673,
+      "step": 3258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006110752932241063,
+      "loss": 2.3842,
+      "step": 3259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006108592369231757,
+      "loss": 2.5107,
+      "step": 3260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006106431588500773,
+      "loss": 2.4593,
+      "step": 3261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006104270590472476,
+      "loss": 2.3932,
+      "step": 3262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000610210937557128,
+      "loss": 2.5353,
+      "step": 3263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006099947944221631,
+      "loss": 2.4696,
+      "step": 3264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006097786296848024,
+      "loss": 2.3334,
+      "step": 3265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006095624433874997,
+      "loss": 2.5708,
+      "step": 3266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006093462355727127,
+      "loss": 2.5191,
+      "step": 3267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006091300062829033,
+      "loss": 2.5495,
+      "step": 3268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006089137555605381,
+      "loss": 2.2429,
+      "step": 3269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006086974834480874,
+      "loss": 2.3733,
+      "step": 3270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006084811899880261,
+      "loss": 2.3195,
+      "step": 3271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006082648752228328,
+      "loss": 2.4581,
+      "step": 3272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006080485391949908,
+      "loss": 2.5033,
+      "step": 3273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006078321819469874,
+      "loss": 2.3852,
+      "step": 3274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006076158035213137,
+      "loss": 2.4171,
+      "step": 3275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006073994039604656,
+      "loss": 2.5382,
+      "step": 3276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006071829833069427,
+      "loss": 2.4218,
+      "step": 3277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006069665416032487,
+      "loss": 2.4509,
+      "step": 3278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000606750078891892,
+      "loss": 2.3832,
+      "step": 3279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006065335952153845,
+      "loss": 2.2851,
+      "step": 3280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006063170906162424,
+      "loss": 2.4038,
+      "step": 3281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006061005651369861,
+      "loss": 2.5137,
+      "step": 3282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006058840188201402,
+      "loss": 2.4752,
+      "step": 3283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000605667451708233,
+      "loss": 2.3911,
+      "step": 3284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006054508638437974,
+      "loss": 2.3803,
+      "step": 3285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006052342552693698,
+      "loss": 2.4156,
+      "step": 3286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006050176260274914,
+      "loss": 2.3986,
+      "step": 3287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006048009761607065,
+      "loss": 2.3419,
+      "step": 3288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006045843057115644,
+      "loss": 2.3858,
+      "step": 3289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000604367614722618,
+      "loss": 2.4624,
+      "step": 3290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006041509032364242,
+      "loss": 2.3603,
+      "step": 3291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006039341712955438,
+      "loss": 2.5012,
+      "step": 3292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006037174189425421,
+      "loss": 2.6615,
+      "step": 3293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000603500646219988,
+      "loss": 2.4278,
+      "step": 3294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006032838531704545,
+      "loss": 2.2853,
+      "step": 3295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006030670398365185,
+      "loss": 2.4373,
+      "step": 3296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006028502062607611,
+      "loss": 2.4043,
+      "step": 3297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006026333524857675,
+      "loss": 2.4273,
+      "step": 3298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006024164785541262,
+      "loss": 2.3707,
+      "step": 3299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006021995845084304,
+      "loss": 2.3247,
+      "step": 3300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006019826703912767,
+      "loss": 2.3784,
+      "step": 3301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006017657362452662,
+      "loss": 2.5208,
+      "step": 3302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006015487821130033,
+      "loss": 2.4466,
+      "step": 3303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000601331808037097,
+      "loss": 2.4166,
+      "step": 3304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006011148140601594,
+      "loss": 2.4368,
+      "step": 3305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006008978002248073,
+      "loss": 2.2785,
+      "step": 3306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006006807665736611,
+      "loss": 2.6144,
+      "step": 3307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006004637131493447,
+      "loss": 2.478,
+      "step": 3308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006002466399944867,
+      "loss": 2.4174,
+      "step": 3309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000600029547151719,
+      "loss": 2.4226,
+      "step": 3310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005998124346636771,
+      "loss": 2.2481,
+      "step": 3311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005995953025730013,
+      "loss": 2.5177,
+      "step": 3312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005993781509223349,
+      "loss": 2.4865,
+      "step": 3313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005991609797543253,
+      "loss": 2.3851,
+      "step": 3314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005989437891116238,
+      "loss": 2.427,
+      "step": 3315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005987265790368856,
+      "loss": 2.4222,
+      "step": 3316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005985093495727696,
+      "loss": 2.4121,
+      "step": 3317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005982921007619383,
+      "loss": 2.3908,
+      "step": 3318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005980748326470585,
+      "loss": 2.5571,
+      "step": 3319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005978575452708003,
+      "loss": 2.3928,
+      "step": 3320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005976402386758381,
+      "loss": 2.5383,
+      "step": 3321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005974229129048493,
+      "loss": 2.2488,
+      "step": 3322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005972055680005159,
+      "loss": 2.4924,
+      "step": 3323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005969882040055231,
+      "loss": 2.3331,
+      "step": 3324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00059677082096256,
+      "loss": 2.3943,
+      "step": 3325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005965534189143199,
+      "loss": 2.3151,
+      "step": 3326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005963359979034989,
+      "loss": 2.3903,
+      "step": 3327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005961185579727975,
+      "loss": 2.4216,
+      "step": 3328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005959010991649199,
+      "loss": 2.4216,
+      "step": 3329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005956836215225736,
+      "loss": 2.4696,
+      "step": 3330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005954661250884703,
+      "loss": 2.3481,
+      "step": 3331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000595248609905325,
+      "loss": 2.5459,
+      "step": 3332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005950310760158566,
+      "loss": 2.527,
+      "step": 3333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005948135234627877,
+      "loss": 2.5008,
+      "step": 3334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005945959522888441,
+      "loss": 2.3467,
+      "step": 3335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005943783625367562,
+      "loss": 2.3751,
+      "step": 3336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000594160754249257,
+      "loss": 2.3016,
+      "step": 3337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005939431274690837,
+      "loss": 2.2731,
+      "step": 3338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005937254822389773,
+      "loss": 2.4546,
+      "step": 3339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005935078186016819,
+      "loss": 2.3472,
+      "step": 3340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005932901365999456,
+      "loss": 2.527,
+      "step": 3341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005930724362765199,
+      "loss": 2.4133,
+      "step": 3342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005928547176741601,
+      "loss": 2.3246,
+      "step": 3343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005926369808356251,
+      "loss": 2.2487,
+      "step": 3344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000592419225803677,
+      "loss": 2.4249,
+      "step": 3345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005922014526210818,
+      "loss": 2.4729,
+      "step": 3346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005919836613306092,
+      "loss": 2.5556,
+      "step": 3347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005917658519750319,
+      "loss": 2.329,
+      "step": 3348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000591548024597127,
+      "loss": 2.5209,
+      "step": 3349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005913301792396743,
+      "loss": 2.4714,
+      "step": 3350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005911123159454574,
+      "loss": 2.4257,
+      "step": 3351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005908944347572635,
+      "loss": 2.232,
+      "step": 3352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005906765357178838,
+      "loss": 2.4544,
+      "step": 3353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000590458618870112,
+      "loss": 2.2764,
+      "step": 3354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005902406842567459,
+      "loss": 2.4708,
+      "step": 3355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000590022731920587,
+      "loss": 2.3903,
+      "step": 3356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005898047619044396,
+      "loss": 2.3445,
+      "step": 3357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005895867742511119,
+      "loss": 2.3793,
+      "step": 3358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000589368769003416,
+      "loss": 2.3837,
+      "step": 3359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005891507462041665,
+      "loss": 2.3867,
+      "step": 3360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000588932705896182,
+      "loss": 2.358,
+      "step": 3361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005887146481222844,
+      "loss": 2.4715,
+      "step": 3362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005884965729252993,
+      "loss": 2.4019,
+      "step": 3363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005882784803480556,
+      "loss": 2.4554,
+      "step": 3364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005880603704333851,
+      "loss": 2.3177,
+      "step": 3365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000587842243224124,
+      "loss": 2.5124,
+      "step": 3366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005876240987631109,
+      "loss": 2.2949,
+      "step": 3367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005874059370931881,
+      "loss": 2.3614,
+      "step": 3368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005871877582572018,
+      "loss": 2.4729,
+      "step": 3369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000586969562298001,
+      "loss": 2.5291,
+      "step": 3370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005867513492584383,
+      "loss": 2.4246,
+      "step": 3371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005865331191813695,
+      "loss": 2.4669,
+      "step": 3372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005863148721096538,
+      "loss": 2.3016,
+      "step": 3373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000586096608086154,
+      "loss": 2.4378,
+      "step": 3374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005858783271537356,
+      "loss": 2.4586,
+      "step": 3375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005856600293552681,
+      "loss": 2.5851,
+      "step": 3376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005854417147336241,
+      "loss": 2.2856,
+      "step": 3377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005852233833316792,
+      "loss": 2.4517,
+      "step": 3378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005850050351923125,
+      "loss": 2.4461,
+      "step": 3379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005847866703584068,
+      "loss": 2.3305,
+      "step": 3380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005845682888728476,
+      "loss": 2.494,
+      "step": 3381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005843498907785236,
+      "loss": 2.3443,
+      "step": 3382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005841314761183275,
+      "loss": 2.3346,
+      "step": 3383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005839130449351544,
+      "loss": 2.3946,
+      "step": 3384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005836945972719032,
+      "loss": 2.5165,
+      "step": 3385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000583476133171476,
+      "loss": 2.3428,
+      "step": 3386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005832576526767776,
+      "loss": 2.4275,
+      "step": 3387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005830391558307168,
+      "loss": 2.3325,
+      "step": 3388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000582820642676205,
+      "loss": 2.2999,
+      "step": 3389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005826021132561572,
+      "loss": 2.4017,
+      "step": 3390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005823835676134913,
+      "loss": 2.3372,
+      "step": 3391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005821650057911283,
+      "loss": 2.3611,
+      "step": 3392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005819464278319931,
+      "loss": 2.4365,
+      "step": 3393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005817278337790129,
+      "loss": 2.4753,
+      "step": 3394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005815092236751184,
+      "loss": 2.3251,
+      "step": 3395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005812905975632437,
+      "loss": 2.3411,
+      "step": 3396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005810719554863258,
+      "loss": 2.5023,
+      "step": 3397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005808532974873044,
+      "loss": 2.3165,
+      "step": 3398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005806346236091231,
+      "loss": 2.5029,
+      "step": 3399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005804159338947284,
+      "loss": 2.2802,
+      "step": 3400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005801972283870697,
+      "loss": 2.2789,
+      "step": 3401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005799785071290995,
+      "loss": 2.5949,
+      "step": 3402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005797597701637737,
+      "loss": 2.3816,
+      "step": 3403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000579541017534051,
+      "loss": 2.5827,
+      "step": 3404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005793222492828931,
+      "loss": 2.396,
+      "step": 3405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005791034654532651,
+      "loss": 2.4292,
+      "step": 3406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000578884666088135,
+      "loss": 2.2852,
+      "step": 3407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005786658512304737,
+      "loss": 2.2667,
+      "step": 3408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005784470209232553,
+      "loss": 2.3495,
+      "step": 3409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005782281752094573,
+      "loss": 2.1892,
+      "step": 3410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005780093141320591,
+      "loss": 2.1708,
+      "step": 3411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005777904377340444,
+      "loss": 2.248,
+      "step": 3412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005775715460583995,
+      "loss": 2.3131,
+      "step": 3413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005773526391481131,
+      "loss": 2.4395,
+      "step": 3414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005771337170461776,
+      "loss": 2.545,
+      "step": 3415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005769147797955881,
+      "loss": 2.3971,
+      "step": 3416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005766958274393427,
+      "loss": 2.3703,
+      "step": 3417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005764768600204425,
+      "loss": 2.3079,
+      "step": 3418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005762578775818916,
+      "loss": 2.3364,
+      "step": 3419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005760388801666971,
+      "loss": 2.4593,
+      "step": 3420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005758198678178687,
+      "loss": 2.4431,
+      "step": 3421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005756008405784192,
+      "loss": 2.2658,
+      "step": 3422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005753817984913646,
+      "loss": 2.4444,
+      "step": 3423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005751627415997237,
+      "loss": 2.2432,
+      "step": 3424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005749436699465177,
+      "loss": 2.3698,
+      "step": 3425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005747245835747715,
+      "loss": 2.3583,
+      "step": 3426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005745054825275124,
+      "loss": 2.4498,
+      "step": 3427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005742863668477708,
+      "loss": 2.526,
+      "step": 3428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005740672365785795,
+      "loss": 2.3979,
+      "step": 3429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005738480917629749,
+      "loss": 2.4508,
+      "step": 3430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005736289324439957,
+      "loss": 2.2675,
+      "step": 3431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005734097586646836,
+      "loss": 2.4181,
+      "step": 3432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005731905704680834,
+      "loss": 2.4284,
+      "step": 3433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005729713678972422,
+      "loss": 2.3703,
+      "step": 3434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005727521509952104,
+      "loss": 2.4638,
+      "step": 3435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000572532919805041,
+      "loss": 2.4921,
+      "step": 3436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005723136743697899,
+      "loss": 2.5198,
+      "step": 3437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005720944147325156,
+      "loss": 2.4013,
+      "step": 3438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005718751409362797,
+      "loss": 2.4029,
+      "step": 3439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005716558530241463,
+      "loss": 2.27,
+      "step": 3440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005714365510391824,
+      "loss": 2.2904,
+      "step": 3441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005712172350244579,
+      "loss": 2.4912,
+      "step": 3442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005709979050230452,
+      "loss": 2.4064,
+      "step": 3443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005707785610780194,
+      "loss": 2.3701,
+      "step": 3444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005705592032324585,
+      "loss": 2.3511,
+      "step": 3445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005703398315294436,
+      "loss": 2.2681,
+      "step": 3446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005701204460120578,
+      "loss": 2.414,
+      "step": 3447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005699010467233874,
+      "loss": 2.4942,
+      "step": 3448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000569681633706521,
+      "loss": 2.4253,
+      "step": 3449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005694622070045506,
+      "loss": 2.2999,
+      "step": 3450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005692427666605703,
+      "loss": 2.3976,
+      "step": 3451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005690233127176765,
+      "loss": 2.3066,
+      "step": 3452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005688038452189696,
+      "loss": 2.2946,
+      "step": 3453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005685843642075512,
+      "loss": 2.2552,
+      "step": 3454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005683648697265267,
+      "loss": 2.5014,
+      "step": 3455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005681453618190033,
+      "loss": 2.2994,
+      "step": 3456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005679258405280914,
+      "loss": 2.505,
+      "step": 3457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005677063058969038,
+      "loss": 2.3626,
+      "step": 3458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005674867579685557,
+      "loss": 2.3846,
+      "step": 3459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005672671967861656,
+      "loss": 2.512,
+      "step": 3460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005670476223928538,
+      "loss": 2.5224,
+      "step": 3461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005668280348317435,
+      "loss": 2.2755,
+      "step": 3462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005666084341459607,
+      "loss": 2.4134,
+      "step": 3463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005663888203786341,
+      "loss": 2.3449,
+      "step": 3464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000566169193572894,
+      "loss": 2.3962,
+      "step": 3465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005659495537718744,
+      "loss": 2.3056,
+      "step": 3466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005657299010187116,
+      "loss": 2.5802,
+      "step": 3467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005655102353565437,
+      "loss": 2.3755,
+      "step": 3468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005652905568285121,
+      "loss": 2.4203,
+      "step": 3469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005650708654777606,
+      "loss": 2.3058,
+      "step": 3470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005648511613474353,
+      "loss": 2.3661,
+      "step": 3471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000564631444480685,
+      "loss": 2.4246,
+      "step": 3472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005644117149206609,
+      "loss": 2.4175,
+      "step": 3473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005641919727105169,
+      "loss": 2.4092,
+      "step": 3474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005639722178934088,
+      "loss": 2.4053,
+      "step": 3475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005637524505124956,
+      "loss": 2.3516,
+      "step": 3476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005635326706109385,
+      "loss": 2.4439,
+      "step": 3477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005633128782319009,
+      "loss": 2.4788,
+      "step": 3478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005630930734185488,
+      "loss": 2.5313,
+      "step": 3479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005628732562140509,
+      "loss": 2.4073,
+      "step": 3480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005626534266615782,
+      "loss": 2.2562,
+      "step": 3481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005624335848043038,
+      "loss": 2.4665,
+      "step": 3482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005622137306854038,
+      "loss": 2.4381,
+      "step": 3483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005619938643480561,
+      "loss": 2.3363,
+      "step": 3484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005617739858354415,
+      "loss": 2.2873,
+      "step": 3485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005615540951907427,
+      "loss": 2.5493,
+      "step": 3486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005613341924571452,
+      "loss": 2.2938,
+      "step": 3487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005611142776778369,
+      "loss": 2.4652,
+      "step": 3488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005608943508960077,
+      "loss": 2.3501,
+      "step": 3489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005606744121548501,
+      "loss": 2.3615,
+      "step": 3490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005604544614975591,
+      "loss": 2.4141,
+      "step": 3491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005602344989673315,
+      "loss": 2.4685,
+      "step": 3492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005600145246073669,
+      "loss": 2.5433,
+      "step": 3493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005597945384608672,
+      "loss": 2.3943,
+      "step": 3494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005595745405710365,
+      "loss": 2.259,
+      "step": 3495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005593545309810811,
+      "loss": 2.5251,
+      "step": 3496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00055913450973421,
+      "loss": 2.4031,
+      "step": 3497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005589144768736338,
+      "loss": 2.2567,
+      "step": 3498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000558694432442566,
+      "loss": 2.4264,
+      "step": 3499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005584743764842223,
+      "loss": 2.4952,
+      "step": 3500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005582543090418202,
+      "loss": 2.4197,
+      "step": 3501
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005580342301585801,
+      "loss": 2.3659,
+      "step": 3502
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005578141398777242,
+      "loss": 2.2304,
+      "step": 3503
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005575940382424771,
+      "loss": 2.505,
+      "step": 3504
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005573739252960656,
+      "loss": 2.3863,
+      "step": 3505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005571538010817185,
+      "loss": 2.3438,
+      "step": 3506
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005569336656426674,
+      "loss": 2.3364,
+      "step": 3507
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005567135190221456,
+      "loss": 2.3147,
+      "step": 3508
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005564933612633884,
+      "loss": 2.2169,
+      "step": 3509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005562731924096342,
+      "loss": 2.4772,
+      "step": 3510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005560530125041224,
+      "loss": 2.3599,
+      "step": 3511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005558328215900957,
+      "loss": 2.3322,
+      "step": 3512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005556126197107981,
+      "loss": 2.421,
+      "step": 3513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005553924069094761,
+      "loss": 2.5156,
+      "step": 3514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005551721832293785,
+      "loss": 2.3111,
+      "step": 3515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005549519487137561,
+      "loss": 2.451,
+      "step": 3516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005547317034058615,
+      "loss": 2.2498,
+      "step": 3517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00055451144734895,
+      "loss": 2.3833,
+      "step": 3518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005542911805862785,
+      "loss": 2.4137,
+      "step": 3519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005540709031611063,
+      "loss": 2.3022,
+      "step": 3520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000553850615116695,
+      "loss": 2.3711,
+      "step": 3521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005536303164963075,
+      "loss": 2.4015,
+      "step": 3522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005534100073432095,
+      "loss": 2.4445,
+      "step": 3523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005531896877006689,
+      "loss": 2.4099,
+      "step": 3524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005529693576119549,
+      "loss": 2.4668,
+      "step": 3525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005527490171203391,
+      "loss": 2.448,
+      "step": 3526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005525286662690956,
+      "loss": 2.4775,
+      "step": 3527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005523083051014998,
+      "loss": 2.4488,
+      "step": 3528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005520879336608296,
+      "loss": 2.5124,
+      "step": 3529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005518675519903647,
+      "loss": 2.4249,
+      "step": 3530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005516471601333872,
+      "loss": 2.391,
+      "step": 3531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005514267581331804,
+      "loss": 2.4868,
+      "step": 3532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005512063460330305,
+      "loss": 2.3969,
+      "step": 3533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005509859238762251,
+      "loss": 2.3641,
+      "step": 3534
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005507654917060541,
+      "loss": 2.4629,
+      "step": 3535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005505450495658089,
+      "loss": 2.401,
+      "step": 3536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005503245974987833,
+      "loss": 2.435,
+      "step": 3537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005501041355482733,
+      "loss": 2.4668,
+      "step": 3538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005498836637575757,
+      "loss": 2.4691,
+      "step": 3539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005496631821699908,
+      "loss": 2.3563,
+      "step": 3540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005494426908288194,
+      "loss": 2.2798,
+      "step": 3541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005492221897773651,
+      "loss": 2.4344,
+      "step": 3542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005490016790589331,
+      "loss": 2.5306,
+      "step": 3543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005487811587168306,
+      "loss": 2.4459,
+      "step": 3544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005485606287943667,
+      "loss": 2.3864,
+      "step": 3545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005483400893348521,
+      "loss": 2.3815,
+      "step": 3546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005481195403815996,
+      "loss": 2.3794,
+      "step": 3547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005478989819779241,
+      "loss": 2.3128,
+      "step": 3548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005476784141671417,
+      "loss": 2.4399,
+      "step": 3549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005474578369925712,
+      "loss": 2.4208,
+      "step": 3550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005472372504975324,
+      "loss": 2.3959,
+      "step": 3551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005470166547253476,
+      "loss": 2.256,
+      "step": 3552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005467960497193404,
+      "loss": 2.3918,
+      "step": 3553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005465754355228368,
+      "loss": 2.4578,
+      "step": 3554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005463548121791639,
+      "loss": 2.2964,
+      "step": 3555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000546134179731651,
+      "loss": 2.3754,
+      "step": 3556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005459135382236295,
+      "loss": 2.317,
+      "step": 3557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005456928876984318,
+      "loss": 2.2822,
+      "step": 3558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005454722281993927,
+      "loss": 2.3404,
+      "step": 3559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005452515597698485,
+      "loss": 2.4532,
+      "step": 3560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005450308824531375,
+      "loss": 2.4678,
+      "step": 3561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005448101962925992,
+      "loss": 2.4702,
+      "step": 3562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005445895013315755,
+      "loss": 2.5813,
+      "step": 3563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005443687976134096,
+      "loss": 2.4313,
+      "step": 3564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005441480851814467,
+      "loss": 2.3438,
+      "step": 3565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005439273640790332,
+      "loss": 2.421,
+      "step": 3566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005437066343495178,
+      "loss": 2.3666,
+      "step": 3567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005434858960362507,
+      "loss": 2.343,
+      "step": 3568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005432651491825836,
+      "loss": 2.3906,
+      "step": 3569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005430443938318702,
+      "loss": 2.3349,
+      "step": 3570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005428236300274655,
+      "loss": 2.295,
+      "step": 3571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005426028578127264,
+      "loss": 2.4497,
+      "step": 3572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005423820772310114,
+      "loss": 2.1663,
+      "step": 3573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005421612883256806,
+      "loss": 2.472,
+      "step": 3574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005419404911400958,
+      "loss": 2.2751,
+      "step": 3575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005417196857176205,
+      "loss": 2.4429,
+      "step": 3576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005414988721016194,
+      "loss": 2.338,
+      "step": 3577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005412780503354596,
+      "loss": 2.3766,
+      "step": 3578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000541057220462509,
+      "loss": 2.6069,
+      "step": 3579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005408363825261373,
+      "loss": 2.5065,
+      "step": 3580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005406155365697164,
+      "loss": 2.3639,
+      "step": 3581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005403946826366188,
+      "loss": 2.3624,
+      "step": 3582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005401738207702193,
+      "loss": 2.4594,
+      "step": 3583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005399529510138938,
+      "loss": 2.4775,
+      "step": 3584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005397320734110201,
+      "loss": 2.4732,
+      "step": 3585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005395111880049775,
+      "loss": 2.3776,
+      "step": 3586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005392902948391465,
+      "loss": 2.3002,
+      "step": 3587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005390693939569095,
+      "loss": 2.2821,
+      "step": 3588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005388484854016501,
+      "loss": 2.3424,
+      "step": 3589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005386275692167537,
+      "loss": 2.3376,
+      "step": 3590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005384066454456072,
+      "loss": 2.4174,
+      "step": 3591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005381857141315984,
+      "loss": 2.354,
+      "step": 3592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005379647753181176,
+      "loss": 2.3703,
+      "step": 3593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005377438290485557,
+      "loss": 2.3954,
+      "step": 3594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005375228753663056,
+      "loss": 2.5458,
+      "step": 3595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000537301914314761,
+      "loss": 2.4473,
+      "step": 3596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005370809459373178,
+      "loss": 2.4176,
+      "step": 3597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005368599702773731,
+      "loss": 2.3891,
+      "step": 3598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005366389873783251,
+      "loss": 2.3566,
+      "step": 3599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005364179972835739,
+      "loss": 2.299,
+      "step": 3600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005361970000365207,
+      "loss": 2.3774,
+      "step": 3601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005359759956805682,
+      "loss": 2.5013,
+      "step": 3602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005357549842591204,
+      "loss": 2.309,
+      "step": 3603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005355339658155828,
+      "loss": 2.2165,
+      "step": 3604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005353129403933624,
+      "loss": 2.4035,
+      "step": 3605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000535091908035867,
+      "loss": 2.3716,
+      "step": 3606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005348708687865069,
+      "loss": 2.3753,
+      "step": 3607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005346498226886926,
+      "loss": 2.4365,
+      "step": 3608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005344287697858362,
+      "loss": 2.3612,
+      "step": 3609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005342077101213516,
+      "loss": 2.3319,
+      "step": 3610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005339866437386538,
+      "loss": 2.4353,
+      "step": 3611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005337655706811589,
+      "loss": 2.3548,
+      "step": 3612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005335444909922845,
+      "loss": 2.2278,
+      "step": 3613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005333234047154495,
+      "loss": 2.2051,
+      "step": 3614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005331023118940742,
+      "loss": 2.3562,
+      "step": 3615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005328812125715798,
+      "loss": 2.4039,
+      "step": 3616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005326601067913893,
+      "loss": 2.3171,
+      "step": 3617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005324389945969267,
+      "loss": 2.3181,
+      "step": 3618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005322178760316171,
+      "loss": 2.4562,
+      "step": 3619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005319967511388871,
+      "loss": 2.4045,
+      "step": 3620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005317756199621642,
+      "loss": 2.3003,
+      "step": 3621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005315544825448779,
+      "loss": 2.3973,
+      "step": 3622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000531333338930458,
+      "loss": 2.2305,
+      "step": 3623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005311121891623363,
+      "loss": 2.2942,
+      "step": 3624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005308910332839451,
+      "loss": 2.4199,
+      "step": 3625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005306698713387183,
+      "loss": 2.3935,
+      "step": 3626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005304487033700911,
+      "loss": 2.3678,
+      "step": 3627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005302275294214997,
+      "loss": 2.3569,
+      "step": 3628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005300063495363811,
+      "loss": 2.4183,
+      "step": 3629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005297851637581746,
+      "loss": 2.3865,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005295639721303192,
+      "loss": 2.4725,
+      "step": 3631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005293427746962564,
+      "loss": 2.5017,
+      "step": 3632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005291215714994276,
+      "loss": 2.4372,
+      "step": 3633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005289003625832762,
+      "loss": 2.4242,
+      "step": 3634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005286791479912468,
+      "loss": 2.3511,
+      "step": 3635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005284579277667839,
+      "loss": 2.3119,
+      "step": 3636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005282367019533349,
+      "loss": 2.3995,
+      "step": 3637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005280154705943471,
+      "loss": 2.3775,
+      "step": 3638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005277942337332689,
+      "loss": 2.3577,
+      "step": 3639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005275729914135502,
+      "loss": 2.3159,
+      "step": 3640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005273517436786419,
+      "loss": 2.3322,
+      "step": 3641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005271304905719958,
+      "loss": 2.5064,
+      "step": 3642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005269092321370649,
+      "loss": 2.4301,
+      "step": 3643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005266879684173031,
+      "loss": 2.4808,
+      "step": 3644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005264666994561657,
+      "loss": 2.4011,
+      "step": 3645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005262454252971082,
+      "loss": 2.3824,
+      "step": 3646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005260241459835883,
+      "loss": 2.597,
+      "step": 3647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005258028615590638,
+      "loss": 2.4929,
+      "step": 3648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005255815720669935,
+      "loss": 2.3084,
+      "step": 3649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005253602775508379,
+      "loss": 2.303,
+      "step": 3650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005251389780540582,
+      "loss": 2.3034,
+      "step": 3651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005249176736201159,
+      "loss": 2.2809,
+      "step": 3652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005246963642924744,
+      "loss": 2.2648,
+      "step": 3653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005244750501145976,
+      "loss": 2.2226,
+      "step": 3654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005242537311299504,
+      "loss": 2.2982,
+      "step": 3655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005240324073819988,
+      "loss": 2.2903,
+      "step": 3656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005238110789142096,
+      "loss": 2.3196,
+      "step": 3657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005235897457700503,
+      "loss": 2.4768,
+      "step": 3658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005233684079929897,
+      "loss": 2.4935,
+      "step": 3659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005231470656264975,
+      "loss": 2.4318,
+      "step": 3660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005229257187140439,
+      "loss": 2.1882,
+      "step": 3661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005227043672991007,
+      "loss": 2.3,
+      "step": 3662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005224830114251394,
+      "loss": 2.3271,
+      "step": 3663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000522261651135634,
+      "loss": 2.366,
+      "step": 3664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005220402864740581,
+      "loss": 2.3426,
+      "step": 3665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005218189174838862,
+      "loss": 2.3357,
+      "step": 3666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005215975442085946,
+      "loss": 2.3017,
+      "step": 3667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005213761666916595,
+      "loss": 2.4788,
+      "step": 3668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005211547849765585,
+      "loss": 2.3601,
+      "step": 3669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005209333991067693,
+      "loss": 2.3569,
+      "step": 3670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005207120091257714,
+      "loss": 2.3226,
+      "step": 3671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005204906150770446,
+      "loss": 2.3545,
+      "step": 3672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005202692170040693,
+      "loss": 2.2367,
+      "step": 3673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005200478149503269,
+      "loss": 2.3502,
+      "step": 3674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005198264089592999,
+      "loss": 2.3854,
+      "step": 3675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005196049990744711,
+      "loss": 2.3956,
+      "step": 3676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000519383585339324,
+      "loss": 2.5238,
+      "step": 3677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005191621677973433,
+      "loss": 2.4608,
+      "step": 3678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005189407464920143,
+      "loss": 2.4903,
+      "step": 3679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005187193214668227,
+      "loss": 2.2811,
+      "step": 3680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005184978927652556,
+      "loss": 2.3646,
+      "step": 3681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005182764604308002,
+      "loss": 2.331,
+      "step": 3682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005180550245069446,
+      "loss": 2.3942,
+      "step": 3683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005178335850371777,
+      "loss": 2.3303,
+      "step": 3684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000517612142064989,
+      "loss": 2.244,
+      "step": 3685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005173906956338687,
+      "loss": 2.2348,
+      "step": 3686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000517169245787308,
+      "loss": 2.2986,
+      "step": 3687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000516947792568798,
+      "loss": 2.3088,
+      "step": 3688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005167263360218315,
+      "loss": 2.2893,
+      "step": 3689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005165048761899007,
+      "loss": 2.4499,
+      "step": 3690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005162834131164999,
+      "loss": 2.2221,
+      "step": 3691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005160619468451228,
+      "loss": 2.4721,
+      "step": 3692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005158404774192641,
+      "loss": 2.3042,
+      "step": 3693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005156190048824196,
+      "loss": 2.3306,
+      "step": 3694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005153975292780852,
+      "loss": 2.2501,
+      "step": 3695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005151760506497575,
+      "loss": 2.3449,
+      "step": 3696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005149545690409336,
+      "loss": 2.5639,
+      "step": 3697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005147330844951117,
+      "loss": 2.4453,
+      "step": 3698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005145115970557897,
+      "loss": 2.4085,
+      "step": 3699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005142901067664668,
+      "loss": 2.378,
+      "step": 3700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005140686136706426,
+      "loss": 2.4704,
+      "step": 3701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005138471178118173,
+      "loss": 2.4414,
+      "step": 3702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005136256192334911,
+      "loss": 2.3988,
+      "step": 3703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005134041179791654,
+      "loss": 2.3298,
+      "step": 3704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005131826140923418,
+      "loss": 2.3881,
+      "step": 3705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005129611076165225,
+      "loss": 2.2765,
+      "step": 3706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005127395985952102,
+      "loss": 2.1949,
+      "step": 3707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005125180870719083,
+      "loss": 2.2631,
+      "step": 3708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005122965730901203,
+      "loss": 2.3465,
+      "step": 3709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005120750566933502,
+      "loss": 2.4954,
+      "step": 3710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005118535379251031,
+      "loss": 2.4262,
+      "step": 3711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005116320168288836,
+      "loss": 2.3543,
+      "step": 3712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005114104934481977,
+      "loss": 2.3933,
+      "step": 3713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005111889678265512,
+      "loss": 2.3197,
+      "step": 3714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005109674400074506,
+      "loss": 2.3724,
+      "step": 3715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005107459100344027,
+      "loss": 2.2741,
+      "step": 3716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005105243779509152,
+      "loss": 2.3689,
+      "step": 3717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005103028438004953,
+      "loss": 2.5064,
+      "step": 3718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005100813076266515,
+      "loss": 2.3574,
+      "step": 3719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005098597694728921,
+      "loss": 2.233,
+      "step": 3720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005096382293827262,
+      "loss": 2.347,
+      "step": 3721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005094166873996632,
+      "loss": 2.3947,
+      "step": 3722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005091951435672122,
+      "loss": 2.4514,
+      "step": 3723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005089735979288839,
+      "loss": 2.3646,
+      "step": 3724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005087520505281886,
+      "loss": 2.5114,
+      "step": 3725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005085305014086367,
+      "loss": 2.3436,
+      "step": 3726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005083089506137394,
+      "loss": 2.1356,
+      "step": 3727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005080873981870082,
+      "loss": 2.358,
+      "step": 3728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005078658441719548,
+      "loss": 2.5071,
+      "step": 3729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005076442886120912,
+      "loss": 2.4767,
+      "step": 3730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005074227315509296,
+      "loss": 2.5212,
+      "step": 3731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005072011730319831,
+      "loss": 2.2885,
+      "step": 3732
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005069796130987642,
+      "loss": 2.4311,
+      "step": 3733
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005067580517947861,
+      "loss": 2.4518,
+      "step": 3734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005065364891635624,
+      "loss": 2.344,
+      "step": 3735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005063149252486069,
+      "loss": 2.4378,
+      "step": 3736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005060933600934335,
+      "loss": 2.4708,
+      "step": 3737
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005058717937415564,
+      "loss": 2.3436,
+      "step": 3738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005056502262364899,
+      "loss": 2.2873,
+      "step": 3739
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000505428657621749,
+      "loss": 2.3239,
+      "step": 3740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005052070879408484,
+      "loss": 2.3208,
+      "step": 3741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005049855172373033,
+      "loss": 2.4472,
+      "step": 3742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000504763945554629,
+      "loss": 2.3647,
+      "step": 3743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005045423729363411,
+      "loss": 2.3777,
+      "step": 3744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000504320799425955,
+      "loss": 2.2973,
+      "step": 3745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005040992250669868,
+      "loss": 2.4352,
+      "step": 3746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005038776499029524,
+      "loss": 2.3386,
+      "step": 3747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005036560739773684,
+      "loss": 2.4407,
+      "step": 3748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005034344973337505,
+      "loss": 2.4591,
+      "step": 3749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005032129200156156,
+      "loss": 2.3922,
+      "step": 3750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005029913420664802,
+      "loss": 2.1664,
+      "step": 3751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005027697635298611,
+      "loss": 2.2964,
+      "step": 3752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005025481844492749,
+      "loss": 2.312,
+      "step": 3753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000502326604868239,
+      "loss": 2.3266,
+      "step": 3754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005021050248302701,
+      "loss": 2.3764,
+      "step": 3755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005018834443788854,
+      "loss": 2.5255,
+      "step": 3756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005016618635576022,
+      "loss": 2.4938,
+      "step": 3757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005014402824099379,
+      "loss": 2.4362,
+      "step": 3758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005012187009794097,
+      "loss": 2.3214,
+      "step": 3759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005009971193095351,
+      "loss": 2.4779,
+      "step": 3760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005007755374438314,
+      "loss": 2.3347,
+      "step": 3761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005005539554258164,
+      "loss": 2.3724,
+      "step": 3762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005003323732990074,
+      "loss": 2.3536,
+      "step": 3763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005001107911069221,
+      "loss": 2.283,
+      "step": 3764
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000499889208893078,
+      "loss": 2.2698,
+      "step": 3765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004996676267009926,
+      "loss": 2.4311,
+      "step": 3766
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004994460445741837,
+      "loss": 2.346,
+      "step": 3767
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004992244625561688,
+      "loss": 2.47,
+      "step": 3768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004990028806904651,
+      "loss": 2.267,
+      "step": 3769
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004987812990205904,
+      "loss": 2.4611,
+      "step": 3770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004985597175900621,
+      "loss": 2.3523,
+      "step": 3771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004983381364423978,
+      "loss": 2.612,
+      "step": 3772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004981165556211146,
+      "loss": 2.1966,
+      "step": 3773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00049789497516973,
+      "loss": 2.4951,
+      "step": 3774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004976733951317612,
+      "loss": 2.285,
+      "step": 3775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000497451815550725,
+      "loss": 2.4168,
+      "step": 3776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000497230236470139,
+      "loss": 2.4883,
+      "step": 3777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004970086579335198,
+      "loss": 2.3249,
+      "step": 3778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004967870799843845,
+      "loss": 2.3457,
+      "step": 3779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004965655026662496,
+      "loss": 2.3085,
+      "step": 3780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004963439260226316,
+      "loss": 2.3299,
+      "step": 3781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004961223500970475,
+      "loss": 2.5008,
+      "step": 3782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004959007749330132,
+      "loss": 2.3301,
+      "step": 3783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004956792005740451,
+      "loss": 2.5218,
+      "step": 3784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004954576270636592,
+      "loss": 2.3062,
+      "step": 3785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004952360544453711,
+      "loss": 2.3254,
+      "step": 3786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004950144827626966,
+      "loss": 2.2711,
+      "step": 3787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004947929120591516,
+      "loss": 2.2954,
+      "step": 3788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004945713423782511,
+      "loss": 2.4197,
+      "step": 3789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004943497737635102,
+      "loss": 2.4331,
+      "step": 3790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000494128206258444,
+      "loss": 2.3368,
+      "step": 3791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004939066399065665,
+      "loss": 2.2916,
+      "step": 3792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004936850747513932,
+      "loss": 2.5323,
+      "step": 3793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004934635108364376,
+      "loss": 2.3366,
+      "step": 3794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004932419482052141,
+      "loss": 2.4876,
+      "step": 3795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004930203869012359,
+      "loss": 2.1681,
+      "step": 3796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004927988269680171,
+      "loss": 2.3963,
+      "step": 3797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004925772684490703,
+      "loss": 2.3089,
+      "step": 3798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000492355711387909,
+      "loss": 2.3469,
+      "step": 3799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004921341558280453,
+      "loss": 2.4953,
+      "step": 3800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004919126018129919,
+      "loss": 2.3563,
+      "step": 3801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004916910493862608,
+      "loss": 2.3339,
+      "step": 3802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004914694985913634,
+      "loss": 2.3272,
+      "step": 3803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004912479494718115,
+      "loss": 2.5224,
+      "step": 3804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004910264020711161,
+      "loss": 2.31,
+      "step": 3805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004908048564327878,
+      "loss": 2.3608,
+      "step": 3806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004905833126003371,
+      "loss": 2.4953,
+      "step": 3807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004903617706172737,
+      "loss": 2.3575,
+      "step": 3808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000490140230527108,
+      "loss": 2.2476,
+      "step": 3809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004899186923733485,
+      "loss": 2.335,
+      "step": 3810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004896971561995048,
+      "loss": 2.328,
+      "step": 3811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000489475622049085,
+      "loss": 2.4414,
+      "step": 3812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004892540899655973,
+      "loss": 2.3647,
+      "step": 3813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004890325599925493,
+      "loss": 2.3907,
+      "step": 3814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004888110321734488,
+      "loss": 2.4125,
+      "step": 3815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004885895065518025,
+      "loss": 2.4441,
+      "step": 3816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004883679831711164,
+      "loss": 2.2726,
+      "step": 3817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004881464620748972,
+      "loss": 2.4356,
+      "step": 3818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004879249433066498,
+      "loss": 2.2783,
+      "step": 3819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048770342690987987,
+      "loss": 2.3583,
+      "step": 3820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048748191292809174,
+      "loss": 2.3877,
+      "step": 3821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004872604014047898,
+      "loss": 2.3279,
+      "step": 3822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004870388923834776,
+      "loss": 2.475,
+      "step": 3823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048681738590765843,
+      "loss": 2.3675,
+      "step": 3824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004865958820208347,
+      "loss": 2.3689,
+      "step": 3825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048637438076650906,
+      "loss": 2.3171,
+      "step": 3826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048615288218818286,
+      "loss": 2.4118,
+      "step": 3827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048593138632935744,
+      "loss": 2.2849,
+      "step": 3828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048570989323353333,
+      "loss": 2.3107,
+      "step": 3829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004854884029442104,
+      "loss": 2.3855,
+      "step": 3830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048526691550488846,
+      "loss": 2.3611,
+      "step": 3831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004850454309590664,
+      "loss": 2.2903,
+      "step": 3832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004848239493502426,
+      "loss": 2.4629,
+      "step": 3833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048460247072191494,
+      "loss": 2.4817,
+      "step": 3834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048438099511758054,
+      "loss": 2.336,
+      "step": 3835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048415952258073586,
+      "loss": 2.422,
+      "step": 3836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048393805315487726,
+      "loss": 2.3776,
+      "step": 3837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048371658688350016,
+      "loss": 2.4556,
+      "step": 3838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004834951238100993,
+      "loss": 2.4141,
+      "step": 3839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048327366397816866,
+      "loss": 2.4975,
+      "step": 3840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048305220743120186,
+      "loss": 2.3333,
+      "step": 3841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004828307542126921,
+      "loss": 2.498,
+      "step": 3842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004826093043661313,
+      "loss": 2.4384,
+      "step": 3843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004823878579350111,
+      "loss": 2.4482,
+      "step": 3844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048216641496282253,
+      "loss": 2.3704,
+      "step": 3845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004819449754930554,
+      "loss": 2.2852,
+      "step": 3846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048172353956919985,
+      "loss": 2.4442,
+      "step": 3847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004815021072347445,
+      "loss": 2.2814,
+      "step": 3848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048128067853317734,
+      "loss": 2.3541,
+      "step": 3849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048105925350798586,
+      "loss": 2.3971,
+      "step": 3850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048083783220265685,
+      "loss": 2.2382,
+      "step": 3851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004806164146606761,
+      "loss": 2.4521,
+      "step": 3852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048039500092552913,
+      "loss": 2.3895,
+      "step": 3853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048017359104070013,
+      "loss": 2.2354,
+      "step": 3854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047995218504967314,
+      "loss": 2.3212,
+      "step": 3855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047973078299593094,
+      "loss": 2.2573,
+      "step": 3856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047950938492295547,
+      "loss": 2.3789,
+      "step": 3857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047928799087422854,
+      "loss": 2.4161,
+      "step": 3858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047906660089323073,
+      "loss": 2.4774,
+      "step": 3859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004788452150234417,
+      "loss": 2.366,
+      "step": 3860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004786238333083406,
+      "loss": 2.3615,
+      "step": 3861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047840245579140555,
+      "loss": 2.2719,
+      "step": 3862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047818108251611373,
+      "loss": 2.352,
+      "step": 3863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000477959713525942,
+      "loss": 2.2403,
+      "step": 3864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047773834886436606,
+      "loss": 2.3236,
+      "step": 3865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004775169885748606,
+      "loss": 2.425,
+      "step": 3866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047729563270089957,
+      "loss": 2.382,
+      "step": 3867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047707428128595603,
+      "loss": 2.5461,
+      "step": 3868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004768529343735026,
+      "loss": 2.1938,
+      "step": 3869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047663159200701045,
+      "loss": 2.5016,
+      "step": 3870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004764102542299499,
+      "loss": 2.4423,
+      "step": 3871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047618892108579065,
+      "loss": 2.2878,
+      "step": 3872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004759675926180012,
+      "loss": 2.4629,
+      "step": 3873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004757462688700495,
+      "loss": 2.4491,
+      "step": 3874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004755249498854024,
+      "loss": 2.4698,
+      "step": 3875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047530363570752574,
+      "loss": 2.4711,
+      "step": 3876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004750823263798842,
+      "loss": 2.4066,
+      "step": 3877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047486102194594206,
+      "loss": 2.3069,
+      "step": 3878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047463972244916197,
+      "loss": 2.3587,
+      "step": 3879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047441842793300646,
+      "loss": 2.3931,
+      "step": 3880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004741971384409364,
+      "loss": 2.3524,
+      "step": 3881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004739758540164118,
+      "loss": 2.2458,
+      "step": 3882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004737545747028918,
+      "loss": 2.3041,
+      "step": 3883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047353330054383433,
+      "loss": 2.3238,
+      "step": 3884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047331203158269684,
+      "loss": 2.3127,
+      "step": 3885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047309076786293514,
+      "loss": 2.3856,
+      "step": 3886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047286950942800423,
+      "loss": 2.296,
+      "step": 3887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047264825632135823,
+      "loss": 2.3139,
+      "step": 3888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047242700858645,
+      "loss": 2.3719,
+      "step": 3889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047220576626673123,
+      "loss": 2.3222,
+      "step": 3890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000471984529405653,
+      "loss": 2.4478,
+      "step": 3891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047176329804666514,
+      "loss": 2.4862,
+      "step": 3892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004715420722332161,
+      "loss": 2.3958,
+      "step": 3893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047132085200875354,
+      "loss": 2.3543,
+      "step": 3894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004710996374167237,
+      "loss": 2.3336,
+      "step": 3895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047087842850057247,
+      "loss": 2.4304,
+      "step": 3896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047065722530374374,
+      "loss": 2.3442,
+      "step": 3897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047043602786968076,
+      "loss": 2.4177,
+      "step": 3898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004702148362418256,
+      "loss": 2.3796,
+      "step": 3899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004699936504636189,
+      "loss": 2.2916,
+      "step": 3900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046977247057850035,
+      "loss": 2.3661,
+      "step": 3901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046955129662990897,
+      "loss": 2.4747,
+      "step": 3902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046933012866128183,
+      "loss": 2.308,
+      "step": 3903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046910896671605506,
+      "loss": 2.2699,
+      "step": 3904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004688878108376639,
+      "loss": 2.1987,
+      "step": 3905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004686666610695419,
+      "loss": 2.3597,
+      "step": 3906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004684455174551222,
+      "loss": 2.4019,
+      "step": 3907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004682243800378357,
+      "loss": 2.2811,
+      "step": 3908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004680032488611131,
+      "loss": 2.4214,
+      "step": 3909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046778212396838306,
+      "loss": 2.3952,
+      "step": 3910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046756100540307324,
+      "loss": 2.3043,
+      "step": 3911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004673398932086106,
+      "loss": 2.277,
+      "step": 3912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004671187874284202,
+      "loss": 2.5082,
+      "step": 3913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004668976881059259,
+      "loss": 2.3262,
+      "step": 3914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046667659528455056,
+      "loss": 2.3049,
+      "step": 3915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004664555090077157,
+      "loss": 2.2418,
+      "step": 3916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004662344293188412,
+      "loss": 2.3437,
+      "step": 3917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046601335626134626,
+      "loss": 2.5121,
+      "step": 3918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046579228987864846,
+      "loss": 2.4746,
+      "step": 3919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004655712302141639,
+      "loss": 2.3752,
+      "step": 3920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004653501773113076,
+      "loss": 2.3526,
+      "step": 3921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004651291312134931,
+      "loss": 2.3111,
+      "step": 3922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004649080919641329,
+      "loss": 2.3006,
+      "step": 3923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004646870596066377,
+      "loss": 2.253,
+      "step": 3924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004644660341844173,
+      "loss": 2.2917,
+      "step": 3925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004642450157408798,
+      "loss": 2.3214,
+      "step": 3926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046402400431943197,
+      "loss": 2.329,
+      "step": 3927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046380299996347924,
+      "loss": 2.4695,
+      "step": 3928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004635820027164261,
+      "loss": 2.3257,
+      "step": 3929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046336101262167494,
+      "loss": 2.444,
+      "step": 3930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000463140029722627,
+      "loss": 2.342,
+      "step": 3931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004629190540626823,
+      "loss": 2.4103,
+      "step": 3932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000462698085685239,
+      "loss": 2.3272,
+      "step": 3933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004624771246336945,
+      "loss": 2.2587,
+      "step": 3934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004622561709514443,
+      "loss": 2.3778,
+      "step": 3935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046203522468188255,
+      "loss": 2.3976,
+      "step": 3936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046181428586840155,
+      "loss": 2.3199,
+      "step": 3937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046159335455439306,
+      "loss": 2.4127,
+      "step": 3938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004613724307832463,
+      "loss": 2.3977,
+      "step": 3939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046115151459834997,
+      "loss": 2.3184,
+      "step": 3940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004609306060430906,
+      "loss": 2.4266,
+      "step": 3941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046070970516085366,
+      "loss": 2.4095,
+      "step": 3942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046048881199502265,
+      "loss": 2.3797,
+      "step": 3943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046026792658897985,
+      "loss": 2.2443,
+      "step": 3944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046004704898610624,
+      "loss": 2.3574,
+      "step": 3945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045982617922978085,
+      "loss": 2.2747,
+      "step": 3946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004596053173633813,
+      "loss": 2.3195,
+      "step": 3947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004593844634302838,
+      "loss": 2.3227,
+      "step": 3948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004591636174738626,
+      "loss": 2.4316,
+      "step": 3949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045894277953749114,
+      "loss": 2.3722,
+      "step": 3950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004587219496645405,
+      "loss": 2.4239,
+      "step": 3951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004585011278983807,
+      "loss": 2.1294,
+      "step": 3952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004582803142823798,
+      "loss": 2.3409,
+      "step": 3953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004580595088599044,
+      "loss": 2.5421,
+      "step": 3954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045783871167431943,
+      "loss": 2.2971,
+      "step": 3955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004576179227689887,
+      "loss": 2.3765,
+      "step": 3956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004573971421872737,
+      "loss": 2.2948,
+      "step": 3957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004571763699725346,
+      "loss": 2.4136,
+      "step": 3958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045695560616813,
+      "loss": 2.3982,
+      "step": 3959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045673485081741634,
+      "loss": 2.4259,
+      "step": 3960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045651410396374927,
+      "loss": 2.1779,
+      "step": 3961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004562933656504822,
+      "loss": 2.2976,
+      "step": 3962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045607263592096695,
+      "loss": 2.3543,
+      "step": 3963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004558519148185535,
+      "loss": 2.2417,
+      "step": 3964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004556312023865905,
+      "loss": 2.4042,
+      "step": 3965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004554104986684245,
+      "loss": 2.3995,
+      "step": 3966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045518980370740087,
+      "loss": 2.363,
+      "step": 3967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045496911754686265,
+      "loss": 2.3854,
+      "step": 3968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004547484402301516,
+      "loss": 2.3664,
+      "step": 3969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004545277718006074,
+      "loss": 2.3772,
+      "step": 3970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004543071123015682,
+      "loss": 2.2949,
+      "step": 3971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004540864617763706,
+      "loss": 2.4595,
+      "step": 3972
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045386582026834903,
+      "loss": 2.3614,
+      "step": 3973
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004536451878208363,
+      "loss": 2.2521,
+      "step": 3974
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004534245644771635,
+      "loss": 2.418,
+      "step": 3975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004532039502806598,
+      "loss": 2.2329,
+      "step": 3976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004529833452746525,
+      "loss": 2.3138,
+      "step": 3977
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004527627495024676,
+      "loss": 2.4078,
+      "step": 3978
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045254216300742896,
+      "loss": 2.4057,
+      "step": 3979
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045232158583285846,
+      "loss": 2.2536,
+      "step": 3980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004521010180220761,
+      "loss": 2.259,
+      "step": 3981
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004518804596184003,
+      "loss": 2.3406,
+      "step": 3982
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004516599106651479,
+      "loss": 2.2932,
+      "step": 3983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045143937120563333,
+      "loss": 2.3467,
+      "step": 3984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045121884128316935,
+      "loss": 2.3758,
+      "step": 3985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045099832094106694,
+      "loss": 2.2559,
+      "step": 3986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004507778102226349,
+      "loss": 2.338,
+      "step": 3987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004505573091711806,
+      "loss": 2.3366,
+      "step": 3988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045033681783000935,
+      "loss": 2.5047,
+      "step": 3989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045011633624242434,
+      "loss": 2.4424,
+      "step": 3990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004498958644517269,
+      "loss": 2.3823,
+      "step": 3991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044967540250121673,
+      "loss": 2.3479,
+      "step": 3992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004494549504341911,
+      "loss": 2.3774,
+      "step": 3993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000449234508293946,
+      "loss": 2.2489,
+      "step": 3994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004490140761237749,
+      "loss": 2.3527,
+      "step": 3995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044879365396696956,
+      "loss": 2.2731,
+      "step": 3996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044857324186681956,
+      "loss": 2.6107,
+      "step": 3997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004483528398666128,
+      "loss": 2.5212,
+      "step": 3998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004481324480096352,
+      "loss": 2.3454,
+      "step": 3999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004479120663391705,
+      "loss": 2.326,
+      "step": 4000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044769169489850024,
+      "loss": 2.2898,
+      "step": 4001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004474713337309046,
+      "loss": 2.2015,
+      "step": 4002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000447250982879661,
+      "loss": 2.3163,
+      "step": 4003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044703064238804527,
+      "loss": 2.2911,
+      "step": 4004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044681031229933116,
+      "loss": 2.2493,
+      "step": 4005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004465899926567905,
+      "loss": 2.4115,
+      "step": 4006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044636968350369256,
+      "loss": 2.4261,
+      "step": 4007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044614938488330525,
+      "loss": 2.368,
+      "step": 4008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044592909683889364,
+      "loss": 2.4067,
+      "step": 4009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004457088194137216,
+      "loss": 2.4063,
+      "step": 4010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004454885526510501,
+      "loss": 2.4994,
+      "step": 4011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004452682965941386,
+      "loss": 2.4311,
+      "step": 4012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044504805128624416,
+      "loss": 2.2786,
+      "step": 4013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004448278167706215,
+      "loss": 2.3238,
+      "step": 4014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004446075930905238,
+      "loss": 2.4054,
+      "step": 4015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000444387380289202,
+      "loss": 2.2979,
+      "step": 4016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004441671784099045,
+      "loss": 2.4224,
+      "step": 4017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004439469874958776,
+      "loss": 2.3477,
+      "step": 4018
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004437268075903661,
+      "loss": 2.2485,
+      "step": 4019
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044350663873661154,
+      "loss": 2.3566,
+      "step": 4020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044328648097785453,
+      "loss": 2.3582,
+      "step": 4021
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044306633435733266,
+      "loss": 2.4225,
+      "step": 4022
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004428461989182815,
+      "loss": 2.3357,
+      "step": 4023
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004426260747039345,
+      "loss": 2.5776,
+      "step": 4024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004424059617575228,
+      "loss": 2.3527,
+      "step": 4025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044218586012227575,
+      "loss": 2.2586,
+      "step": 4026
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004419657698414199,
+      "loss": 2.4817,
+      "step": 4027
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004417456909581798,
+      "loss": 2.426,
+      "step": 4028
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044152562351577787,
+      "loss": 2.4482,
+      "step": 4029
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044130556755743406,
+      "loss": 2.443,
+      "step": 4030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004410855231263663,
+      "loss": 2.3924,
+      "step": 4031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004408654902657901,
+      "loss": 2.4292,
+      "step": 4032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000440645469018919,
+      "loss": 2.3247,
+      "step": 4033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004404254594289636,
+      "loss": 2.2934,
+      "step": 4034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004402054615391329,
+      "loss": 2.3589,
+      "step": 4035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043998547539263313,
+      "loss": 2.2423,
+      "step": 4036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043976550103266864,
+      "loss": 2.3043,
+      "step": 4037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043954553850244105,
+      "loss": 2.2717,
+      "step": 4038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043932558784514996,
+      "loss": 2.5028,
+      "step": 4039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043910564910399247,
+      "loss": 2.2838,
+      "step": 4040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004388857223221632,
+      "loss": 2.4944,
+      "step": 4041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004386658075428547,
+      "loss": 2.4119,
+      "step": 4042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043844590480925745,
+      "loss": 2.3278,
+      "step": 4043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043822601416455863,
+      "loss": 2.1609,
+      "step": 4044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000438006135651944,
+      "loss": 2.3278,
+      "step": 4045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004377862693145964,
+      "loss": 2.3856,
+      "step": 4046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004375664151956962,
+      "loss": 2.3979,
+      "step": 4047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004373465733384218,
+      "loss": 2.5311,
+      "step": 4048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004371267437859491,
+      "loss": 2.3981,
+      "step": 4049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043690692658145136,
+      "loss": 2.3768,
+      "step": 4050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043668712176809935,
+      "loss": 2.26,
+      "step": 4051
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043646732938906155,
+      "loss": 2.3931,
+      "step": 4052
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004362475494875044,
+      "loss": 2.4636,
+      "step": 4053
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043602778210659126,
+      "loss": 2.2806,
+      "step": 4054
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004358080272894832,
+      "loss": 2.2028,
+      "step": 4055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004355882850793392,
+      "loss": 2.5465,
+      "step": 4056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004353685555193151,
+      "loss": 2.4587,
+      "step": 4057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004351488386525646,
+      "loss": 2.4153,
+      "step": 4058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043492913452223945,
+      "loss": 2.4769,
+      "step": 4059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043470944317148803,
+      "loss": 2.5143,
+      "step": 4060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043448976464345644,
+      "loss": 2.2603,
+      "step": 4061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043427009898128866,
+      "loss": 2.438,
+      "step": 4062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043405044622812553,
+      "loss": 2.2395,
+      "step": 4063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043383080642710604,
+      "loss": 2.408,
+      "step": 4064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004336111796213661,
+      "loss": 2.3056,
+      "step": 4065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043339156585403934,
+      "loss": 2.4875,
+      "step": 4066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043317196516825667,
+      "loss": 2.3519,
+      "step": 4067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043295237760714646,
+      "loss": 2.3212,
+      "step": 4068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043273280321383444,
+      "loss": 2.344,
+      "step": 4069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004325132420314443,
+      "loss": 2.3626,
+      "step": 4070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004322936941030963,
+      "loss": 2.2844,
+      "step": 4071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004320741594719087,
+      "loss": 2.2641,
+      "step": 4072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043185463818099683,
+      "loss": 2.2046,
+      "step": 4073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043163513027347336,
+      "loss": 2.4233,
+      "step": 4074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004314156357924487,
+      "loss": 2.2919,
+      "step": 4075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004311961547810305,
+      "loss": 2.3372,
+      "step": 4076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004309766872823235,
+      "loss": 2.3666,
+      "step": 4077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043075723333942995,
+      "loss": 2.32,
+      "step": 4078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043053779299544957,
+      "loss": 2.5283,
+      "step": 4079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043031836629347885,
+      "loss": 2.4362,
+      "step": 4080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004300989532766127,
+      "loss": 2.3477,
+      "step": 4081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004298795539879422,
+      "loss": 2.5132,
+      "step": 4082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004296601684705565,
+      "loss": 2.2454,
+      "step": 4083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004294407967675415,
+      "loss": 2.4985,
+      "step": 4084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004292214389219806,
+      "loss": 2.2712,
+      "step": 4085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004290020949769549,
+      "loss": 2.2603,
+      "step": 4086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004287827649755422,
+      "loss": 2.2789,
+      "step": 4087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004285634489608176,
+      "loss": 2.3081,
+      "step": 4088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042834414697585384,
+      "loss": 2.396,
+      "step": 4089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042812485906372033,
+      "loss": 2.4638,
+      "step": 4090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042790558526748445,
+      "loss": 2.3298,
+      "step": 4091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004276863256302102,
+      "loss": 2.4041,
+      "step": 4092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042746708019495913,
+      "loss": 2.2907,
+      "step": 4093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004272478490047897,
+      "loss": 2.3615,
+      "step": 4094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042702863210275797,
+      "loss": 2.4229,
+      "step": 4095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042680942953191665,
+      "loss": 2.3915,
+      "step": 4096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042659024133531644,
+      "loss": 2.296,
+      "step": 4097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042637106755600434,
+      "loss": 2.2805,
+      "step": 4098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004261519082370252,
+      "loss": 2.4254,
+      "step": 4099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004259327634214206,
+      "loss": 2.4426,
+      "step": 4100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004257136331522293,
+      "loss": 2.3488,
+      "step": 4101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004254945174724875,
+      "loss": 2.3194,
+      "step": 4102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042527541642522847,
+      "loss": 2.2705,
+      "step": 4103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004250563300534824,
+      "loss": 2.2874,
+      "step": 4104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042483725840027656,
+      "loss": 2.4928,
+      "step": 4105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042461820150863555,
+      "loss": 2.2139,
+      "step": 4106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042439915942158086,
+      "loss": 2.3942,
+      "step": 4107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004241801321821314,
+      "loss": 2.3684,
+      "step": 4108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042396111983330305,
+      "loss": 2.4082,
+      "step": 4109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042374212241810843,
+      "loss": 2.4446,
+      "step": 4110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042352313997955754,
+      "loss": 2.3976,
+      "step": 4111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004233041725606572,
+      "loss": 2.3535,
+      "step": 4112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042308522020441196,
+      "loss": 2.2628,
+      "step": 4113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004228662829538226,
+      "loss": 2.4396,
+      "step": 4114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000422647360851887,
+      "loss": 2.3555,
+      "step": 4115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004224284539416007,
+      "loss": 2.4174,
+      "step": 4116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004222095622659556,
+      "loss": 2.3918,
+      "step": 4117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004219906858679409,
+      "loss": 2.4428,
+      "step": 4118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042177182479054287,
+      "loss": 2.4361,
+      "step": 4119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004215529790767447,
+      "loss": 2.4506,
+      "step": 4120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004213341487695264,
+      "loss": 2.3372,
+      "step": 4121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004211153339118652,
+      "loss": 2.341,
+      "step": 4122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004208965345467349,
+      "loss": 2.2874,
+      "step": 4123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000420677750717107,
+      "loss": 2.3715,
+      "step": 4124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004204589824659491,
+      "loss": 2.2601,
+      "step": 4125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004202402298362264,
+      "loss": 2.3336,
+      "step": 4126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042002149287090064,
+      "loss": 2.3984,
+      "step": 4127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041980277161293033,
+      "loss": 2.4172,
+      "step": 4128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041958406610527157,
+      "loss": 2.4668,
+      "step": 4129
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041936537639087696,
+      "loss": 2.4185,
+      "step": 4130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004191467025126957,
+      "loss": 2.3445,
+      "step": 4131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004189280445136745,
+      "loss": 2.2741,
+      "step": 4132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041870940243675646,
+      "loss": 2.4058,
+      "step": 4133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041849077632488154,
+      "loss": 2.4352,
+      "step": 4134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004182721662209871,
+      "loss": 2.3889,
+      "step": 4135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004180535721680069,
+      "loss": 2.3871,
+      "step": 4136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004178349942088717,
+      "loss": 2.4939,
+      "step": 4137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041761643238650886,
+      "loss": 2.3567,
+      "step": 4138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004173978867438428,
+      "loss": 2.3371,
+      "step": 4139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041717935732379495,
+      "loss": 2.3714,
+      "step": 4140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004169608441692833,
+      "loss": 2.3329,
+      "step": 4141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041674234732322243,
+      "loss": 2.4268,
+      "step": 4142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041652386682852424,
+      "loss": 2.4391,
+      "step": 4143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041630540272809686,
+      "loss": 2.2066,
+      "step": 4144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004160869550648455,
+      "loss": 2.3564,
+      "step": 4145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004158685238816726,
+      "loss": 2.307,
+      "step": 4146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004156501092214764,
+      "loss": 2.3815,
+      "step": 4147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041543171112715257,
+      "loss": 2.3077,
+      "step": 4148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004152133296415933,
+      "loss": 2.4515,
+      "step": 4149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004149949648076874,
+      "loss": 2.2278,
+      "step": 4150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041477661666832095,
+      "loss": 2.3791,
+      "step": 4151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041455828526637604,
+      "loss": 2.418,
+      "step": 4152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000414339970644732,
+      "loss": 2.3089,
+      "step": 4153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041412167284626453,
+      "loss": 2.4581,
+      "step": 4154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004139033919138463,
+      "loss": 2.4406,
+      "step": 4155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041368512789034616,
+      "loss": 2.2962,
+      "step": 4156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041346688081863057,
+      "loss": 2.3137,
+      "step": 4157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004132486507415617,
+      "loss": 2.4125,
+      "step": 4158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041303043770199904,
+      "loss": 2.3161,
+      "step": 4159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004128122417427983,
+      "loss": 2.4173,
+      "step": 4160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004125940629068119,
+      "loss": 2.3854,
+      "step": 4161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041237590123688924,
+      "loss": 2.4735,
+      "step": 4162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041215775677587614,
+      "loss": 2.3197,
+      "step": 4163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041193962956661493,
+      "loss": 2.3172,
+      "step": 4164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041172151965194456,
+      "loss": 2.3862,
+      "step": 4165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041150342707470055,
+      "loss": 2.3667,
+      "step": 4166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004112853518777156,
+      "loss": 2.3331,
+      "step": 4167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004110672941038181,
+      "loss": 2.134,
+      "step": 4168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004108492537958337,
+      "loss": 2.3412,
+      "step": 4169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004106312309965842,
+      "loss": 2.4411,
+      "step": 4170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004104132257488881,
+      "loss": 2.3662,
+      "step": 4171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041019523809556044,
+      "loss": 2.4002,
+      "step": 4172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040997726807941317,
+      "loss": 2.3477,
+      "step": 4173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004097593157432542,
+      "loss": 2.3293,
+      "step": 4174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040954138112988815,
+      "loss": 2.4132,
+      "step": 4175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004093234642821164,
+      "loss": 2.4084,
+      "step": 4176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004091055652427364,
+      "loss": 2.4961,
+      "step": 4177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040888768405454275,
+      "loss": 2.3602,
+      "step": 4178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004086698207603259,
+      "loss": 2.2619,
+      "step": 4179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040845197540287316,
+      "loss": 2.2951,
+      "step": 4180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040823414802496804,
+      "loss": 2.2244,
+      "step": 4181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040801633866939096,
+      "loss": 2.4367,
+      "step": 4182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040779854737891814,
+      "loss": 2.5085,
+      "step": 4183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004075807741963231,
+      "loss": 2.3053,
+      "step": 4184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040736301916437494,
+      "loss": 2.4784,
+      "step": 4185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040714528232583993,
+      "loss": 2.4262,
+      "step": 4186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004069275637234803,
+      "loss": 2.3488,
+      "step": 4187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004067098634000545,
+      "loss": 2.3996,
+      "step": 4188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004064921813983182,
+      "loss": 2.4336,
+      "step": 4189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004062745177610229,
+      "loss": 2.3347,
+      "step": 4190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040605687253091646,
+      "loss": 2.2942,
+      "step": 4191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004058392457507432,
+      "loss": 2.2891,
+      "step": 4192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004056216374632438,
+      "loss": 2.3004,
+      "step": 4193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004054040477111559,
+      "loss": 2.2366,
+      "step": 4194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040518647653721234,
+      "loss": 2.3193,
+      "step": 4195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040496892398414345,
+      "loss": 2.4228,
+      "step": 4196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004047513900946751,
+      "loss": 2.4353,
+      "step": 4197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040453387491152983,
+      "loss": 2.3454,
+      "step": 4198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040431637847742637,
+      "loss": 2.3761,
+      "step": 4199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004040989008350802,
+      "loss": 2.2411,
+      "step": 4200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004038814420272026,
+      "loss": 2.3598,
+      "step": 4201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004036640020965012,
+      "loss": 2.4705,
+      "step": 4202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004034465810856803,
+      "loss": 2.2932,
+      "step": 4203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040322917903743983,
+      "loss": 2.4278,
+      "step": 4204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000403011795994477,
+      "loss": 2.3223,
+      "step": 4205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004027944319994842,
+      "loss": 2.3472,
+      "step": 4206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004025770870951508,
+      "loss": 2.3795,
+      "step": 4207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040235976132416205,
+      "loss": 2.3697,
+      "step": 4208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040214245472919974,
+      "loss": 2.3296,
+      "step": 4209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004019251673529415,
+      "loss": 2.4527,
+      "step": 4210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040170789923806173,
+      "loss": 2.3942,
+      "step": 4211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004014906504272305,
+      "loss": 2.4049,
+      "step": 4212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004012734209631145,
+      "loss": 2.307,
+      "step": 4213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004010562108883764,
+      "loss": 2.3604,
+      "step": 4214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004008390202456748,
+      "loss": 2.3248,
+      "step": 4215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004006218490776652,
+      "loss": 2.2252,
+      "step": 4216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004004046974269988,
+      "loss": 2.4613,
+      "step": 4217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004001875653363229,
+      "loss": 2.3423,
+      "step": 4218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003999704528482812,
+      "loss": 2.4765,
+      "step": 4219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039975336000551344,
+      "loss": 2.3387,
+      "step": 4220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039953628685065524,
+      "loss": 2.2428,
+      "step": 4221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000399319233426339,
+      "loss": 2.4488,
+      "step": 4222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003991021997751928,
+      "loss": 2.396,
+      "step": 4223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003988851859398408,
+      "loss": 2.3511,
+      "step": 4224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003986681919629033,
+      "loss": 2.3475,
+      "step": 4225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039845121788699666,
+      "loss": 2.4721,
+      "step": 4226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003982342637547339,
+      "loss": 2.4289,
+      "step": 4227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039801732960872337,
+      "loss": 2.2306,
+      "step": 4228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039780041549156977,
+      "loss": 2.2062,
+      "step": 4229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039758352144587397,
+      "loss": 2.4579,
+      "step": 4230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039736664751423257,
+      "loss": 2.205,
+      "step": 4231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039714979373923874,
+      "loss": 2.3442,
+      "step": 4232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003969329601634815,
+      "loss": 2.3088,
+      "step": 4233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039671614682954573,
+      "loss": 2.2821,
+      "step": 4234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003964993537800121,
+      "loss": 2.3809,
+      "step": 4235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039628258105745803,
+      "loss": 2.3582,
+      "step": 4236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039606582870445616,
+      "loss": 2.4164,
+      "step": 4237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003958490967635759,
+      "loss": 2.5044,
+      "step": 4238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000395632385277382,
+      "loss": 2.3409,
+      "step": 4239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039541569428843564,
+      "loss": 2.4704,
+      "step": 4240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003951990238392935,
+      "loss": 2.451,
+      "step": 4241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003949823739725088,
+      "loss": 2.2183,
+      "step": 4242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003947657447306302,
+      "loss": 2.2423,
+      "step": 4243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003945491361562028,
+      "loss": 2.3692,
+      "step": 4244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003943325482917671,
+      "loss": 2.2367,
+      "step": 4245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039411598117986,
+      "loss": 2.3146,
+      "step": 4246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003938994348630141,
+      "loss": 2.3748,
+      "step": 4247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039368290938375766,
+      "loss": 2.3368,
+      "step": 4248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003934664047846156,
+      "loss": 2.3354,
+      "step": 4249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003932499211081081,
+      "loss": 2.3285,
+      "step": 4250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039303345839675143,
+      "loss": 2.1712,
+      "step": 4251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039281701669305753,
+      "loss": 2.2873,
+      "step": 4252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039260059603953446,
+      "loss": 2.3154,
+      "step": 4253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003923841964786864,
+      "loss": 2.3184,
+      "step": 4254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003921678180530127,
+      "loss": 2.3248,
+      "step": 4255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039195146080500925,
+      "loss": 2.2524,
+      "step": 4256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003917351247771673,
+      "loss": 2.3935,
+      "step": 4257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003915188100119741,
+      "loss": 2.2582,
+      "step": 4258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003913025165519125,
+      "loss": 2.387,
+      "step": 4259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039108624443946196,
+      "loss": 2.4091,
+      "step": 4260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003908699937170968,
+      "loss": 2.4215,
+      "step": 4261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003906537644272875,
+      "loss": 2.399,
+      "step": 4262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039043755661250055,
+      "loss": 2.4264,
+      "step": 4263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003902213703151976,
+      "loss": 2.4556,
+      "step": 4264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039000520557783705,
+      "loss": 2.3944,
+      "step": 4265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003897890624428721,
+      "loss": 2.4262,
+      "step": 4266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038957294095275243,
+      "loss": 2.5265,
+      "step": 4267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038935684114992287,
+      "loss": 2.3784,
+      "step": 4268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003891407630768243,
+      "loss": 2.421,
+      "step": 4269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003889247067758936,
+      "loss": 2.2929,
+      "step": 4270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038870867228956287,
+      "loss": 2.3046,
+      "step": 4271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038849265966026017,
+      "loss": 2.4242,
+      "step": 4272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038827666893040923,
+      "loss": 2.1722,
+      "step": 4273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038806070014242957,
+      "loss": 2.3585,
+      "step": 4274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038784475333873594,
+      "loss": 2.5397,
+      "step": 4275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038762882856173967,
+      "loss": 2.2483,
+      "step": 4276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000387412925853847,
+      "loss": 2.3443,
+      "step": 4277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003871970452574602,
+      "loss": 2.3891,
+      "step": 4278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038698118681497685,
+      "loss": 2.3691,
+      "step": 4279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038676535056879043,
+      "loss": 2.3009,
+      "step": 4280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038654953656129033,
+      "loss": 2.4099,
+      "step": 4281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000386333744834861,
+      "loss": 2.3651,
+      "step": 4282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000386117975431883,
+      "loss": 2.515,
+      "step": 4283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003859022283947322,
+      "loss": 2.2726,
+      "step": 4284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038568650376578,
+      "loss": 2.4098,
+      "step": 4285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038547080158739375,
+      "loss": 2.4048,
+      "step": 4286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003852551219019363,
+      "loss": 2.402,
+      "step": 4287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038503946475176593,
+      "loss": 2.2349,
+      "step": 4288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038482383017923646,
+      "loss": 2.3684,
+      "step": 4289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038460821822669755,
+      "loss": 2.2313,
+      "step": 4290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000384392628936494,
+      "loss": 2.3689,
+      "step": 4291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038417706235096665,
+      "loss": 2.3668,
+      "step": 4292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038396151851245155,
+      "loss": 2.2954,
+      "step": 4293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038374599746328044,
+      "loss": 2.3173,
+      "step": 4294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038353049924578033,
+      "loss": 2.2736,
+      "step": 4295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003833150239022741,
+      "loss": 2.3545,
+      "step": 4296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038309957147507974,
+      "loss": 2.4239,
+      "step": 4297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003828841420065113,
+      "loss": 2.2336,
+      "step": 4298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003826687355388778,
+      "loss": 2.2838,
+      "step": 4299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000382453352114484,
+      "loss": 2.3737,
+      "step": 4300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038223799177563005,
+      "loss": 2.3931,
+      "step": 4301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003820226545646113,
+      "loss": 2.1984,
+      "step": 4302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038180734052371926,
+      "loss": 2.3727,
+      "step": 4303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038159204969524037,
+      "loss": 2.2514,
+      "step": 4304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003813767821214564,
+      "loss": 2.4168,
+      "step": 4305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003811615378446451,
+      "loss": 2.4743,
+      "step": 4306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003809463169070788,
+      "loss": 2.3678,
+      "step": 4307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038073111935102633,
+      "loss": 2.3298,
+      "step": 4308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003805159452187509,
+      "loss": 2.5217,
+      "step": 4309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000380300794552512,
+      "loss": 2.261,
+      "step": 4310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003800856673945637,
+      "loss": 2.3266,
+      "step": 4311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037987056378715594,
+      "loss": 2.249,
+      "step": 4312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003796554837725338,
+      "loss": 2.3753,
+      "step": 4313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003794404273929383,
+      "loss": 2.4164,
+      "step": 4314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003792253946906051,
+      "loss": 2.2867,
+      "step": 4315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003790103857077653,
+      "loss": 2.3031,
+      "step": 4316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003787954004866458,
+      "loss": 2.3064,
+      "step": 4317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003785804390694684,
+      "loss": 2.5152,
+      "step": 4318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037836550149845043,
+      "loss": 2.3051,
+      "step": 4319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003781505878158047,
+      "loss": 2.4311,
+      "step": 4320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003779356980637389,
+      "loss": 2.3768,
+      "step": 4321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037772083228445615,
+      "loss": 2.3643,
+      "step": 4322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003775059905201552,
+      "loss": 2.27,
+      "step": 4323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003772911728130295,
+      "loss": 2.2787,
+      "step": 4324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037707637920526854,
+      "loss": 2.1943,
+      "step": 4325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003768616097390564,
+      "loss": 2.546,
+      "step": 4326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003766468644565727,
+      "loss": 2.2791,
+      "step": 4327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003764321433999922,
+      "loss": 2.4,
+      "step": 4328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000376217446611485,
+      "loss": 2.5051,
+      "step": 4329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003760027741332166,
+      "loss": 2.3734,
+      "step": 4330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037578812600734746,
+      "loss": 2.3547,
+      "step": 4331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003755735022760332,
+      "loss": 2.4166,
+      "step": 4332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037535890298142494,
+      "loss": 2.4067,
+      "step": 4333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003751443281656688,
+      "loss": 2.1909,
+      "step": 4334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037492977787090595,
+      "loss": 2.4138,
+      "step": 4335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037471525213927313,
+      "loss": 2.3056,
+      "step": 4336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037450075101290214,
+      "loss": 2.4661,
+      "step": 4337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037428627453391984,
+      "loss": 2.4012,
+      "step": 4338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000374071822744448,
+      "loss": 2.2583,
+      "step": 4339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037385739568660393,
+      "loss": 2.4149,
+      "step": 4340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037364299340250027,
+      "loss": 2.3947,
+      "step": 4341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003734286159342442,
+      "loss": 2.292,
+      "step": 4342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003732142633239384,
+      "loss": 2.3745,
+      "step": 4343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037299993561368066,
+      "loss": 2.4348,
+      "step": 4344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037278563284556354,
+      "loss": 2.2568,
+      "step": 4345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003725713550616752,
+      "loss": 2.3886,
+      "step": 4346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037235710230409876,
+      "loss": 2.4084,
+      "step": 4347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003721428746149122,
+      "loss": 2.2363,
+      "step": 4348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003719286720361886,
+      "loss": 2.2603,
+      "step": 4349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003717144946099964,
+      "loss": 2.3894,
+      "step": 4350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037150034237839856,
+      "loss": 2.3257,
+      "step": 4351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000371286215383454,
+      "loss": 2.2685,
+      "step": 4352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037107211366721574,
+      "loss": 2.3516,
+      "step": 4353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003708580372717324,
+      "loss": 2.3188,
+      "step": 4354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003706439862390472,
+      "loss": 2.2905,
+      "step": 4355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037042996061119885,
+      "loss": 2.4214,
+      "step": 4356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003702159604302209,
+      "loss": 2.3295,
+      "step": 4357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037000198573814173,
+      "loss": 2.317,
+      "step": 4358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036978803657698484,
+      "loss": 2.3141,
+      "step": 4359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036957411298876885,
+      "loss": 2.4327,
+      "step": 4360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003693602150155071,
+      "loss": 2.3887,
+      "step": 4361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036914634269920776,
+      "loss": 2.3187,
+      "step": 4362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003689324960818747,
+      "loss": 2.3587,
+      "step": 4363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003687186752055061,
+      "loss": 2.4389,
+      "step": 4364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036850488011209515,
+      "loss": 2.6451,
+      "step": 4365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003682911108436302,
+      "loss": 2.5197,
+      "step": 4366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003680773674420941,
+      "loss": 2.4574,
+      "step": 4367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003678636499494654,
+      "loss": 2.3011,
+      "step": 4368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036764995840771674,
+      "loss": 2.2683,
+      "step": 4369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003674362928588162,
+      "loss": 2.4644,
+      "step": 4370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036722265334472647,
+      "loss": 2.3364,
+      "step": 4371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003670090399074051,
+      "loss": 2.1988,
+      "step": 4372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003667954525888048,
+      "loss": 2.1145,
+      "step": 4373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036658189143087307,
+      "loss": 2.2375,
+      "step": 4374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003663683564755521,
+      "loss": 2.2456,
+      "step": 4375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003661548477647789,
+      "loss": 2.2354,
+      "step": 4376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003659413653404856,
+      "loss": 2.3782,
+      "step": 4377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036572790924459876,
+      "loss": 2.4276,
+      "step": 4378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036551447951904036,
+      "loss": 2.3165,
+      "step": 4379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036530107620572695,
+      "loss": 2.2336,
+      "step": 4380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003650876993465695,
+      "loss": 2.3677,
+      "step": 4381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003648743489834742,
+      "loss": 2.2569,
+      "step": 4382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003646610251583418,
+      "loss": 2.2991,
+      "step": 4383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036444772791306826,
+      "loss": 2.3854,
+      "step": 4384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003642344572895439,
+      "loss": 2.327,
+      "step": 4385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036402121332965387,
+      "loss": 2.3236,
+      "step": 4386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036380799607527826,
+      "loss": 2.3737,
+      "step": 4387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036359480556829183,
+      "loss": 2.295,
+      "step": 4388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003633816418505637,
+      "loss": 2.397,
+      "step": 4389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003631685049639586,
+      "loss": 2.3608,
+      "step": 4390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003629553949503353,
+      "loss": 2.3977,
+      "step": 4391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003627423118515474,
+      "loss": 2.29,
+      "step": 4392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003625292557094434,
+      "loss": 2.3265,
+      "step": 4393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003623162265658662,
+      "loss": 2.3317,
+      "step": 4394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000362103224462654,
+      "loss": 2.2989,
+      "step": 4395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036189024944163886,
+      "loss": 2.2765,
+      "step": 4396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003616773015446483,
+      "loss": 2.5431,
+      "step": 4397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003614643808135039,
+      "loss": 2.4305,
+      "step": 4398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036125148729002223,
+      "loss": 2.3639,
+      "step": 4399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036103862101601427,
+      "loss": 2.2776,
+      "step": 4400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036082578203328633,
+      "loss": 2.2465,
+      "step": 4401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036061297038363847,
+      "loss": 2.3183,
+      "step": 4402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003604001861088658,
+      "loss": 2.4916,
+      "step": 4403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003601874292507583,
+      "loss": 2.2806,
+      "step": 4404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035997469985109975,
+      "loss": 2.3442,
+      "step": 4405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035976199795166954,
+      "loss": 2.3124,
+      "step": 4406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003595493235942412,
+      "loss": 2.3554,
+      "step": 4407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003593366768205827,
+      "loss": 2.3484,
+      "step": 4408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003591240576724567,
+      "loss": 2.3426,
+      "step": 4409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035891146619162036,
+      "loss": 2.278,
+      "step": 4410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035869890241982595,
+      "loss": 2.2772,
+      "step": 4411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003584863663988196,
+      "loss": 2.2885,
+      "step": 4412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003582738581703422,
+      "loss": 2.4778,
+      "step": 4413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035806137777612935,
+      "loss": 2.1379,
+      "step": 4414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000357848925257911,
+      "loss": 2.3209,
+      "step": 4415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035763650065741173,
+      "loss": 2.3059,
+      "step": 4416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003574241040163506,
+      "loss": 2.3468,
+      "step": 4417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035721173537644135,
+      "loss": 2.2616,
+      "step": 4418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003569993947793918,
+      "loss": 2.3381,
+      "step": 4419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035678708226690483,
+      "loss": 2.1798,
+      "step": 4420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035657479788067697,
+      "loss": 2.3103,
+      "step": 4421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035636254166240036,
+      "loss": 2.2634,
+      "step": 4422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003561503136537606,
+      "loss": 2.3147,
+      "step": 4423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003559381138964383,
+      "loss": 2.36,
+      "step": 4424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003557259424321084,
+      "loss": 2.3853,
+      "step": 4425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003555137993024401,
+      "loss": 2.362,
+      "step": 4426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003553016845490972,
+      "loss": 2.1878,
+      "step": 4427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035508959821373815,
+      "loss": 2.3448,
+      "step": 4428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035487754033801535,
+      "loss": 2.2649,
+      "step": 4429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000354665510963576,
+      "loss": 2.4124,
+      "step": 4430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003544535101320615,
+      "loss": 2.4272,
+      "step": 4431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003542415378851075,
+      "loss": 2.4775,
+      "step": 4432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035402959426434445,
+      "loss": 2.338,
+      "step": 4433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000353817679311397,
+      "loss": 2.2802,
+      "step": 4434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000353605793067884,
+      "loss": 2.406,
+      "step": 4435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035339393557541875,
+      "loss": 2.2354,
+      "step": 4436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035318210687560906,
+      "loss": 2.3277,
+      "step": 4437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035297030701005673,
+      "loss": 2.4477,
+      "step": 4438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003527585360203586,
+      "loss": 2.367,
+      "step": 4439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003525467939481049,
+      "loss": 2.1308,
+      "step": 4440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035233508083488084,
+      "loss": 2.3067,
+      "step": 4441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003521233967222658,
+      "loss": 2.3126,
+      "step": 4442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003519117416518332,
+      "loss": 2.4559,
+      "step": 4443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003517001156651514,
+      "loss": 2.3,
+      "step": 4444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003514885188037823,
+      "loss": 2.4854,
+      "step": 4445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003512769511092825,
+      "loss": 2.314,
+      "step": 4446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035106541262320276,
+      "loss": 2.3878,
+      "step": 4447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003508539033870879,
+      "loss": 2.3791,
+      "step": 4448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035064242344247766,
+      "loss": 2.4263,
+      "step": 4449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003504309728309052,
+      "loss": 2.4077,
+      "step": 4450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003502195515938986,
+      "loss": 2.2234,
+      "step": 4451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035000815977297964,
+      "loss": 2.293,
+      "step": 4452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003497967974096647,
+      "loss": 2.2203,
+      "step": 4453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034958546454546397,
+      "loss": 2.398,
+      "step": 4454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034937416122188253,
+      "loss": 2.3208,
+      "step": 4455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034916288748041894,
+      "loss": 2.2283,
+      "step": 4456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034895164336256636,
+      "loss": 2.3551,
+      "step": 4457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034874042890981196,
+      "loss": 2.3342,
+      "step": 4458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034852924416363694,
+      "loss": 2.4728,
+      "step": 4459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003483180891655173,
+      "loss": 2.3261,
+      "step": 4460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003481069639569225,
+      "loss": 2.3421,
+      "step": 4461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003478958685793165,
+      "loss": 2.3183,
+      "step": 4462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034768480307415726,
+      "loss": 2.2531,
+      "step": 4463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000347473767482897,
+      "loss": 2.3332,
+      "step": 4464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003472627618469817,
+      "loss": 2.4687,
+      "step": 4465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034705178620785215,
+      "loss": 2.3967,
+      "step": 4466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003468408406069428,
+      "loss": 2.2315,
+      "step": 4467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034662992508568204,
+      "loss": 2.3026,
+      "step": 4468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003464190396854926,
+      "loss": 2.2116,
+      "step": 4469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003462081844477912,
+      "loss": 2.3004,
+      "step": 4470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034599735941398904,
+      "loss": 2.3867,
+      "step": 4471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034578656462549084,
+      "loss": 2.2531,
+      "step": 4472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034557580012369536,
+      "loss": 2.3619,
+      "step": 4473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034536506594999605,
+      "loss": 2.3689,
+      "step": 4474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003451543621457797,
+      "loss": 2.3129,
+      "step": 4475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034494368875242733,
+      "loss": 2.2725,
+      "step": 4476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003447330458113143,
+      "loss": 2.4303,
+      "step": 4477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003445224333638098,
+      "loss": 2.2576,
+      "step": 4478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034431185145127676,
+      "loss": 2.4234,
+      "step": 4479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003441013001150727,
+      "loss": 2.3426,
+      "step": 4480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034389077939654835,
+      "loss": 2.3543,
+      "step": 4481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034368028933704924,
+      "loss": 2.1361,
+      "step": 4482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003434698299779143,
+      "loss": 2.3877,
+      "step": 4483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034325940136047686,
+      "loss": 2.33,
+      "step": 4484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003430490035260638,
+      "loss": 2.372,
+      "step": 4485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034283863651599593,
+      "loss": 2.3364,
+      "step": 4486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003426283003715886,
+      "loss": 2.2848,
+      "step": 4487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034241799513415067,
+      "loss": 2.2317,
+      "step": 4488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034220772084498486,
+      "loss": 2.3273,
+      "step": 4489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003419974775453879,
+      "loss": 2.358,
+      "step": 4490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003417872652766506,
+      "loss": 2.4146,
+      "step": 4491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034157708408005723,
+      "loss": 2.358,
+      "step": 4492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034136693399688657,
+      "loss": 2.1908,
+      "step": 4493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000341156815068411,
+      "loss": 2.443,
+      "step": 4494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003409467273358967,
+      "loss": 2.4792,
+      "step": 4495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003407366708406037,
+      "loss": 2.4021,
+      "step": 4496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003405266456237858,
+      "loss": 2.2308,
+      "step": 4497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034031665172669134,
+      "loss": 2.2428,
+      "step": 4498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003401066891905618,
+      "loss": 2.4043,
+      "step": 4499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003398967580566325,
+      "loss": 2.4242,
+      "step": 4500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003396868583661331,
+      "loss": 2.2694,
+      "step": 4501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033947699016028657,
+      "loss": 2.2924,
+      "step": 4502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033926715348031,
+      "loss": 2.4244,
+      "step": 4503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033905734836741416,
+      "loss": 2.3643,
+      "step": 4504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003388475748628038,
+      "loss": 2.3587,
+      "step": 4505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003386378330076772,
+      "loss": 2.155,
+      "step": 4506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003384281228432267,
+      "loss": 2.2362,
+      "step": 4507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003382184444106379,
+      "loss": 2.1859,
+      "step": 4508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033800879775109093,
+      "loss": 2.3849,
+      "step": 4509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003377991829057591,
+      "loss": 2.2233,
+      "step": 4510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033758959991580986,
+      "loss": 2.4031,
+      "step": 4511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000337380048822404,
+      "loss": 2.3371,
+      "step": 4512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033717052966669605,
+      "loss": 2.3838,
+      "step": 4513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003369610424898348,
+      "loss": 2.6432,
+      "step": 4514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033675158733296247,
+      "loss": 2.2518,
+      "step": 4515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033654216423721465,
+      "loss": 2.4846,
+      "step": 4516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033633277324372123,
+      "loss": 2.3867,
+      "step": 4517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033612341439360533,
+      "loss": 2.4368,
+      "step": 4518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003359140877279837,
+      "loss": 2.3479,
+      "step": 4519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003357047932879673,
+      "loss": 2.2491,
+      "step": 4520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003354955311146606,
+      "loss": 2.2731,
+      "step": 4521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003352863012491612,
+      "loss": 2.2972,
+      "step": 4522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003350771037325608,
+      "loss": 2.4907,
+      "step": 4523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003348679386059448,
+      "loss": 2.3116,
+      "step": 4524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033465880591039223,
+      "loss": 2.4217,
+      "step": 4525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003344497056869755,
+      "loss": 2.4201,
+      "step": 4526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033424063797676075,
+      "loss": 2.4077,
+      "step": 4527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003340316028208079,
+      "loss": 2.3909,
+      "step": 4528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033382260026017024,
+      "loss": 2.3736,
+      "step": 4529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033361363033589466,
+      "loss": 2.2832,
+      "step": 4530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003334046930890221,
+      "loss": 2.1664,
+      "step": 4531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003331957885605865,
+      "loss": 2.2208,
+      "step": 4532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003329869167916155,
+      "loss": 2.3607,
+      "step": 4533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033277807782313074,
+      "loss": 2.433,
+      "step": 4534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003325692716961466,
+      "loss": 2.3001,
+      "step": 4535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033236049845167205,
+      "loss": 2.2228,
+      "step": 4536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033215175813070864,
+      "loss": 2.2941,
+      "step": 4537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003319430507742521,
+      "loss": 2.4109,
+      "step": 4538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003317343764232913,
+      "loss": 2.3176,
+      "step": 4539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003315257351188089,
+      "loss": 2.573,
+      "step": 4540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003313171269017807,
+      "loss": 2.3698,
+      "step": 4541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003311085518131767,
+      "loss": 2.2841,
+      "step": 4542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003309000098939595,
+      "loss": 2.3142,
+      "step": 4543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033069150118508593,
+      "loss": 2.3418,
+      "step": 4544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033048302572750586,
+      "loss": 2.3866,
+      "step": 4545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033027458356216265,
+      "loss": 2.1802,
+      "step": 4546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033006617472999345,
+      "loss": 2.3644,
+      "step": 4547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003298577992719286,
+      "loss": 2.3025,
+      "step": 4548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032964945722889206,
+      "loss": 2.3214,
+      "step": 4549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003294411486418007,
+      "loss": 2.5148,
+      "step": 4550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003292328735515655,
+      "loss": 2.5164,
+      "step": 4551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003290246319990908,
+      "loss": 2.3797,
+      "step": 4552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003288164240252737,
+      "loss": 2.3292,
+      "step": 4553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032860824967100553,
+      "loss": 2.1509,
+      "step": 4554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032840010897717045,
+      "loss": 2.3745,
+      "step": 4555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032819200198464605,
+      "loss": 2.2955,
+      "step": 4556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032798392873430355,
+      "loss": 2.2257,
+      "step": 4557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003277758892670076,
+      "loss": 2.4093,
+      "step": 4558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000327567883623616,
+      "loss": 2.2184,
+      "step": 4559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003273599118449798,
+      "loss": 2.3108,
+      "step": 4560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003271519739719437,
+      "loss": 2.3216,
+      "step": 4561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003269440700453453,
+      "loss": 2.2376,
+      "step": 4562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032673620010601634,
+      "loss": 2.4098,
+      "step": 4563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032652836419478097,
+      "loss": 2.3734,
+      "step": 4564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003263205623524573,
+      "loss": 2.5075,
+      "step": 4565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032611279461985634,
+      "loss": 2.414,
+      "step": 4566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003259050610377827,
+      "loss": 2.4615,
+      "step": 4567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000325697361647034,
+      "loss": 2.2768,
+      "step": 4568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032548969648840155,
+      "loss": 2.4082,
+      "step": 4569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032528206560266944,
+      "loss": 2.2878,
+      "step": 4570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032507446903061547,
+      "loss": 2.326,
+      "step": 4571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032486690681301047,
+      "loss": 2.2613,
+      "step": 4572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003246593789906183,
+      "loss": 2.1872,
+      "step": 4573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003244518856041966,
+      "loss": 2.2801,
+      "step": 4574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000324244426694496,
+      "loss": 2.4314,
+      "step": 4575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032403700230226016,
+      "loss": 2.3129,
+      "step": 4576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032382961246822634,
+      "loss": 2.2407,
+      "step": 4577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032362225723312464,
+      "loss": 2.3163,
+      "step": 4578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003234149366376784,
+      "loss": 2.4765,
+      "step": 4579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003232076507226046,
+      "loss": 2.2034,
+      "step": 4580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032300039952861304,
+      "loss": 2.3575,
+      "step": 4581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003227931830964067,
+      "loss": 2.195,
+      "step": 4582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003225860014666816,
+      "loss": 2.2502,
+      "step": 4583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003223788546801273,
+      "loss": 2.4227,
+      "step": 4584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032217174277742643,
+      "loss": 2.402,
+      "step": 4585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032196466579925454,
+      "loss": 2.4417,
+      "step": 4586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032175762378628043,
+      "loss": 2.3854,
+      "step": 4587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003215506167791662,
+      "loss": 2.296,
+      "step": 4588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003213436448185666,
+      "loss": 2.2822,
+      "step": 4589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032113670794513017,
+      "loss": 2.3512,
+      "step": 4590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032092980619949824,
+      "loss": 2.1944,
+      "step": 4591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032072293962230505,
+      "loss": 2.4085,
+      "step": 4592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003205161082541781,
+      "loss": 2.2191,
+      "step": 4593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000320309312135738,
+      "loss": 2.3272,
+      "step": 4594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003201025513075983,
+      "loss": 2.3381,
+      "step": 4595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031989582581036605,
+      "loss": 2.3587,
+      "step": 4596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003196891356846408,
+      "loss": 2.3822,
+      "step": 4597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031948248097101546,
+      "loss": 2.3049,
+      "step": 4598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031927586171007596,
+      "loss": 2.2394,
+      "step": 4599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000319069277942401,
+      "loss": 2.3709,
+      "step": 4600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003188627297085628,
+      "loss": 2.4997,
+      "step": 4601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031865621704912635,
+      "loss": 2.2818,
+      "step": 4602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003184497400046494,
+      "loss": 2.4032,
+      "step": 4603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031824329861568335,
+      "loss": 2.3203,
+      "step": 4604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031803689292277186,
+      "loss": 2.2942,
+      "step": 4605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003178305229664519,
+      "loss": 2.3883,
+      "step": 4606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031762418878725354,
+      "loss": 2.3422,
+      "step": 4607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031741789042569993,
+      "loss": 2.334,
+      "step": 4608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003172116279223066,
+      "loss": 2.2532,
+      "step": 4609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031700540131758266,
+      "loss": 2.2536,
+      "step": 4610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031679921065202966,
+      "loss": 2.4178,
+      "step": 4611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031659305596614285,
+      "loss": 2.2769,
+      "step": 4612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003163869373004096,
+      "loss": 2.2715,
+      "step": 4613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031618085469531035,
+      "loss": 2.3185,
+      "step": 4614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031597480819131894,
+      "loss": 2.3697,
+      "step": 4615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031576879782890165,
+      "loss": 2.3592,
+      "step": 4616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003155628236485178,
+      "loss": 2.2762,
+      "step": 4617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003153568856906199,
+      "loss": 2.3981,
+      "step": 4618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031515098399565285,
+      "loss": 2.328,
+      "step": 4619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031494511860405454,
+      "loss": 2.1731,
+      "step": 4620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031473928955625617,
+      "loss": 2.3086,
+      "step": 4621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031453349689268106,
+      "loss": 2.3359,
+      "step": 4622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003143277406537463,
+      "loss": 2.4068,
+      "step": 4623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031412202087986084,
+      "loss": 2.3258,
+      "step": 4624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003139163376114274,
+      "loss": 2.1849,
+      "step": 4625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031371069088884074,
+      "loss": 2.3151,
+      "step": 4626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031350508075248884,
+      "loss": 2.3341,
+      "step": 4627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003132995072427526,
+      "loss": 2.4575,
+      "step": 4628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031309397040000554,
+      "loss": 2.2192,
+      "step": 4629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031288847026461386,
+      "loss": 2.3755,
+      "step": 4630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003126830068769369,
+      "loss": 2.288,
+      "step": 4631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003124775802773265,
+      "loss": 2.2417,
+      "step": 4632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003122721905061271,
+      "loss": 2.3201,
+      "step": 4633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031206683760367647,
+      "loss": 2.4055,
+      "step": 4634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003118615216103049,
+      "loss": 2.4659,
+      "step": 4635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003116562425663353,
+      "loss": 2.2368,
+      "step": 4636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003114510005120831,
+      "loss": 2.4744,
+      "step": 4637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031124579548785693,
+      "loss": 2.3628,
+      "step": 4638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003110406275339582,
+      "loss": 2.3904,
+      "step": 4639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003108354966906805,
+      "loss": 2.3754,
+      "step": 4640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031063040299831067,
+      "loss": 2.277,
+      "step": 4641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031042534649712794,
+      "loss": 2.2315,
+      "step": 4642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003102203272274042,
+      "loss": 2.4252,
+      "step": 4643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003100153452294041,
+      "loss": 2.1805,
+      "step": 4644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030981040054338545,
+      "loss": 2.309,
+      "step": 4645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030960549320959796,
+      "loss": 2.4804,
+      "step": 4646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030940062326828435,
+      "loss": 2.4843,
+      "step": 4647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030919579075968013,
+      "loss": 2.4614,
+      "step": 4648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030899099572401304,
+      "loss": 2.2027,
+      "step": 4649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003087862382015042,
+      "loss": 2.3489,
+      "step": 4650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030858151823236646,
+      "loss": 2.3663,
+      "step": 4651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003083768358568061,
+      "loss": 2.3701,
+      "step": 4652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003081721911150214,
+      "loss": 2.336,
+      "step": 4653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030796758404720364,
+      "loss": 2.3419,
+      "step": 4654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030776301469353633,
+      "loss": 2.3734,
+      "step": 4655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003075584830941962,
+      "loss": 2.4495,
+      "step": 4656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003073539892893519,
+      "loss": 2.2653,
+      "step": 4657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003071495333191651,
+      "loss": 2.3694,
+      "step": 4658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030694511522378964,
+      "loss": 2.1954,
+      "step": 4659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003067407350433722,
+      "loss": 2.3793,
+      "step": 4660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003065363928180521,
+      "loss": 2.3322,
+      "step": 4661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003063320885879612,
+      "loss": 2.1718,
+      "step": 4662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030612782239322346,
+      "loss": 2.329,
+      "step": 4663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003059235942739559,
+      "loss": 2.2813,
+      "step": 4664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003057194042702676,
+      "loss": 2.2807,
+      "step": 4665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030551525242226063,
+      "loss": 2.4342,
+      "step": 4666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030531113877002927,
+      "loss": 2.4045,
+      "step": 4667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030510706335366036,
+      "loss": 2.4262,
+      "step": 4668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003049030262132333,
+      "loss": 2.2865,
+      "step": 4669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030469902738881964,
+      "loss": 2.175,
+      "step": 4670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003044950669204838,
+      "loss": 2.1667,
+      "step": 4671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003042911448482828,
+      "loss": 2.3216,
+      "step": 4672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003040872612122656,
+      "loss": 2.3346,
+      "step": 4673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030388341605247385,
+      "loss": 2.3384,
+      "step": 4674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030367960940894177,
+      "loss": 2.2698,
+      "step": 4675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030347584132169573,
+      "loss": 2.3928,
+      "step": 4676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030327211183075494,
+      "loss": 2.3991,
+      "step": 4677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003030684209761307,
+      "loss": 2.3431,
+      "step": 4678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003028647687978269,
+      "loss": 2.3699,
+      "step": 4679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030266115533583954,
+      "loss": 2.3539,
+      "step": 4680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030245758063015746,
+      "loss": 2.3874,
+      "step": 4681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003022540447207613,
+      "loss": 2.3995,
+      "step": 4682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030205054764762484,
+      "loss": 2.2554,
+      "step": 4683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003018470894507136,
+      "loss": 2.274,
+      "step": 4684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003016436701699859,
+      "loss": 2.2373,
+      "step": 4685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030144028984539203,
+      "loss": 2.3896,
+      "step": 4686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030123694851687466,
+      "loss": 2.3425,
+      "step": 4687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003010336462243692,
+      "loss": 2.2733,
+      "step": 4688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003008303830078032,
+      "loss": 2.4707,
+      "step": 4689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030062715890709623,
+      "loss": 2.2375,
+      "step": 4690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003004239739621608,
+      "loss": 2.2771,
+      "step": 4691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030022082821290074,
+      "loss": 2.4946,
+      "step": 4692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003000177216992135,
+      "loss": 2.3612,
+      "step": 4693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029981465446098765,
+      "loss": 2.2504,
+      "step": 4694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002996116265381048,
+      "loss": 2.3995,
+      "step": 4695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029940863797043844,
+      "loss": 2.1696,
+      "step": 4696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029920568879785434,
+      "loss": 2.3317,
+      "step": 4697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002990027790602107,
+      "loss": 2.3233,
+      "step": 4698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002987999087973581,
+      "loss": 2.2866,
+      "step": 4699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029859707804913904,
+      "loss": 2.3299,
+      "step": 4700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002983942868553884,
+      "loss": 2.2978,
+      "step": 4701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029819153525593333,
+      "loss": 2.4046,
+      "step": 4702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002979888232905931,
+      "loss": 2.2746,
+      "step": 4703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002977861509991793,
+      "loss": 2.2669,
+      "step": 4704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002975835184214959,
+      "loss": 2.3622,
+      "step": 4705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029738092559733866,
+      "loss": 2.4179,
+      "step": 4706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029717837256649566,
+      "loss": 2.2919,
+      "step": 4707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969758593687475,
+      "loss": 2.2869,
+      "step": 4708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029677338604386627,
+      "loss": 2.3163,
+      "step": 4709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029657095263161727,
+      "loss": 2.3546,
+      "step": 4710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002963685591717569,
+      "loss": 2.2655,
+      "step": 4711
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002961662057040343,
+      "loss": 2.2284,
+      "step": 4712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029596389226819054,
+      "loss": 2.2758,
+      "step": 4713
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000295761618903959,
+      "loss": 2.1956,
+      "step": 4714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002955593856510651,
+      "loss": 2.5365,
+      "step": 4715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029535719254922645,
+      "loss": 2.2844,
+      "step": 4716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002951550396381525,
+      "loss": 2.2014,
+      "step": 4717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002949529269575453,
+      "loss": 2.308,
+      "step": 4718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002947508545470985,
+      "loss": 2.2153,
+      "step": 4719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029454882244649784,
+      "loss": 2.3367,
+      "step": 4720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029434683069542176,
+      "loss": 2.3835,
+      "step": 4721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029414487933354026,
+      "loss": 2.2458,
+      "step": 4722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029394296840051554,
+      "loss": 2.329,
+      "step": 4723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002937410979360017,
+      "loss": 2.5205,
+      "step": 4724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002935392679796449,
+      "loss": 2.3486,
+      "step": 4725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002933374785710839,
+      "loss": 2.169,
+      "step": 4726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002931357297499489,
+      "loss": 2.2141,
+      "step": 4727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029293402155586225,
+      "loss": 2.3731,
+      "step": 4728
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002927323540284384,
+      "loss": 2.2762,
+      "step": 4729
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002925307272072836,
+      "loss": 2.2283,
+      "step": 4730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029232914113199647,
+      "loss": 2.2902,
+      "step": 4731
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029212759584216764,
+      "loss": 2.3418,
+      "step": 4732
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029192609137737927,
+      "loss": 2.292,
+      "step": 4733
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029172462777720586,
+      "loss": 2.3303,
+      "step": 4734
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029152320508121353,
+      "loss": 2.3978,
+      "step": 4735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000291321823328961,
+      "loss": 2.3471,
+      "step": 4736
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029112048255999843,
+      "loss": 2.2297,
+      "step": 4737
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002909191828138683,
+      "loss": 2.3553,
+      "step": 4738
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029071792413010454,
+      "loss": 2.2369,
+      "step": 4739
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002905167065482334,
+      "loss": 2.1283,
+      "step": 4740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029031553010777265,
+      "loss": 2.3935,
+      "step": 4741
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002901143948482329,
+      "loss": 2.2575,
+      "step": 4742
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002899133008091154,
+      "loss": 2.1012,
+      "step": 4743
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028971224802991444,
+      "loss": 2.2911,
+      "step": 4744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002895112365501156,
+      "loss": 2.3688,
+      "step": 4745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028931026640919633,
+      "loss": 2.2063,
+      "step": 4746
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028910933764662606,
+      "loss": 2.2628,
+      "step": 4747
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002889084503018664,
+      "loss": 2.3604,
+      "step": 4748
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028870760441437025,
+      "loss": 2.3904,
+      "step": 4749
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028850680002358303,
+      "loss": 2.3022,
+      "step": 4750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002883060371689415,
+      "loss": 2.1946,
+      "step": 4751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002881053158898742,
+      "loss": 2.166,
+      "step": 4752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002879046362258021,
+      "loss": 2.3386,
+      "step": 4753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028770399821613745,
+      "loss": 2.3395,
+      "step": 4754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002875034019002846,
+      "loss": 2.3446,
+      "step": 4755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002873028473176392,
+      "loss": 2.3616,
+      "step": 4756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002871023345075896,
+      "loss": 2.4312,
+      "step": 4757
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028690186350951514,
+      "loss": 2.4755,
+      "step": 4758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028670143436278754,
+      "loss": 2.2247,
+      "step": 4759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002865010471067699,
+      "loss": 2.2987,
+      "step": 4760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002863007017808172,
+      "loss": 2.2093,
+      "step": 4761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028610039842427595,
+      "loss": 2.4987,
+      "step": 4762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000285900137076485,
+      "loss": 2.2861,
+      "step": 4763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002856999177767746,
+      "loss": 2.4587,
+      "step": 4764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028549974056446683,
+      "loss": 2.464,
+      "step": 4765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028529960547887524,
+      "loss": 2.2952,
+      "step": 4766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002850995125593053,
+      "loss": 2.4011,
+      "step": 4767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028489946184505414,
+      "loss": 2.3143,
+      "step": 4768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000284699453375411,
+      "loss": 2.2465,
+      "step": 4769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028449948718965604,
+      "loss": 2.2681,
+      "step": 4770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000284299563327062,
+      "loss": 2.4023,
+      "step": 4771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002840996818268926,
+      "loss": 2.336,
+      "step": 4772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002838998427284036,
+      "loss": 2.2402,
+      "step": 4773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002837000460708422,
+      "loss": 2.2768,
+      "step": 4774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002835002918934476,
+      "loss": 2.3592,
+      "step": 4775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028330058023545024,
+      "loss": 2.233,
+      "step": 4776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028310091113607284,
+      "loss": 2.4715,
+      "step": 4777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000282901284634529,
+      "loss": 2.3262,
+      "step": 4778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002827017007700242,
+      "loss": 2.31,
+      "step": 4779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002825021595817561,
+      "loss": 2.3551,
+      "step": 4780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028230266110891323,
+      "loss": 2.4336,
+      "step": 4781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002821032053906759,
+      "loss": 2.3545,
+      "step": 4782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002819037924662166,
+      "loss": 2.3023,
+      "step": 4783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028170442237469866,
+      "loss": 2.3982,
+      "step": 4784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028150509515527725,
+      "loss": 2.4695,
+      "step": 4785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028130581084709946,
+      "loss": 2.2552,
+      "step": 4786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002811065694893035,
+      "loss": 2.2976,
+      "step": 4787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002809073711210194,
+      "loss": 2.3209,
+      "step": 4788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028070821578136837,
+      "loss": 2.3396,
+      "step": 4789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028050910350946366,
+      "loss": 2.2937,
+      "step": 4790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028031003434441016,
+      "loss": 2.3168,
+      "step": 4791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028011100832530376,
+      "loss": 2.3415,
+      "step": 4792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027991202549123206,
+      "loss": 2.4487,
+      "step": 4793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002797130858812743,
+      "loss": 2.2611,
+      "step": 4794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027951418953450096,
+      "loss": 2.2543,
+      "step": 4795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002793153364899744,
+      "loss": 2.2677,
+      "step": 4796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027911652678674854,
+      "loss": 2.3036,
+      "step": 4797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002789177604638683,
+      "loss": 2.2426,
+      "step": 4798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002787190375603704,
+      "loss": 2.2993,
+      "step": 4799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027852035811528296,
+      "loss": 2.3128,
+      "step": 4800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027832172216762534,
+      "loss": 2.2319,
+      "step": 4801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027812312975640903,
+      "loss": 2.2601,
+      "step": 4802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027792458092063606,
+      "loss": 2.3245,
+      "step": 4803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002777260756993008,
+      "loss": 2.2773,
+      "step": 4804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002775276141313885,
+      "loss": 2.4616,
+      "step": 4805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002773291962558756,
+      "loss": 2.3802,
+      "step": 4806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002771308221117309,
+      "loss": 2.1916,
+      "step": 4807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002769324917379137,
+      "loss": 2.4142,
+      "step": 4808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002767342051733749,
+      "loss": 2.2606,
+      "step": 4809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027653596245705736,
+      "loss": 2.3673,
+      "step": 4810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002763377636278946,
+      "loss": 2.3175,
+      "step": 4811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002761396087248118,
+      "loss": 2.2691,
+      "step": 4812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002759414977867257,
+      "loss": 2.3258,
+      "step": 4813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002757434308525442,
+      "loss": 2.1638,
+      "step": 4814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027554540796116656,
+      "loss": 2.2534,
+      "step": 4815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002753474291514832,
+      "loss": 2.3881,
+      "step": 4816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027514949446237635,
+      "loss": 2.3717,
+      "step": 4817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027495160393271954,
+      "loss": 2.3538,
+      "step": 4818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002747537576013772,
+      "loss": 2.2093,
+      "step": 4819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027455595550720526,
+      "loss": 2.3677,
+      "step": 4820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002743581976890511,
+      "loss": 2.3402,
+      "step": 4821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027416048418575303,
+      "loss": 2.2488,
+      "step": 4822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027396281503614116,
+      "loss": 2.3099,
+      "step": 4823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002737651902790369,
+      "loss": 2.2511,
+      "step": 4824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002735676099532525,
+      "loss": 2.4045,
+      "step": 4825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027337007409759166,
+      "loss": 2.3036,
+      "step": 4826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002731725827508494,
+      "loss": 2.2961,
+      "step": 4827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002729751359518117,
+      "loss": 2.3119,
+      "step": 4828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002727777337392566,
+      "loss": 2.4012,
+      "step": 4829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002725803761519524,
+      "loss": 2.3224,
+      "step": 4830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027238306322865935,
+      "loss": 2.3348,
+      "step": 4831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002721857950081287,
+      "loss": 2.3458,
+      "step": 4832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002719885715291027,
+      "loss": 2.3837,
+      "step": 4833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027179139283031484,
+      "loss": 2.4134,
+      "step": 4834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027159425895049045,
+      "loss": 2.2877,
+      "step": 4835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027139716992834515,
+      "loss": 2.232,
+      "step": 4836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002712001258025865,
+      "loss": 2.3583,
+      "step": 4837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027100312661191287,
+      "loss": 2.4633,
+      "step": 4838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002708061723950135,
+      "loss": 2.3244,
+      "step": 4839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027060926319056954,
+      "loss": 2.2213,
+      "step": 4840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027041239903725293,
+      "loss": 2.3928,
+      "step": 4841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027021557997372657,
+      "loss": 2.2284,
+      "step": 4842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002700188060386446,
+      "loss": 2.4131,
+      "step": 4843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002698220772706525,
+      "loss": 2.2497,
+      "step": 4844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026962539370838697,
+      "loss": 2.3744,
+      "step": 4845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002694287553904754,
+      "loss": 2.411,
+      "step": 4846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026923216235553657,
+      "loss": 2.3257,
+      "step": 4847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002690356146421803,
+      "loss": 2.4488,
+      "step": 4848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026883911228900724,
+      "loss": 2.4,
+      "step": 4849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026864265533460975,
+      "loss": 2.1863,
+      "step": 4850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000268446243817571,
+      "loss": 2.3234,
+      "step": 4851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026824987777646504,
+      "loss": 2.358,
+      "step": 4852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026805355724985713,
+      "loss": 2.2826,
+      "step": 4853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026785728227630355,
+      "loss": 2.3433,
+      "step": 4854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002676610528943516,
+      "loss": 2.2789,
+      "step": 4855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026746486914253987,
+      "loss": 2.401,
+      "step": 4856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026726873105939756,
+      "loss": 2.3211,
+      "step": 4857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002670726386834456,
+      "loss": 2.4667,
+      "step": 4858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026687659205319503,
+      "loss": 2.2207,
+      "step": 4859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026668059120714865,
+      "loss": 2.2868,
+      "step": 4860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002664846361837997,
+      "loss": 2.3538,
+      "step": 4861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002662887270216331,
+      "loss": 2.2318,
+      "step": 4862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026609286375912396,
+      "loss": 2.3216,
+      "step": 4863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026589704643473916,
+      "loss": 2.2461,
+      "step": 4864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002657012750869361,
+      "loss": 2.252,
+      "step": 4865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002655055497541629,
+      "loss": 2.1799,
+      "step": 4866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002653098704748595,
+      "loss": 2.4322,
+      "step": 4867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000265114237287456,
+      "loss": 2.5183,
+      "step": 4868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026491865023037356,
+      "loss": 2.3753,
+      "step": 4869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026472310934202477,
+      "loss": 2.3601,
+      "step": 4870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026452761466081256,
+      "loss": 2.1116,
+      "step": 4871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002643321662251313,
+      "loss": 2.1349,
+      "step": 4872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026413676407336593,
+      "loss": 2.3936,
+      "step": 4873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026394140824389237,
+      "loss": 2.2268,
+      "step": 4874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002637460987750775,
+      "loss": 2.261,
+      "step": 4875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026355083570527883,
+      "loss": 2.4808,
+      "step": 4876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026335561907284524,
+      "loss": 2.2638,
+      "step": 4877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026316044891611627,
+      "loss": 2.3532,
+      "step": 4878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002629653252734224,
+      "loss": 2.311,
+      "step": 4879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002627702481830846,
+      "loss": 2.4125,
+      "step": 4880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002625752176834152,
+      "loss": 2.2996,
+      "step": 4881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002623802338127168,
+      "loss": 2.3128,
+      "step": 4882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026218529660928353,
+      "loss": 2.3472,
+      "step": 4883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026199040611140013,
+      "loss": 2.1899,
+      "step": 4884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026179556235734184,
+      "loss": 2.3183,
+      "step": 4885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026160076538537505,
+      "loss": 2.4493,
+      "step": 4886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002614060152337568,
+      "loss": 2.4058,
+      "step": 4887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026121131194073476,
+      "loss": 2.3551,
+      "step": 4888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026101665554454804,
+      "loss": 2.4105,
+      "step": 4889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002608220460834258,
+      "loss": 2.3023,
+      "step": 4890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026062748359558863,
+      "loss": 2.4022,
+      "step": 4891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026043296811924737,
+      "loss": 2.3096,
+      "step": 4892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026023849969260374,
+      "loss": 2.2365,
+      "step": 4893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026004407835385045,
+      "loss": 2.3538,
+      "step": 4894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002598497041411709,
+      "loss": 2.4398,
+      "step": 4895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002596553770927389,
+      "loss": 2.2964,
+      "step": 4896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002594610972467196,
+      "loss": 2.4253,
+      "step": 4897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025926686464126827,
+      "loss": 2.1776,
+      "step": 4898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025907267931453106,
+      "loss": 2.2883,
+      "step": 4899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025887854130464535,
+      "loss": 2.3642,
+      "step": 4900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002586844506497386,
+      "loss": 2.2244,
+      "step": 4901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025849040738792906,
+      "loss": 2.355,
+      "step": 4902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025829641155732585,
+      "loss": 2.1202,
+      "step": 4903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002581024631960287,
+      "loss": 2.4955,
+      "step": 4904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025790856234212843,
+      "loss": 2.3656,
+      "step": 4905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025771470903370575,
+      "loss": 2.2067,
+      "step": 4906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002575209033088326,
+      "loss": 2.3655,
+      "step": 4907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002573271452055712,
+      "loss": 2.4017,
+      "step": 4908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002571334347619746,
+      "loss": 2.4072,
+      "step": 4909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002569397720160867,
+      "loss": 2.457,
+      "step": 4910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025674615700594196,
+      "loss": 2.2752,
+      "step": 4911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025655258976956506,
+      "loss": 2.3632,
+      "step": 4912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002563590703449718,
+      "loss": 2.4499,
+      "step": 4913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002561655987701682,
+      "loss": 2.3807,
+      "step": 4914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000255972175083151,
+      "loss": 2.4374,
+      "step": 4915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002557787993219078,
+      "loss": 2.2541,
+      "step": 4916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025558547152441626,
+      "loss": 2.3496,
+      "step": 4917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025539219172864534,
+      "loss": 2.3244,
+      "step": 4918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002551989599725541,
+      "loss": 2.4433,
+      "step": 4919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002550057762940918,
+      "loss": 2.3529,
+      "step": 4920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002548126407311993,
+      "loss": 2.1807,
+      "step": 4921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002546195533218071,
+      "loss": 2.434,
+      "step": 4922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002544265141038363,
+      "loss": 2.1156,
+      "step": 4923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025423352311519925,
+      "loss": 2.2734,
+      "step": 4924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002540405803937982,
+      "loss": 2.4762,
+      "step": 4925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025384768597752573,
+      "loss": 2.2328,
+      "step": 4926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025365483990426584,
+      "loss": 2.2674,
+      "step": 4927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002534620422118922,
+      "loss": 2.3065,
+      "step": 4928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025326929293826916,
+      "loss": 2.3824,
+      "step": 4929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002530765921212519,
+      "loss": 2.5382,
+      "step": 4930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002528839397986855,
+      "loss": 2.2831,
+      "step": 4931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002526913360084063,
+      "loss": 2.306,
+      "step": 4932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002524987807882404,
+      "loss": 2.2013,
+      "step": 4933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002523062741760047,
+      "loss": 2.3308,
+      "step": 4934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025211381620950644,
+      "loss": 2.4063,
+      "step": 4935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002519214069265432,
+      "loss": 2.4322,
+      "step": 4936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025172904636490325,
+      "loss": 2.3022,
+      "step": 4937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025153673456236543,
+      "loss": 2.3993,
+      "step": 4938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025134447155669855,
+      "loss": 2.1668,
+      "step": 4939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025115225738566203,
+      "loss": 2.2317,
+      "step": 4940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002509600920870058,
+      "loss": 2.3436,
+      "step": 4941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002507679756984699,
+      "loss": 2.2415,
+      "step": 4942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000250575908257785,
+      "loss": 2.2817,
+      "step": 4943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002503838898026726,
+      "loss": 2.2997,
+      "step": 4944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002501919203708437,
+      "loss": 2.1394,
+      "step": 4945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 2.3058,
+      "step": 4946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002498081287278338,
+      "loss": 2.3215,
+      "step": 4947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002496163065920278,
+      "loss": 2.2536,
+      "step": 4948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002494245336302545,
+      "loss": 2.4102,
+      "step": 4949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024923280988017714,
+      "loss": 2.1284,
+      "step": 4950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024904113537944954,
+      "loss": 2.337,
+      "step": 4951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002488495101657154,
+      "loss": 2.2246,
+      "step": 4952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002486579342766086,
+      "loss": 2.2378,
+      "step": 4953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024846640774975413,
+      "loss": 2.3557,
+      "step": 4954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024827493062276655,
+      "loss": 2.2743,
+      "step": 4955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024808350293325083,
+      "loss": 2.4752,
+      "step": 4956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002478921247188026,
+      "loss": 2.1767,
+      "step": 4957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024770079601700725,
+      "loss": 2.469,
+      "step": 4958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002475095168654411,
+      "loss": 2.5084,
+      "step": 4959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002473182873016701,
+      "loss": 2.3667,
+      "step": 4960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002471271073632508,
+      "loss": 2.3067,
+      "step": 4961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024693597708772995,
+      "loss": 2.3954,
+      "step": 4962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002467448965126443,
+      "loss": 2.2389,
+      "step": 4963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002465538656755212,
+      "loss": 2.2567,
+      "step": 4964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002463628846138783,
+      "loss": 2.2928,
+      "step": 4965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024617195336522324,
+      "loss": 2.4593,
+      "step": 4966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002459810719670537,
+      "loss": 2.3824,
+      "step": 4967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002457902404568579,
+      "loss": 2.3707,
+      "step": 4968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002455994588721138,
+      "loss": 2.3861,
+      "step": 4969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002454087272502903,
+      "loss": 2.1905,
+      "step": 4970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002452180456288461,
+      "loss": 2.2301,
+      "step": 4971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024502741404523,
+      "loss": 2.4119,
+      "step": 4972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024483683253688086,
+      "loss": 2.3572,
+      "step": 4973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002446463011412281,
+      "loss": 2.3008,
+      "step": 4974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002444558198956907,
+      "loss": 2.3767,
+      "step": 4975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024426538883767866,
+      "loss": 2.2107,
+      "step": 4976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002440750080045912,
+      "loss": 2.3508,
+      "step": 4977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024388467743381853,
+      "loss": 2.3702,
+      "step": 4978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024369439716274033,
+      "loss": 2.3465,
+      "step": 4979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024350416722872654,
+      "loss": 2.1966,
+      "step": 4980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024331398766913755,
+      "loss": 2.3678,
+      "step": 4981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024312385852132351,
+      "loss": 2.2859,
+      "step": 4982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002429337798226246,
+      "loss": 2.3159,
+      "step": 4983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024274375161037164,
+      "loss": 2.3017,
+      "step": 4984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024255377392188476,
+      "loss": 2.2339,
+      "step": 4985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024236384679447498,
+      "loss": 2.2465,
+      "step": 4986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002421739702654428,
+      "loss": 2.2053,
+      "step": 4987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024198414437207893,
+      "loss": 2.33,
+      "step": 4988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024179436915166424,
+      "loss": 2.2158,
+      "step": 4989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002416046446414693,
+      "loss": 2.2683,
+      "step": 4990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024141497087875526,
+      "loss": 2.4517,
+      "step": 4991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024122534790077315,
+      "loss": 2.2116,
+      "step": 4992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024103577574476388,
+      "loss": 2.1653,
+      "step": 4993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024084625444795827,
+      "loss": 2.3678,
+      "step": 4994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024065678404757736,
+      "loss": 2.2297,
+      "step": 4995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002404673645808319,
+      "loss": 2.307,
+      "step": 4996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024027799608492318,
+      "loss": 2.3411,
+      "step": 4997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002400886785970422,
+      "loss": 2.3086,
+      "step": 4998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023989941215436984,
+      "loss": 2.2623,
+      "step": 4999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000239710196794077,
+      "loss": 2.4068,
+      "step": 5000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002395210325533246,
+      "loss": 2.4356,
+      "step": 5001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002393319194692632,
+      "loss": 2.3133,
+      "step": 5002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002391428575790341,
+      "loss": 2.2056,
+      "step": 5003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023895384691976763,
+      "loss": 2.1729,
+      "step": 5004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002387648875285849,
+      "loss": 2.165,
+      "step": 5005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002385759794425963,
+      "loss": 2.2996,
+      "step": 5006
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023838712269890218,
+      "loss": 2.4317,
+      "step": 5007
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023819831733459346,
+      "loss": 2.3448,
+      "step": 5008
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023800956338675022,
+      "loss": 2.4547,
+      "step": 5009
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023782086089244264,
+      "loss": 2.4029,
+      "step": 5010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023763220988873125,
+      "loss": 2.293,
+      "step": 5011
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023744361041266572,
+      "loss": 2.4305,
+      "step": 5012
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002372550625012863,
+      "loss": 2.3799,
+      "step": 5013
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023706656619162274,
+      "loss": 2.3852,
+      "step": 5014
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002368781215206946,
+      "loss": 2.3068,
+      "step": 5015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023668972852551134,
+      "loss": 2.2744,
+      "step": 5016
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023650138724307264,
+      "loss": 2.1804,
+      "step": 5017
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023631309771036734,
+      "loss": 2.4375,
+      "step": 5018
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023612485996437498,
+      "loss": 2.3109,
+      "step": 5019
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002359366740420642,
+      "loss": 2.3017,
+      "step": 5020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023574853998039374,
+      "loss": 2.277,
+      "step": 5021
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023556045781631212,
+      "loss": 2.4778,
+      "step": 5022
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023537242758675747,
+      "loss": 2.3589,
+      "step": 5023
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023518444932865823,
+      "loss": 2.1876,
+      "step": 5024
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023499652307893237,
+      "loss": 2.3076,
+      "step": 5025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023480864887448754,
+      "loss": 2.3458,
+      "step": 5026
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023462082675222118,
+      "loss": 2.3372,
+      "step": 5027
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002344330567490206,
+      "loss": 2.2779,
+      "step": 5028
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023424533890176264,
+      "loss": 2.1685,
+      "step": 5029
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023405767324731426,
+      "loss": 2.2748,
+      "step": 5030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002338700598225322,
+      "loss": 2.4194,
+      "step": 5031
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023368249866426256,
+      "loss": 2.2649,
+      "step": 5032
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023349498980934137,
+      "loss": 2.4286,
+      "step": 5033
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002333075332945941,
+      "loss": 2.2273,
+      "step": 5034
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023312012915683672,
+      "loss": 2.4121,
+      "step": 5035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023293277743287417,
+      "loss": 2.2677,
+      "step": 5036
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023274547815950114,
+      "loss": 2.3538,
+      "step": 5037
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023255823137350264,
+      "loss": 2.3606,
+      "step": 5038
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002323710371116527,
+      "loss": 2.2628,
+      "step": 5039
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023218389541071504,
+      "loss": 2.2348,
+      "step": 5040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023199680630744385,
+      "loss": 2.1799,
+      "step": 5041
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023180976983858215,
+      "loss": 2.2469,
+      "step": 5042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023162278604086267,
+      "loss": 2.4135,
+      "step": 5043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023143585495100856,
+      "loss": 2.3295,
+      "step": 5044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023124897660573157,
+      "loss": 2.3806,
+      "step": 5045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023106215104173406,
+      "loss": 2.3296,
+      "step": 5046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023087537829570744,
+      "loss": 2.2556,
+      "step": 5047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023068865840433284,
+      "loss": 2.3473,
+      "step": 5048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023050199140428102,
+      "loss": 2.3368,
+      "step": 5049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002303153773322123,
+      "loss": 2.2496,
+      "step": 5050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023012881622477678,
+      "loss": 2.4278,
+      "step": 5051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022994230811861432,
+      "loss": 2.3894,
+      "step": 5052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022975585305035396,
+      "loss": 2.3158,
+      "step": 5053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022956945105661447,
+      "loss": 2.3228,
+      "step": 5054
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002293831021740042,
+      "loss": 2.2522,
+      "step": 5055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002291968064391209,
+      "loss": 2.2666,
+      "step": 5056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022901056388855223,
+      "loss": 2.4652,
+      "step": 5057
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002288243745588755,
+      "loss": 2.3491,
+      "step": 5058
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022863823848665705,
+      "loss": 2.1228,
+      "step": 5059
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022845215570845302,
+      "loss": 2.4283,
+      "step": 5060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000228266126260809,
+      "loss": 2.3293,
+      "step": 5061
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002280801501802604,
+      "loss": 2.4017,
+      "step": 5062
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022789422750333195,
+      "loss": 2.3374,
+      "step": 5063
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022770835826653747,
+      "loss": 2.2251,
+      "step": 5064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022752254250638126,
+      "loss": 2.4515,
+      "step": 5065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022733678025935622,
+      "loss": 2.4659,
+      "step": 5066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022715107156194504,
+      "loss": 2.3995,
+      "step": 5067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022696541645062014,
+      "loss": 2.2858,
+      "step": 5068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022677981496184313,
+      "loss": 2.2328,
+      "step": 5069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022659426713206494,
+      "loss": 2.3609,
+      "step": 5070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002264087729977265,
+      "loss": 2.2858,
+      "step": 5071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022622333259525757,
+      "loss": 2.2916,
+      "step": 5072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022603794596107803,
+      "loss": 2.2513,
+      "step": 5073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022585261313159662,
+      "loss": 2.3644,
+      "step": 5074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022566733414321172,
+      "loss": 2.3789,
+      "step": 5075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002254821090323112,
+      "loss": 2.2634,
+      "step": 5076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022529693783527205,
+      "loss": 2.2416,
+      "step": 5077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022511182058846115,
+      "loss": 2.2319,
+      "step": 5078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002249267573282347,
+      "loss": 2.5264,
+      "step": 5079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022474174809093795,
+      "loss": 2.3389,
+      "step": 5080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022455679291290577,
+      "loss": 2.3662,
+      "step": 5081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022437189183046235,
+      "loss": 2.3169,
+      "step": 5082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022418704487992108,
+      "loss": 2.2274,
+      "step": 5083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022400225209758519,
+      "loss": 2.4103,
+      "step": 5084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002238175135197471,
+      "loss": 2.2433,
+      "step": 5085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022363282918268833,
+      "loss": 2.3374,
+      "step": 5086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022344819912267994,
+      "loss": 2.3203,
+      "step": 5087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022326362337598204,
+      "loss": 2.239,
+      "step": 5088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022307910197884485,
+      "loss": 2.4171,
+      "step": 5089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000222894634967507,
+      "loss": 2.2992,
+      "step": 5090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022271022237819682,
+      "loss": 2.3527,
+      "step": 5091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022252586424713228,
+      "loss": 2.3808,
+      "step": 5092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002223415606105202,
+      "loss": 2.2616,
+      "step": 5093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022215731150455653,
+      "loss": 2.3137,
+      "step": 5094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022197311696542726,
+      "loss": 2.2771,
+      "step": 5095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022178897702930712,
+      "loss": 2.2739,
+      "step": 5096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022160489173235994,
+      "loss": 2.4718,
+      "step": 5097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022142086111073944,
+      "loss": 2.26,
+      "step": 5098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000221236885200588,
+      "loss": 2.3286,
+      "step": 5099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022105296403803775,
+      "loss": 2.2454,
+      "step": 5100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022086909765920982,
+      "loss": 2.306,
+      "step": 5101
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022068528610021443,
+      "loss": 2.3002,
+      "step": 5102
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022050152939715113,
+      "loss": 2.318,
+      "step": 5103
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022031782758610903,
+      "loss": 2.3243,
+      "step": 5104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022013418070316592,
+      "loss": 2.3344,
+      "step": 5105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002199505887843894,
+      "loss": 2.3489,
+      "step": 5106
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021976705186583573,
+      "loss": 2.1885,
+      "step": 5107
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021958356998355068,
+      "loss": 2.2462,
+      "step": 5108
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002194001431735691,
+      "loss": 2.3405,
+      "step": 5109
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021921677147191478,
+      "loss": 2.2129,
+      "step": 5110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002190334549146012,
+      "loss": 2.1413,
+      "step": 5111
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021885019353763097,
+      "loss": 2.2451,
+      "step": 5112
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021866698737699536,
+      "loss": 2.1622,
+      "step": 5113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021848383646867525,
+      "loss": 2.484,
+      "step": 5114
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021830074084864038,
+      "loss": 2.3757,
+      "step": 5115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021811770055284968,
+      "loss": 2.2652,
+      "step": 5116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021793471561725136,
+      "loss": 2.4171,
+      "step": 5117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021775178607778302,
+      "loss": 2.4358,
+      "step": 5118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002175689119703707,
+      "loss": 2.367,
+      "step": 5119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021738609333093006,
+      "loss": 2.5009,
+      "step": 5120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021720333019536542,
+      "loss": 2.2579,
+      "step": 5121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021702062259957094,
+      "loss": 2.2808,
+      "step": 5122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021683797057942922,
+      "loss": 2.3767,
+      "step": 5123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000216655374170812,
+      "loss": 2.4424,
+      "step": 5124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021647283340958052,
+      "loss": 2.245,
+      "step": 5125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002162903483315845,
+      "loss": 2.391,
+      "step": 5126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002161079189726634,
+      "loss": 2.3487,
+      "step": 5127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002159255453686453,
+      "loss": 2.3246,
+      "step": 5128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021574322755534724,
+      "loss": 2.4776,
+      "step": 5129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002155609655685754,
+      "loss": 2.2182,
+      "step": 5130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021537875944412544,
+      "loss": 2.3118,
+      "step": 5131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021519660921778127,
+      "loss": 2.2944,
+      "step": 5132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021501451492531666,
+      "loss": 2.3313,
+      "step": 5133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021483247660249367,
+      "loss": 2.2259,
+      "step": 5134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002146504942850637,
+      "loss": 2.3669,
+      "step": 5135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021446856800876724,
+      "loss": 2.2675,
+      "step": 5136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021428669780933335,
+      "loss": 2.3762,
+      "step": 5137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021410488372248054,
+      "loss": 2.3026,
+      "step": 5138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021392312578391638,
+      "loss": 2.3495,
+      "step": 5139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021374142402933694,
+      "loss": 2.2051,
+      "step": 5140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021355977849442753,
+      "loss": 2.294,
+      "step": 5141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002133781892148623,
+      "loss": 2.2029,
+      "step": 5142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021319665622630436,
+      "loss": 2.5058,
+      "step": 5143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021301517956440587,
+      "loss": 2.4202,
+      "step": 5144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021283375926480814,
+      "loss": 2.2856,
+      "step": 5145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000212652395363141,
+      "loss": 2.156,
+      "step": 5146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021247108789502334,
+      "loss": 2.2811,
+      "step": 5147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021228983689606279,
+      "loss": 2.2035,
+      "step": 5148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021210864240185645,
+      "loss": 2.2736,
+      "step": 5149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002119275044479898,
+      "loss": 2.1603,
+      "step": 5150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021174642307003716,
+      "loss": 2.3272,
+      "step": 5151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021156539830356237,
+      "loss": 2.2779,
+      "step": 5152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002113844301841175,
+      "loss": 2.1892,
+      "step": 5153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002112035187472436,
+      "loss": 2.2321,
+      "step": 5154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021102266402847104,
+      "loss": 2.3389,
+      "step": 5155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021084186606331858,
+      "loss": 2.3425,
+      "step": 5156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021066112488729377,
+      "loss": 2.3379,
+      "step": 5157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002104804405358936,
+      "loss": 2.2387,
+      "step": 5158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021029981304460316,
+      "loss": 2.3124,
+      "step": 5159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021011924244889713,
+      "loss": 2.2688,
+      "step": 5160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020993872878423837,
+      "loss": 2.3049,
+      "step": 5161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020975827208607884,
+      "loss": 2.2088,
+      "step": 5162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020957787238985932,
+      "loss": 2.4357,
+      "step": 5163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020939752973100906,
+      "loss": 2.3288,
+      "step": 5164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020921724414494676,
+      "loss": 2.1753,
+      "step": 5165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002090370156670796,
+      "loss": 2.2443,
+      "step": 5166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020885684433280332,
+      "loss": 2.3303,
+      "step": 5167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020867673017750272,
+      "loss": 2.2078,
+      "step": 5168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002084966732365512,
+      "loss": 2.3051,
+      "step": 5169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020831667354531076,
+      "loss": 2.3732,
+      "step": 5170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020813673113913268,
+      "loss": 2.2036,
+      "step": 5171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002079568460533568,
+      "loss": 2.2053,
+      "step": 5172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020777701832331143,
+      "loss": 2.4301,
+      "step": 5173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002075972479843138,
+      "loss": 2.2405,
+      "step": 5174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020741753507166967,
+      "loss": 2.3172,
+      "step": 5175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020723787962067402,
+      "loss": 2.3855,
+      "step": 5176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002070582816666101,
+      "loss": 2.2792,
+      "step": 5177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020687874124474987,
+      "loss": 2.4027,
+      "step": 5178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020669925839035435,
+      "loss": 2.3084,
+      "step": 5179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020651983313867297,
+      "loss": 2.3321,
+      "step": 5180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020634046552494363,
+      "loss": 2.2443,
+      "step": 5181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020616115558439357,
+      "loss": 2.3244,
+      "step": 5182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020598190335223817,
+      "loss": 2.2456,
+      "step": 5183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020580270886368134,
+      "loss": 2.2883,
+      "step": 5184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020562357215391641,
+      "loss": 2.2466,
+      "step": 5185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020544449325812447,
+      "loss": 2.1847,
+      "step": 5186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020526547221147602,
+      "loss": 2.3124,
+      "step": 5187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020508650904912967,
+      "loss": 2.3649,
+      "step": 5188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002049076038062328,
+      "loss": 2.3633,
+      "step": 5189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020472875651792138,
+      "loss": 2.4867,
+      "step": 5190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020454996721932034,
+      "loss": 2.2367,
+      "step": 5191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020437123594554263,
+      "loss": 2.3428,
+      "step": 5192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020419256273169047,
+      "loss": 2.3343,
+      "step": 5193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020401394761285414,
+      "loss": 2.2954,
+      "step": 5194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002038353906241127,
+      "loss": 2.347,
+      "step": 5195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020365689180053387,
+      "loss": 2.2547,
+      "step": 5196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020347845117717363,
+      "loss": 2.2324,
+      "step": 5197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020330006878907698,
+      "loss": 2.2978,
+      "step": 5198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020312174467127742,
+      "loss": 2.381,
+      "step": 5199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020294347885879672,
+      "loss": 2.3519,
+      "step": 5200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020276527138664535,
+      "loss": 2.2245,
+      "step": 5201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020258712228982207,
+      "loss": 2.1811,
+      "step": 5202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020240903160331487,
+      "loss": 2.3166,
+      "step": 5203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020223099936209944,
+      "loss": 2.2612,
+      "step": 5204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002020530256011407,
+      "loss": 2.3261,
+      "step": 5205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020187511035539153,
+      "loss": 2.3178,
+      "step": 5206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020169725365979364,
+      "loss": 2.2701,
+      "step": 5207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020151945554927692,
+      "loss": 2.3867,
+      "step": 5208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020134171605876034,
+      "loss": 2.3316,
+      "step": 5209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002011640352231508,
+      "loss": 2.1882,
+      "step": 5210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002009864130773437,
+      "loss": 2.2139,
+      "step": 5211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002008088496562235,
+      "loss": 2.321,
+      "step": 5212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020063134499466228,
+      "loss": 2.3581,
+      "step": 5213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020045389912752145,
+      "loss": 2.193,
+      "step": 5214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002002765120896502,
+      "loss": 2.1865,
+      "step": 5215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002000991839158864,
+      "loss": 2.3563,
+      "step": 5216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019992191464105624,
+      "loss": 2.3324,
+      "step": 5217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019974470429997483,
+      "loss": 2.2723,
+      "step": 5218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019956755292744487,
+      "loss": 2.2985,
+      "step": 5219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001993904605582585,
+      "loss": 2.2065,
+      "step": 5220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019921342722719538,
+      "loss": 2.1528,
+      "step": 5221
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000199036452969024,
+      "loss": 2.4437,
+      "step": 5222
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019885953781850115,
+      "loss": 2.4087,
+      "step": 5223
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019868268181037185,
+      "loss": 2.2267,
+      "step": 5224
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001985058849793699,
+      "loss": 2.1983,
+      "step": 5225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019832914736021734,
+      "loss": 2.2624,
+      "step": 5226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019815246898762445,
+      "loss": 2.4118,
+      "step": 5227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001979758498962898,
+      "loss": 2.3439,
+      "step": 5228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019779929012090042,
+      "loss": 2.2963,
+      "step": 5229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001976227896961319,
+      "loss": 2.3694,
+      "step": 5230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001974463486566478,
+      "loss": 2.1517,
+      "step": 5231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019726996703710043,
+      "loss": 2.3367,
+      "step": 5232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001970936448721301,
+      "loss": 2.2064,
+      "step": 5233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019691738219636552,
+      "loss": 2.2738,
+      "step": 5234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019674117904442363,
+      "loss": 2.2642,
+      "step": 5235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019656503545091003,
+      "loss": 2.2979,
+      "step": 5236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019638895145041835,
+      "loss": 2.3661,
+      "step": 5237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019621292707753025,
+      "loss": 2.208,
+      "step": 5238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019603696236681646,
+      "loss": 2.4285,
+      "step": 5239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001958610573528351,
+      "loss": 2.3388,
+      "step": 5240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019568521207013334,
+      "loss": 2.3468,
+      "step": 5241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001955094265532461,
+      "loss": 2.2149,
+      "step": 5242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001953337008366967,
+      "loss": 2.3896,
+      "step": 5243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019515803495499662,
+      "loss": 2.2212,
+      "step": 5244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000194982428942646,
+      "loss": 2.4093,
+      "step": 5245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019480688283413263,
+      "loss": 2.2445,
+      "step": 5246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019463139666393316,
+      "loss": 2.3358,
+      "step": 5247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019445597046651198,
+      "loss": 2.3304,
+      "step": 5248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019428060427632181,
+      "loss": 2.2868,
+      "step": 5249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019410529812780375,
+      "loss": 2.2292,
+      "step": 5250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019393005205538678,
+      "loss": 2.32,
+      "step": 5251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001937548660934884,
+      "loss": 2.292,
+      "step": 5252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019357974027651452,
+      "loss": 2.3381,
+      "step": 5253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019340467463885863,
+      "loss": 2.3969,
+      "step": 5254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019322966921490276,
+      "loss": 2.3514,
+      "step": 5255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001930547240390171,
+      "loss": 2.389,
+      "step": 5256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019287983914555962,
+      "loss": 2.3493,
+      "step": 5257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019270501456887708,
+      "loss": 2.3115,
+      "step": 5258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019253025034330425,
+      "loss": 2.1419,
+      "step": 5259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019235554650316372,
+      "loss": 2.318,
+      "step": 5260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001921809030827664,
+      "loss": 2.2856,
+      "step": 5261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019200632011641123,
+      "loss": 2.3817,
+      "step": 5262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001918317976383856,
+      "loss": 2.2286,
+      "step": 5263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019165733568296474,
+      "loss": 2.3146,
+      "step": 5264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019148293428441178,
+      "loss": 2.3882,
+      "step": 5265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019130859347697867,
+      "loss": 2.3079,
+      "step": 5266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019113431329490467,
+      "loss": 2.4628,
+      "step": 5267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019096009377241773,
+      "loss": 2.3874,
+      "step": 5268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019078593494373358,
+      "loss": 2.3539,
+      "step": 5269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001906118368430561,
+      "loss": 2.3243,
+      "step": 5270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019043779950457697,
+      "loss": 2.3724,
+      "step": 5271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019026382296247657,
+      "loss": 2.2881,
+      "step": 5272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019008990725092275,
+      "loss": 2.267,
+      "step": 5273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001899160524040718,
+      "loss": 2.253,
+      "step": 5274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001897422584560678,
+      "loss": 2.2422,
+      "step": 5275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018956852544104304,
+      "loss": 2.3018,
+      "step": 5276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001893948533931174,
+      "loss": 2.3298,
+      "step": 5277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001892212423463997,
+      "loss": 2.2121,
+      "step": 5278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001890476923349858,
+      "loss": 2.3661,
+      "step": 5279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018887420339296036,
+      "loss": 2.3899,
+      "step": 5280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018870077555439548,
+      "loss": 2.2879,
+      "step": 5281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001885274088533515,
+      "loss": 2.2658,
+      "step": 5282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001883541033238767,
+      "loss": 2.2511,
+      "step": 5283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018818085900000725,
+      "loss": 2.2821,
+      "step": 5284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018800767591576757,
+      "loss": 2.2488,
+      "step": 5285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018783455410517004,
+      "loss": 2.2708,
+      "step": 5286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018766149360221468,
+      "loss": 2.4007,
+      "step": 5287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018748849444088977,
+      "loss": 2.2696,
+      "step": 5288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018731555665517114,
+      "loss": 2.2639,
+      "step": 5289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018714268027902325,
+      "loss": 2.2683,
+      "step": 5290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018696986534639777,
+      "loss": 2.1158,
+      "step": 5291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000186797111891235,
+      "loss": 2.2225,
+      "step": 5292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018662441994746255,
+      "loss": 2.3567,
+      "step": 5293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018645178954899627,
+      "loss": 2.2813,
+      "step": 5294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018627922072973963,
+      "loss": 2.3727,
+      "step": 5295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018610671352358466,
+      "loss": 2.3242,
+      "step": 5296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018593426796441066,
+      "loss": 2.2915,
+      "step": 5297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018576188408608486,
+      "loss": 2.1755,
+      "step": 5298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001855895619224628,
+      "loss": 2.255,
+      "step": 5299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001854173015073874,
+      "loss": 2.2547,
+      "step": 5300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018524510287468997,
+      "loss": 2.2007,
+      "step": 5301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018507296605818936,
+      "loss": 2.4416,
+      "step": 5302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018490089109169218,
+      "loss": 2.2657,
+      "step": 5303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018472887800899295,
+      "loss": 2.3202,
+      "step": 5304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018455692684387427,
+      "loss": 2.2047,
+      "step": 5305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018438503763010662,
+      "loss": 2.2712,
+      "step": 5306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000184213210401448,
+      "loss": 2.2358,
+      "step": 5307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001840414451916443,
+      "loss": 2.1969,
+      "step": 5308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018386974203442934,
+      "loss": 2.3231,
+      "step": 5309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018369810096352475,
+      "loss": 2.3306,
+      "step": 5310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018352652201263965,
+      "loss": 2.3073,
+      "step": 5311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018335500521547148,
+      "loss": 2.3398,
+      "step": 5312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001831835506057054,
+      "loss": 2.3288,
+      "step": 5313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018301215821701406,
+      "loss": 2.2709,
+      "step": 5314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001828408280830579,
+      "loss": 2.1164,
+      "step": 5315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018266956023748522,
+      "loss": 2.4028,
+      "step": 5316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018249835471393234,
+      "loss": 2.3461,
+      "step": 5317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018232721154602284,
+      "loss": 2.3507,
+      "step": 5318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001821561307673687,
+      "loss": 2.2595,
+      "step": 5319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000181985112411569,
+      "loss": 2.2918,
+      "step": 5320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000181814156512211,
+      "loss": 2.2777,
+      "step": 5321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001816432631028692,
+      "loss": 2.4013,
+      "step": 5322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018147243221710663,
+      "loss": 2.2961,
+      "step": 5323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018130166388847336,
+      "loss": 2.2371,
+      "step": 5324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018113095815050717,
+      "loss": 2.4054,
+      "step": 5325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018096031503673417,
+      "loss": 2.3921,
+      "step": 5326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018078973458066734,
+      "loss": 2.4337,
+      "step": 5327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018061921681580823,
+      "loss": 2.2715,
+      "step": 5328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001804487617756454,
+      "loss": 2.2291,
+      "step": 5329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018027836949365534,
+      "loss": 2.2888,
+      "step": 5330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018010804000330194,
+      "loss": 2.3073,
+      "step": 5331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001799377733380373,
+      "loss": 2.3527,
+      "step": 5332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017976756953130097,
+      "loss": 2.2181,
+      "step": 5333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017959742861651994,
+      "loss": 2.2086,
+      "step": 5334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017942735062710901,
+      "loss": 2.3604,
+      "step": 5335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001792573355964706,
+      "loss": 2.3834,
+      "step": 5336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017908738355799453,
+      "loss": 2.1913,
+      "step": 5337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001789174945450589,
+      "loss": 2.2619,
+      "step": 5338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017874766859102864,
+      "loss": 2.3183,
+      "step": 5339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017857790572925697,
+      "loss": 2.1898,
+      "step": 5340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017840820599308426,
+      "loss": 2.2657,
+      "step": 5341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017823856941583872,
+      "loss": 2.3431,
+      "step": 5342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017806899603083583,
+      "loss": 2.245,
+      "step": 5343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017789948587137926,
+      "loss": 2.2257,
+      "step": 5344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017773003897075956,
+      "loss": 2.3188,
+      "step": 5345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017756065536225557,
+      "loss": 2.2239,
+      "step": 5346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017739133507913314,
+      "loss": 2.2966,
+      "step": 5347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017722207815464591,
+      "loss": 2.3745,
+      "step": 5348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017705288462203483,
+      "loss": 2.2411,
+      "step": 5349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017688375451452897,
+      "loss": 2.4199,
+      "step": 5350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017671468786534434,
+      "loss": 2.168,
+      "step": 5351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001765456847076849,
+      "loss": 2.2218,
+      "step": 5352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017637674507474193,
+      "loss": 2.231,
+      "step": 5353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001762078689996941,
+      "loss": 2.4799,
+      "step": 5354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017603905651570806,
+      "loss": 2.3346,
+      "step": 5355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001758703076559376,
+      "loss": 2.3678,
+      "step": 5356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017570162245352406,
+      "loss": 2.1156,
+      "step": 5357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017553300094159613,
+      "loss": 2.2687,
+      "step": 5358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017536444315327055,
+      "loss": 2.4389,
+      "step": 5359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017519594912165083,
+      "loss": 2.3176,
+      "step": 5360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017502751887982866,
+      "loss": 2.1128,
+      "step": 5361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017485915246088262,
+      "loss": 2.2501,
+      "step": 5362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001746908498978791,
+      "loss": 2.3358,
+      "step": 5363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017452261122387152,
+      "loss": 2.2726,
+      "step": 5364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017435443647190136,
+      "loss": 2.2145,
+      "step": 5365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017418632567499736,
+      "loss": 2.4008,
+      "step": 5366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017401827886617544,
+      "loss": 2.1729,
+      "step": 5367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001738502960784391,
+      "loss": 2.2638,
+      "step": 5368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017368237734477927,
+      "loss": 2.2417,
+      "step": 5369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017351452269817407,
+      "loss": 2.2422,
+      "step": 5370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017334673217158974,
+      "loss": 2.2141,
+      "step": 5371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017317900579797902,
+      "loss": 2.1708,
+      "step": 5372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017301134361028282,
+      "loss": 2.2913,
+      "step": 5373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017284374564142897,
+      "loss": 2.1881,
+      "step": 5374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017267621192433274,
+      "loss": 2.2829,
+      "step": 5375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017250874249189686,
+      "loss": 2.3477,
+      "step": 5376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017234133737701162,
+      "loss": 2.2302,
+      "step": 5377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017217399661255428,
+      "loss": 2.3919,
+      "step": 5378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001720067202313899,
+      "loss": 2.3422,
+      "step": 5379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017183950826637063,
+      "loss": 2.2181,
+      "step": 5380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017167236075033576,
+      "loss": 2.4699,
+      "step": 5381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017150527771611258,
+      "loss": 2.1828,
+      "step": 5382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017133825919651507,
+      "loss": 2.3714,
+      "step": 5383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001711713052243448,
+      "loss": 2.3237,
+      "step": 5384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017100441583239057,
+      "loss": 2.2778,
+      "step": 5385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017083759105342882,
+      "loss": 2.1604,
+      "step": 5386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017067083092022273,
+      "loss": 2.2361,
+      "step": 5387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017050413546552346,
+      "loss": 2.3703,
+      "step": 5388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001703375047220689,
+      "loss": 2.3596,
+      "step": 5389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017017093872258453,
+      "loss": 2.3406,
+      "step": 5390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001700044374997828,
+      "loss": 2.2863,
+      "step": 5391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001698380010863639,
+      "loss": 2.3667,
+      "step": 5392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016967162951501513,
+      "loss": 2.4022,
+      "step": 5393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016950532281841092,
+      "loss": 2.237,
+      "step": 5394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016933908102921292,
+      "loss": 2.3565,
+      "step": 5395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016917290418007027,
+      "loss": 2.3071,
+      "step": 5396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001690067923036191,
+      "loss": 2.4451,
+      "step": 5397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001688407454324828,
+      "loss": 2.2292,
+      "step": 5398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016867476359927225,
+      "loss": 2.1517,
+      "step": 5399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016850884683658563,
+      "loss": 2.229,
+      "step": 5400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016834299517700784,
+      "loss": 2.2447,
+      "step": 5401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016817720865311132,
+      "loss": 2.2916,
+      "step": 5402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016801148729745553,
+      "loss": 2.2973,
+      "step": 5403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016784583114258755,
+      "loss": 2.302,
+      "step": 5404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016768024022104105,
+      "loss": 2.3344,
+      "step": 5405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016751471456533763,
+      "loss": 2.3858,
+      "step": 5406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001673492542079853,
+      "loss": 2.3555,
+      "step": 5407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001671838591814796,
+      "loss": 2.3149,
+      "step": 5408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016701852951830353,
+      "loss": 2.166,
+      "step": 5409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016685326525092676,
+      "loss": 2.3395,
+      "step": 5410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016668806641180633,
+      "loss": 2.3623,
+      "step": 5411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016652293303338629,
+      "loss": 2.3752,
+      "step": 5412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016635786514809824,
+      "loss": 2.3017,
+      "step": 5413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016619286278836038,
+      "loss": 2.234,
+      "step": 5414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001660279259865785,
+      "loss": 2.2862,
+      "step": 5415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016586305477514528,
+      "loss": 2.3161,
+      "step": 5416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001656982491864405,
+      "loss": 2.4115,
+      "step": 5417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016553350925283095,
+      "loss": 2.3373,
+      "step": 5418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016536883500667087,
+      "loss": 2.2015,
+      "step": 5419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016520422648030154,
+      "loss": 2.2401,
+      "step": 5420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000165039683706051,
+      "loss": 2.2926,
+      "step": 5421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001648752067162347,
+      "loss": 2.2207,
+      "step": 5422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016471079554315494,
+      "loss": 2.2322,
+      "step": 5423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016454645021910102,
+      "loss": 2.4936,
+      "step": 5424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016438217077634998,
+      "loss": 2.3217,
+      "step": 5425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016421795724716497,
+      "loss": 2.3465,
+      "step": 5426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016405380966379702,
+      "loss": 2.2812,
+      "step": 5427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001638897280584837,
+      "loss": 2.3848,
+      "step": 5428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016372571246344974,
+      "loss": 2.2584,
+      "step": 5429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016356176291090679,
+      "loss": 2.3946,
+      "step": 5430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016339787943305402,
+      "loss": 2.3135,
+      "step": 5431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016323406206207686,
+      "loss": 2.4579,
+      "step": 5432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016307031083014867,
+      "loss": 2.388,
+      "step": 5433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000162906625769429,
+      "loss": 2.3885,
+      "step": 5434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016274300691206478,
+      "loss": 2.287,
+      "step": 5435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016257945429018977,
+      "loss": 2.2761,
+      "step": 5436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016241596793592516,
+      "loss": 2.1916,
+      "step": 5437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016225254788137844,
+      "loss": 2.2632,
+      "step": 5438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016208919415864475,
+      "loss": 2.37,
+      "step": 5439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016192590679980578,
+      "loss": 2.2586,
+      "step": 5440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016176268583693016,
+      "loss": 2.3959,
+      "step": 5441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016159953130207388,
+      "loss": 2.3139,
+      "step": 5442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016143644322727958,
+      "loss": 2.215,
+      "step": 5443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016127342164457676,
+      "loss": 2.1494,
+      "step": 5444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016111046658598188,
+      "loss": 2.22,
+      "step": 5445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016094757808349865,
+      "loss": 2.2912,
+      "step": 5446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016078475616911758,
+      "loss": 2.3153,
+      "step": 5447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016062200087481606,
+      "loss": 2.2316,
+      "step": 5448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016045931223255817,
+      "loss": 2.3277,
+      "step": 5449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016029669027429517,
+      "loss": 2.3321,
+      "step": 5450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000160134135031965,
+      "loss": 2.2281,
+      "step": 5451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015997164653749286,
+      "loss": 2.1693,
+      "step": 5452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001598092248227907,
+      "loss": 2.1659,
+      "step": 5453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015964686991975718,
+      "loss": 2.5109,
+      "step": 5454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015948458186027792,
+      "loss": 2.3602,
+      "step": 5455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015932236067622542,
+      "loss": 2.0019,
+      "step": 5456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015916020639945895,
+      "loss": 2.2975,
+      "step": 5457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015899811906182505,
+      "loss": 2.2925,
+      "step": 5458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015883609869515652,
+      "loss": 2.3078,
+      "step": 5459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015867414533127355,
+      "loss": 2.194,
+      "step": 5460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015851225900198284,
+      "loss": 2.3459,
+      "step": 5461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015835043973907797,
+      "loss": 2.1703,
+      "step": 5462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015818868757433929,
+      "loss": 2.3274,
+      "step": 5463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001580270025395343,
+      "loss": 2.32,
+      "step": 5464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015786538466641676,
+      "loss": 2.2159,
+      "step": 5465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000157703833986728,
+      "loss": 2.3957,
+      "step": 5466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015754235053219546,
+      "loss": 2.3884,
+      "step": 5467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015738093433453348,
+      "loss": 2.1953,
+      "step": 5468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015721958542544375,
+      "loss": 2.1068,
+      "step": 5469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015705830383661406,
+      "loss": 2.1277,
+      "step": 5470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015689708959971933,
+      "loss": 2.2683,
+      "step": 5471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015673594274642105,
+      "loss": 2.2715,
+      "step": 5472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015657486330836784,
+      "loss": 2.2973,
+      "step": 5473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015641385131719466,
+      "loss": 2.2413,
+      "step": 5474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015625290680452365,
+      "loss": 2.4386,
+      "step": 5475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015609202980196323,
+      "loss": 2.3781,
+      "step": 5476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015593122034110896,
+      "loss": 2.282,
+      "step": 5477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001557704784535427,
+      "loss": 2.2264,
+      "step": 5478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001556098041708335,
+      "loss": 2.2558,
+      "step": 5479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000155449197524537,
+      "loss": 2.4138,
+      "step": 5480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001552886585461955,
+      "loss": 2.3119,
+      "step": 5481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001551281872673379,
+      "loss": 2.3516,
+      "step": 5482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015496778371947996,
+      "loss": 2.3383,
+      "step": 5483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015480744793412393,
+      "loss": 2.3015,
+      "step": 5484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015464717994275918,
+      "loss": 2.1364,
+      "step": 5485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015448697977686116,
+      "loss": 2.1276,
+      "step": 5486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015432684746789272,
+      "loss": 2.2619,
+      "step": 5487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015416678304730285,
+      "loss": 2.2727,
+      "step": 5488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015400678654652728,
+      "loss": 2.2168,
+      "step": 5489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001538468579969884,
+      "loss": 2.1448,
+      "step": 5490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015368699743009557,
+      "loss": 2.2783,
+      "step": 5491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001535272048772443,
+      "loss": 2.1763,
+      "step": 5492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015336748036981729,
+      "loss": 2.142,
+      "step": 5493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001532078239391834,
+      "loss": 2.2355,
+      "step": 5494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015304823561669822,
+      "loss": 2.1854,
+      "step": 5495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015288871543370436,
+      "loss": 2.355,
+      "step": 5496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015272926342153054,
+      "loss": 2.401,
+      "step": 5497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015256987961149234,
+      "loss": 2.3561,
+      "step": 5498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015241056403489167,
+      "loss": 2.4263,
+      "step": 5499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001522513167230176,
+      "loss": 2.2839,
+      "step": 5500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015209213770714524,
+      "loss": 2.2937,
+      "step": 5501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015193302701853672,
+      "loss": 2.3441,
+      "step": 5502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001517739846884404,
+      "loss": 2.3144,
+      "step": 5503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015161501074809137,
+      "loss": 2.3436,
+      "step": 5504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015145610522871106,
+      "loss": 2.2709,
+      "step": 5505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015129726816150792,
+      "loss": 2.4165,
+      "step": 5506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015113849957767682,
+      "loss": 2.3225,
+      "step": 5507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015097979950839892,
+      "loss": 2.3317,
+      "step": 5508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015082116798484203,
+      "loss": 2.3309,
+      "step": 5509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001506626050381606,
+      "loss": 2.408,
+      "step": 5510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015050411069949537,
+      "loss": 2.2511,
+      "step": 5511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015034568499997414,
+      "loss": 2.2388,
+      "step": 5512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001501873279707105,
+      "loss": 2.3598,
+      "step": 5513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015002903964280535,
+      "loss": 2.404,
+      "step": 5514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014987082004734538,
+      "loss": 2.2236,
+      "step": 5515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014971266921540415,
+      "loss": 2.3313,
+      "step": 5516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001495545871780416,
+      "loss": 2.3473,
+      "step": 5517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014939657396630435,
+      "loss": 2.3816,
+      "step": 5518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014923862961122508,
+      "loss": 2.1716,
+      "step": 5519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014908075414382362,
+      "loss": 2.3593,
+      "step": 5520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014892294759510566,
+      "loss": 2.3219,
+      "step": 5521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001487652099960634,
+      "loss": 2.2973,
+      "step": 5522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014860754137767607,
+      "loss": 2.4496,
+      "step": 5523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001484499417709087,
+      "loss": 2.3283,
+      "step": 5524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001482924112067129,
+      "loss": 2.3077,
+      "step": 5525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014813494971602714,
+      "loss": 2.4132,
+      "step": 5526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014797755732977593,
+      "loss": 2.4762,
+      "step": 5527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014782023407887,
+      "loss": 2.3056,
+      "step": 5528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001476629799942073,
+      "loss": 2.1068,
+      "step": 5529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014750579510667145,
+      "loss": 2.2618,
+      "step": 5530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001473486794471327,
+      "loss": 2.2082,
+      "step": 5531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001471916330464476,
+      "loss": 2.4143,
+      "step": 5532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014703465593545945,
+      "loss": 2.2088,
+      "step": 5533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014687774814499776,
+      "loss": 2.3282,
+      "step": 5534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014672090970587837,
+      "loss": 2.3493,
+      "step": 5535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014656414064890345,
+      "loss": 2.2851,
+      "step": 5536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001464074410048616,
+      "loss": 2.3375,
+      "step": 5537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001462508108045277,
+      "loss": 2.3495,
+      "step": 5538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014609425007866317,
+      "loss": 2.3633,
+      "step": 5539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001459377588580159,
+      "loss": 2.2397,
+      "step": 5540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014578133717331983,
+      "loss": 2.2983,
+      "step": 5541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014562498505529532,
+      "loss": 2.2204,
+      "step": 5542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014546870253464905,
+      "loss": 2.2678,
+      "step": 5543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000145312489642074,
+      "loss": 2.2162,
+      "step": 5544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001451563464082498,
+      "loss": 2.2726,
+      "step": 5545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014500027286384187,
+      "loss": 2.2497,
+      "step": 5546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001448442690395026,
+      "loss": 2.2999,
+      "step": 5547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014468833496587003,
+      "loss": 2.3403,
+      "step": 5548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014453247067356878,
+      "loss": 2.2711,
+      "step": 5549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014437667619320992,
+      "loss": 2.2485,
+      "step": 5550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014422095155539066,
+      "loss": 2.3126,
+      "step": 5551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001440652967906943,
+      "loss": 2.313,
+      "step": 5552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014390971192969088,
+      "loss": 2.1951,
+      "step": 5553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001437541970029363,
+      "loss": 2.3241,
+      "step": 5554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014359875204097273,
+      "loss": 2.2125,
+      "step": 5555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014344337707432902,
+      "loss": 2.394,
+      "step": 5556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014328807213351985,
+      "loss": 2.1755,
+      "step": 5557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001431328372490463,
+      "loss": 2.2395,
+      "step": 5558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014297767245139553,
+      "loss": 2.2612,
+      "step": 5559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014282257777104124,
+      "loss": 2.3651,
+      "step": 5560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001426675532384434,
+      "loss": 2.2296,
+      "step": 5561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001425125988840477,
+      "loss": 2.2008,
+      "step": 5562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001423577147382865,
+      "loss": 2.3164,
+      "step": 5563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014220290083157821,
+      "loss": 2.4212,
+      "step": 5564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014204815719432728,
+      "loss": 2.4462,
+      "step": 5565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001418934838569247,
+      "loss": 2.1836,
+      "step": 5566
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001417388808497476,
+      "loss": 2.37,
+      "step": 5567
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014158434820315914,
+      "loss": 2.3235,
+      "step": 5568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014142988594750865,
+      "loss": 2.2884,
+      "step": 5569
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014127549411313174,
+      "loss": 2.386,
+      "step": 5570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001411211727303499,
+      "loss": 2.3381,
+      "step": 5571
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014096692182947147,
+      "loss": 2.4793,
+      "step": 5572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001408127414407901,
+      "loss": 2.1491,
+      "step": 5573
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001406586315945864,
+      "loss": 2.3663,
+      "step": 5574
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001405045923211265,
+      "loss": 2.2724,
+      "step": 5575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014035062365066293,
+      "loss": 2.1932,
+      "step": 5576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014019672561343415,
+      "loss": 2.2774,
+      "step": 5577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014004289823966526,
+      "loss": 2.264,
+      "step": 5578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013988914155956683,
+      "loss": 2.2945,
+      "step": 5579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001397354556033361,
+      "loss": 2.3545,
+      "step": 5580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001395818404011561,
+      "loss": 2.3172,
+      "step": 5581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013942829598319594,
+      "loss": 2.1932,
+      "step": 5582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001392748223796111,
+      "loss": 2.2897,
+      "step": 5583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013912141962054293,
+      "loss": 2.2008,
+      "step": 5584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013896808773611885,
+      "loss": 2.1332,
+      "step": 5585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013881482675645235,
+      "loss": 2.3462,
+      "step": 5586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013866163671164317,
+      "loss": 2.4007,
+      "step": 5587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001385085176317772,
+      "loss": 2.2526,
+      "step": 5588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013835546954692608,
+      "loss": 2.3575,
+      "step": 5589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001382024924871476,
+      "loss": 2.3322,
+      "step": 5590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013804958648248566,
+      "loss": 2.424,
+      "step": 5591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001378967515629701,
+      "loss": 2.3008,
+      "step": 5592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013774398775861702,
+      "loss": 2.4001,
+      "step": 5593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001375912950994285,
+      "loss": 2.334,
+      "step": 5594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013743867361539248,
+      "loss": 2.1928,
+      "step": 5595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013728612333648298,
+      "loss": 2.3969,
+      "step": 5596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001371336442926601,
+      "loss": 2.2777,
+      "step": 5597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001369812365138698,
+      "loss": 2.1633,
+      "step": 5598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001368289000300444,
+      "loss": 2.4156,
+      "step": 5599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013667663487110173,
+      "loss": 2.2232,
+      "step": 5600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013652444106694605,
+      "loss": 2.244,
+      "step": 5601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013637231864746735,
+      "loss": 2.2831,
+      "step": 5602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001362202676425417,
+      "loss": 2.2477,
+      "step": 5603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013606828808203093,
+      "loss": 2.4274,
+      "step": 5604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013591637999578332,
+      "loss": 2.1662,
+      "step": 5605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013576454341363253,
+      "loss": 2.2528,
+      "step": 5606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001356127783653987,
+      "loss": 2.2703,
+      "step": 5607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013546108488088766,
+      "loss": 2.2581,
+      "step": 5608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013530946298989088,
+      "loss": 2.2764,
+      "step": 5609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013515791272218652,
+      "loss": 2.1916,
+      "step": 5610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001350064341075381,
+      "loss": 2.3426,
+      "step": 5611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013485502717569504,
+      "loss": 2.1616,
+      "step": 5612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001347036919563932,
+      "loss": 2.3332,
+      "step": 5613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013455242847935377,
+      "loss": 2.2593,
+      "step": 5614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013440123677428402,
+      "loss": 2.3035,
+      "step": 5615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013425011687087752,
+      "loss": 2.2966,
+      "step": 5616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013409906879881324,
+      "loss": 2.2145,
+      "step": 5617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013394809258775615,
+      "loss": 2.234,
+      "step": 5618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013379718826735714,
+      "loss": 2.3281,
+      "step": 5619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013364635586725316,
+      "loss": 2.269,
+      "step": 5620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013349559541706696,
+      "loss": 2.338,
+      "step": 5621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013334490694640706,
+      "loss": 2.2085,
+      "step": 5622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013319429048486782,
+      "loss": 2.191,
+      "step": 5623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013304374606202946,
+      "loss": 2.4,
+      "step": 5624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013289327370745801,
+      "loss": 2.4281,
+      "step": 5625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013274287345070563,
+      "loss": 2.279,
+      "step": 5626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013259254532131028,
+      "loss": 2.2759,
+      "step": 5627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001324422893487954,
+      "loss": 2.3145,
+      "step": 5628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013229210556267045,
+      "loss": 2.328,
+      "step": 5629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013214199399243088,
+      "loss": 2.1769,
+      "step": 5630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013199195466755747,
+      "loss": 2.2342,
+      "step": 5631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013184198761751758,
+      "loss": 2.2186,
+      "step": 5632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001316920928717636,
+      "loss": 2.3398,
+      "step": 5633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013154227045973437,
+      "loss": 2.3376,
+      "step": 5634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001313925204108541,
+      "loss": 2.302,
+      "step": 5635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001312428427545327,
+      "loss": 2.2231,
+      "step": 5636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001310932375201664,
+      "loss": 2.297,
+      "step": 5637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013094370473713669,
+      "loss": 2.3902,
+      "step": 5638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013079424443481093,
+      "loss": 2.2773,
+      "step": 5639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013064485664254262,
+      "loss": 2.2535,
+      "step": 5640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001304955413896705,
+      "loss": 2.1639,
+      "step": 5641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001303462987055193,
+      "loss": 2.3909,
+      "step": 5642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013019712861939964,
+      "loss": 2.2779,
+      "step": 5643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013004803116060766,
+      "loss": 2.3606,
+      "step": 5644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012989900635842534,
+      "loss": 2.3119,
+      "step": 5645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001297500542421201,
+      "loss": 2.2984,
+      "step": 5646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012960117484094564,
+      "loss": 2.1608,
+      "step": 5647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012945236818414114,
+      "loss": 2.1839,
+      "step": 5648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012930363430093135,
+      "loss": 2.1808,
+      "step": 5649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012915497322052677,
+      "loss": 2.2604,
+      "step": 5650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012900638497212363,
+      "loss": 2.2004,
+      "step": 5651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001288578695849038,
+      "loss": 2.3054,
+      "step": 5652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000128709427088035,
+      "loss": 2.3598,
+      "step": 5653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012856105751067083,
+      "loss": 2.1546,
+      "step": 5654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012841276088194992,
+      "loss": 2.2174,
+      "step": 5655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012826453723099713,
+      "loss": 2.1115,
+      "step": 5656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012811638658692272,
+      "loss": 2.2265,
+      "step": 5657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012796830897882257,
+      "loss": 2.4576,
+      "step": 5658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012782030443577857,
+      "loss": 2.2453,
+      "step": 5659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012767237298685785,
+      "loss": 2.3613,
+      "step": 5660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001275245146611136,
+      "loss": 2.1884,
+      "step": 5661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012737672948758423,
+      "loss": 2.2712,
+      "step": 5662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012722901749529387,
+      "loss": 2.2396,
+      "step": 5663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012708137871325265,
+      "loss": 2.5159,
+      "step": 5664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012693381317045588,
+      "loss": 2.1699,
+      "step": 5665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012678632089588453,
+      "loss": 2.2751,
+      "step": 5666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012663890191850564,
+      "loss": 2.0731,
+      "step": 5667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012649155626727127,
+      "loss": 2.2217,
+      "step": 5668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012634428397111924,
+      "loss": 2.1641,
+      "step": 5669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012619708505897336,
+      "loss": 2.4128,
+      "step": 5670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012604995955974252,
+      "loss": 2.2722,
+      "step": 5671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012590290750232142,
+      "loss": 2.299,
+      "step": 5672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001257559289155901,
+      "loss": 2.418,
+      "step": 5673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012560902382841466,
+      "loss": 2.3363,
+      "step": 5674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001254621922696465,
+      "loss": 2.3256,
+      "step": 5675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012531543426812247,
+      "loss": 2.1414,
+      "step": 5676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001251687498526651,
+      "loss": 2.3287,
+      "step": 5677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012502213905208237,
+      "loss": 2.3085,
+      "step": 5678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012487560189516773,
+      "loss": 2.1357,
+      "step": 5679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012472913841070043,
+      "loss": 2.306,
+      "step": 5680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012458274862744528,
+      "loss": 2.2089,
+      "step": 5681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012443643257415237,
+      "loss": 2.3687,
+      "step": 5682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001242901902795574,
+      "loss": 2.3352,
+      "step": 5683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001241440217723815,
+      "loss": 2.2841,
+      "step": 5684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012399792708133128,
+      "loss": 2.3879,
+      "step": 5685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001238519062350993,
+      "loss": 2.1798,
+      "step": 5686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001237059592623629,
+      "loss": 2.3593,
+      "step": 5687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012356008619178566,
+      "loss": 2.2845,
+      "step": 5688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012341428705201613,
+      "loss": 2.2375,
+      "step": 5689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012326856187168834,
+      "loss": 2.259,
+      "step": 5690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012312291067942215,
+      "loss": 2.2541,
+      "step": 5691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012297733350382267,
+      "loss": 2.2397,
+      "step": 5692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001228318303734803,
+      "loss": 2.3341,
+      "step": 5693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001226864013169713,
+      "loss": 2.2375,
+      "step": 5694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012254104636285712,
+      "loss": 2.1802,
+      "step": 5695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001223957655396845,
+      "loss": 2.2408,
+      "step": 5696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012225055887598612,
+      "loss": 2.4317,
+      "step": 5697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001221054264002797,
+      "loss": 2.3881,
+      "step": 5698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012196036814106831,
+      "loss": 2.3634,
+      "step": 5699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012181538412684096,
+      "loss": 2.2181,
+      "step": 5700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012167047438607138,
+      "loss": 2.1865,
+      "step": 5701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012152563894721935,
+      "loss": 2.4219,
+      "step": 5702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012138087783872975,
+      "loss": 2.2338,
+      "step": 5703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012123619108903284,
+      "loss": 2.3319,
+      "step": 5704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001210915787265443,
+      "loss": 2.2505,
+      "step": 5705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012094704077966512,
+      "loss": 2.4037,
+      "step": 5706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012080257727678196,
+      "loss": 2.2958,
+      "step": 5707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012065818824626673,
+      "loss": 2.3424,
+      "step": 5708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012051387371647665,
+      "loss": 2.1604,
+      "step": 5709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012036963371575427,
+      "loss": 2.1911,
+      "step": 5710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012022546827242758,
+      "loss": 2.1197,
+      "step": 5711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012008137741480974,
+      "loss": 2.1617,
+      "step": 5712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011993736117119964,
+      "loss": 2.0654,
+      "step": 5713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011979341956988133,
+      "loss": 2.1548,
+      "step": 5714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011964955263912414,
+      "loss": 2.0934,
+      "step": 5715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011950576040718275,
+      "loss": 2.0747,
+      "step": 5716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011936204290229725,
+      "loss": 2.0658,
+      "step": 5717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001192184001526927,
+      "loss": 2.0619,
+      "step": 5718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011907483218658022,
+      "loss": 1.9608,
+      "step": 5719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011893133903215542,
+      "loss": 2.1561,
+      "step": 5720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011878792071759991,
+      "loss": 2.1997,
+      "step": 5721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011864457727108014,
+      "loss": 1.9484,
+      "step": 5722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011850130872074788,
+      "loss": 2.0036,
+      "step": 5723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011835811509474059,
+      "loss": 2.0626,
+      "step": 5724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011821499642118061,
+      "loss": 2.102,
+      "step": 5725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011807195272817556,
+      "loss": 2.0993,
+      "step": 5726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011792898404381869,
+      "loss": 2.1706,
+      "step": 5727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011778609039618804,
+      "loss": 1.9485,
+      "step": 5728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011764327181334744,
+      "loss": 1.9655,
+      "step": 5729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011750052832334557,
+      "loss": 1.9247,
+      "step": 5730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011735785995421656,
+      "loss": 2.0157,
+      "step": 5731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011721526673397964,
+      "loss": 2.0295,
+      "step": 5732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011707274869063922,
+      "loss": 1.9822,
+      "step": 5733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011693030585218529,
+      "loss": 2.1255,
+      "step": 5734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011678793824659295,
+      "loss": 1.9046,
+      "step": 5735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011664564590182236,
+      "loss": 1.9793,
+      "step": 5736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000116503428845819,
+      "loss": 1.9749,
+      "step": 5737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011636128710651344,
+      "loss": 2.0015,
+      "step": 5738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001162192207118215,
+      "loss": 2.0746,
+      "step": 5739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001160772296896444,
+      "loss": 2.0896,
+      "step": 5740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011593531406786862,
+      "loss": 2.0359,
+      "step": 5741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001157934738743654,
+      "loss": 1.9605,
+      "step": 5742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011565170913699142,
+      "loss": 2.0175,
+      "step": 5743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011551001988358861,
+      "loss": 1.9248,
+      "step": 5744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011536840614198374,
+      "loss": 1.9898,
+      "step": 5745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011522686793998938,
+      "loss": 2.2308,
+      "step": 5746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011508540530540251,
+      "loss": 1.9775,
+      "step": 5747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011494401826600598,
+      "loss": 2.0779,
+      "step": 5748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011480270684956728,
+      "loss": 1.9684,
+      "step": 5749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011466147108383918,
+      "loss": 2.0127,
+      "step": 5750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001145203109965598,
+      "loss": 2.1591,
+      "step": 5751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011437922661545214,
+      "loss": 2.092,
+      "step": 5752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011423821796822425,
+      "loss": 1.9521,
+      "step": 5753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011409728508256983,
+      "loss": 1.9711,
+      "step": 5754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011395642798616718,
+      "loss": 2.0197,
+      "step": 5755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011381564670667976,
+      "loss": 1.9526,
+      "step": 5756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011367494127175654,
+      "loss": 2.0642,
+      "step": 5757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011353431170903123,
+      "loss": 1.9445,
+      "step": 5758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011339375804612262,
+      "loss": 1.9668,
+      "step": 5759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011325328031063464,
+      "loss": 2.0813,
+      "step": 5760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011311287853015651,
+      "loss": 1.9585,
+      "step": 5761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011297255273226254,
+      "loss": 1.9747,
+      "step": 5762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001128323029445118,
+      "loss": 2.0005,
+      "step": 5763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011269212919444866,
+      "loss": 2.037,
+      "step": 5764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011255203150960247,
+      "loss": 1.9116,
+      "step": 5765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001124120099174875,
+      "loss": 1.9324,
+      "step": 5766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001122720644456035,
+      "loss": 2.032,
+      "step": 5767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011213219512143502,
+      "loss": 2.0169,
+      "step": 5768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011199240197245164,
+      "loss": 2.039,
+      "step": 5769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001118526850261079,
+      "loss": 2.0787,
+      "step": 5770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011171304430984352,
+      "loss": 2.03,
+      "step": 5771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011157347985108307,
+      "loss": 2.1585,
+      "step": 5772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001114339916772364,
+      "loss": 2.0334,
+      "step": 5773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011129457981569835,
+      "loss": 2.0566,
+      "step": 5774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011115524429384854,
+      "loss": 1.9492,
+      "step": 5775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011101598513905182,
+      "loss": 2.0806,
+      "step": 5776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011087680237865766,
+      "loss": 1.9773,
+      "step": 5777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011073769604000122,
+      "loss": 1.948,
+      "step": 5778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011059866615040204,
+      "loss": 2.0847,
+      "step": 5779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011045971273716476,
+      "loss": 2.1178,
+      "step": 5780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011032083582757935,
+      "loss": 2.0707,
+      "step": 5781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011018203544892042,
+      "loss": 2.049,
+      "step": 5782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011004331162844739,
+      "loss": 2.0184,
+      "step": 5783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010990466439340518,
+      "loss": 2.1043,
+      "step": 5784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010976609377102331,
+      "loss": 1.9657,
+      "step": 5785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010962759978851616,
+      "loss": 2.03,
+      "step": 5786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010948918247308343,
+      "loss": 1.8732,
+      "step": 5787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001093508418519093,
+      "loss": 2.0026,
+      "step": 5788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010921257795216343,
+      "loss": 2.0856,
+      "step": 5789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000109074390801,
+      "loss": 2.0249,
+      "step": 5790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001089362804255582,
+      "loss": 1.9073,
+      "step": 5791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001087982468529622,
+      "loss": 2.0494,
+      "step": 5792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010866029011032086,
+      "loss": 1.9385,
+      "step": 5793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010852241022472836,
+      "loss": 1.9435,
+      "step": 5794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010838460722326371,
+      "loss": 1.9009,
+      "step": 5795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010824688113299053,
+      "loss": 2.0492,
+      "step": 5796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010810923198095757,
+      "loss": 1.937,
+      "step": 5797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010797165979419827,
+      "loss": 2.0429,
+      "step": 5798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000107834164599731,
+      "loss": 2.0845,
+      "step": 5799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001076967464245593,
+      "loss": 1.9321,
+      "step": 5800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010755940529567138,
+      "loss": 1.9732,
+      "step": 5801
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010742214124004024,
+      "loss": 1.9601,
+      "step": 5802
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010728495428462381,
+      "loss": 2.0176,
+      "step": 5803
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010714784445636478,
+      "loss": 2.0571,
+      "step": 5804
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010701081178219102,
+      "loss": 1.9507,
+      "step": 5805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010687385628901492,
+      "loss": 2.055,
+      "step": 5806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010673697800373361,
+      "loss": 1.9792,
+      "step": 5807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001066001769532296,
+      "loss": 2.018,
+      "step": 5808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010646345316436979,
+      "loss": 2.096,
+      "step": 5809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010632680666400573,
+      "loss": 1.9958,
+      "step": 5810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001061902374789745,
+      "loss": 2.0751,
+      "step": 5811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010605374563609732,
+      "loss": 1.9961,
+      "step": 5812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010591733116218044,
+      "loss": 1.9856,
+      "step": 5813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010578099408401509,
+      "loss": 2.0158,
+      "step": 5814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010564473442837697,
+      "loss": 2.0055,
+      "step": 5815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010550855222202693,
+      "loss": 1.9365,
+      "step": 5816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010537244749171043,
+      "loss": 2.0282,
+      "step": 5817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010523642026415764,
+      "loss": 1.9073,
+      "step": 5818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010510047056608357,
+      "loss": 2.0535,
+      "step": 5819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010496459842418792,
+      "loss": 2.0064,
+      "step": 5820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010482880386515542,
+      "loss": 2.1013,
+      "step": 5821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010469308691565543,
+      "loss": 2.03,
+      "step": 5822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001045574476023421,
+      "loss": 1.9309,
+      "step": 5823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010442188595185403,
+      "loss": 2.0577,
+      "step": 5824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010428640199081502,
+      "loss": 1.9763,
+      "step": 5825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001041509957458332,
+      "loss": 1.9931,
+      "step": 5826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001040156672435017,
+      "loss": 1.8546,
+      "step": 5827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010388041651039859,
+      "loss": 1.9719,
+      "step": 5828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010374524357308613,
+      "loss": 1.9513,
+      "step": 5829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010361014845811168,
+      "loss": 1.9236,
+      "step": 5830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010347513119200696,
+      "loss": 1.9274,
+      "step": 5831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010334019180128901,
+      "loss": 1.9104,
+      "step": 5832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010320533031245905,
+      "loss": 1.9816,
+      "step": 5833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010307054675200306,
+      "loss": 2.1299,
+      "step": 5834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010293584114639199,
+      "loss": 1.8902,
+      "step": 5835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010280121352208128,
+      "loss": 1.953,
+      "step": 5836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010266666390551089,
+      "loss": 1.8241,
+      "step": 5837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010253219232310595,
+      "loss": 2.0241,
+      "step": 5838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001023977988012758,
+      "loss": 2.0188,
+      "step": 5839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010226348336641456,
+      "loss": 1.9318,
+      "step": 5840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010212924604490121,
+      "loss": 1.9612,
+      "step": 5841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010199508686309911,
+      "loss": 2.0489,
+      "step": 5842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010186100584735658,
+      "loss": 1.9507,
+      "step": 5843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010172700302400634,
+      "loss": 1.9011,
+      "step": 5844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010159307841936582,
+      "loss": 1.8663,
+      "step": 5845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000101459232059737,
+      "loss": 1.9436,
+      "step": 5846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010132546397140685,
+      "loss": 1.902,
+      "step": 5847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010119177418064646,
+      "loss": 1.8538,
+      "step": 5848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010105816271371199,
+      "loss": 2.0734,
+      "step": 5849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000100924629596844,
+      "loss": 2.1016,
+      "step": 5850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001007911748562676,
+      "loss": 2.0774,
+      "step": 5851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010065779851819267,
+      "loss": 2.0206,
+      "step": 5852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010052450060881341,
+      "loss": 1.8696,
+      "step": 5853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010039128115430902,
+      "loss": 2.0171,
+      "step": 5854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010025814018084317,
+      "loss": 1.7664,
+      "step": 5855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010012507771456397,
+      "loss": 1.8851,
+      "step": 5856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.999209378160412e-05,
+      "loss": 1.904,
+      "step": 5857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.9859188408081e-05,
+      "loss": 2.0571,
+      "step": 5858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.97263616200964e-05,
+      "loss": 2.0306,
+      "step": 5859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.959361344373696e-05,
+      "loss": 1.9281,
+      "step": 5860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.946094390507376e-05,
+      "loss": 2.0659,
+      "step": 5861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.93283530301623e-05,
+      "loss": 1.983,
+      "step": 5862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.919584084504279e-05,
+      "loss": 1.8753,
+      "step": 5863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.90634073757397e-05,
+      "loss": 2.0038,
+      "step": 5864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.893105264826257e-05,
+      "loss": 1.946,
+      "step": 5865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.879877668860504e-05,
+      "loss": 1.9277,
+      "step": 5866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.866657952274527e-05,
+      "loss": 2.0334,
+      "step": 5867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.853446117664638e-05,
+      "loss": 2.0015,
+      "step": 5868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840242167625535e-05,
+      "loss": 2.0454,
+      "step": 5869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.827046104750442e-05,
+      "loss": 2.0107,
+      "step": 5870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.813857931630977e-05,
+      "loss": 1.9218,
+      "step": 5871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.800677650857226e-05,
+      "loss": 1.9768,
+      "step": 5872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.787505265017721e-05,
+      "loss": 1.9319,
+      "step": 5873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.774340776699464e-05,
+      "loss": 1.9842,
+      "step": 5874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.76118418848787e-05,
+      "loss": 1.9816,
+      "step": 5875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.748035502966857e-05,
+      "loss": 2.0435,
+      "step": 5876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.734894722718729e-05,
+      "loss": 1.8872,
+      "step": 5877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.721761850324279e-05,
+      "loss": 1.9873,
+      "step": 5878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.708636888362727e-05,
+      "loss": 1.9757,
+      "step": 5879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.695519839411732e-05,
+      "loss": 2.0379,
+      "step": 5880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.682410706047429e-05,
+      "loss": 1.9495,
+      "step": 5881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.669309490844397e-05,
+      "loss": 1.9231,
+      "step": 5882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.656216196375628e-05,
+      "loss": 1.97,
+      "step": 5883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.643130825212581e-05,
+      "loss": 1.927,
+      "step": 5884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.630053379925153e-05,
+      "loss": 2.1313,
+      "step": 5885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.616983863081663e-05,
+      "loss": 1.9542,
+      "step": 5886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.603922277248922e-05,
+      "loss": 1.9093,
+      "step": 5887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.590868624992154e-05,
+      "loss": 1.9021,
+      "step": 5888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.577822908875028e-05,
+      "loss": 1.988,
+      "step": 5889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.564785131459647e-05,
+      "loss": 2.0631,
+      "step": 5890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.551755295306552e-05,
+      "loss": 1.9839,
+      "step": 5891
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.538733402974753e-05,
+      "loss": 1.9243,
+      "step": 5892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.525719457021675e-05,
+      "loss": 1.9832,
+      "step": 5893
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.512713460003175e-05,
+      "loss": 1.9512,
+      "step": 5894
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.49971541447358e-05,
+      "loss": 2.0177,
+      "step": 5895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.486725322985634e-05,
+      "loss": 2.0876,
+      "step": 5896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.473743188090506e-05,
+      "loss": 1.9684,
+      "step": 5897
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.460769012337839e-05,
+      "loss": 2.0586,
+      "step": 5898
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.447802798275684e-05,
+      "loss": 1.9342,
+      "step": 5899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.434844548450527e-05,
+      "loss": 1.8763,
+      "step": 5900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.421894265407322e-05,
+      "loss": 2.0743,
+      "step": 5901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.408951951689409e-05,
+      "loss": 1.8863,
+      "step": 5902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.39601760983862e-05,
+      "loss": 1.9319,
+      "step": 5903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.383091242395176e-05,
+      "loss": 2.1125,
+      "step": 5904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.370172851897746e-05,
+      "loss": 1.9764,
+      "step": 5905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.357262440883435e-05,
+      "loss": 1.9634,
+      "step": 5906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.344360011887771e-05,
+      "loss": 1.9804,
+      "step": 5907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.331465567444724e-05,
+      "loss": 1.905,
+      "step": 5908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.318579110086717e-05,
+      "loss": 1.8979,
+      "step": 5909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.305700642344567e-05,
+      "loss": 1.8738,
+      "step": 5910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.29283016674754e-05,
+      "loss": 1.9032,
+      "step": 5911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.27996768582332e-05,
+      "loss": 2.006,
+      "step": 5912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.267113202098027e-05,
+      "loss": 1.9181,
+      "step": 5913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.254266718096222e-05,
+      "loss": 2.0612,
+      "step": 5914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.241428236340903e-05,
+      "loss": 1.9276,
+      "step": 5915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.228597759353458e-05,
+      "loss": 1.9504,
+      "step": 5916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.215775289653738e-05,
+      "loss": 1.8998,
+      "step": 5917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.202960829759981e-05,
+      "loss": 1.9592,
+      "step": 5918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.19015438218892e-05,
+      "loss": 2.0897,
+      "step": 5919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.177355949455651e-05,
+      "loss": 2.08,
+      "step": 5920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.164565534073705e-05,
+      "loss": 1.9125,
+      "step": 5921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.151783138555082e-05,
+      "loss": 1.9874,
+      "step": 5922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.139008765410162e-05,
+      "loss": 1.9703,
+      "step": 5923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.12624241714775e-05,
+      "loss": 1.8429,
+      "step": 5924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.113484096275115e-05,
+      "loss": 2.0233,
+      "step": 5925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.100733805297917e-05,
+      "loss": 1.9074,
+      "step": 5926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.087991546720226e-05,
+      "loss": 1.8915,
+      "step": 5927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.075257323044583e-05,
+      "loss": 1.9219,
+      "step": 5928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.062531136771896e-05,
+      "loss": 1.9873,
+      "step": 5929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.049812990401546e-05,
+      "loss": 1.9549,
+      "step": 5930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.037102886431298e-05,
+      "loss": 1.9911,
+      "step": 5931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.024400827357343e-05,
+      "loss": 2.0086,
+      "step": 5932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.011706815674298e-05,
+      "loss": 1.9718,
+      "step": 5933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.999020853875211e-05,
+      "loss": 1.8621,
+      "step": 5934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.98634294445152e-05,
+      "loss": 1.9764,
+      "step": 5935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.973673089893131e-05,
+      "loss": 1.9768,
+      "step": 5936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.961011292688309e-05,
+      "loss": 1.984,
+      "step": 5937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.948357555323772e-05,
+      "loss": 1.9239,
+      "step": 5938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.935711880284653e-05,
+      "loss": 2.0647,
+      "step": 5939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.923074270054471e-05,
+      "loss": 2.0093,
+      "step": 5940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.910444727115202e-05,
+      "loss": 1.9812,
+      "step": 5941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.89782325394724e-05,
+      "loss": 1.9551,
+      "step": 5942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.885209853029358e-05,
+      "loss": 1.8858,
+      "step": 5943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.872604526838768e-05,
+      "loss": 1.9781,
+      "step": 5944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.860007277851062e-05,
+      "loss": 2.0452,
+      "step": 5945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.847418108540312e-05,
+      "loss": 1.9731,
+      "step": 5946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.834837021378933e-05,
+      "loss": 1.9347,
+      "step": 5947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.822264018837811e-05,
+      "loss": 2.1036,
+      "step": 5948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.809699103386204e-05,
+      "loss": 1.8891,
+      "step": 5949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.797142277491798e-05,
+      "loss": 2.0435,
+      "step": 5950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.784593543620672e-05,
+      "loss": 2.0309,
+      "step": 5951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.772052904237349e-05,
+      "loss": 1.9147,
+      "step": 5952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.759520361804746e-05,
+      "loss": 1.9219,
+      "step": 5953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.746995918784162e-05,
+      "loss": 1.9323,
+      "step": 5954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.734479577635368e-05,
+      "loss": 1.9528,
+      "step": 5955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.72197134081647e-05,
+      "loss": 1.9608,
+      "step": 5956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.70947121078406e-05,
+      "loss": 1.9936,
+      "step": 5957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.696979189993065e-05,
+      "loss": 1.9904,
+      "step": 5958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.68449528089687e-05,
+      "loss": 1.8318,
+      "step": 5959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.67201948594723e-05,
+      "loss": 2.07,
+      "step": 5960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.659551807594345e-05,
+      "loss": 1.9346,
+      "step": 5961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.647092248286781e-05,
+      "loss": 1.9705,
+      "step": 5962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.634640810471555e-05,
+      "loss": 1.9716,
+      "step": 5963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.622197496594047e-05,
+      "loss": 1.9569,
+      "step": 5964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.609762309098063e-05,
+      "loss": 2.0069,
+      "step": 5965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.597335250425808e-05,
+      "loss": 2.0503,
+      "step": 5966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.584916323017872e-05,
+      "loss": 2.0169,
+      "step": 5967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.572505529313285e-05,
+      "loss": 1.9614,
+      "step": 5968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.560102871749475e-05,
+      "loss": 1.8636,
+      "step": 5969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.547708352762234e-05,
+      "loss": 2.105,
+      "step": 5970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.535321974785798e-05,
+      "loss": 1.9067,
+      "step": 5971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.52294374025277e-05,
+      "loss": 1.8487,
+      "step": 5972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.510573651594167e-05,
+      "loss": 1.9044,
+      "step": 5973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.498211711239417e-05,
+      "loss": 1.946,
+      "step": 5974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.485857921616358e-05,
+      "loss": 1.9459,
+      "step": 5975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.473512285151186e-05,
+      "loss": 1.8322,
+      "step": 5976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.461174804268528e-05,
+      "loss": 1.9476,
+      "step": 5977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.448845481391382e-05,
+      "loss": 2.036,
+      "step": 5978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.436524318941191e-05,
+      "loss": 1.9998,
+      "step": 5979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.424211319337755e-05,
+      "loss": 1.8941,
+      "step": 5980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.41190648499926e-05,
+      "loss": 1.9232,
+      "step": 5981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.399609818342341e-05,
+      "loss": 1.925,
+      "step": 5982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.387321321781977e-05,
+      "loss": 1.9635,
+      "step": 5983
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.375040997731587e-05,
+      "loss": 2.0361,
+      "step": 5984
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.362768848602948e-05,
+      "loss": 2.0136,
+      "step": 5985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.350504876806242e-05,
+      "loss": 1.9469,
+      "step": 5986
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.338249084750038e-05,
+      "loss": 1.904,
+      "step": 5987
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.326001474841339e-05,
+      "loss": 1.9735,
+      "step": 5988
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.313762049485474e-05,
+      "loss": 1.7571,
+      "step": 5989
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.301530811086239e-05,
+      "loss": 1.9855,
+      "step": 5990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.289307762045761e-05,
+      "loss": 1.8628,
+      "step": 5991
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.277092904764594e-05,
+      "loss": 2.1061,
+      "step": 5992
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.264886241641661e-05,
+      "loss": 1.9131,
+      "step": 5993
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.25268777507428e-05,
+      "loss": 1.9796,
+      "step": 5994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.240497507458172e-05,
+      "loss": 1.9855,
+      "step": 5995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.228315441187461e-05,
+      "loss": 2.1389,
+      "step": 5996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.216141578654624e-05,
+      "loss": 2.0396,
+      "step": 5997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.20397592225054e-05,
+      "loss": 1.927,
+      "step": 5998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.191818474364482e-05,
+      "loss": 1.8822,
+      "step": 5999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.179669237384096e-05,
+      "loss": 1.8483,
+      "step": 6000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.167528213695447e-05,
+      "loss": 2.0409,
+      "step": 6001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.15539540568297e-05,
+      "loss": 2.1373,
+      "step": 6002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.143270815729476e-05,
+      "loss": 1.8988,
+      "step": 6003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.131154446216166e-05,
+      "loss": 1.8658,
+      "step": 6004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.11904629952263e-05,
+      "loss": 1.9992,
+      "step": 6005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.106946378026858e-05,
+      "loss": 2.0878,
+      "step": 6006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.094854684105201e-05,
+      "loss": 2.0129,
+      "step": 6007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.082771220132396e-05,
+      "loss": 1.9428,
+      "step": 6008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.070695988481592e-05,
+      "loss": 2.0072,
+      "step": 6009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.058628991524275e-05,
+      "loss": 2.0335,
+      "step": 6010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.046570231630362e-05,
+      "loss": 1.9141,
+      "step": 6011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.03451971116812e-05,
+      "loss": 2.0369,
+      "step": 6012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.022477432504216e-05,
+      "loss": 1.9053,
+      "step": 6013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.010443398003664e-05,
+      "loss": 1.7681,
+      "step": 6014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.998417610029912e-05,
+      "loss": 1.88,
+      "step": 6015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.986400070944744e-05,
+      "loss": 1.9174,
+      "step": 6016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.97439078310836e-05,
+      "loss": 2.0602,
+      "step": 6017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.962389748879312e-05,
+      "loss": 1.7871,
+      "step": 6018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.950396970614531e-05,
+      "loss": 2.0268,
+      "step": 6019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.93841245066933e-05,
+      "loss": 1.9395,
+      "step": 6020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.926436191397435e-05,
+      "loss": 1.9861,
+      "step": 6021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.914468195150881e-05,
+      "loss": 2.0141,
+      "step": 6022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.902508464280162e-05,
+      "loss": 1.9397,
+      "step": 6023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.890557001134074e-05,
+      "loss": 2.006,
+      "step": 6024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.878613808059837e-05,
+      "loss": 1.9656,
+      "step": 6025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.866678887403022e-05,
+      "loss": 1.9984,
+      "step": 6026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.854752241507574e-05,
+      "loss": 1.955,
+      "step": 6027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.842833872715837e-05,
+      "loss": 1.9622,
+      "step": 6028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.830923783368527e-05,
+      "loss": 1.9665,
+      "step": 6029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.819021975804708e-05,
+      "loss": 1.9002,
+      "step": 6030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.807128452361833e-05,
+      "loss": 2.0081,
+      "step": 6031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.795243215375725e-05,
+      "loss": 1.9548,
+      "step": 6032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.783366267180592e-05,
+      "loss": 1.8624,
+      "step": 6033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.77149761010898e-05,
+      "loss": 2.0571,
+      "step": 6034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.759637246491863e-05,
+      "loss": 1.896,
+      "step": 6035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.74778517865854e-05,
+      "loss": 2.0051,
+      "step": 6036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.735941408936692e-05,
+      "loss": 1.8824,
+      "step": 6037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.724105939652359e-05,
+      "loss": 1.9042,
+      "step": 6038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.712278773129988e-05,
+      "loss": 1.9602,
+      "step": 6039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.70045991169237e-05,
+      "loss": 1.9998,
+      "step": 6040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.688649357660643e-05,
+      "loss": 1.9353,
+      "step": 6041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.676847113354363e-05,
+      "loss": 2.0277,
+      "step": 6042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.665053181091413e-05,
+      "loss": 1.9989,
+      "step": 6043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.653267563188072e-05,
+      "loss": 1.978,
+      "step": 6044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.641490261958972e-05,
+      "loss": 2.1393,
+      "step": 6045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.6297212797171e-05,
+      "loss": 2.0107,
+      "step": 6046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.617960618773822e-05,
+      "loss": 1.9266,
+      "step": 6047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.60620828143887e-05,
+      "loss": 2.0219,
+      "step": 6048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.594464270020362e-05,
+      "loss": 2.0504,
+      "step": 6049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.58272858682475e-05,
+      "loss": 2.0421,
+      "step": 6050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.571001234156849e-05,
+      "loss": 1.9957,
+      "step": 6051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.559282214319858e-05,
+      "loss": 2.0146,
+      "step": 6052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.547571529615322e-05,
+      "loss": 1.9929,
+      "step": 6053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.535869182343158e-05,
+      "loss": 1.9825,
+      "step": 6054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.524175174801645e-05,
+      "loss": 2.0284,
+      "step": 6055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.512489509287441e-05,
+      "loss": 2.0209,
+      "step": 6056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.50081218809554e-05,
+      "loss": 2.0194,
+      "step": 6057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.489143213519301e-05,
+      "loss": 1.9737,
+      "step": 6058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.477482587850436e-05,
+      "loss": 1.9457,
+      "step": 6059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.465830313379063e-05,
+      "loss": 2.0227,
+      "step": 6060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.454186392393597e-05,
+      "loss": 2.0726,
+      "step": 6061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.442550827180867e-05,
+      "loss": 1.9657,
+      "step": 6062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.43092362002602e-05,
+      "loss": 1.8831,
+      "step": 6063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.419304773212587e-05,
+      "loss": 2.0504,
+      "step": 6064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.407694289022426e-05,
+      "loss": 2.0253,
+      "step": 6065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.396092169735807e-05,
+      "loss": 1.9745,
+      "step": 6066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.384498417631314e-05,
+      "loss": 2.0296,
+      "step": 6067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.372913034985878e-05,
+      "loss": 2.0978,
+      "step": 6068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.361336024074833e-05,
+      "loss": 1.9495,
+      "step": 6069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.349767387171824e-05,
+      "loss": 1.963,
+      "step": 6070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.338207126548896e-05,
+      "loss": 1.9779,
+      "step": 6071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.326655244476399e-05,
+      "loss": 2.0374,
+      "step": 6072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.315111743223068e-05,
+      "loss": 2.0423,
+      "step": 6073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.30357662505598e-05,
+      "loss": 2.0064,
+      "step": 6074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.292049892240582e-05,
+      "loss": 1.9219,
+      "step": 6075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.280531547040653e-05,
+      "loss": 1.8938,
+      "step": 6076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.26902159171835e-05,
+      "loss": 1.9357,
+      "step": 6077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.257520028534159e-05,
+      "loss": 1.8725,
+      "step": 6078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.246026859746924e-05,
+      "loss": 2.0491,
+      "step": 6079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.234542087613849e-05,
+      "loss": 1.912,
+      "step": 6080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.223065714390459e-05,
+      "loss": 2.0901,
+      "step": 6081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.211597742330677e-05,
+      "loss": 1.9071,
+      "step": 6082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.200138173686754e-05,
+      "loss": 1.9025,
+      "step": 6083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.188687010709283e-05,
+      "loss": 2.018,
+      "step": 6084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.177244255647209e-05,
+      "loss": 1.9307,
+      "step": 6085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.165809910747818e-05,
+      "loss": 1.9161,
+      "step": 6086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.154383978256773e-05,
+      "loss": 1.7847,
+      "step": 6087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.142966460418055e-05,
+      "loss": 1.9219,
+      "step": 6088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.131557359474022e-05,
+      "loss": 2.0899,
+      "step": 6089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.12015667766534e-05,
+      "loss": 1.9465,
+      "step": 6090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.108764417231056e-05,
+      "loss": 2.0723,
+      "step": 6091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.097380580408535e-05,
+      "loss": 2.0632,
+      "step": 6092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.086005169433524e-05,
+      "loss": 2.1281,
+      "step": 6093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.074638186540078e-05,
+      "loss": 1.9248,
+      "step": 6094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0632796339606e-05,
+      "loss": 1.891,
+      "step": 6095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.051929513925886e-05,
+      "loss": 1.9803,
+      "step": 6096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.040587828665002e-05,
+      "loss": 1.8701,
+      "step": 6097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.02925458040542e-05,
+      "loss": 1.868,
+      "step": 6098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.017929771372927e-05,
+      "loss": 1.9195,
+      "step": 6099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.006613403791651e-05,
+      "loss": 1.8553,
+      "step": 6100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.995305479884051e-05,
+      "loss": 1.9702,
+      "step": 6101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.984006001870974e-05,
+      "loss": 1.9869,
+      "step": 6102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.97271497197155e-05,
+      "loss": 1.963,
+      "step": 6103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9614323924033e-05,
+      "loss": 1.9614,
+      "step": 6104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.95015826538205e-05,
+      "loss": 1.9206,
+      "step": 6105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.938892593121982e-05,
+      "loss": 1.8955,
+      "step": 6106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.927635377835606e-05,
+      "loss": 1.9205,
+      "step": 6107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9163866217338e-05,
+      "loss": 1.9522,
+      "step": 6108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.905146327025735e-05,
+      "loss": 1.8956,
+      "step": 6109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.893914495918968e-05,
+      "loss": 1.9745,
+      "step": 6110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.882691130619361e-05,
+      "loss": 2.0127,
+      "step": 6111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.871476233331126e-05,
+      "loss": 1.8041,
+      "step": 6112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.860269806256808e-05,
+      "loss": 1.884,
+      "step": 6113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.849071851597277e-05,
+      "loss": 1.9896,
+      "step": 6114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.837882371551762e-05,
+      "loss": 2.0868,
+      "step": 6115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.826701368317833e-05,
+      "loss": 2.0263,
+      "step": 6116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.815528844091368e-05,
+      "loss": 1.9988,
+      "step": 6117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.804364801066593e-05,
+      "loss": 2.0843,
+      "step": 6118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.793209241436049e-05,
+      "loss": 2.0403,
+      "step": 6119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.782062167390662e-05,
+      "loss": 1.8889,
+      "step": 6120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.770923581119625e-05,
+      "loss": 2.0719,
+      "step": 6121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.759793484810528e-05,
+      "loss": 1.9998,
+      "step": 6122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.748671880649243e-05,
+      "loss": 2.0416,
+      "step": 6123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.737558770819996e-05,
+      "loss": 1.9411,
+      "step": 6124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.726454157505352e-05,
+      "loss": 2.0747,
+      "step": 6125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.715358042886194e-05,
+      "loss": 2.0799,
+      "step": 6126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.70427042914174e-05,
+      "loss": 2.0231,
+      "step": 6127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.693191318449527e-05,
+      "loss": 1.9544,
+      "step": 6128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.682120712985457e-05,
+      "loss": 1.9867,
+      "step": 6129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.671058614923714e-05,
+      "loss": 1.9943,
+      "step": 6130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.66000502643686e-05,
+      "loss": 2.1614,
+      "step": 6131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.648959949695748e-05,
+      "loss": 1.8717,
+      "step": 6132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.637923386869571e-05,
+      "loss": 2.0802,
+      "step": 6133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.626895340125844e-05,
+      "loss": 1.9298,
+      "step": 6134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.615875811630428e-05,
+      "loss": 2.064,
+      "step": 6135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.604864803547511e-05,
+      "loss": 1.9884,
+      "step": 6136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.593862318039584e-05,
+      "loss": 1.9016,
+      "step": 6137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.582868357267479e-05,
+      "loss": 2.0144,
+      "step": 6138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.571882923390349e-05,
+      "loss": 1.9242,
+      "step": 6139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.560906018565677e-05,
+      "loss": 2.0038,
+      "step": 6140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.549937644949255e-05,
+      "loss": 2.0619,
+      "step": 6141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.538977804695234e-05,
+      "loss": 1.942,
+      "step": 6142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.528026499956069e-05,
+      "loss": 1.9926,
+      "step": 6143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.517083732882534e-05,
+      "loss": 1.8512,
+      "step": 6144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.506149505623721e-05,
+      "loss": 1.9875,
+      "step": 6145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.495223820327057e-05,
+      "loss": 2.1136,
+      "step": 6146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.484306679138301e-05,
+      "loss": 1.9485,
+      "step": 6147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.473398084201504e-05,
+      "loss": 1.9135,
+      "step": 6148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.462498037659076e-05,
+      "loss": 2.0536,
+      "step": 6149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.451606541651722e-05,
+      "loss": 1.8842,
+      "step": 6150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.440723598318465e-05,
+      "loss": 1.9524,
+      "step": 6151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.429849209796657e-05,
+      "loss": 1.944,
+      "step": 6152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.418983378221988e-05,
+      "loss": 1.8743,
+      "step": 6153
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.408126105728435e-05,
+      "loss": 2.018,
+      "step": 6154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.397277394448304e-05,
+      "loss": 1.8969,
+      "step": 6155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.386437246512245e-05,
+      "loss": 1.8653,
+      "step": 6156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.375605664049178e-05,
+      "loss": 1.9214,
+      "step": 6157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.364782649186396e-05,
+      "loss": 1.8864,
+      "step": 6158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.353968204049476e-05,
+      "loss": 2.0121,
+      "step": 6159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.343162330762308e-05,
+      "loss": 1.8849,
+      "step": 6160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.332365031447102e-05,
+      "loss": 1.8858,
+      "step": 6161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.321576308224397e-05,
+      "loss": 2.0316,
+      "step": 6162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.310796163213056e-05,
+      "loss": 2.0394,
+      "step": 6163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.300024598530234e-05,
+      "loss": 1.9462,
+      "step": 6164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.289261616291409e-05,
+      "loss": 1.9179,
+      "step": 6165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.278507218610368e-05,
+      "loss": 1.8228,
+      "step": 6166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.26776140759922e-05,
+      "loss": 1.7659,
+      "step": 6167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.257024185368381e-05,
+      "loss": 2.1014,
+      "step": 6168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.246295554026588e-05,
+      "loss": 1.9507,
+      "step": 6169
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.235575515680897e-05,
+      "loss": 1.9226,
+      "step": 6170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.224864072436665e-05,
+      "loss": 1.9729,
+      "step": 6171
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.214161226397552e-05,
+      "loss": 1.8907,
+      "step": 6172
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.203466979665534e-05,
+      "loss": 2.134,
+      "step": 6173
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.19278133434093e-05,
+      "loss": 1.9035,
+      "step": 6174
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.182104292522322e-05,
+      "loss": 1.9542,
+      "step": 6175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.17143585630664e-05,
+      "loss": 2.0762,
+      "step": 6176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.160776027789105e-05,
+      "loss": 1.8799,
+      "step": 6177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.150124809063246e-05,
+      "loss": 1.8848,
+      "step": 6178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.139482202220892e-05,
+      "loss": 2.0932,
+      "step": 6179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.128848209352228e-05,
+      "loss": 2.0819,
+      "step": 6180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.118222832545689e-05,
+      "loss": 1.951,
+      "step": 6181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.107606073888045e-05,
+      "loss": 1.9434,
+      "step": 6182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.096997935464382e-05,
+      "loss": 1.9802,
+      "step": 6183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0863984193580634e-05,
+      "loss": 1.951,
+      "step": 6184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.075807527650806e-05,
+      "loss": 1.9089,
+      "step": 6185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0652252624225915e-05,
+      "loss": 1.9304,
+      "step": 6186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.054651625751717e-05,
+      "loss": 1.9341,
+      "step": 6187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.044086619714778e-05,
+      "loss": 1.9068,
+      "step": 6188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.033530246386704e-05,
+      "loss": 1.9782,
+      "step": 6189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.022982507840713e-05,
+      "loss": 1.9806,
+      "step": 6190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0124434061483157e-05,
+      "loss": 1.9875,
+      "step": 6191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.00191294337934e-05,
+      "loss": 2.0437,
+      "step": 6192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9913911216019076e-05,
+      "loss": 1.9631,
+      "step": 6193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9808779428824455e-05,
+      "loss": 1.9889,
+      "step": 6194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.970373409285695e-05,
+      "loss": 2.0489,
+      "step": 6195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.959877522874696e-05,
+      "loss": 1.9412,
+      "step": 6196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.949390285710776e-05,
+      "loss": 1.9891,
+      "step": 6197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9389116998535775e-05,
+      "loss": 1.9062,
+      "step": 6198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.928441767361037e-05,
+      "loss": 2.0079,
+      "step": 6199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.917980490289382e-05,
+      "loss": 2.0677,
+      "step": 6200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.907527870693169e-05,
+      "loss": 1.9744,
+      "step": 6201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.897083910625228e-05,
+      "loss": 1.9885,
+      "step": 6202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.886648612136708e-05,
+      "loss": 1.8954,
+      "step": 6203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.876221977277041e-05,
+      "loss": 1.9571,
+      "step": 6204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.865804008093961e-05,
+      "loss": 1.8925,
+      "step": 6205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8553947066334935e-05,
+      "loss": 1.9721,
+      "step": 6206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.844994074939991e-05,
+      "loss": 1.95,
+      "step": 6207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.834602115056059e-05,
+      "loss": 1.9513,
+      "step": 6208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.824218829022643e-05,
+      "loss": 2.0243,
+      "step": 6209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.813844218878961e-05,
+      "loss": 1.8992,
+      "step": 6210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.803478286662511e-05,
+      "loss": 1.8979,
+      "step": 6211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7931210344091433e-05,
+      "loss": 2.002,
+      "step": 6212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.782772464152947e-05,
+      "loss": 1.8945,
+      "step": 6213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.772432577926323e-05,
+      "loss": 2.0219,
+      "step": 6214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.762101377759971e-05,
+      "loss": 1.915,
+      "step": 6215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.751778865682894e-05,
+      "loss": 1.8563,
+      "step": 6216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7414650437223656e-05,
+      "loss": 2.1411,
+      "step": 6217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.731159913903988e-05,
+      "loss": 1.9519,
+      "step": 6218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.720863478251615e-05,
+      "loss": 1.8061,
+      "step": 6219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7105757387874154e-05,
+      "loss": 2.0629,
+      "step": 6220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.700296697531843e-05,
+      "loss": 1.9726,
+      "step": 6221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.690026356503647e-05,
+      "loss": 1.8571,
+      "step": 6222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6797647177198874e-05,
+      "loss": 1.9478,
+      "step": 6223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6695117831958795e-05,
+      "loss": 1.9669,
+      "step": 6224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.659267554945247e-05,
+      "loss": 1.9781,
+      "step": 6225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6490320349799066e-05,
+      "loss": 1.8759,
+      "step": 6226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.638805225310045e-05,
+      "loss": 1.9157,
+      "step": 6227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6285871279441736e-05,
+      "loss": 1.9466,
+      "step": 6228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6183777448890524e-05,
+      "loss": 2.0758,
+      "step": 6229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6081770781497785e-05,
+      "loss": 1.9518,
+      "step": 6230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.597985129729688e-05,
+      "loss": 1.8523,
+      "step": 6231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5878019016304295e-05,
+      "loss": 2.0325,
+      "step": 6232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.577627395851925e-05,
+      "loss": 2.0026,
+      "step": 6233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.567461614392416e-05,
+      "loss": 1.8278,
+      "step": 6234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.557304559248383e-05,
+      "loss": 1.8539,
+      "step": 6235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.547156232414646e-05,
+      "loss": 1.9429,
+      "step": 6236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5370166358842677e-05,
+      "loss": 1.8881,
+      "step": 6237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5268857716485984e-05,
+      "loss": 2.1642,
+      "step": 6238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5167636416973085e-05,
+      "loss": 1.9401,
+      "step": 6239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5066502480183246e-05,
+      "loss": 2.0157,
+      "step": 6240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4965455925978625e-05,
+      "loss": 2.1482,
+      "step": 6241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.486449677420408e-05,
+      "loss": 2.0238,
+      "step": 6242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.476362504468768e-05,
+      "loss": 1.9293,
+      "step": 6243
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.466284075723993e-05,
+      "loss": 1.9324,
+      "step": 6244
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.456214393165448e-05,
+      "loss": 1.8847,
+      "step": 6245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.446153458770764e-05,
+      "loss": 2.0002,
+      "step": 6246
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.436101274515842e-05,
+      "loss": 1.9037,
+      "step": 6247
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.426057842374882e-05,
+      "loss": 1.9154,
+      "step": 6248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4160231643203605e-05,
+      "loss": 2.0707,
+      "step": 6249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.405997242323052e-05,
+      "loss": 2.0045,
+      "step": 6250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.395980078351975e-05,
+      "loss": 1.9157,
+      "step": 6251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.385971674374457e-05,
+      "loss": 2.1324,
+      "step": 6252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.375972032356097e-05,
+      "loss": 1.971,
+      "step": 6253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.365981154260752e-05,
+      "loss": 1.959,
+      "step": 6254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3559990420506025e-05,
+      "loss": 1.9213,
+      "step": 6255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.346025697686069e-05,
+      "loss": 1.9222,
+      "step": 6256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.336061123125874e-05,
+      "loss": 1.8725,
+      "step": 6257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.326105320326996e-05,
+      "loss": 2.0543,
+      "step": 6258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3161582912447106e-05,
+      "loss": 1.9744,
+      "step": 6259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3062200378325485e-05,
+      "loss": 1.8028,
+      "step": 6260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.296290562042349e-05,
+      "loss": 1.9836,
+      "step": 6261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.286369865824192e-05,
+      "loss": 1.8593,
+      "step": 6262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2764579511264624e-05,
+      "loss": 1.8893,
+      "step": 6263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.266554819895808e-05,
+      "loss": 1.9188,
+      "step": 6264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.25666047407713e-05,
+      "loss": 1.9078,
+      "step": 6265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2467749156136504e-05,
+      "loss": 1.9517,
+      "step": 6266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2368981464468336e-05,
+      "loss": 2.0208,
+      "step": 6267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.227030168516417e-05,
+      "loss": 1.9939,
+      "step": 6268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2171709837604276e-05,
+      "loss": 2.0781,
+      "step": 6269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.207320594115161e-05,
+      "loss": 2.0071,
+      "step": 6270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.19747900151516e-05,
+      "loss": 1.934,
+      "step": 6271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.187646207893287e-05,
+      "loss": 1.914,
+      "step": 6272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.177822215180639e-05,
+      "loss": 1.9533,
+      "step": 6273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.168007025306604e-05,
+      "loss": 1.9611,
+      "step": 6274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1582006401988066e-05,
+      "loss": 1.9117,
+      "step": 6275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.148403061783197e-05,
+      "loss": 1.8773,
+      "step": 6276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.138614291983967e-05,
+      "loss": 1.8889,
+      "step": 6277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.12883433272357e-05,
+      "loss": 1.7956,
+      "step": 6278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.11906318592274e-05,
+      "loss": 1.9231,
+      "step": 6279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1093008535004816e-05,
+      "loss": 1.9154,
+      "step": 6280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.09954733737405e-05,
+      "loss": 1.9112,
+      "step": 6281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.089802639459001e-05,
+      "loss": 1.9112,
+      "step": 6282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0800667616691444e-05,
+      "loss": 1.8706,
+      "step": 6283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.070339705916543e-05,
+      "loss": 1.9441,
+      "step": 6284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0606214741115484e-05,
+      "loss": 1.8955,
+      "step": 6285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.050912068162766e-05,
+      "loss": 2.0235,
+      "step": 6286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.041211489977066e-05,
+      "loss": 1.9384,
+      "step": 6287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0315197414596005e-05,
+      "loss": 1.8977,
+      "step": 6288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.021836824513759e-05,
+      "loss": 1.8583,
+      "step": 6289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0121627410412486e-05,
+      "loss": 1.9332,
+      "step": 6290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.002497492941982e-05,
+      "loss": 1.9004,
+      "step": 6291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.992841082114169e-05,
+      "loss": 2.0319,
+      "step": 6292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9831935104542706e-05,
+      "loss": 1.8667,
+      "step": 6293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.973554779857037e-05,
+      "loss": 1.9037,
+      "step": 6294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.963924892215443e-05,
+      "loss": 2.0015,
+      "step": 6295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9543038494207716e-05,
+      "loss": 1.9363,
+      "step": 6296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.944691653362526e-05,
+      "loss": 1.9681,
+      "step": 6297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.935088305928487e-05,
+      "loss": 2.03,
+      "step": 6298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9254938090047164e-05,
+      "loss": 2.1404,
+      "step": 6299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.915908164475519e-05,
+      "loss": 2.0187,
+      "step": 6300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.906331374223466e-05,
+      "loss": 1.9457,
+      "step": 6301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8967634401293705e-05,
+      "loss": 2.0009,
+      "step": 6302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.887204364072334e-05,
+      "loss": 2.0565,
+      "step": 6303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8776541479297275e-05,
+      "loss": 1.8696,
+      "step": 6304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.868112793577145e-05,
+      "loss": 1.935,
+      "step": 6305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.858580302888466e-05,
+      "loss": 1.8517,
+      "step": 6306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8490566777358084e-05,
+      "loss": 1.9805,
+      "step": 6307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8395419199895655e-05,
+      "loss": 1.8373,
+      "step": 6308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8300360315183867e-05,
+      "loss": 2.0286,
+      "step": 6309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.82053901418919e-05,
+      "loss": 1.9575,
+      "step": 6310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8110508698671286e-05,
+      "loss": 1.9789,
+      "step": 6311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8015716004156274e-05,
+      "loss": 1.9603,
+      "step": 6312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.792101207696365e-05,
+      "loss": 1.914,
+      "step": 6313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.782639693569263e-05,
+      "loss": 1.7516,
+      "step": 6314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.773187059892531e-05,
+      "loss": 1.9,
+      "step": 6315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7637433085226e-05,
+      "loss": 1.9272,
+      "step": 6316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7543084413141924e-05,
+      "loss": 1.9524,
+      "step": 6317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.744882460120259e-05,
+      "loss": 2.0304,
+      "step": 6318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.735465366792008e-05,
+      "loss": 2.239,
+      "step": 6319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7260571631788926e-05,
+      "loss": 1.8442,
+      "step": 6320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.716657851128664e-05,
+      "loss": 1.9188,
+      "step": 6321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7072674324872775e-05,
+      "loss": 1.9383,
+      "step": 6322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.69788590909897e-05,
+      "loss": 1.9591,
+      "step": 6323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6885132828062225e-05,
+      "loss": 1.9833,
+      "step": 6324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.679149555449763e-05,
+      "loss": 1.8568,
+      "step": 6325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.669794728868587e-05,
+      "loss": 1.9514,
+      "step": 6326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.660448804899925e-05,
+      "loss": 1.9795,
+      "step": 6327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6511117853792754e-05,
+      "loss": 2.0299,
+      "step": 6328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6417836721403595e-05,
+      "loss": 1.9461,
+      "step": 6329
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.63246446701519e-05,
+      "loss": 2.1041,
+      "step": 6330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.62315417183401e-05,
+      "loss": 2.05,
+      "step": 6331
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6138527884253045e-05,
+      "loss": 1.9384,
+      "step": 6332
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6045603186158136e-05,
+      "loss": 2.0097,
+      "step": 6333
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.595276764230538e-05,
+      "loss": 1.9461,
+      "step": 6334
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5860021270927034e-05,
+      "loss": 1.9571,
+      "step": 6335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5767364090238126e-05,
+      "loss": 1.9284,
+      "step": 6336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567479611843606e-05,
+      "loss": 1.9174,
+      "step": 6337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.558231737370067e-05,
+      "loss": 1.9852,
+      "step": 6338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5489927874194204e-05,
+      "loss": 1.9454,
+      "step": 6339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.53976276380616e-05,
+      "loss": 1.9569,
+      "step": 6340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5305416683429915e-05,
+      "loss": 1.9573,
+      "step": 6341
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.521329502840915e-05,
+      "loss": 2.0014,
+      "step": 6342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.512126269109135e-05,
+      "loss": 1.9455,
+      "step": 6343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.502931968955132e-05,
+      "loss": 2.0063,
+      "step": 6344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4937466041846065e-05,
+      "loss": 1.9857,
+      "step": 6345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4845701766015214e-05,
+      "loss": 1.9354,
+      "step": 6346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4754026880080625e-05,
+      "loss": 2.0388,
+      "step": 6347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.466244140204701e-05,
+      "loss": 1.94,
+      "step": 6348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4570945349901084e-05,
+      "loss": 1.9109,
+      "step": 6349
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4479538741612315e-05,
+      "loss": 1.9778,
+      "step": 6350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4388221595132396e-05,
+      "loss": 1.9041,
+      "step": 6351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.429699392839554e-05,
+      "loss": 1.9235,
+      "step": 6352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.42058557593184e-05,
+      "loss": 2.0063,
+      "step": 6353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.411480710580007e-05,
+      "loss": 1.9811,
+      "step": 6354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.40238479857219e-05,
+      "loss": 1.9479,
+      "step": 6355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.393297841694793e-05,
+      "loss": 1.9324,
+      "step": 6356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3842198417324344e-05,
+      "loss": 1.9106,
+      "step": 6357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.375150800467986e-05,
+      "loss": 2.0841,
+      "step": 6358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.366090719682575e-05,
+      "loss": 1.8757,
+      "step": 6359
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.357039601155538e-05,
+      "loss": 2.0207,
+      "step": 6360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.34799744666447e-05,
+      "loss": 1.9425,
+      "step": 6361
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.338964257985195e-05,
+      "loss": 1.9874,
+      "step": 6362
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.329940036891788e-05,
+      "loss": 2.0912,
+      "step": 6363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.320924785156577e-05,
+      "loss": 1.8087,
+      "step": 6364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3119185045500886e-05,
+      "loss": 2.0724,
+      "step": 6365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.30292119684112e-05,
+      "loss": 1.9426,
+      "step": 6366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2939328637966846e-05,
+      "loss": 2.0409,
+      "step": 6367
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.284953507182043e-05,
+      "loss": 1.9794,
+      "step": 6368
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.275983128760696e-05,
+      "loss": 1.8852,
+      "step": 6369
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2670217302943905e-05,
+      "loss": 1.8343,
+      "step": 6370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.258069313543089e-05,
+      "loss": 1.9655,
+      "step": 6371
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.249125880264998e-05,
+      "loss": 2.0647,
+      "step": 6372
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.240191432216562e-05,
+      "loss": 1.887,
+      "step": 6373
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.231265971152448e-05,
+      "loss": 1.8633,
+      "step": 6374
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.222349498825584e-05,
+      "loss": 2.0307,
+      "step": 6375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2134420169871125e-05,
+      "loss": 2.0078,
+      "step": 6376
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2045435273864175e-05,
+      "loss": 1.7464,
+      "step": 6377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1956540317711176e-05,
+      "loss": 1.9655,
+      "step": 6378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.18677353188705e-05,
+      "loss": 1.9461,
+      "step": 6379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.177902029478314e-05,
+      "loss": 2.042,
+      "step": 6380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1690395262872273e-05,
+      "loss": 1.9682,
+      "step": 6381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.160186024054313e-05,
+      "loss": 1.9416,
+      "step": 6382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.151341524518382e-05,
+      "loss": 1.8487,
+      "step": 6383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.142506029416432e-05,
+      "loss": 2.0063,
+      "step": 6384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.133679540483703e-05,
+      "loss": 1.9059,
+      "step": 6385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1248620594536876e-05,
+      "loss": 2.0525,
+      "step": 6386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.116053588058083e-05,
+      "loss": 1.8896,
+      "step": 6387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.107254128026827e-05,
+      "loss": 1.7978,
+      "step": 6388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.098463681088071e-05,
+      "loss": 1.8863,
+      "step": 6389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.089682248968235e-05,
+      "loss": 2.0023,
+      "step": 6390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.080909833391944e-05,
+      "loss": 1.9831,
+      "step": 6391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.072146436082047e-05,
+      "loss": 1.928,
+      "step": 6392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.063392058759635e-05,
+      "loss": 1.9102,
+      "step": 6393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0546467031440106e-05,
+      "loss": 1.8626,
+      "step": 6394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.045910370952716e-05,
+      "loss": 1.9623,
+      "step": 6395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.037183063901523e-05,
+      "loss": 1.9926,
+      "step": 6396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.028464783704444e-05,
+      "loss": 1.8729,
+      "step": 6397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.019755532073688e-05,
+      "loss": 1.9584,
+      "step": 6398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0110553107197135e-05,
+      "loss": 1.8613,
+      "step": 6399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.002364121351187e-05,
+      "loss": 1.9487,
+      "step": 6400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.993681965675011e-05,
+      "loss": 2.1127,
+      "step": 6401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.985008845396332e-05,
+      "loss": 2.0087,
+      "step": 6402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.976344762218487e-05,
+      "loss": 1.9563,
+      "step": 6403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.967689717843071e-05,
+      "loss": 2.0,
+      "step": 6404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.959043713969879e-05,
+      "loss": 1.9392,
+      "step": 6405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.950406752296942e-05,
+      "loss": 1.8439,
+      "step": 6406
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9417788345205155e-05,
+      "loss": 1.9162,
+      "step": 6407
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.933159962335081e-05,
+      "loss": 1.8677,
+      "step": 6408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9245501374333246e-05,
+      "loss": 1.8853,
+      "step": 6409
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9159493615061934e-05,
+      "loss": 2.0069,
+      "step": 6410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.90735763624282e-05,
+      "loss": 1.8447,
+      "step": 6411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8987749633305644e-05,
+      "loss": 1.929,
+      "step": 6412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.890201344455041e-05,
+      "loss": 2.0754,
+      "step": 6413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8816367813000464e-05,
+      "loss": 2.0971,
+      "step": 6414
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.873081275547624e-05,
+      "loss": 1.9233,
+      "step": 6415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8645348288780134e-05,
+      "loss": 1.8838,
+      "step": 6416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.855997442969711e-05,
+      "loss": 1.9718,
+      "step": 6417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.847469119499408e-05,
+      "loss": 1.8967,
+      "step": 6418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.838949860142027e-05,
+      "loss": 2.0582,
+      "step": 6419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8304396665707006e-05,
+      "loss": 1.9095,
+      "step": 6420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.821938540456782e-05,
+      "loss": 1.9699,
+      "step": 6421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8134464834698454e-05,
+      "loss": 1.8025,
+      "step": 6422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8049634972776915e-05,
+      "loss": 1.836,
+      "step": 6423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7964895835463364e-05,
+      "loss": 1.8453,
+      "step": 6424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.788024743940016e-05,
+      "loss": 1.8928,
+      "step": 6425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7795689801211675e-05,
+      "loss": 1.7927,
+      "step": 6426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.771122293750467e-05,
+      "loss": 1.95,
+      "step": 6427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.762684686486789e-05,
+      "loss": 1.9382,
+      "step": 6428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.754256159987246e-05,
+      "loss": 1.9547,
+      "step": 6429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.745836715907147e-05,
+      "loss": 2.0145,
+      "step": 6430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.737426355900037e-05,
+      "loss": 1.8595,
+      "step": 6431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7290250816176595e-05,
+      "loss": 1.9241,
+      "step": 6432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.72063289470998e-05,
+      "loss": 1.8953,
+      "step": 6433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.712249796825173e-05,
+      "loss": 1.9166,
+      "step": 6434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.703875789609656e-05,
+      "loss": 1.8543,
+      "step": 6435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6955108747080135e-05,
+      "loss": 1.8423,
+      "step": 6436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.687155053763092e-05,
+      "loss": 1.8807,
+      "step": 6437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6788083284159165e-05,
+      "loss": 2.0306,
+      "step": 6438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.670470700305745e-05,
+      "loss": 2.0086,
+      "step": 6439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.662142171070043e-05,
+      "loss": 1.8948,
+      "step": 6440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.653822742344498e-05,
+      "loss": 1.8457,
+      "step": 6441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6455124157629804e-05,
+      "loss": 1.9467,
+      "step": 6442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.637211192957618e-05,
+      "loss": 1.8813,
+      "step": 6443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6289190755587064e-05,
+      "loss": 1.9334,
+      "step": 6444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6206360651947934e-05,
+      "loss": 2.0339,
+      "step": 6445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6123621634926104e-05,
+      "loss": 1.9659,
+      "step": 6446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6040973720771033e-05,
+      "loss": 1.9713,
+      "step": 6447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.595841692571439e-05,
+      "loss": 1.9531,
+      "step": 6448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.587595126596976e-05,
+      "loss": 1.8397,
+      "step": 6449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.579357675773309e-05,
+      "loss": 1.8658,
+      "step": 6450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.571129341718238e-05,
+      "loss": 1.9189,
+      "step": 6451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.562910126047747e-05,
+      "loss": 1.8284,
+      "step": 6452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.554700030376062e-05,
+      "loss": 1.9641,
+      "step": 6453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5464990563155885e-05,
+      "loss": 1.8722,
+      "step": 6454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538307205476954e-05,
+      "loss": 1.9018,
+      "step": 6455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530124479468999e-05,
+      "loss": 1.9818,
+      "step": 6456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5219508798987774e-05,
+      "loss": 1.8841,
+      "step": 6457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.513786408371533e-05,
+      "loss": 1.9228,
+      "step": 6458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5056310664907276e-05,
+      "loss": 1.8816,
+      "step": 6459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.497484855858019e-05,
+      "loss": 2.0235,
+      "step": 6460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4893477780732817e-05,
+      "loss": 1.8625,
+      "step": 6461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4812198347346006e-05,
+      "loss": 2.0024,
+      "step": 6462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4731010274382524e-05,
+      "loss": 1.8677,
+      "step": 6463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.464991357778746e-05,
+      "loss": 1.8755,
+      "step": 6464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4568908273487574e-05,
+      "loss": 1.9421,
+      "step": 6465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.448799437739192e-05,
+      "loss": 1.9223,
+      "step": 6466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4407171905391666e-05,
+      "loss": 2.0332,
+      "step": 6467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.43264408733599e-05,
+      "loss": 1.9404,
+      "step": 6468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4245801297151605e-05,
+      "loss": 1.9447,
+      "step": 6469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4165253192604106e-05,
+      "loss": 1.9845,
+      "step": 6470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4084796575536646e-05,
+      "loss": 1.8877,
+      "step": 6471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.400443146175036e-05,
+      "loss": 1.9356,
+      "step": 6472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.392415786702868e-05,
+      "loss": 1.8364,
+      "step": 6473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.384397580713683e-05,
+      "loss": 1.9108,
+      "step": 6474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.376388529782215e-05,
+      "loss": 2.0509,
+      "step": 6475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3683886354813956e-05,
+      "loss": 1.9117,
+      "step": 6476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.360397899382361e-05,
+      "loss": 1.8782,
+      "step": 6477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3524163230544634e-05,
+      "loss": 2.1087,
+      "step": 6478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.344443908065231e-05,
+      "loss": 1.7663,
+      "step": 6479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.336480655980406e-05,
+      "loss": 1.8563,
+      "step": 6480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.328526568363932e-05,
+      "loss": 1.9506,
+      "step": 6481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3205816467779316e-05,
+      "loss": 1.9434,
+      "step": 6482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.312645892782767e-05,
+      "loss": 2.0297,
+      "step": 6483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3047193079369764e-05,
+      "loss": 1.8595,
+      "step": 6484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.296801893797297e-05,
+      "loss": 1.89,
+      "step": 6485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.288893651918662e-05,
+      "loss": 1.9492,
+      "step": 6486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2809945838542074e-05,
+      "loss": 1.8131,
+      "step": 6487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2731046911552654e-05,
+      "loss": 1.797,
+      "step": 6488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2652239753713897e-05,
+      "loss": 1.9329,
+      "step": 6489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2573524380502814e-05,
+      "loss": 1.9682,
+      "step": 6490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.249490080737894e-05,
+      "loss": 1.7457,
+      "step": 6491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.241636904978346e-05,
+      "loss": 1.9746,
+      "step": 6492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2337929123139434e-05,
+      "loss": 2.0142,
+      "step": 6493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.225958104285232e-05,
+      "loss": 2.0135,
+      "step": 6494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2181324824309075e-05,
+      "loss": 2.0343,
+      "step": 6495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.210316048287881e-05,
+      "loss": 1.9998,
+      "step": 6496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2025088033912685e-05,
+      "loss": 1.9218,
+      "step": 6497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.194710749274371e-05,
+      "loss": 1.9132,
+      "step": 6498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.186921887468669e-05,
+      "loss": 1.96,
+      "step": 6499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.179142219503878e-05,
+      "loss": 1.905,
+      "step": 6500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.171371746907864e-05,
+      "loss": 1.9272,
+      "step": 6501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.163610471206718e-05,
+      "loss": 1.8917,
+      "step": 6502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1558583939247046e-05,
+      "loss": 2.0397,
+      "step": 6503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1481155165842935e-05,
+      "loss": 2.1128,
+      "step": 6504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1403818407061525e-05,
+      "loss": 1.9613,
+      "step": 6505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.132657367809133e-05,
+      "loss": 1.9699,
+      "step": 6506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.124942099410277e-05,
+      "loss": 2.0472,
+      "step": 6507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1172360370248175e-05,
+      "loss": 1.9509,
+      "step": 6508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.109539182166182e-05,
+      "loss": 1.7575,
+      "step": 6509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1018515363460065e-05,
+      "loss": 2.0255,
+      "step": 6510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0941731010740934e-05,
+      "loss": 1.8186,
+      "step": 6511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0865038778584595e-05,
+      "loss": 1.9601,
+      "step": 6512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.078843868205283e-05,
+      "loss": 1.8432,
+      "step": 6513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.071193073618961e-05,
+      "loss": 1.9302,
+      "step": 6514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.063551495602057e-05,
+      "loss": 1.9945,
+      "step": 6515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.05591913565535e-05,
+      "loss": 1.8181,
+      "step": 6516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0482959952777778e-05,
+      "loss": 1.8672,
+      "step": 6517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0406820759665056e-05,
+      "loss": 1.8954,
+      "step": 6518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0330773792168588e-05,
+      "loss": 1.9262,
+      "step": 6519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.025481906522354e-05,
+      "loss": 1.9154,
+      "step": 6520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0178956593747087e-05,
+      "loss": 1.9433,
+      "step": 6521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0103186392638204e-05,
+      "loss": 2.0201,
+      "step": 6522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0027508476777653e-05,
+      "loss": 1.8809,
+      "step": 6523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.995192286102838e-05,
+      "loss": 1.9248,
+      "step": 6524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9876429560234896e-05,
+      "loss": 1.9721,
+      "step": 6525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9801028589223567e-05,
+      "loss": 2.027,
+      "step": 6526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.972571996280293e-05,
+      "loss": 1.9559,
+      "step": 6527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.965050369576311e-05,
+      "loss": 2.0113,
+      "step": 6528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.957537980287617e-05,
+      "loss": 1.8544,
+      "step": 6529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.950034829889614e-05,
+      "loss": 1.9417,
+      "step": 6530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9425409198558684e-05,
+      "loss": 1.966,
+      "step": 6531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9350562516581524e-05,
+      "loss": 1.8532,
+      "step": 6532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9275808267664184e-05,
+      "loss": 1.8545,
+      "step": 6533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9201146466487915e-05,
+      "loss": 1.8519,
+      "step": 6534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.912657712771599e-05,
+      "loss": 1.8762,
+      "step": 6535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.905210026599331e-05,
+      "loss": 1.7473,
+      "step": 6536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8977715895946775e-05,
+      "loss": 1.8802,
+      "step": 6537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8903424032185156e-05,
+      "loss": 1.8661,
+      "step": 6538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8829224689298994e-05,
+      "loss": 1.9679,
+      "step": 6539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8755117881860582e-05,
+      "loss": 1.8159,
+      "step": 6540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.868110362442411e-05,
+      "loss": 1.8931,
+      "step": 6541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8607181931525562e-05,
+      "loss": 1.9165,
+      "step": 6542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8533352817682767e-05,
+      "loss": 1.9875,
+      "step": 6543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.845961629739546e-05,
+      "loss": 1.9745,
+      "step": 6544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8385972385145064e-05,
+      "loss": 1.9806,
+      "step": 6545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8312421095394837e-05,
+      "loss": 2.1433,
+      "step": 6546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8238962442589834e-05,
+      "loss": 1.9854,
+      "step": 6547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.816559644115696e-05,
+      "loss": 1.8962,
+      "step": 6548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8092323105505023e-05,
+      "loss": 1.8904,
+      "step": 6549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8019142450024294e-05,
+      "loss": 1.9067,
+      "step": 6550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.794605448908727e-05,
+      "loss": 1.9179,
+      "step": 6551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.787305923704797e-05,
+      "loss": 2.0433,
+      "step": 6552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.780015670824221e-05,
+      "loss": 1.9681,
+      "step": 6553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.772734691698775e-05,
+      "loss": 1.8181,
+      "step": 6554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7654629877584047e-05,
+      "loss": 2.021,
+      "step": 6555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7582005604312232e-05,
+      "loss": 1.9278,
+      "step": 6556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7509474111435506e-05,
+      "loss": 1.9551,
+      "step": 6557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7437035413198475e-05,
+      "loss": 1.9729,
+      "step": 6558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7364689523827925e-05,
+      "loss": 1.9307,
+      "step": 6559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7292436457532044e-05,
+      "loss": 1.9161,
+      "step": 6560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.722027622850104e-05,
+      "loss": 1.8871,
+      "step": 6561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.714820885090674e-05,
+      "loss": 1.978,
+      "step": 6562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.707623433890277e-05,
+      "loss": 1.9883,
+      "step": 6563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7004352706624602e-05,
+      "loss": 1.9838,
+      "step": 6564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6932563968189493e-05,
+      "loss": 1.8113,
+      "step": 6565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6860868137696225e-05,
+      "loss": 1.8808,
+      "step": 6566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6789265229225536e-05,
+      "loss": 2.013,
+      "step": 6567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6717755256839892e-05,
+      "loss": 1.9999,
+      "step": 6568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6646338234583345e-05,
+      "loss": 1.9936,
+      "step": 6569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6575014176481948e-05,
+      "loss": 1.9708,
+      "step": 6570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6503783096543333e-05,
+      "loss": 1.8197,
+      "step": 6571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6432645008756974e-05,
+      "loss": 1.9265,
+      "step": 6572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6361599927093914e-05,
+      "loss": 1.8728,
+      "step": 6573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6290647865507112e-05,
+      "loss": 1.8829,
+      "step": 6574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6219788837931026e-05,
+      "loss": 1.9334,
+      "step": 6575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6149022858282134e-05,
+      "loss": 2.0734,
+      "step": 6576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6078349940458433e-05,
+      "loss": 1.9487,
+      "step": 6577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6007770098339824e-05,
+      "loss": 1.9232,
+      "step": 6578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.593728334578771e-05,
+      "loss": 1.8672,
+      "step": 6579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.58668896966453e-05,
+      "loss": 1.9001,
+      "step": 6580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5796589164737594e-05,
+      "loss": 1.9209,
+      "step": 6581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5726381763871266e-05,
+      "loss": 1.8946,
+      "step": 6582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.565626750783456e-05,
+      "loss": 2.0556,
+      "step": 6583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5586246410397697e-05,
+      "loss": 1.956,
+      "step": 6584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.551631848531233e-05,
+      "loss": 1.8892,
+      "step": 6585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5446483746311976e-05,
+      "loss": 1.9337,
+      "step": 6586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.537674220711189e-05,
+      "loss": 1.9593,
+      "step": 6587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530709388140878e-05,
+      "loss": 1.9841,
+      "step": 6588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5237538782881376e-05,
+      "loss": 2.0809,
+      "step": 6589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5168076925189743e-05,
+      "loss": 1.9406,
+      "step": 6590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5098708321975915e-05,
+      "loss": 1.8659,
+      "step": 6591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5029432986863608e-05,
+      "loss": 1.946,
+      "step": 6592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4960250933458052e-05,
+      "loss": 1.9506,
+      "step": 6593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4891162175346204e-05,
+      "loss": 2.0519,
+      "step": 6594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4822166726096773e-05,
+      "loss": 1.8451,
+      "step": 6595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4753264599260083e-05,
+      "loss": 1.9101,
+      "step": 6596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.468445580836809e-05,
+      "loss": 1.8938,
+      "step": 6597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4615740366934647e-05,
+      "loss": 1.9263,
+      "step": 6598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4547118288454963e-05,
+      "loss": 1.8917,
+      "step": 6599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.447858958640614e-05,
+      "loss": 1.9851,
+      "step": 6600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4410154274246754e-05,
+      "loss": 2.0134,
+      "step": 6601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.434181236541716e-05,
+      "loss": 2.0673,
+      "step": 6602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4273563873339398e-05,
+      "loss": 1.8474,
+      "step": 6603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.420540881141703e-05,
+      "loss": 1.9616,
+      "step": 6604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4137347193035453e-05,
+      "loss": 1.8392,
+      "step": 6605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.406937903156159e-05,
+      "loss": 1.8976,
+      "step": 6606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.400150434034387e-05,
+      "loss": 1.9381,
+      "step": 6607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3933723132712804e-05,
+      "loss": 1.9906,
+      "step": 6608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.386603542198007e-05,
+      "loss": 1.9881,
+      "step": 6609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3798441221439093e-05,
+      "loss": 1.9936,
+      "step": 6610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3730940544365252e-05,
+      "loss": 1.9695,
+      "step": 6611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3663533404015226e-05,
+      "loss": 1.8813,
+      "step": 6612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3596219813627362e-05,
+      "loss": 2.0355,
+      "step": 6613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3528999786421755e-05,
+      "loss": 1.8423,
+      "step": 6614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3461873335600016e-05,
+      "loss": 1.8874,
+      "step": 6615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3394840474345424e-05,
+      "loss": 2.0342,
+      "step": 6616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3327901215823013e-05,
+      "loss": 1.9613,
+      "step": 6617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.326105557317909e-05,
+      "loss": 1.9048,
+      "step": 6618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.319430355954194e-05,
+      "loss": 1.8109,
+      "step": 6619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.312764518802124e-05,
+      "loss": 2.0355,
+      "step": 6620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3061080471708407e-05,
+      "loss": 1.9509,
+      "step": 6621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2994609423676315e-05,
+      "loss": 1.9547,
+      "step": 6622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.292823205697947e-05,
+      "loss": 1.9992,
+      "step": 6623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.286194838465416e-05,
+      "loss": 1.9595,
+      "step": 6624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.279575841971815e-05,
+      "loss": 1.9199,
+      "step": 6625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2729662175170808e-05,
+      "loss": 1.8222,
+      "step": 6626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2663659663992975e-05,
+      "loss": 1.8347,
+      "step": 6627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.259775089914734e-05,
+      "loss": 1.8809,
+      "step": 6628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2531935893577827e-05,
+      "loss": 1.9721,
+      "step": 6629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2466214660210317e-05,
+      "loss": 1.8205,
+      "step": 6630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2400587211952094e-05,
+      "loss": 1.8782,
+      "step": 6631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2335053561692074e-05,
+      "loss": 1.9208,
+      "step": 6632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2269613722300685e-05,
+      "loss": 1.9255,
+      "step": 6633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2204267706629867e-05,
+      "loss": 1.9031,
+      "step": 6634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.213901552751335e-05,
+      "loss": 1.9396,
+      "step": 6635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2073857197766277e-05,
+      "loss": 1.8827,
+      "step": 6636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2008792730185413e-05,
+      "loss": 1.9212,
+      "step": 6637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1943822137549086e-05,
+      "loss": 1.8031,
+      "step": 6638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.187894543261715e-05,
+      "loss": 1.8876,
+      "step": 6639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.181416262813102e-05,
+      "loss": 1.8276,
+      "step": 6640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1749473736813797e-05,
+      "loss": 1.7627,
+      "step": 6641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1684878771369933e-05,
+      "loss": 1.926,
+      "step": 6642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1620377744485553e-05,
+      "loss": 1.988,
+      "step": 6643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.155597066882836e-05,
+      "loss": 1.8557,
+      "step": 6644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1491657557047517e-05,
+      "loss": 1.9653,
+      "step": 6645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.142743842177386e-05,
+      "loss": 1.8697,
+      "step": 6646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1363313275619633e-05,
+      "loss": 1.8717,
+      "step": 6647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.129928213117871e-05,
+      "loss": 1.7725,
+      "step": 6648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1235345001026363e-05,
+      "loss": 1.8099,
+      "step": 6649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1171501897719546e-05,
+      "loss": 1.9714,
+      "step": 6650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1107752833796733e-05,
+      "loss": 1.9023,
+      "step": 6651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.104409782177802e-05,
+      "loss": 2.0602,
+      "step": 6652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0980536874164736e-05,
+      "loss": 1.9636,
+      "step": 6653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.091707000344001e-05,
+      "loss": 1.8512,
+      "step": 6654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0853697222068368e-05,
+      "loss": 1.9541,
+      "step": 6655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.079041854249586e-05,
+      "loss": 1.8815,
+      "step": 6656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0727233977150094e-05,
+      "loss": 1.9724,
+      "step": 6657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0664143538440263e-05,
+      "loss": 1.9563,
+      "step": 6658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0601147238756945e-05,
+      "loss": 1.8157,
+      "step": 6659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0538245090472252e-05,
+      "loss": 1.8561,
+      "step": 6660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0475437105939798e-05,
+      "loss": 2.0359,
+      "step": 6661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0412723297494883e-05,
+      "loss": 2.053,
+      "step": 6662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0350103677454046e-05,
+      "loss": 1.8815,
+      "step": 6663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0287578258115447e-05,
+      "loss": 1.9329,
+      "step": 6664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0225147051758876e-05,
+      "loss": 1.9459,
+      "step": 6665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0162810070645353e-05,
+      "loss": 1.8967,
+      "step": 6666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.010056732701754e-05,
+      "loss": 1.9541,
+      "step": 6667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.003841883309965e-05,
+      "loss": 1.9463,
+      "step": 6668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.997636460109731e-05,
+      "loss": 1.8523,
+      "step": 6669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9914404643197604e-05,
+      "loss": 1.8505,
+      "step": 6670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9852538971569133e-05,
+      "loss": 1.8065,
+      "step": 6671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9790767598362014e-05,
+      "loss": 2.0095,
+      "step": 6672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.972909053570787e-05,
+      "loss": 1.9714,
+      "step": 6673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9667507795719685e-05,
+      "loss": 1.8302,
+      "step": 6674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9606019390492e-05,
+      "loss": 1.9497,
+      "step": 6675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9544625332100718e-05,
+      "loss": 1.866,
+      "step": 6676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9483325632603523e-05,
+      "loss": 1.9195,
+      "step": 6677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9422120304039115e-05,
+      "loss": 1.9658,
+      "step": 6678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.936100935842805e-05,
+      "loss": 1.8349,
+      "step": 6679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.929999280777217e-05,
+      "loss": 1.8504,
+      "step": 6680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9239070664054835e-05,
+      "loss": 1.9419,
+      "step": 6681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9178242939240697e-05,
+      "loss": 1.7926,
+      "step": 6682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.911750964527609e-05,
+      "loss": 1.8809,
+      "step": 6683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9056870794088687e-05,
+      "loss": 1.9456,
+      "step": 6684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8996326397587695e-05,
+      "loss": 1.9299,
+      "step": 6685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8935876467663703e-05,
+      "loss": 1.9956,
+      "step": 6686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.887552101618867e-05,
+      "loss": 1.7925,
+      "step": 6687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.881526005501616e-05,
+      "loss": 1.8089,
+      "step": 6688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.87550935959811e-05,
+      "loss": 1.6575,
+      "step": 6689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8695021650899813e-05,
+      "loss": 1.9802,
+      "step": 6690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8635044231570253e-05,
+      "loss": 1.8738,
+      "step": 6691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8575161349771496e-05,
+      "loss": 2.0309,
+      "step": 6692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.851537301726436e-05,
+      "loss": 1.897,
+      "step": 6693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8455679245790834e-05,
+      "loss": 1.8633,
+      "step": 6694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8396080047074547e-05,
+      "loss": 1.9673,
+      "step": 6695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8336575432820414e-05,
+      "loss": 1.8269,
+      "step": 6696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.827716541471486e-05,
+      "loss": 1.9843,
+      "step": 6697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8217850004425717e-05,
+      "loss": 1.897,
+      "step": 6698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8158629213602118e-05,
+      "loss": 2.0069,
+      "step": 6699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8099503053874867e-05,
+      "loss": 1.8014,
+      "step": 6700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8040471536855894e-05,
+      "loss": 1.9037,
+      "step": 6701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7981534674138822e-05,
+      "loss": 1.9436,
+      "step": 6702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.792269247729833e-05,
+      "loss": 1.9481,
+      "step": 6703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.78639449578909e-05,
+      "loss": 1.9513,
+      "step": 6704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7805292127454076e-05,
+      "loss": 1.9075,
+      "step": 6705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.774673399750715e-05,
+      "loss": 1.8654,
+      "step": 6706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7688270579550526e-05,
+      "loss": 1.9451,
+      "step": 6707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7629901885066134e-05,
+      "loss": 1.8801,
+      "step": 6708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7571627925517197e-05,
+      "loss": 1.9728,
+      "step": 6709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7513448712348446e-05,
+      "loss": 1.7914,
+      "step": 6710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7455364256986016e-05,
+      "loss": 1.9432,
+      "step": 6711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7397374570837454e-05,
+      "loss": 1.8713,
+      "step": 6712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.733947966529148e-05,
+      "loss": 1.9404,
+      "step": 6713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7281679551718444e-05,
+      "loss": 2.0593,
+      "step": 6714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7223974241469932e-05,
+      "loss": 2.02,
+      "step": 6715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7166363745878932e-05,
+      "loss": 1.859,
+      "step": 6716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7108848076259897e-05,
+      "loss": 1.8763,
+      "step": 6717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7051427243908624e-05,
+      "loss": 1.9655,
+      "step": 6718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6994101260102256e-05,
+      "loss": 1.9011,
+      "step": 6719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.693687013609929e-05,
+      "loss": 2.0428,
+      "step": 6720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6879733883139513e-05,
+      "loss": 1.9406,
+      "step": 6721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6822692512444337e-05,
+      "loss": 1.9358,
+      "step": 6722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6765746035216357e-05,
+      "loss": 1.8918,
+      "step": 6723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6708894462639457e-05,
+      "loss": 1.9944,
+      "step": 6724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6652137805879154e-05,
+      "loss": 1.7996,
+      "step": 6725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.659547607608197e-05,
+      "loss": 1.9283,
+      "step": 6726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6538909284376126e-05,
+      "loss": 1.817,
+      "step": 6727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6482437441870947e-05,
+      "loss": 1.839,
+      "step": 6728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.642606055965723e-05,
+      "loss": 1.9329,
+      "step": 6729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6369778648807065e-05,
+      "loss": 1.9229,
+      "step": 6730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6313591720374055e-05,
+      "loss": 1.9743,
+      "step": 6731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625749978539276e-05,
+      "loss": 1.959,
+      "step": 6732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6201502854879646e-05,
+      "loss": 1.9237,
+      "step": 6733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6145600939831973e-05,
+      "loss": 2.0271,
+      "step": 6734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6089794051228736e-05,
+      "loss": 1.9936,
+      "step": 6735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6034082200030053e-05,
+      "loss": 1.8614,
+      "step": 6736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5978465397177344e-05,
+      "loss": 2.0153,
+      "step": 6737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5922943653593537e-05,
+      "loss": 1.88,
+      "step": 6738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.586751698018285e-05,
+      "loss": 1.9726,
+      "step": 6739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5812185387830803e-05,
+      "loss": 1.9553,
+      "step": 6740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5756948887404143e-05,
+      "loss": 1.8744,
+      "step": 6741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5701807489751084e-05,
+      "loss": 2.0042,
+      "step": 6742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.564676120570102e-05,
+      "loss": 2.0057,
+      "step": 6743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5591810046064802e-05,
+      "loss": 1.7967,
+      "step": 6744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5536954021634575e-05,
+      "loss": 1.8913,
+      "step": 6745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.548219314318383e-05,
+      "loss": 2.001,
+      "step": 6746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.542752742146719e-05,
+      "loss": 1.9602,
+      "step": 6747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5372956867220676e-05,
+      "loss": 1.8742,
+      "step": 6748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5318481491161884e-05,
+      "loss": 1.8623,
+      "step": 6749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5264101303989253e-05,
+      "loss": 1.8264,
+      "step": 6750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5209816316382852e-05,
+      "loss": 1.8058,
+      "step": 6751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5155626539004041e-05,
+      "loss": 2.0301,
+      "step": 6752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5101531982495309e-05,
+      "loss": 1.9737,
+      "step": 6753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.504753265748049e-05,
+      "loss": 1.9059,
+      "step": 6754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4993628574564933e-05,
+      "loss": 1.8166,
+      "step": 6755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4939819744335004e-05,
+      "loss": 1.8791,
+      "step": 6756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4886106177358471e-05,
+      "loss": 1.8711,
+      "step": 6757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4832487884184453e-05,
+      "loss": 1.7895,
+      "step": 6758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4778964875343192e-05,
+      "loss": 1.8409,
+      "step": 6759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4725537161346503e-05,
+      "loss": 1.9757,
+      "step": 6760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4672204752687157e-05,
+      "loss": 1.9607,
+      "step": 6761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4618967659839389e-05,
+      "loss": 1.9193,
+      "step": 6762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4565825893258666e-05,
+      "loss": 1.9498,
+      "step": 6763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4512779463381808e-05,
+      "loss": 1.9443,
+      "step": 6764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.445982838062676e-05,
+      "loss": 1.8218,
+      "step": 6765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.440697265539298e-05,
+      "loss": 1.9157,
+      "step": 6766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4354212298060943e-05,
+      "loss": 1.9617,
+      "step": 6767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4301547318992525e-05,
+      "loss": 1.8066,
+      "step": 6768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4248977728530843e-05,
+      "loss": 2.0003,
+      "step": 6769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4196503537000305e-05,
+      "loss": 1.9473,
+      "step": 6770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.41441247547065e-05,
+      "loss": 1.8514,
+      "step": 6771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4091841391936478e-05,
+      "loss": 1.8528,
+      "step": 6772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4039653458958301e-05,
+      "loss": 1.958,
+      "step": 6773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3987560966021495e-05,
+      "loss": 1.8792,
+      "step": 6774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3935563923356597e-05,
+      "loss": 1.9551,
+      "step": 6775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3883662341175718e-05,
+      "loss": 1.8439,
+      "step": 6776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3831856229671923e-05,
+      "loss": 1.9164,
+      "step": 6777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.37801455990198e-05,
+      "loss": 1.8383,
+      "step": 6778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3728530459374944e-05,
+      "loss": 1.8042,
+      "step": 6779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3677010820874304e-05,
+      "loss": 1.8026,
+      "step": 6780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3625586693636004e-05,
+      "loss": 1.8869,
+      "step": 6781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3574258087759639e-05,
+      "loss": 1.9476,
+      "step": 6782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3523025013325752e-05,
+      "loss": 1.9141,
+      "step": 6783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3471887480396239e-05,
+      "loss": 1.9724,
+      "step": 6784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3420845499014289e-05,
+      "loss": 1.989,
+      "step": 6785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3369899079204273e-05,
+      "loss": 1.8978,
+      "step": 6786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3319048230971797e-05,
+      "loss": 1.9428,
+      "step": 6787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3268292964303708e-05,
+      "loss": 2.0394,
+      "step": 6788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.321763328916803e-05,
+      "loss": 1.8529,
+      "step": 6789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3167069215514027e-05,
+      "loss": 1.8475,
+      "step": 6790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3116600753272367e-05,
+      "loss": 1.9261,
+      "step": 6791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3066227912354623e-05,
+      "loss": 1.8927,
+      "step": 6792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3015950702653822e-05,
+      "loss": 1.7924,
+      "step": 6793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2965769134044181e-05,
+      "loss": 1.9145,
+      "step": 6794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2915683216381091e-05,
+      "loss": 1.8137,
+      "step": 6795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2865692959501075e-05,
+      "loss": 1.822,
+      "step": 6796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2815798373221942e-05,
+      "loss": 1.7924,
+      "step": 6797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.276599946734286e-05,
+      "loss": 2.0063,
+      "step": 6798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2716296251644e-05,
+      "loss": 1.9746,
+      "step": 6799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2666688735886777e-05,
+      "loss": 1.8544,
+      "step": 6800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2617176929813956e-05,
+      "loss": 1.9498,
+      "step": 6801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.25677608431492e-05,
+      "loss": 1.8998,
+      "step": 6802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2518440485597749e-05,
+      "loss": 1.8135,
+      "step": 6803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2469215866845795e-05,
+      "loss": 1.9258,
+      "step": 6804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2420086996560776e-05,
+      "loss": 1.9233,
+      "step": 6805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.237105388439136e-05,
+      "loss": 1.9864,
+      "step": 6806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2322116539967453e-05,
+      "loss": 1.9428,
+      "step": 6807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2273274972899928e-05,
+      "loss": 1.8636,
+      "step": 6808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2224529192781164e-05,
+      "loss": 1.9373,
+      "step": 6809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2175879209184504e-05,
+      "loss": 1.9677,
+      "step": 6810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2127325031664526e-05,
+      "loss": 1.8327,
+      "step": 6811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2078866669757104e-05,
+      "loss": 2.0186,
+      "step": 6812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.203050413297907e-05,
+      "loss": 2.085,
+      "step": 6813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1982237430828768e-05,
+      "loss": 2.0303,
+      "step": 6814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1934066572785341e-05,
+      "loss": 1.8294,
+      "step": 6815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1885991568309385e-05,
+      "loss": 1.8295,
+      "step": 6816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1838012426842459e-05,
+      "loss": 1.8057,
+      "step": 6817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1790129157807583e-05,
+      "loss": 1.8973,
+      "step": 6818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1742341770608678e-05,
+      "loss": 1.9225,
+      "step": 6819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1694650274630958e-05,
+      "loss": 1.8548,
+      "step": 6820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1647054679240765e-05,
+      "loss": 1.8868,
+      "step": 6821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1599554993785677e-05,
+      "loss": 2.0353,
+      "step": 6822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1552151227594287e-05,
+      "loss": 2.0215,
+      "step": 6823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1504843389976427e-05,
+      "loss": 1.8653,
+      "step": 6824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1457631490223218e-05,
+      "loss": 1.8313,
+      "step": 6825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1410515537606802e-05,
+      "loss": 1.8434,
+      "step": 6826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1363495541380497e-05,
+      "loss": 1.8514,
+      "step": 6827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.131657151077875e-05,
+      "loss": 1.9857,
+      "step": 6828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.126974345501719e-05,
+      "loss": 1.8888,
+      "step": 6829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1223011383292625e-05,
+      "loss": 1.7787,
+      "step": 6830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1176375304782938e-05,
+      "loss": 1.8662,
+      "step": 6831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.11298352286473e-05,
+      "loss": 1.8625,
+      "step": 6832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1083391164025902e-05,
+      "loss": 1.9628,
+      "step": 6833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1037043120040058e-05,
+      "loss": 1.834,
+      "step": 6834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.099079110579232e-05,
+      "loss": 1.9193,
+      "step": 6835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0944635130366364e-05,
+      "loss": 1.9347,
+      "step": 6836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0898575202826943e-05,
+      "loss": 1.8868,
+      "step": 6837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.085261133221993e-05,
+      "loss": 2.0113,
+      "step": 6838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0806743527572493e-05,
+      "loss": 2.0368,
+      "step": 6839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0760971797892761e-05,
+      "loss": 1.9101,
+      "step": 6840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0715296152170095e-05,
+      "loss": 1.8498,
+      "step": 6841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0669716599374934e-05,
+      "loss": 1.9409,
+      "step": 6842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0624233148458895e-05,
+      "loss": 1.8795,
+      "step": 6843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0578845808354554e-05,
+      "loss": 1.966,
+      "step": 6844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0533554587975946e-05,
+      "loss": 1.8594,
+      "step": 6845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0488359496217847e-05,
+      "loss": 1.866,
+      "step": 6846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0443260541956434e-05,
+      "loss": 1.929,
+      "step": 6847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0398257734048899e-05,
+      "loss": 1.9986,
+      "step": 6848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0353351081333561e-05,
+      "loss": 1.8509,
+      "step": 6849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0308540592629757e-05,
+      "loss": 1.8821,
+      "step": 6850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0263826276738164e-05,
+      "loss": 2.0304,
+      "step": 6851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0219208142440317e-05,
+      "loss": 1.7575,
+      "step": 6852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.017468619849904e-05,
+      "loss": 1.9815,
+      "step": 6853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0130260453658225e-05,
+      "loss": 1.8502,
+      "step": 6854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0085930916642838e-05,
+      "loss": 1.8833,
+      "step": 6855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0041697596158972e-05,
+      "loss": 1.8356,
+      "step": 6856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.997560500893732e-06,
+      "loss": 1.9682,
+      "step": 6857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.953519639515519e-06,
+      "loss": 1.9429,
+      "step": 6858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.90957502067369e-06,
+      "loss": 1.9138,
+      "step": 6859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.86572665299873e-06,
+      "loss": 2.0239,
+      "step": 6860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.821974545102252e-06,
+      "loss": 1.8858,
+      "step": 6861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.778318705576883e-06,
+      "loss": 1.9218,
+      "step": 6862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.73475914299643e-06,
+      "loss": 1.7473,
+      "step": 6863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.691295865915773e-06,
+      "loss": 2.0386,
+      "step": 6864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.64792888287086e-06,
+      "loss": 1.9343,
+      "step": 6865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.604658202378768e-06,
+      "loss": 1.9113,
+      "step": 6866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.561483832937535e-06,
+      "loss": 1.8945,
+      "step": 6867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.518405783026485e-06,
+      "loss": 1.993,
+      "step": 6868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.475424061105909e-06,
+      "loss": 1.8286,
+      "step": 6869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.432538675617108e-06,
+      "loss": 1.8927,
+      "step": 6870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.389749634982569e-06,
+      "loss": 1.7727,
+      "step": 6871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.347056947605903e-06,
+      "loss": 1.8757,
+      "step": 6872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.30446062187168e-06,
+      "loss": 1.867,
+      "step": 6873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.261960666145596e-06,
+      "loss": 1.8934,
+      "step": 6874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.219557088774421e-06,
+      "loss": 1.9971,
+      "step": 6875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.177249898085993e-06,
+      "loss": 1.8668,
+      "step": 6876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.135039102389165e-06,
+      "loss": 1.9313,
+      "step": 6877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.092924709973971e-06,
+      "loss": 2.069,
+      "step": 6878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.050906729111519e-06,
+      "loss": 1.9788,
+      "step": 6879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.008985168053819e-06,
+      "loss": 1.8701,
+      "step": 6880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.96716003503406e-06,
+      "loss": 1.8707,
+      "step": 6881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.92543133826651e-06,
+      "loss": 1.8019,
+      "step": 6882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.883799085946443e-06,
+      "loss": 1.7966,
+      "step": 6883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.842263286250208e-06,
+      "loss": 1.9699,
+      "step": 6884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.800823947335169e-06,
+      "loss": 1.9129,
+      "step": 6885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.759481077339927e-06,
+      "loss": 2.0223,
+      "step": 6886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.718234684383875e-06,
+      "loss": 1.9373,
+      "step": 6887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.6770847765677e-06,
+      "loss": 1.8558,
+      "step": 6888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.636031361972884e-06,
+      "loss": 1.8056,
+      "step": 6889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.595074448662199e-06,
+      "loss": 1.8582,
+      "step": 6890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.554214044679321e-06,
+      "loss": 2.0216,
+      "step": 6891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.513450158049108e-06,
+      "loss": 1.9842,
+      "step": 6892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.472782796777268e-06,
+      "loss": 1.8681,
+      "step": 6893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.432211968850744e-06,
+      "loss": 2.0676,
+      "step": 6894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.391737682237277e-06,
+      "loss": 2.0043,
+      "step": 6895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.351359944886006e-06,
+      "loss": 1.8662,
+      "step": 6896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.311078764726755e-06,
+      "loss": 1.8295,
+      "step": 6897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.270894149670582e-06,
+      "loss": 1.7661,
+      "step": 6898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.230806107609512e-06,
+      "loss": 1.876,
+      "step": 6899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.190814646416688e-06,
+      "loss": 1.879,
+      "step": 6900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.150919773946164e-06,
+      "loss": 2.0432,
+      "step": 6901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.111121498033113e-06,
+      "loss": 1.8065,
+      "step": 6902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.07141982649373e-06,
+      "loss": 2.0229,
+      "step": 6903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.031814767125112e-06,
+      "loss": 1.8731,
+      "step": 6904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.992306327705534e-06,
+      "loss": 1.8026,
+      "step": 6905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.95289451599429e-06,
+      "loss": 2.0604,
+      "step": 6906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.913579339731636e-06,
+      "loss": 2.0131,
+      "step": 6907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.874360806638836e-06,
+      "loss": 1.8486,
+      "step": 6908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.835238924418231e-06,
+      "loss": 1.8461,
+      "step": 6909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.79621370075312e-06,
+      "loss": 2.0728,
+      "step": 6910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.757285143307814e-06,
+      "loss": 1.8651,
+      "step": 6911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.718453259727699e-06,
+      "loss": 1.8477,
+      "step": 6912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.679718057639284e-06,
+      "loss": 1.9298,
+      "step": 6913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.64107954464982e-06,
+      "loss": 2.0118,
+      "step": 6914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.602537728347736e-06,
+      "loss": 1.9845,
+      "step": 6915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.56409261630242e-06,
+      "loss": 1.9424,
+      "step": 6916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.525744216064334e-06,
+      "loss": 1.8586,
+      "step": 6917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.4874925351648415e-06,
+      "loss": 1.9882,
+      "step": 6918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.449337581116489e-06,
+      "loss": 1.9791,
+      "step": 6919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.411279361412615e-06,
+      "loss": 2.0701,
+      "step": 6920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.37331788352763e-06,
+      "loss": 1.8533,
+      "step": 6921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.33545315491696e-06,
+      "loss": 1.9532,
+      "step": 6922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.2976851830171556e-06,
+      "loss": 1.8792,
+      "step": 6923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.260013975245561e-06,
+      "loss": 1.8578,
+      "step": 6924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.222439539000536e-06,
+      "loss": 1.8956,
+      "step": 6925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.1849618816616224e-06,
+      "loss": 1.8424,
+      "step": 6926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.14758101058921e-06,
+      "loss": 1.9401,
+      "step": 6927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.1102969331246495e-06,
+      "loss": 1.9771,
+      "step": 6928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.073109656590415e-06,
+      "loss": 1.9602,
+      "step": 6929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.0360191882898324e-06,
+      "loss": 1.9076,
+      "step": 6930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.999025535507242e-06,
+      "loss": 1.9336,
+      "step": 6931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.962128705508053e-06,
+      "loss": 1.8333,
+      "step": 6932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.925328705538636e-06,
+      "loss": 1.7702,
+      "step": 6933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.8886255428262655e-06,
+      "loss": 1.8699,
+      "step": 6934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.852019224579287e-06,
+      "loss": 1.8894,
+      "step": 6935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.815509757986948e-06,
+      "loss": 1.8967,
+      "step": 6936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.7790971502195154e-06,
+      "loss": 2.0144,
+      "step": 6937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.742781408428267e-06,
+      "loss": 1.9454,
+      "step": 6938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.7065625397453314e-06,
+      "loss": 1.8154,
+      "step": 6939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.670440551284074e-06,
+      "loss": 1.8204,
+      "step": 6940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.634415450138543e-06,
+      "loss": 1.9734,
+      "step": 6941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.598487243383855e-06,
+      "loss": 1.9507,
+      "step": 6942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.562655938076201e-06,
+      "loss": 1.8515,
+      "step": 6943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.526921541252618e-06,
+      "loss": 1.8219,
+      "step": 6944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.491284059931102e-06,
+      "loss": 1.845,
+      "step": 6945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.4557435011107775e-06,
+      "loss": 1.9304,
+      "step": 6946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.4202998717715625e-06,
+      "loss": 1.9352,
+      "step": 6947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.384953178874386e-06,
+      "loss": 2.0006,
+      "step": 6948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.349703429361142e-06,
+      "loss": 1.9671,
+      "step": 6949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.314550630154791e-06,
+      "loss": 1.882,
+      "step": 6950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.279494788159035e-06,
+      "loss": 1.7786,
+      "step": 6951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2445359102586975e-06,
+      "loss": 1.8066,
+      "step": 6952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.209674003319565e-06,
+      "loss": 1.8955,
+      "step": 6953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.174909074188217e-06,
+      "loss": 1.774,
+      "step": 6954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.1402411296924694e-06,
+      "loss": 1.8039,
+      "step": 6955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.105670176640765e-06,
+      "loss": 1.6669,
+      "step": 6956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.071196221822783e-06,
+      "loss": 1.9641,
+      "step": 6957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.036819272008886e-06,
+      "loss": 1.9411,
+      "step": 6958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.002539333950618e-06,
+      "loss": 1.8482,
+      "step": 6959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.968356414380371e-06,
+      "loss": 1.8945,
+      "step": 6960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9342705200114975e-06,
+      "loss": 1.847,
+      "step": 6961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.900281657538309e-06,
+      "loss": 1.9733,
+      "step": 6962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.866389833635965e-06,
+      "loss": 1.9337,
+      "step": 6963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.832595054960643e-06,
+      "loss": 1.9056,
+      "step": 6964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.798897328149477e-06,
+      "loss": 1.7699,
+      "step": 6965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.765296659820618e-06,
+      "loss": 1.8831,
+      "step": 6966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7317930565728985e-06,
+      "loss": 1.8277,
+      "step": 6967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.698386524986388e-06,
+      "loss": 1.8731,
+      "step": 6968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.665077071621894e-06,
+      "loss": 1.9418,
+      "step": 6969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.631864703021183e-06,
+      "loss": 1.9588,
+      "step": 6970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.598749425707039e-06,
+      "loss": 1.9794,
+      "step": 6971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.565731246183037e-06,
+      "loss": 1.8376,
+      "step": 6972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.532810170933933e-06,
+      "loss": 1.9491,
+      "step": 6973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.499986206425167e-06,
+      "loss": 1.9013,
+      "step": 6974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.467259359103194e-06,
+      "loss": 1.8653,
+      "step": 6975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.434629635395316e-06,
+      "loss": 1.8375,
+      "step": 6976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.402097041709964e-06,
+      "loss": 1.9126,
+      "step": 6977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.369661584436247e-06,
+      "loss": 1.9438,
+      "step": 6978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3373232699444605e-06,
+      "loss": 2.0162,
+      "step": 6979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.305082104585579e-06,
+      "loss": 1.8247,
+      "step": 6980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.272938094691648e-06,
+      "loss": 1.8876,
+      "step": 6981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.240891246575508e-06,
+      "loss": 1.9416,
+      "step": 6982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.208941566531067e-06,
+      "loss": 1.8408,
+      "step": 6983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.177089060833029e-06,
+      "loss": 1.8652,
+      "step": 6984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1453337357370565e-06,
+      "loss": 1.9783,
+      "step": 6985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.113675597479772e-06,
+      "loss": 1.884,
+      "step": 6986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.082114652278591e-06,
+      "loss": 1.85,
+      "step": 6987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.050650906332e-06,
+      "loss": 1.8211,
+      "step": 6988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.019284365819277e-06,
+      "loss": 1.7588,
+      "step": 6989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.988015036900661e-06,
+      "loss": 2.0354,
+      "step": 6990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9568429257171865e-06,
+      "loss": 1.9126,
+      "step": 6991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.92576803839101e-06,
+      "loss": 1.8136,
+      "step": 6992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.894790381025027e-06,
+      "loss": 1.8795,
+      "step": 6993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8639099597030945e-06,
+      "loss": 1.8641,
+      "step": 6994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.833126780489972e-06,
+      "loss": 1.8302,
+      "step": 6995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.802440849431322e-06,
+      "loss": 1.8032,
+      "step": 6996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.771852172553659e-06,
+      "loss": 1.8657,
+      "step": 6997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.741360755864455e-06,
+      "loss": 1.8726,
+      "step": 6998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.71096660535203e-06,
+      "loss": 1.9249,
+      "step": 6999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.680669726985665e-06,
+      "loss": 1.9066,
+      "step": 7000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.650470126715545e-06,
+      "loss": 1.9387,
+      "step": 7001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6203678104727034e-06,
+      "loss": 1.9487,
+      "step": 7002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.590362784169022e-06,
+      "loss": 2.0153,
+      "step": 7003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.560455053697343e-06,
+      "loss": 1.9357,
+      "step": 7004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.530644624931412e-06,
+      "loss": 1.8159,
+      "step": 7005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.500931503725825e-06,
+      "loss": 2.0223,
+      "step": 7006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4713156959160786e-06,
+      "loss": 1.8086,
+      "step": 7007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.441797207318576e-06,
+      "loss": 1.9243,
+      "step": 7008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.41237604373057e-06,
+      "loss": 1.8806,
+      "step": 7009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.383052210930271e-06,
+      "loss": 1.9293,
+      "step": 7010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.353825714676685e-06,
+      "loss": 1.9567,
+      "step": 7011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3246965607097176e-06,
+      "loss": 1.7306,
+      "step": 7012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.295664754750184e-06,
+      "loss": 1.8472,
+      "step": 7013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2667303024998e-06,
+      "loss": 1.8849,
+      "step": 7014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.237893209641186e-06,
+      "loss": 1.9104,
+      "step": 7015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2091534818377575e-06,
+      "loss": 1.9008,
+      "step": 7016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.180511124733832e-06,
+      "loss": 1.8948,
+      "step": 7017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.151966143954577e-06,
+      "loss": 1.9653,
+      "step": 7018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.12351854510612e-06,
+      "loss": 1.9241,
+      "step": 7019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0951683337754345e-06,
+      "loss": 1.911,
+      "step": 7020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.066915515530345e-06,
+      "loss": 1.896,
+      "step": 7021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0387600959195804e-06,
+      "loss": 1.9582,
+      "step": 7022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.010702080472662e-06,
+      "loss": 1.7969,
+      "step": 7023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.982741474700014e-06,
+      "loss": 1.919,
+      "step": 7024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954878284093022e-06,
+      "loss": 1.9709,
+      "step": 7025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9271125141238095e-06,
+      "loss": 1.825,
+      "step": 7026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8994441702455144e-06,
+      "loss": 1.8955,
+      "step": 7027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.871873257891955e-06,
+      "loss": 1.9128,
+      "step": 7028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.844399782478025e-06,
+      "loss": 1.9676,
+      "step": 7029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.817023749399184e-06,
+      "loss": 1.9115,
+      "step": 7030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7897451640321325e-06,
+      "loss": 1.7648,
+      "step": 7031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7625640317341394e-06,
+      "loss": 1.8344,
+      "step": 7032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.735480357843435e-06,
+      "loss": 1.8569,
+      "step": 7033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.708494147679209e-06,
+      "loss": 1.9211,
+      "step": 7034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6816054065412773e-06,
+      "loss": 1.9835,
+      "step": 7035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6548141397104716e-06,
+      "loss": 1.9688,
+      "step": 7036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.628120352448583e-06,
+      "loss": 1.9807,
+      "step": 7037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6015240499979175e-06,
+      "loss": 1.8223,
+      "step": 7038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5750252375820747e-06,
+      "loss": 2.0611,
+      "step": 7039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.548623920405114e-06,
+      "loss": 1.8744,
+      "step": 7040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5223201036521657e-06,
+      "loss": 1.9349,
+      "step": 7041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4961137924892085e-06,
+      "loss": 1.998,
+      "step": 7042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4700049920629584e-06,
+      "loss": 1.9459,
+      "step": 7043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.443993707501036e-06,
+      "loss": 1.7619,
+      "step": 7044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.418079943911967e-06,
+      "loss": 1.8475,
+      "step": 7045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.392263706385068e-06,
+      "loss": 1.9001,
+      "step": 7046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.366544999990506e-06,
+      "loss": 1.8329,
+      "step": 7047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.340923829779352e-06,
+      "loss": 1.8594,
+      "step": 7048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3154002007833584e-06,
+      "loss": 1.9523,
+      "step": 7049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.289974118015293e-06,
+      "loss": 1.7657,
+      "step": 7050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2646455864687176e-06,
+      "loss": 1.8695,
+      "step": 7051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2394146111179858e-06,
+      "loss": 1.9231,
+      "step": 7052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.214281196918356e-06,
+      "loss": 1.8618,
+      "step": 7053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1892453488058803e-06,
+      "loss": 1.7983,
+      "step": 7054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1643070716974586e-06,
+      "loss": 1.9023,
+      "step": 7055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1394663704908954e-06,
+      "loss": 1.825,
+      "step": 7056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.114723250064677e-06,
+      "loss": 1.8401,
+      "step": 7057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0900777152783054e-06,
+      "loss": 1.986,
+      "step": 7058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0655297709720198e-06,
+      "loss": 1.912,
+      "step": 7059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0410794219668523e-06,
+      "loss": 1.9683,
+      "step": 7060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.016726673064796e-06,
+      "loss": 1.9058,
+      "step": 7061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9924715290485237e-06,
+      "loss": 1.9845,
+      "step": 7062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9683139946816707e-06,
+      "loss": 1.9547,
+      "step": 7063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.944254074708608e-06,
+      "loss": 2.1124,
+      "step": 7064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.920291773854611e-06,
+      "loss": 1.8639,
+      "step": 7065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8964270968258046e-06,
+      "loss": 1.9878,
+      "step": 7066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.872660048308995e-06,
+      "loss": 1.9655,
+      "step": 7067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.848990632971893e-06,
+      "loss": 1.9018,
+      "step": 7068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.825418855463113e-06,
+      "loss": 2.0207,
+      "step": 7069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8019447204120085e-06,
+      "loss": 1.9636,
+      "step": 7070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.77856823242878e-06,
+      "loss": 1.9082,
+      "step": 7071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7552893961043657e-06,
+      "loss": 2.0033,
+      "step": 7072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7321082160107203e-06,
+      "loss": 1.9038,
+      "step": 7073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.709024696700424e-06,
+      "loss": 2.1055,
+      "step": 7074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6860388427070725e-06,
+      "loss": 1.8805,
+      "step": 7075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.663150658544833e-06,
+      "loss": 1.8605,
+      "step": 7076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.640360148708887e-06,
+      "loss": 1.9705,
+      "step": 7077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.617667317675154e-06,
+      "loss": 1.8266,
+      "step": 7078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.595072169900348e-06,
+      "loss": 1.7372,
+      "step": 7079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5725747098221397e-06,
+      "loss": 1.897,
+      "step": 7080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5501749418588294e-06,
+      "loss": 1.8801,
+      "step": 7081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5278728704096756e-06,
+      "loss": 2.0932,
+      "step": 7082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5056684998546187e-06,
+      "loss": 1.9487,
+      "step": 7083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4835618345545043e-06,
+      "loss": 1.9597,
+      "step": 7084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.46155287885097e-06,
+      "loss": 1.8347,
+      "step": 7085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.439641637066392e-06,
+      "loss": 1.9426,
+      "step": 7086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4178281135041614e-06,
+      "loss": 1.9752,
+      "step": 7087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3961123124482397e-06,
+      "loss": 2.0034,
+      "step": 7088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3744942381635493e-06,
+      "loss": 1.9146,
+      "step": 7089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3529738948956934e-06,
+      "loss": 1.9131,
+      "step": 7090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3315512868712364e-06,
+      "loss": 1.9131,
+      "step": 7091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.310226418297423e-06,
+      "loss": 1.8986,
+      "step": 7092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.288999293362293e-06,
+      "loss": 1.9651,
+      "step": 7093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2678699162348438e-06,
+      "loss": 1.9087,
+      "step": 7094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2468382910647014e-06,
+      "loss": 2.0645,
+      "step": 7095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2259044219823943e-06,
+      "loss": 1.9741,
+      "step": 7096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2050683130992456e-06,
+      "loss": 1.8771,
+      "step": 7097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.184329968507259e-06,
+      "loss": 1.8795,
+      "step": 7098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.163689392279511e-06,
+      "loss": 1.8158,
+      "step": 7099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1431465884695356e-06,
+      "loss": 1.997,
+      "step": 7100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.122701561111884e-06,
+      "loss": 1.8446,
+      "step": 7101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.102354314221844e-06,
+      "loss": 2.0127,
+      "step": 7102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0821048517955523e-06,
+      "loss": 1.9264,
+      "step": 7103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.061953177809883e-06,
+      "loss": 1.7942,
+      "step": 7104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0418992962224494e-06,
+      "loss": 1.9566,
+      "step": 7105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0219432109718216e-06,
+      "loss": 1.8007,
+      "step": 7106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0020849259771987e-06,
+      "loss": 1.7845,
+      "step": 7107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9823244451386835e-06,
+      "loss": 1.9289,
+      "step": 7108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.962661772337171e-06,
+      "loss": 1.7452,
+      "step": 7109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9430969114341836e-06,
+      "loss": 1.8185,
+      "step": 7110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9236298662722586e-06,
+      "loss": 1.9342,
+      "step": 7111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.90426064067456e-06,
+      "loss": 1.9075,
+      "step": 7112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8849892384451561e-06,
+      "loss": 1.7947,
+      "step": 7113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8658156633687973e-06,
+      "loss": 1.818,
+      "step": 7114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8467399192111379e-06,
+      "loss": 1.9673,
+      "step": 7115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8277620097184589e-06,
+      "loss": 1.8029,
+      "step": 7116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.808881938618001e-06,
+      "loss": 1.8045,
+      "step": 7117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7900997096176875e-06,
+      "loss": 1.94,
+      "step": 7118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7714153264062338e-06,
+      "loss": 1.9285,
+      "step": 7119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7528287926531495e-06,
+      "loss": 2.0376,
+      "step": 7120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.734340112008792e-06,
+      "loss": 1.8614,
+      "step": 7121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7159492881041461e-06,
+      "loss": 1.9202,
+      "step": 7122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6976563245511557e-06,
+      "loss": 1.9194,
+      "step": 7123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6794612249424469e-06,
+      "loss": 1.8829,
+      "step": 7124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6613639928513834e-06,
+      "loss": 1.9428,
+      "step": 7125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6433646318321782e-06,
+      "loss": 1.8691,
+      "step": 7126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.625463145419892e-06,
+      "loss": 1.9304,
+      "step": 7127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6076595371302128e-06,
+      "loss": 1.9033,
+      "step": 7128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5899538104596767e-06,
+      "loss": 1.9842,
+      "step": 7129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723459688856135e-06,
+      "loss": 1.829,
+      "step": 7130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5548360158661456e-06,
+      "loss": 2.0489,
+      "step": 7131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5374239548400226e-06,
+      "loss": 1.935,
+      "step": 7132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5201097892269533e-06,
+      "loss": 1.9105,
+      "step": 7133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5028935224273843e-06,
+      "loss": 1.8021,
+      "step": 7134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4857751578225e-06,
+      "loss": 1.8726,
+      "step": 7135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4687546987741663e-06,
+      "loss": 1.8863,
+      "step": 7136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.451832148625154e-06,
+      "loss": 1.8866,
+      "step": 7137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4350075106989713e-06,
+      "loss": 1.9047,
+      "step": 7138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4182807882999193e-06,
+      "loss": 1.8075,
+      "step": 7139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4016519847129817e-06,
+      "loss": 1.8004,
+      "step": 7140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3851211032040457e-06,
+      "loss": 1.9008,
+      "step": 7141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.36868814701957e-06,
+      "loss": 1.9555,
+      "step": 7142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3523531193870287e-06,
+      "loss": 1.8599,
+      "step": 7143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.336116023514411e-06,
+      "loss": 2.0628,
+      "step": 7144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3199768625906661e-06,
+      "loss": 1.9017,
+      "step": 7145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3039356397854807e-06,
+      "loss": 1.9876,
+      "step": 7146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.287992358249168e-06,
+      "loss": 1.8233,
+      "step": 7147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2721470211130015e-06,
+      "loss": 1.9391,
+      "step": 7148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2563996314888248e-06,
+      "loss": 1.8981,
+      "step": 7149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.240750192469442e-06,
+      "loss": 1.917,
+      "step": 7150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2251987071282279e-06,
+      "loss": 1.941,
+      "step": 7151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2097451785194614e-06,
+      "loss": 1.8353,
+      "step": 7152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1943896096781037e-06,
+      "loss": 1.9342,
+      "step": 7153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1791320036199648e-06,
+      "loss": 1.9714,
+      "step": 7154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1639723633415366e-06,
+      "loss": 1.8539,
+      "step": 7155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1489106918200486e-06,
+      "loss": 1.816,
+      "step": 7156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1339469920135793e-06,
+      "loss": 1.7076,
+      "step": 7157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1190812668608885e-06,
+      "loss": 1.8282,
+      "step": 7158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1043135192815302e-06,
+      "loss": 1.9492,
+      "step": 7159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0896437521757951e-06,
+      "loss": 1.8393,
+      "step": 7160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0750719684248233e-06,
+      "loss": 1.9389,
+      "step": 7161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0605981708903811e-06,
+      "loss": 1.9475,
+      "step": 7162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0462223624150835e-06,
+      "loss": 1.949,
+      "step": 7163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.031944545822172e-06,
+      "loss": 1.8428,
+      "step": 7164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0177647239158483e-06,
+      "loss": 1.8848,
+      "step": 7165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0036828994808845e-06,
+      "loss": 1.9146,
+      "step": 7166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.89699075282846e-07,
+      "loss": 1.932,
+      "step": 7167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.75813254068203e-07,
+      "loss": 1.863,
+      "step": 7168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.620254385639404e-07,
+      "loss": 1.872,
+      "step": 7169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.483356314779478e-07,
+      "loss": 2.0262,
+      "step": 7170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.347438354988524e-07,
+      "loss": 1.9853,
+      "step": 7171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.212500532960189e-07,
+      "loss": 1.8674,
+      "step": 7172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.07854287519494e-07,
+      "loss": 1.8658,
+      "step": 7173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.945565408002287e-07,
+      "loss": 1.9333,
+      "step": 7174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.813568157496899e-07,
+      "loss": 1.8105,
+      "step": 7175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.682551149604146e-07,
+      "loss": 1.8084,
+      "step": 7176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.552514410053447e-07,
+      "loss": 1.8962,
+      "step": 7177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.423457964384373e-07,
+      "loss": 1.8727,
+      "step": 7178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.295381837942206e-07,
+      "loss": 1.938,
+      "step": 7179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.168286055880713e-07,
+      "loss": 1.9046,
+      "step": 7180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.042170643161039e-07,
+      "loss": 1.9239,
+      "step": 7181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.917035624551705e-07,
+      "loss": 1.8648,
+      "step": 7182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.792881024628051e-07,
+      "loss": 1.9546,
+      "step": 7183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.669706867773906e-07,
+      "loss": 2.013,
+      "step": 7184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.547513178179366e-07,
+      "loss": 1.8303,
+      "step": 7185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.42629997984301e-07,
+      "loss": 1.8128,
+      "step": 7186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.306067296570796e-07,
+      "loss": 1.9385,
+      "step": 7187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.186815151975501e-07,
+      "loss": 1.8679,
+      "step": 7188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.068543569477281e-07,
+      "loss": 1.9012,
+      "step": 7189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.951252572304223e-07,
+      "loss": 2.0461,
+      "step": 7190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.834942183491788e-07,
+      "loss": 1.843,
+      "step": 7191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.719612425882815e-07,
+      "loss": 1.9763,
+      "step": 7192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.605263322127519e-07,
+      "loss": 1.939,
+      "step": 7193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.491894894683492e-07,
+      "loss": 1.9766,
+      "step": 7194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.379507165815146e-07,
+      "loss": 1.8296,
+      "step": 7195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.26810015759538e-07,
+      "loss": 1.8525,
+      "step": 7196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.157673891903359e-07,
+      "loss": 1.8334,
+      "step": 7197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.048228390426736e-07,
+      "loss": 1.8034,
+      "step": 7198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.939763674659981e-07,
+      "loss": 1.8114,
+      "step": 7199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.832279765904947e-07,
+      "loss": 1.8222,
+      "step": 7200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.725776685270855e-07,
+      "loss": 1.9374,
+      "step": 7201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.620254453674312e-07,
+      "loss": 1.7956,
+      "step": 7202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.515713091839291e-07,
+      "loss": 1.9409,
+      "step": 7203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.412152620297151e-07,
+      "loss": 1.6932,
+      "step": 7204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.309573059387174e-07,
+      "loss": 1.8418,
+      "step": 7205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.207974429254359e-07,
+      "loss": 1.9855,
+      "step": 7206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.107356749853298e-07,
+      "loss": 1.8527,
+      "step": 7207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.007720040943742e-07,
+      "loss": 1.9176,
+      "step": 7208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.909064322094481e-07,
+      "loss": 1.8526,
+      "step": 7209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.811389612681127e-07,
+      "loss": 1.8957,
+      "step": 7210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.714695931886115e-07,
+      "loss": 1.92,
+      "step": 7211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.618983298699253e-07,
+      "loss": 1.8957,
+      "step": 7212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.524251731918283e-07,
+      "loss": 1.7864,
+      "step": 7213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.430501250148322e-07,
+      "loss": 1.8498,
+      "step": 7214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3377318718013095e-07,
+      "loss": 1.8857,
+      "step": 7215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.245943615096004e-07,
+      "loss": 1.8148,
+      "step": 7216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.155136498060208e-07,
+      "loss": 1.907,
+      "step": 7217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0653105385274336e-07,
+      "loss": 1.9679,
+      "step": 7218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9764657541385694e-07,
+      "loss": 1.9324,
+      "step": 7219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.888602162342991e-07,
+      "loss": 1.8666,
+      "step": 7220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8017197803963397e-07,
+      "loss": 1.8243,
+      "step": 7221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.715818625361633e-07,
+      "loss": 1.8475,
+      "step": 7222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6308987141103757e-07,
+      "loss": 1.9449,
+      "step": 7223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5469600633192266e-07,
+      "loss": 1.9192,
+      "step": 7224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4640026894733335e-07,
+      "loss": 1.7745,
+      "step": 7225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.382026608865774e-07,
+      "loss": 1.8129,
+      "step": 7226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3010318375958916e-07,
+      "loss": 1.9714,
+      "step": 7227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2210183915709624e-07,
+      "loss": 1.7711,
+      "step": 7228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.141986286504528e-07,
+      "loss": 1.8397,
+      "step": 7229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.063935537918616e-07,
+      "loss": 1.9219,
+      "step": 7230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9868661611415213e-07,
+      "loss": 1.9323,
+      "step": 7231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9107781713100244e-07,
+      "loss": 1.7619,
+      "step": 7232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8356715833666167e-07,
+      "loss": 1.7666,
+      "step": 7233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.761546412062277e-07,
+      "loss": 1.8655,
+      "step": 7234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6884026719548036e-07,
+      "loss": 1.8711,
+      "step": 7235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6162403774088184e-07,
+      "loss": 1.9333,
+      "step": 7236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5450595425974277e-07,
+      "loss": 1.8743,
+      "step": 7237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.474860181498895e-07,
+      "loss": 1.8422,
+      "step": 7238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.405642307901079e-07,
+      "loss": 1.8576,
+      "step": 7239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.337405935398107e-07,
+      "loss": 1.9013,
+      "step": 7240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.27015107739037e-07,
+      "loss": 1.9059,
+      "step": 7241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2038777470867467e-07,
+      "loss": 1.86,
+      "step": 7242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1385859575029365e-07,
+      "loss": 1.7859,
+      "step": 7243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0742757214625708e-07,
+      "loss": 1.9346,
+      "step": 7244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0109470515949911e-07,
+      "loss": 1.7844,
+      "step": 7245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9485999603374717e-07,
+      "loss": 1.9742,
+      "step": 7246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8872344599357715e-07,
+      "loss": 1.9366,
+      "step": 7247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.826850562440252e-07,
+      "loss": 1.9214,
+      "step": 7248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7674482797114256e-07,
+      "loss": 1.9237,
+      "step": 7249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7090276234149604e-07,
+      "loss": 1.9104,
+      "step": 7250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6515886050239014e-07,
+      "loss": 1.8359,
+      "step": 7251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.59513123581978e-07,
+      "loss": 1.732,
+      "step": 7252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5396555268898383e-07,
+      "loss": 1.8751,
+      "step": 7253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4851614891298048e-07,
+      "loss": 1.8335,
+      "step": 7254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4316491332416748e-07,
+      "loss": 1.8579,
+      "step": 7255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3791184697353742e-07,
+      "loss": 1.9651,
+      "step": 7256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3275695089270955e-07,
+      "loss": 1.8142,
+      "step": 7257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2770022609409627e-07,
+      "loss": 1.9393,
+      "step": 7258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2274167357079203e-07,
+      "loss": 2.003,
+      "step": 7259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1788129429668448e-07,
+      "loss": 1.9259,
+      "step": 7260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1311908922628788e-07,
+      "loss": 1.9913,
+      "step": 7261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0845505929490962e-07,
+      "loss": 1.8185,
+      "step": 7262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0388920541848368e-07,
+      "loss": 1.8122,
+      "step": 7263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.942152849379272e-08,
+      "loss": 1.9606,
+      "step": 7264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.505202939819046e-08,
+      "loss": 1.9063,
+      "step": 7265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.078070898987933e-08,
+      "loss": 1.9714,
+      "step": 7266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.660756810768833e-08,
+      "loss": 1.9323,
+      "step": 7267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.25326075712396e-08,
+      "loss": 1.8233,
+      "step": 7268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.855582818072637e-08,
+      "loss": 1.7039,
+      "step": 7269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.46772307173571e-08,
+      "loss": 1.9356,
+      "step": 7270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.089681594268926e-08,
+      "loss": 1.8862,
+      "step": 7271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.721458459924002e-08,
+      "loss": 1.8625,
+      "step": 7272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.363053741020863e-08,
+      "loss": 1.9826,
+      "step": 7273
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.014467507947652e-08,
+      "loss": 1.9031,
+      "step": 7274
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.675699829160719e-08,
+      "loss": 1.7275,
+      "step": 7275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3467507711957296e-08,
+      "loss": 2.0801,
+      "step": 7276
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.0276203986565626e-08,
+      "loss": 1.8285,
+      "step": 7277
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.7183087742153074e-08,
+      "loss": 1.8604,
+      "step": 7278
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.418815958623368e-08,
+      "loss": 1.9562,
+      "step": 7279
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1291420106948084e-08,
+      "loss": 1.9572,
+      "step": 7280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8492869873285596e-08,
+      "loss": 1.9278,
+      "step": 7281
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5792509434751096e-08,
+      "loss": 1.8431,
+      "step": 7282
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.319033932180915e-08,
+      "loss": 1.7114,
+      "step": 7283
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0686360045384386e-08,
+      "loss": 1.8046,
+      "step": 7284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8280572097361123e-08,
+      "loss": 1.7998,
+      "step": 7285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5972975950083742e-08,
+      "loss": 1.8388,
+      "step": 7286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.376357205691182e-08,
+      "loss": 1.894,
+      "step": 7287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1652360851665e-08,
+      "loss": 1.8235,
+      "step": 7288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.963934274895607e-08,
+      "loss": 1.9087,
+      "step": 7289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.772451814419096e-08,
+      "loss": 1.9018,
+      "step": 7290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5907887413457722e-08,
+      "loss": 1.8349,
+      "step": 7291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4189450913415502e-08,
+      "loss": 1.8986,
+      "step": 7292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2569208981627611e-08,
+      "loss": 1.8569,
+      "step": 7293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1047161936339479e-08,
+      "loss": 1.8878,
+      "step": 7294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.623310076423141e-09,
+      "loss": 1.958,
+      "step": 7295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.29765368148827e-09,
+      "loss": 1.7767,
+      "step": 7296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.0701930119376755e-09,
+      "loss": 1.8904,
+      "step": 7297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.940928308856286e-09,
+      "loss": 1.7869,
+      "step": 7298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.90985979395564e-09,
+      "loss": 1.8561,
+      "step": 7299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9769876697959284e-09,
+      "loss": 1.8938,
+      "step": 7300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1423121195639505e-09,
+      "loss": 1.768,
+      "step": 7301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4058333071286243e-09,
+      "loss": 1.8844,
+      "step": 7302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7675513772075213e-09,
+      "loss": 1.8752,
+      "step": 7303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.227466455144821e-09,
+      "loss": 1.9529,
+      "step": 7304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.855786470223336e-10,
+      "loss": 1.8795,
+      "step": 7305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.41888039548477e-10,
+      "loss": 1.8405,
+      "step": 7306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.963947003358335e-10,
+      "loss": 1.8125,
+      "step": 7307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9098677512571245e-11,
+      "loss": 1.9765,
+      "step": 7308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.3754,
+      "step": 7309
+    },
+    {
+      "epoch": 1.0,
+      "step": 7309,
+      "total_flos": 9.941323478361702e+17,
+      "train_loss": 2.4119914807312988,
+      "train_runtime": 20054.373,
+      "train_samples_per_second": 34.984,
+      "train_steps_per_second": 0.364
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7309,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 9.941323478361702e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}