diff --git "a/finetuned_ckpt/trainer_state.json" "b/finetuned_ckpt/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/finetuned_ckpt/trainer_state.json"
@@ -0,0 +1,98818 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.999240929102778,
+  "eval_steps": 500,
+  "global_step": 16465,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.818181818181818e-06,
+      "loss": 1.0437,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.636363636363636e-06,
+      "loss": 1.3714,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.454545454545454e-06,
+      "loss": 1.3415,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.272727272727272e-06,
+      "loss": 1.0657,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.0078,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0909090909090907e-05,
+      "loss": 1.0692,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2727272727272727e-05,
+      "loss": 1.137,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4545454545454545e-05,
+      "loss": 1.1838,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6363636363636363e-05,
+      "loss": 0.9552,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.2488,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999999999999998e-05,
+      "loss": 1.1938,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1818181818181814e-05,
+      "loss": 0.9991,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3636363636363634e-05,
+      "loss": 0.8313,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5454545454545454e-05,
+      "loss": 1.1667,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.727272727272727e-05,
+      "loss": 1.1672,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.909090909090909e-05,
+      "loss": 1.0052,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.090909090909091e-05,
+      "loss": 0.9519,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2727272727272725e-05,
+      "loss": 1.2473,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.454545454545454e-05,
+      "loss": 1.0741,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.9493,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8181818181818174e-05,
+      "loss": 1.0745,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9999999999999996e-05,
+      "loss": 1.0815,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.181818181818182e-05,
+      "loss": 0.9671,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.363636363636363e-05,
+      "loss": 1.0986,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.545454545454545e-05,
+      "loss": 0.7583,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.727272727272727e-05,
+      "loss": 0.8695,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9090909090909084e-05,
+      "loss": 0.8064,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.090909090909091e-05,
+      "loss": 1.005,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.272727272727272e-05,
+      "loss": 0.7028,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.454545454545454e-05,
+      "loss": 0.6968,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6363636363636356e-05,
+      "loss": 0.7165,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.818181818181818e-05,
+      "loss": 0.6678,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 0.9222,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.181818181818182e-05,
+      "loss": 0.7907,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.363636363636363e-05,
+      "loss": 0.7851,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.545454545454545e-05,
+      "loss": 0.7579,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.727272727272727e-05,
+      "loss": 0.6105,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.909090909090908e-05,
+      "loss": 0.64,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.09090909090909e-05,
+      "loss": 0.5102,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 0.8755,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.454545454545454e-05,
+      "loss": 0.8683,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.636363636363635e-05,
+      "loss": 0.4924,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.818181818181818e-05,
+      "loss": 0.3446,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.999999999999999e-05,
+      "loss": 0.4318,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.18181818181818e-05,
+      "loss": 0.2453,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.363636363636364e-05,
+      "loss": 0.3709,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.545454545454545e-05,
+      "loss": 0.4465,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.727272727272726e-05,
+      "loss": 0.4609,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.90909090909091e-05,
+      "loss": 0.5246,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.09090909090909e-05,
+      "loss": 0.3516,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.272727272727271e-05,
+      "loss": 0.5925,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.454545454545454e-05,
+      "loss": 0.3211,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.636363636363636e-05,
+      "loss": 0.3742,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.818181818181817e-05,
+      "loss": 0.6121,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999999999999999e-05,
+      "loss": 0.3881,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010181818181818181,
+      "loss": 0.3437,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010363636363636362,
+      "loss": 0.5659,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010545454545454545,
+      "loss": 0.3096,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010727272727272727,
+      "loss": 0.5505,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010909090909090908,
+      "loss": 0.5087,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001109090909090909,
+      "loss": 0.7122,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011272727272727271,
+      "loss": 0.2805,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011454545454545453,
+      "loss": 0.6002,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011636363636363636,
+      "loss": 0.6795,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011818181818181817,
+      "loss": 0.395,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 0.6266,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012181818181818181,
+      "loss": 0.4259,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012363636363636364,
+      "loss": 0.5388,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012545454545454543,
+      "loss": 0.3304,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012727272727272725,
+      "loss": 0.3686,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012909090909090908,
+      "loss": 0.2292,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001309090909090909,
+      "loss": 0.4417,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013272727272727272,
+      "loss": 0.4536,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013454545454545455,
+      "loss": 0.1587,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013636363636363634,
+      "loss": 0.5313,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013818181818181816,
+      "loss": 0.2263,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014,
+      "loss": 0.3606,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001418181818181818,
+      "loss": 0.1051,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001436363636363636,
+      "loss": 0.4122,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014545454545454546,
+      "loss": 0.2199,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014727272727272725,
+      "loss": 0.3692,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014909090909090908,
+      "loss": 0.4072,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001509090909090909,
+      "loss": 0.4437,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001527272727272727,
+      "loss": 0.2313,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015454545454545452,
+      "loss": 0.5631,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015636363636363637,
+      "loss": 0.3966,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015818181818181816,
+      "loss": 0.3592,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015999999999999999,
+      "loss": 0.6545,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001618181818181818,
+      "loss": 0.6373,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001636363636363636,
+      "loss": 0.5364,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016545454545454543,
+      "loss": 0.5372,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016727272727272728,
+      "loss": 0.4225,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016909090909090907,
+      "loss": 0.3679,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001709090909090909,
+      "loss": 0.4635,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017272727272727272,
+      "loss": 0.3707,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017454545454545452,
+      "loss": 0.3252,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017636363636363634,
+      "loss": 0.5124,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001781818181818182,
+      "loss": 0.4837,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.6666,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001818181818181818,
+      "loss": 0.274,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001836363636363636,
+      "loss": 0.4441,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018545454545454543,
+      "loss": 0.539,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018727272727272728,
+      "loss": 0.3286,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018909090909090907,
+      "loss": 0.5618,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001909090909090909,
+      "loss": 0.3391,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019272727272727272,
+      "loss": 0.5338,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019454545454545451,
+      "loss": 0.474,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019636363636363634,
+      "loss": 0.1881,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001981818181818182,
+      "loss": 0.3942,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999999999999998,
+      "loss": 0.4205,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002018181818181818,
+      "loss": 0.4837,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020363636363636363,
+      "loss": 0.5874,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020545454545454542,
+      "loss": 0.3381,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00020727272727272725,
+      "loss": 0.5398,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002090909090909091,
+      "loss": 0.5314,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002109090909090909,
+      "loss": 0.5965,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021272727272727272,
+      "loss": 0.4296,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021454545454545454,
+      "loss": 0.4317,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021636363636363633,
+      "loss": 0.4354,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021818181818181816,
+      "loss": 0.4298,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00021999999999999995,
+      "loss": 0.4279,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002218181818181818,
+      "loss": 0.3701,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022363636363636363,
+      "loss": 0.4132,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022545454545454542,
+      "loss": 0.6167,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022727272727272725,
+      "loss": 0.5235,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00022909090909090907,
+      "loss": 0.4865,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023090909090909086,
+      "loss": 0.4657,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023272727272727271,
+      "loss": 0.6469,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023454545454545454,
+      "loss": 0.1927,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023636363636363633,
+      "loss": 0.3842,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023818181818181816,
+      "loss": 0.4845,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.3187,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002418181818181818,
+      "loss": 0.3742,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00024363636363636362,
+      "loss": 0.5076,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00024545454545454545,
+      "loss": 0.5817,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00024727272727272727,
+      "loss": 0.5747,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002490909090909091,
+      "loss": 0.4138,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025090909090909086,
+      "loss": 0.2683,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002527272727272727,
+      "loss": 0.4361,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002545454545454545,
+      "loss": 0.2895,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025636363636363633,
+      "loss": 0.623,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00025818181818181815,
+      "loss": 0.3903,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00026,
+      "loss": 0.2136,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002618181818181818,
+      "loss": 0.2805,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002636363636363636,
+      "loss": 0.5483,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00026545454545454545,
+      "loss": 0.2948,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002672727272727272,
+      "loss": 0.2427,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002690909090909091,
+      "loss": 0.4136,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002709090909090909,
+      "loss": 0.3174,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002727272727272727,
+      "loss": 0.4177,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002745454545454545,
+      "loss": 0.5528,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00027636363636363633,
+      "loss": 0.5296,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00027818181818181815,
+      "loss": 0.3643,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00028,
+      "loss": 0.3344,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002818181818181818,
+      "loss": 0.3172,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002836363636363636,
+      "loss": 0.5335,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00028545454545454544,
+      "loss": 0.2319,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002872727272727272,
+      "loss": 0.4038,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00028909090909090904,
+      "loss": 0.3582,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002909090909090909,
+      "loss": 0.417,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002927272727272727,
+      "loss": 0.1654,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002945454545454545,
+      "loss": 0.3233,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029636363636363633,
+      "loss": 0.4184,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029818181818181815,
+      "loss": 0.509,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0003,
+      "loss": 0.2922,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999999972139699,
+      "loss": 0.2725,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999999888558798,
+      "loss": 0.6556,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999997492573,
+      "loss": 0.3606,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000299999955423521,
+      "loss": 0.2122,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999993034925354,
+      "loss": 0.6728,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999989970292854,
+      "loss": 0.4107,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999998634845471,
+      "loss": 0.3136,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999982169411066,
+      "loss": 0.2579,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999977433162065,
+      "loss": 0.3546,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999997213970789,
+      "loss": 0.4242,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999996628904874,
+      "loss": 0.4885,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999995988118483,
+      "loss": 0.3881,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000299999529161164,
+      "loss": 0.2773,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000299999453938437,
+      "loss": 0.5948,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00029999937314367016,
+      "loss": 0.3865,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002999992867768665,
+      "loss": 0.5225,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999991948380292,
+      "loss": 0.2902,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999990973271617,
+      "loss": 0.2447,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999989942442676,
+      "loss": 0.4913,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999888558935065,
+      "loss": 0.2725,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999877136241513,
+      "loss": 0.2087,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999865156346496,
+      "loss": 0.3319,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999985261925049,
+      "loss": 0.4781,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999983952495393,
+      "loss": 0.5818,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999982587345733,
+      "loss": 0.4772,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999981166476118,
+      "loss": 0.3338,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999979689886601,
+      "loss": 0.1476,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999781575772383,
+      "loss": 0.3143,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999976569548085,
+      "loss": 0.5108,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999974925799201,
+      "loss": 0.4619,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999732263306467,
+      "loss": 0.3785,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999971471142486,
+      "loss": 0.6238,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999696602347835,
+      "loss": 0.3606,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999677936076073,
+      "loss": 0.2799,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999965871261026,
+      "loss": 0.3467,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999963893195111,
+      "loss": 0.4611,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999618594099363,
+      "loss": 0.2833,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999959769905577,
+      "loss": 0.3678,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000299995762468211,
+      "loss": 0.354,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999554237396166,
+      "loss": 0.3824,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999953167078178,
+      "loss": 0.3294,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999508546978777,
+      "loss": 0.2361,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999484865988015,
+      "loss": 0.2253,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999460627810377,
+      "loss": 0.5142,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999435832446763,
+      "loss": 0.3787,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002999941047989809,
+      "loss": 0.5363,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999384570165305,
+      "loss": 0.2896,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999358103249374,
+      "loss": 0.3938,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029999331079151267,
+      "loss": 0.6212,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999303497872,
+      "loss": 0.3271,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999927535941259,
+      "loss": 0.3866,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999246663774083,
+      "loss": 0.4239,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999217410957556,
+      "loss": 0.2669,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999918760096408,
+      "loss": 0.4804,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999915723379477,
+      "loss": 0.3723,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999126309450755,
+      "loss": 0.4723,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029999094827933183,
+      "loss": 0.2585,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999906278924322,
+      "loss": 0.4599,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999903019338206,
+      "loss": 0.1762,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998997040350914,
+      "loss": 0.6406,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999896333015101,
+      "loss": 0.3653,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998929062783605,
+      "loss": 0.4757,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999889423824997,
+      "loss": 0.3712,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998858856551393,
+      "loss": 0.4202,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000299988229176892,
+      "loss": 0.3619,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999878642166471,
+      "loss": 0.4158,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998749368479296,
+      "loss": 0.2779,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999871175813432,
+      "loss": 0.6044,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999867359063119,
+      "loss": 0.3736,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999863486597132,
+      "loss": 0.2485,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998595584156145,
+      "loss": 0.3877,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999855574518713,
+      "loss": 0.4692,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998515349065753,
+      "loss": 0.2173,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998474395793507,
+      "loss": 0.517,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998432885371926,
+      "loss": 0.377,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998390817802544,
+      "loss": 0.2557,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998348193086927,
+      "loss": 0.2506,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999830501122665,
+      "loss": 0.5651,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999826127222333,
+      "loss": 0.2954,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999821697607859,
+      "loss": 0.3545,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029998172122794065,
+      "loss": 0.3171,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002999812671237143,
+      "loss": 0.0741,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029998080744812365,
+      "loss": 0.2749,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999803422011859,
+      "loss": 0.3257,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999798713829182,
+      "loss": 0.3338,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999793949933381,
+      "loss": 0.4425,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999789130324632,
+      "loss": 0.3899,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997842550031164,
+      "loss": 0.2719,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997793239690127,
+      "loss": 0.2064,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997743372225054,
+      "loss": 0.2475,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000299976929476378,
+      "loss": 0.3733,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999764196593023,
+      "loss": 0.3868,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999759042710424,
+      "loss": 0.402,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997538331161743,
+      "loss": 0.4802,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999748567810468,
+      "loss": 0.3362,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997432467935014,
+      "loss": 0.2669,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000299973787006547,
+      "loss": 0.4333,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997324376265745,
+      "loss": 0.3182,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999726949477018,
+      "loss": 0.2728,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999721405617002,
+      "loss": 0.3919,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997158060467345,
+      "loss": 0.2064,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999710150766422,
+      "loss": 0.3011,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029997044397762755,
+      "loss": 0.4111,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996986730765075,
+      "loss": 0.3339,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999692850667331,
+      "loss": 0.2085,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999686972548963,
+      "loss": 0.179,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999681038721622,
+      "loss": 0.3568,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996750491855277,
+      "loss": 0.3835,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999669003940904,
+      "loss": 0.3924,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996629029879735,
+      "loss": 0.2393,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996567463269646,
+      "loss": 0.28,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996505339581047,
+      "loss": 0.2806,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002999644265881626,
+      "loss": 0.1921,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029996379420977595,
+      "loss": 0.3723,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029996315626067415,
+      "loss": 0.4365,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029996251274088087,
+      "loss": 0.3499,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029996186365042,
+      "loss": 0.146,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029996120898931566,
+      "loss": 0.3259,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029996054875759215,
+      "loss": 0.0345,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995988295527403,
+      "loss": 0.3186,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000299959211582386,
+      "loss": 0.3591,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000299958534638953,
+      "loss": 0.2167,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999578521250002,
+      "loss": 0.3219,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995716404055295,
+      "loss": 0.5844,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999564703856368,
+      "loss": 0.1811,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995577116027753,
+      "loss": 0.4347,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995506636450105,
+      "loss": 0.3219,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995435599833363,
+      "loss": 0.4494,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999536400618016,
+      "loss": 0.3272,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999529185549316,
+      "loss": 0.3801,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999521914777504,
+      "loss": 0.4697,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029995145883028503,
+      "loss": 0.5078,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999507206125627,
+      "loss": 0.1618,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999499768246108,
+      "loss": 0.2617,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000299949227466457,
+      "loss": 0.3804,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994847253812915,
+      "loss": 0.4561,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999477120396552,
+      "loss": 0.4977,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999469459710635,
+      "loss": 0.2029,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999461743323825,
+      "loss": 0.305,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999453971236408,
+      "loss": 0.5059,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999446143448673,
+      "loss": 0.3725,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994382599609113,
+      "loss": 0.6375,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994303207734154,
+      "loss": 0.1973,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002999422325886479,
+      "loss": 0.2211,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029994142753004015,
+      "loss": 0.3343,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000299940616901548,
+      "loss": 0.1735,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00029993980070320165,
+      "loss": 0.3668,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999389789350314,
+      "loss": 0.3392,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993815159706777,
+      "loss": 0.1918,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999373186893415,
+      "loss": 0.609,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993648021188355,
+      "loss": 0.4783,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000299935636164725,
+      "loss": 0.3132,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999347865478973,
+      "loss": 0.3271,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993393136143194,
+      "loss": 0.3123,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993307060536065,
+      "loss": 0.2013,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999322042797155,
+      "loss": 0.3008,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993133238452867,
+      "loss": 0.2861,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029993045491983245,
+      "loss": 0.2908,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992957188565954,
+      "loss": 0.1871,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992868328204274,
+      "loss": 0.2227,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992778910901496,
+      "loss": 0.4098,
+      "step": 326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992688936660944,
+      "loss": 0.2893,
+      "step": 327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999259840548597,
+      "loss": 0.2273,
+      "step": 328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999250731737993,
+      "loss": 0.2629,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999241567234621,
+      "loss": 0.297,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992323470388204,
+      "loss": 0.4238,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992230711509355,
+      "loss": 0.4894,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992137395713097,
+      "loss": 0.4424,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029992043523002895,
+      "loss": 0.2386,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991949093382246,
+      "loss": 0.2839,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991854106854645,
+      "loss": 0.6452,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991758563423633,
+      "loss": 0.2609,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991662463092745,
+      "loss": 0.42,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991565805865573,
+      "loss": 0.3616,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991468591745687,
+      "loss": 0.3846,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000299913708207367,
+      "loss": 0.3083,
+      "step": 341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999127249284225,
+      "loss": 0.2291,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029991173608065995,
+      "loss": 0.3025,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000299910741664116,
+      "loss": 0.3424,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002999097416788275,
+      "loss": 0.3925,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999087361248318,
+      "loss": 0.3662,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999077250021662,
+      "loss": 0.1759,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990670831086814,
+      "loss": 0.2067,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990568605097547,
+      "loss": 0.2343,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990465822252616,
+      "loss": 0.2792,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990362482555844,
+      "loss": 0.3089,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999025858601106,
+      "loss": 0.3161,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029990154132622125,
+      "loss": 0.431,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002999004912239293,
+      "loss": 0.4195,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989943555327363,
+      "loss": 0.3545,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989837431429353,
+      "loss": 0.3528,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998973075070284,
+      "loss": 0.4253,
+      "step": 357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998962351315178,
+      "loss": 0.1749,
+      "step": 358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989515718780173,
+      "loss": 0.2562,
+      "step": 359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989407367592006,
+      "loss": 0.3759,
+      "step": 360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998929845959131,
+      "loss": 0.204,
+      "step": 361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029989188994782143,
+      "loss": 0.264,
+      "step": 362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998907897316855,
+      "loss": 0.5466,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998896839475463,
+      "loss": 0.2671,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998885725954449,
+      "loss": 0.2672,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998874556754226,
+      "loss": 0.1921,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998863331875209,
+      "loss": 0.2238,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029988520513178136,
+      "loss": 0.1988,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000299884071508246,
+      "loss": 0.3544,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000299882932316957,
+      "loss": 0.3866,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998817875579565,
+      "loss": 0.4072,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029988063723128714,
+      "loss": 0.4454,
+      "step": 372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987948133699166,
+      "loss": 0.2996,
+      "step": 373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998783198751129,
+      "loss": 0.3298,
+      "step": 374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987715284569415,
+      "loss": 0.4764,
+      "step": 375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987598024877864,
+      "loss": 0.3028,
+      "step": 376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00029987480208440995,
+      "loss": 0.4967,
+      "step": 377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002998736183526319,
+      "loss": 0.5879,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029987242905348845,
+      "loss": 0.3786,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998712341870237,
+      "loss": 0.1929,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029987003375328217,
+      "loss": 0.3589,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029986882775230833,
+      "loss": 0.2731,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000299867616184147,
+      "loss": 0.1609,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029986639904884326,
+      "loss": 0.2304,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998651763464423,
+      "loss": 0.2167,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029986394807698946,
+      "loss": 0.3218,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998627142405304,
+      "loss": 0.277,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029986147483711104,
+      "loss": 0.5421,
+      "step": 388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998602298667773,
+      "loss": 0.3409,
+      "step": 389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029985897932957546,
+      "loss": 0.4092,
+      "step": 390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998577232255521,
+      "loss": 0.3223,
+      "step": 391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998564615547537,
+      "loss": 0.3169,
+      "step": 392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998551943172272,
+      "loss": 0.3379,
+      "step": 393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998539215130197,
+      "loss": 0.307,
+      "step": 394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029985264314217843,
+      "loss": 0.3132,
+      "step": 395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029985135920475097,
+      "loss": 0.2442,
+      "step": 396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998500697007849,
+      "loss": 0.5513,
+      "step": 397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998487746303282,
+      "loss": 0.0905,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984747399342893,
+      "loss": 0.3658,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998461677901354,
+      "loss": 0.3121,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984485602049625,
+      "loss": 0.4953,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984353868456006,
+      "loss": 0.3584,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998422157823758,
+      "loss": 0.4015,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029984088731399263,
+      "loss": 0.2617,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998395532794599,
+      "loss": 0.0714,
+      "step": 405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002998382136788272,
+      "loss": 0.3164,
+      "step": 406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029983686851214425,
+      "loss": 0.1292,
+      "step": 407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000299835517779461,
+      "loss": 0.4262,
+      "step": 408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029983416148082764,
+      "loss": 0.2923,
+      "step": 409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029983279961629466,
+      "loss": 0.327,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00029983143218591246,
+      "loss": 0.2678,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029983005918973195,
+      "loss": 0.2347,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982868062780415,
+      "loss": 0.3129,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982729650018027,
+      "loss": 0.5026,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998259068069116,
+      "loss": 0.5613,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982451154804987,
+      "loss": 0.3302,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982311072364695,
+      "loss": 0.328,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029982170433375476,
+      "loss": 0.3657,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998202923784256,
+      "loss": 0.2838,
+      "step": 419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029981887485771197,
+      "loss": 0.3928,
+      "step": 420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029981745177166643,
+      "loss": 0.2999,
+      "step": 421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029981602312034196,
+      "loss": 0.4211,
+      "step": 422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998145889037915,
+      "loss": 0.3738,
+      "step": 423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998131491220684,
+      "loss": 0.4139,
+      "step": 424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998117037752261,
+      "loss": 0.296,
+      "step": 425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998102528633184,
+      "loss": 0.4458,
+      "step": 426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998087963863991,
+      "loss": 0.2338,
+      "step": 427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998073343445223,
+      "loss": 0.2631,
+      "step": 428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029980586673774233,
+      "loss": 0.3304,
+      "step": 429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998043935661137,
+      "loss": 0.2876,
+      "step": 430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998029148296912,
+      "loss": 0.1625,
+      "step": 431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002998014305285297,
+      "loss": 0.3688,
+      "step": 432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997999406626843,
+      "loss": 0.3722,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029979844523221047,
+      "loss": 0.3131,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997969442371636,
+      "loss": 0.3079,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997954376775996,
+      "loss": 0.2827,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997939255535743,
+      "loss": 0.2785,
+      "step": 437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029979240786514395,
+      "loss": 0.4122,
+      "step": 438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997908846123649,
+      "loss": 0.2062,
+      "step": 439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029978935579529374,
+      "loss": 0.4249,
+      "step": 440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029978782141398727,
+      "loss": 0.5095,
+      "step": 441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029978628146850255,
+      "loss": 0.2382,
+      "step": 442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002997847359588966,
+      "loss": 0.2783,
+      "step": 443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00029978318488522705,
+      "loss": 0.342,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029978162824755136,
+      "loss": 0.2918,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029978006604592744,
+      "loss": 0.2921,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997784982804133,
+      "loss": 0.2273,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029977692495106715,
+      "loss": 0.4954,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029977534605794744,
+      "loss": 0.2556,
+      "step": 449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997737616011129,
+      "loss": 0.3203,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029977217158062223,
+      "loss": 0.1777,
+      "step": 451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997705759965347,
+      "loss": 0.2739,
+      "step": 452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997689748489094,
+      "loss": 0.3715,
+      "step": 453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997673681378059,
+      "loss": 0.1215,
+      "step": 454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997657558632838,
+      "loss": 0.3623,
+      "step": 455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997641380254031,
+      "loss": 0.3727,
+      "step": 456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029976251462422384,
+      "loss": 0.4168,
+      "step": 457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997608856598064,
+      "loss": 0.4712,
+      "step": 458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029975925113221114,
+      "loss": 0.2771,
+      "step": 459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997576110414989,
+      "loss": 0.1335,
+      "step": 460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997559653877306,
+      "loss": 0.1498,
+      "step": 461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029975431417096735,
+      "loss": 0.3372,
+      "step": 462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029975265739127046,
+      "loss": 0.4225,
+      "step": 463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997509950487014,
+      "loss": 0.404,
+      "step": 464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997493271433222,
+      "loss": 0.3955,
+      "step": 465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997476536751945,
+      "loss": 0.6078,
+      "step": 466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997459746443806,
+      "loss": 0.2999,
+      "step": 467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997442900509429,
+      "loss": 0.3965,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997425998949439,
+      "loss": 0.3885,
+      "step": 469
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997409041764465,
+      "loss": 0.1915,
+      "step": 470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029973920289551363,
+      "loss": 0.2949,
+      "step": 471
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029973749605220843,
+      "loss": 0.2247,
+      "step": 472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029973578364659437,
+      "loss": 0.2409,
+      "step": 473
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029973406567873503,
+      "loss": 0.4252,
+      "step": 474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029973234214869423,
+      "loss": 0.4664,
+      "step": 475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002997306130565361,
+      "loss": 0.3716,
+      "step": 476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029972887840232465,
+      "loss": 0.1988,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997271381861245,
+      "loss": 0.282,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997253924080003,
+      "loss": 0.2726,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997236410680167,
+      "loss": 0.2146,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029972188416623895,
+      "loss": 0.2979,
+      "step": 481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997201217027323,
+      "loss": 0.1986,
+      "step": 482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029971835367756215,
+      "loss": 0.2256,
+      "step": 483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029971658009079425,
+      "loss": 0.3377,
+      "step": 484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029971480094249437,
+      "loss": 0.3734,
+      "step": 485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997130162327287,
+      "loss": 0.2543,
+      "step": 486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029971122596156356,
+      "loss": 0.4601,
+      "step": 487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029970943012906534,
+      "loss": 0.1657,
+      "step": 488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997076287353008,
+      "loss": 0.1734,
+      "step": 489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997058217803369,
+      "loss": 0.3586,
+      "step": 490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002997040092642407,
+      "loss": 0.3834,
+      "step": 491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029970219118707957,
+      "loss": 0.2606,
+      "step": 492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029970036754892106,
+      "loss": 0.3879,
+      "step": 493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969853834983285,
+      "loss": 0.3575,
+      "step": 494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969670358988293,
+      "loss": 0.3558,
+      "step": 495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969486326913953,
+      "loss": 0.5673,
+      "step": 496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029969301738767086,
+      "loss": 0.351,
+      "step": 497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996911659455456,
+      "loss": 0.321,
+      "step": 498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996893089428324,
+      "loss": 0.2921,
+      "step": 499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996874463796004,
+      "loss": 0.1399,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996855782559188,
+      "loss": 0.2705,
+      "step": 501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029968370457185683,
+      "loss": 0.2905,
+      "step": 502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996818253274842,
+      "loss": 0.2765,
+      "step": 503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029967994052287065,
+      "loss": 0.1394,
+      "step": 504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996780501580863,
+      "loss": 0.1822,
+      "step": 505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996761542332013,
+      "loss": 0.2529,
+      "step": 506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029967425274828605,
+      "loss": 0.3515,
+      "step": 507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00029967234570341133,
+      "loss": 0.3273,
+      "step": 508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996704330986478,
+      "loss": 0.3779,
+      "step": 509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002996685149340666,
+      "loss": 0.1201,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029966659120973896,
+      "loss": 0.4215,
+      "step": 511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996646619257364,
+      "loss": 0.4635,
+      "step": 512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029966272708213047,
+      "loss": 0.1463,
+      "step": 513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996607866789931,
+      "loss": 0.4531,
+      "step": 514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996588407163964,
+      "loss": 0.1818,
+      "step": 515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996568891944127,
+      "loss": 0.3224,
+      "step": 516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996549321131144,
+      "loss": 0.3723,
+      "step": 517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996529694725742,
+      "loss": 0.2467,
+      "step": 518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000299651001272865,
+      "loss": 0.2826,
+      "step": 519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029964902751406007,
+      "loss": 0.5684,
+      "step": 520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029964704819623254,
+      "loss": 0.2713,
+      "step": 521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000299645063319456,
+      "loss": 0.2355,
+      "step": 522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029964307288380423,
+      "loss": 0.3478,
+      "step": 523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029964107688935113,
+      "loss": 0.3571,
+      "step": 524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029963907533617077,
+      "loss": 0.4088,
+      "step": 525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996370682243376,
+      "loss": 0.1707,
+      "step": 526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996350555539262,
+      "loss": 0.2044,
+      "step": 527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029963303732501124,
+      "loss": 0.1962,
+      "step": 528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029963101353766775,
+      "loss": 0.2835,
+      "step": 529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029962898419197095,
+      "loss": 0.2094,
+      "step": 530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996269492879961,
+      "loss": 0.1623,
+      "step": 531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996249088258188,
+      "loss": 0.3834,
+      "step": 532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029962286280551506,
+      "loss": 0.1057,
+      "step": 533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029962081122716065,
+      "loss": 0.092,
+      "step": 534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029961875409083186,
+      "loss": 0.4468,
+      "step": 535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029961669139660517,
+      "loss": 0.2699,
+      "step": 536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996146231445571,
+      "loss": 0.2229,
+      "step": 537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029961254933476446,
+      "loss": 0.384,
+      "step": 538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996104699673044,
+      "loss": 0.4219,
+      "step": 539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029960838504225414,
+      "loss": 0.2765,
+      "step": 540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996062945596911,
+      "loss": 0.4226,
+      "step": 541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00029960419851969284,
+      "loss": 0.2395,
+      "step": 542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002996020969223374,
+      "loss": 0.3721,
+      "step": 543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029959998976770275,
+      "loss": 0.5709,
+      "step": 544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995978770558672,
+      "loss": 0.2719,
+      "step": 545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029959575878690917,
+      "loss": 0.2691,
+      "step": 546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029959363496090744,
+      "loss": 0.2402,
+      "step": 547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995915055779408,
+      "loss": 0.1479,
+      "step": 548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995893706380884,
+      "loss": 0.2172,
+      "step": 549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995872301414296,
+      "loss": 0.2597,
+      "step": 550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029958508408804376,
+      "loss": 0.3459,
+      "step": 551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995829324780108,
+      "loss": 0.3623,
+      "step": 552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995807753114105,
+      "loss": 0.2283,
+      "step": 553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029957861258832306,
+      "loss": 0.358,
+      "step": 554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995764443088288,
+      "loss": 0.2108,
+      "step": 555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029957427047300824,
+      "loss": 0.1743,
+      "step": 556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029957209108094217,
+      "loss": 0.0729,
+      "step": 557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995699061327115,
+      "loss": 0.3481,
+      "step": 558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029956771562839746,
+      "loss": 0.3423,
+      "step": 559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995655195680814,
+      "loss": 0.386,
+      "step": 560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029956331795184484,
+      "loss": 0.2053,
+      "step": 561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029956111077976964,
+      "loss": 0.1772,
+      "step": 562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995588980519378,
+      "loss": 0.2538,
+      "step": 563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995566797684314,
+      "loss": 0.2372,
+      "step": 564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995544559293329,
+      "loss": 0.2269,
+      "step": 565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029955222653472503,
+      "loss": 0.4162,
+      "step": 566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995499915846904,
+      "loss": 0.364,
+      "step": 567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029954775107931224,
+      "loss": 0.432,
+      "step": 568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995455050186736,
+      "loss": 0.3256,
+      "step": 569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000299543253402858,
+      "loss": 0.4145,
+      "step": 570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995409962319491,
+      "loss": 0.3369,
+      "step": 571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029953873350603067,
+      "loss": 0.3079,
+      "step": 572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029953646522518687,
+      "loss": 0.4847,
+      "step": 573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029953419138950186,
+      "loss": 0.3155,
+      "step": 574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995319119990601,
+      "loss": 0.2454,
+      "step": 575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002995296270539463,
+      "loss": 0.3962,
+      "step": 576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029952733655424547,
+      "loss": 0.2893,
+      "step": 577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995250405000424,
+      "loss": 0.4459,
+      "step": 578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995227388914227,
+      "loss": 0.4131,
+      "step": 579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029952043172847166,
+      "loss": 0.3026,
+      "step": 580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995181190112751,
+      "loss": 0.321,
+      "step": 581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029951580073991885,
+      "loss": 0.248,
+      "step": 582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000299513476914489,
+      "loss": 0.1454,
+      "step": 583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029951114753507197,
+      "loss": 0.2537,
+      "step": 584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029950881260175426,
+      "loss": 0.233,
+      "step": 585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029950647211462257,
+      "loss": 0.4671,
+      "step": 586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029950412607376387,
+      "loss": 0.2935,
+      "step": 587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002995017744792653,
+      "loss": 0.2588,
+      "step": 588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029949941733121424,
+      "loss": 0.1436,
+      "step": 589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994970546296982,
+      "loss": 0.38,
+      "step": 590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000299494686374805,
+      "loss": 0.3124,
+      "step": 591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994923125666226,
+      "loss": 0.2337,
+      "step": 592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029948993320523916,
+      "loss": 0.4887,
+      "step": 593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994875482907431,
+      "loss": 0.2111,
+      "step": 594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994851578232229,
+      "loss": 0.3063,
+      "step": 595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994827618027675,
+      "loss": 0.1798,
+      "step": 596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994803602294659,
+      "loss": 0.2053,
+      "step": 597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994779531034072,
+      "loss": 0.2124,
+      "step": 598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994755404246809,
+      "loss": 0.1837,
+      "step": 599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994731221933766,
+      "loss": 0.224,
+      "step": 600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994706984095842,
+      "loss": 0.0651,
+      "step": 601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994682690733936,
+      "loss": 0.383,
+      "step": 602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029946583418489516,
+      "loss": 0.1943,
+      "step": 603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029946339374417927,
+      "loss": 0.2282,
+      "step": 604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994609477513366,
+      "loss": 0.2863,
+      "step": 605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029945849620645803,
+      "loss": 0.4083,
+      "step": 606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029945603910963456,
+      "loss": 0.212,
+      "step": 607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994535764609575,
+      "loss": 0.2627,
+      "step": 608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002994511082605184,
+      "loss": 0.1652,
+      "step": 609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994486345084089,
+      "loss": 0.5203,
+      "step": 610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994461552047208,
+      "loss": 0.3043,
+      "step": 611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994436703495464,
+      "loss": 0.242,
+      "step": 612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029944117994297784,
+      "loss": 0.1837,
+      "step": 613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029943868398510766,
+      "loss": 0.2564,
+      "step": 614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994361824760286,
+      "loss": 0.2774,
+      "step": 615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994336754158335,
+      "loss": 0.2654,
+      "step": 616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994311628046157,
+      "loss": 0.2832,
+      "step": 617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029942864464246833,
+      "loss": 0.258,
+      "step": 618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029942612092948506,
+      "loss": 0.2167,
+      "step": 619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994235916657595,
+      "loss": 0.2065,
+      "step": 620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029942105685138576,
+      "loss": 0.2448,
+      "step": 621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994185164864579,
+      "loss": 0.1594,
+      "step": 622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994159705710704,
+      "loss": 0.1468,
+      "step": 623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994134191053176,
+      "loss": 0.1639,
+      "step": 624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994108620892946,
+      "loss": 0.1069,
+      "step": 625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994082995230961,
+      "loss": 0.1347,
+      "step": 626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002994057314068175,
+      "loss": 0.2825,
+      "step": 627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000299403157740554,
+      "loss": 0.3625,
+      "step": 628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029940057852440136,
+      "loss": 0.3256,
+      "step": 629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993979937584553,
+      "loss": 0.2062,
+      "step": 630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029939540344281197,
+      "loss": 0.1888,
+      "step": 631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993928075775674,
+      "loss": 0.2005,
+      "step": 632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029939020616281817,
+      "loss": 0.139,
+      "step": 633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029938759919866093,
+      "loss": 0.0329,
+      "step": 634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029938498668519235,
+      "loss": 0.222,
+      "step": 635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029938236862250963,
+      "loss": 0.3898,
+      "step": 636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029937974501071,
+      "loss": 0.3655,
+      "step": 637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029937711584989085,
+      "loss": 0.156,
+      "step": 638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029937448114014995,
+      "loss": 0.2562,
+      "step": 639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000299371840881585,
+      "loss": 0.2923,
+      "step": 640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002993691950742943,
+      "loss": 0.4049,
+      "step": 641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00029936654371837595,
+      "loss": 0.3547,
+      "step": 642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993638868139286,
+      "loss": 0.3683,
+      "step": 643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993612243610508,
+      "loss": 0.3134,
+      "step": 644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993585563598415,
+      "loss": 0.1229,
+      "step": 645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993558828103998,
+      "loss": 0.4321,
+      "step": 646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993532037128251,
+      "loss": 0.3215,
+      "step": 647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993505190672169,
+      "loss": 0.1549,
+      "step": 648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029934782887367477,
+      "loss": 0.2787,
+      "step": 649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993451331322988,
+      "loss": 0.3185,
+      "step": 650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993424318431891,
+      "loss": 0.3658,
+      "step": 651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029933972500644597,
+      "loss": 0.4424,
+      "step": 652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029933701262217,
+      "loss": 0.2291,
+      "step": 653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000299334294690462,
+      "loss": 0.3363,
+      "step": 654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993315712114228,
+      "loss": 0.2799,
+      "step": 655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029932884218515366,
+      "loss": 0.3773,
+      "step": 656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993261076117559,
+      "loss": 0.4417,
+      "step": 657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029932336749133125,
+      "loss": 0.3764,
+      "step": 658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993206218239813,
+      "loss": 0.1566,
+      "step": 659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993178706098081,
+      "loss": 0.2881,
+      "step": 660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993151138489139,
+      "loss": 0.2519,
+      "step": 661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993123515414011,
+      "loss": 0.3047,
+      "step": 662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029930958368737235,
+      "loss": 0.3065,
+      "step": 663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993068102869303,
+      "loss": 0.4237,
+      "step": 664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002993040313401782,
+      "loss": 0.1637,
+      "step": 665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029930124684721907,
+      "loss": 0.2781,
+      "step": 666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992984568081565,
+      "loss": 0.2007,
+      "step": 667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029929566122309406,
+      "loss": 0.2458,
+      "step": 668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992928600921356,
+      "loss": 0.0155,
+      "step": 669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992900534153852,
+      "loss": 0.4149,
+      "step": 670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992872411929471,
+      "loss": 0.2661,
+      "step": 671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992844234249258,
+      "loss": 0.2798,
+      "step": 672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029928160011142587,
+      "loss": 0.4266,
+      "step": 673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002992787712525523,
+      "loss": 0.2037,
+      "step": 674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00029927593684841015,
+      "loss": 0.4644,
+      "step": 675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992730968991047,
+      "loss": 0.3323,
+      "step": 676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992702514047414,
+      "loss": 0.3219,
+      "step": 677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029926740036542596,
+      "loss": 0.4154,
+      "step": 678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992645437812644,
+      "loss": 0.154,
+      "step": 679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029926168165236275,
+      "loss": 0.2073,
+      "step": 680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992588139788273,
+      "loss": 0.3378,
+      "step": 681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992559407607646,
+      "loss": 0.4548,
+      "step": 682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029925306199828143,
+      "loss": 0.2912,
+      "step": 683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029925017769148466,
+      "loss": 0.4363,
+      "step": 684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992472878404815,
+      "loss": 0.2653,
+      "step": 685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992443924453792,
+      "loss": 0.3229,
+      "step": 686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992414915062854,
+      "loss": 0.0778,
+      "step": 687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029923858502330786,
+      "loss": 0.2847,
+      "step": 688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992356729965545,
+      "loss": 0.1495,
+      "step": 689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992327554261335,
+      "loss": 0.4434,
+      "step": 690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029922983231215326,
+      "loss": 0.2614,
+      "step": 691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029922690365472243,
+      "loss": 0.2328,
+      "step": 692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029922396945394967,
+      "loss": 0.3486,
+      "step": 693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029922102970994405,
+      "loss": 0.3103,
+      "step": 694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029921808442281473,
+      "loss": 0.3266,
+      "step": 695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029921513359267124,
+      "loss": 0.4676,
+      "step": 696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029921217721962307,
+      "loss": 0.5129,
+      "step": 697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029920921530378005,
+      "loss": 0.4523,
+      "step": 698
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029920624784525224,
+      "loss": 0.3135,
+      "step": 699
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029920327484414983,
+      "loss": 0.2785,
+      "step": 700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002992002963005834,
+      "loss": 0.419,
+      "step": 701
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991973122146634,
+      "loss": 0.3511,
+      "step": 702
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991943225865008,
+      "loss": 0.2734,
+      "step": 703
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991913274162066,
+      "loss": 0.2312,
+      "step": 704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991883267038921,
+      "loss": 0.105,
+      "step": 705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029918532044966885,
+      "loss": 0.3721,
+      "step": 706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991823086536483,
+      "loss": 0.4562,
+      "step": 707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002991792913159425,
+      "loss": 0.3046,
+      "step": 708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991762684366635,
+      "loss": 0.2957,
+      "step": 709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029917324001592355,
+      "loss": 0.3137,
+      "step": 710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991702060538352,
+      "loss": 0.2167,
+      "step": 711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029916716655051115,
+      "loss": 0.2974,
+      "step": 712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991641215060643,
+      "loss": 0.324,
+      "step": 713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991610709206077,
+      "loss": 0.266,
+      "step": 714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991580147942548,
+      "loss": 0.3172,
+      "step": 715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000299154953127119,
+      "loss": 0.2799,
+      "step": 716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991518859193141,
+      "loss": 0.3701,
+      "step": 717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000299148813170954,
+      "loss": 0.342,
+      "step": 718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991457348821529,
+      "loss": 0.205,
+      "step": 719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991426510530251,
+      "loss": 0.2479,
+      "step": 720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029913956168368516,
+      "loss": 0.1703,
+      "step": 721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991364667742479,
+      "loss": 0.427,
+      "step": 722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029913336632482815,
+      "loss": 0.2069,
+      "step": 723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029913026033554126,
+      "loss": 0.1666,
+      "step": 724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991271488065025,
+      "loss": 0.1995,
+      "step": 725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029912403173782743,
+      "loss": 0.3251,
+      "step": 726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991209091296319,
+      "loss": 0.2823,
+      "step": 727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002991177809820319,
+      "loss": 0.1423,
+      "step": 728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029911464729514366,
+      "loss": 0.182,
+      "step": 729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029911150806908346,
+      "loss": 0.201,
+      "step": 730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029910836330396804,
+      "loss": 0.3141,
+      "step": 731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029910521299991423,
+      "loss": 0.1276,
+      "step": 732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029910205715703896,
+      "loss": 0.3491,
+      "step": 733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990988957754595,
+      "loss": 0.2206,
+      "step": 734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029909572885529327,
+      "loss": 0.2723,
+      "step": 735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000299092556396658,
+      "loss": 0.3119,
+      "step": 736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029908937839967146,
+      "loss": 0.2072,
+      "step": 737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029908619486445167,
+      "loss": 0.2549,
+      "step": 738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029908300579111694,
+      "loss": 0.3454,
+      "step": 739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002990798111797857,
+      "loss": 0.2484,
+      "step": 740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029907661103057673,
+      "loss": 0.2278,
+      "step": 741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990734053436088,
+      "loss": 0.3896,
+      "step": 742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029907019411900095,
+      "loss": 0.1961,
+      "step": 743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029906697735687253,
+      "loss": 0.5036,
+      "step": 744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029906375505734317,
+      "loss": 0.501,
+      "step": 745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990605272205323,
+      "loss": 0.2001,
+      "step": 746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029905729384656,
+      "loss": 0.3499,
+      "step": 747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029905405493554633,
+      "loss": 0.324,
+      "step": 748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990508104876117,
+      "loss": 0.2427,
+      "step": 749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029904756050287646,
+      "loss": 0.3883,
+      "step": 750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029904430498146146,
+      "loss": 0.4183,
+      "step": 751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029904104392348757,
+      "loss": 0.4166,
+      "step": 752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000299037777329076,
+      "loss": 0.2737,
+      "step": 753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000299034505198348,
+      "loss": 0.4867,
+      "step": 754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990312275314252,
+      "loss": 0.3071,
+      "step": 755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990279443284293,
+      "loss": 0.1181,
+      "step": 756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029902465558948233,
+      "loss": 0.1702,
+      "step": 757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990213613147064,
+      "loss": 0.2976,
+      "step": 758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029901806150422393,
+      "loss": 0.3961,
+      "step": 759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990147561581574,
+      "loss": 0.3832,
+      "step": 760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990114452766297,
+      "loss": 0.2617,
+      "step": 761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002990081288597638,
+      "loss": 0.3486,
+      "step": 762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029900480690768283,
+      "loss": 0.3362,
+      "step": 763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029900147942051025,
+      "loss": 0.3107,
+      "step": 764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989981463983697,
+      "loss": 0.2914,
+      "step": 765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989948078413849,
+      "loss": 0.1994,
+      "step": 766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989914637496799,
+      "loss": 0.4098,
+      "step": 767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029898811412337896,
+      "loss": 0.4923,
+      "step": 768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989847589626065,
+      "loss": 0.1955,
+      "step": 769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989813982674871,
+      "loss": 0.3696,
+      "step": 770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00029897803203814566,
+      "loss": 0.3544,
+      "step": 771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002989746602747072,
+      "loss": 0.3265,
+      "step": 772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000298971282977297,
+      "loss": 0.4474,
+      "step": 773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989679001460404,
+      "loss": 0.1689,
+      "step": 774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029896451178106325,
+      "loss": 0.2232,
+      "step": 775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029896111788249135,
+      "loss": 0.1775,
+      "step": 776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989577184504507,
+      "loss": 0.3358,
+      "step": 777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029895431348506753,
+      "loss": 0.1955,
+      "step": 778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029895090298646856,
+      "loss": 0.4043,
+      "step": 779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029894748695478025,
+      "loss": 0.2092,
+      "step": 780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989440653901296,
+      "loss": 0.3913,
+      "step": 781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029894063829264376,
+      "loss": 0.3509,
+      "step": 782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029893720566244995,
+      "loss": 0.2773,
+      "step": 783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029893376749967566,
+      "loss": 0.1448,
+      "step": 784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989303238044487,
+      "loss": 0.2642,
+      "step": 785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989268745768969,
+      "loss": 0.2806,
+      "step": 786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989234198171484,
+      "loss": 0.2678,
+      "step": 787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029891995952533167,
+      "loss": 0.1944,
+      "step": 788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989164937015751,
+      "loss": 0.3416,
+      "step": 789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029891302234600747,
+      "loss": 0.1723,
+      "step": 790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029890954545875773,
+      "loss": 0.5298,
+      "step": 791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002989060630399551,
+      "loss": 0.3644,
+      "step": 792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029890257508972885,
+      "loss": 0.1947,
+      "step": 793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988990816082086,
+      "loss": 0.2186,
+      "step": 794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988955825955241,
+      "loss": 0.1525,
+      "step": 795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988920780518054,
+      "loss": 0.4187,
+      "step": 796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029888856797718257,
+      "loss": 0.2263,
+      "step": 797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029888505237178603,
+      "loss": 0.2353,
+      "step": 798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029888153123574645,
+      "loss": 0.2691,
+      "step": 799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029887800456919456,
+      "loss": 0.3462,
+      "step": 800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029887447237226134,
+      "loss": 0.2493,
+      "step": 801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988709346450781,
+      "loss": 0.3273,
+      "step": 802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988673913877761,
+      "loss": 0.5431,
+      "step": 803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002988638426004872,
+      "loss": 0.1533,
+      "step": 804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000298860288283343,
+      "loss": 0.5905,
+      "step": 805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00029885672843647565,
+      "loss": 0.321,
+      "step": 806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029885316306001733,
+      "loss": 0.4048,
+      "step": 807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988495921541005,
+      "loss": 0.2949,
+      "step": 808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988460157188579,
+      "loss": 0.3131,
+      "step": 809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029884243375442225,
+      "loss": 0.1388,
+      "step": 810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988388462609266,
+      "loss": 0.0826,
+      "step": 811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988352532385044,
+      "loss": 0.3406,
+      "step": 812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988316546872889,
+      "loss": 0.354,
+      "step": 813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029882805060741394,
+      "loss": 0.1797,
+      "step": 814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988244409990133,
+      "loss": 0.2594,
+      "step": 815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988208258622211,
+      "loss": 0.3113,
+      "step": 816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988172051971717,
+      "loss": 0.3167,
+      "step": 817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988135790039994,
+      "loss": 0.2237,
+      "step": 818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002988099472828391,
+      "loss": 0.2831,
+      "step": 819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029880631003382564,
+      "loss": 0.3461,
+      "step": 820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029880266725709417,
+      "loss": 0.3189,
+      "step": 821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987990189527799,
+      "loss": 0.2852,
+      "step": 822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987953651210185,
+      "loss": 0.2256,
+      "step": 823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029879170576194554,
+      "loss": 0.2242,
+      "step": 824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987880408756971,
+      "loss": 0.4494,
+      "step": 825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987843704624093,
+      "loss": 0.4021,
+      "step": 826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029878069452221834,
+      "loss": 0.4495,
+      "step": 827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029877701305526094,
+      "loss": 0.2871,
+      "step": 828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029877332606167376,
+      "loss": 0.2543,
+      "step": 829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987696335415938,
+      "loss": 0.2894,
+      "step": 830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987659354951582,
+      "loss": 0.3151,
+      "step": 831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987622319225044,
+      "loss": 0.3337,
+      "step": 832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987585228237699,
+      "loss": 0.2579,
+      "step": 833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987548081990925,
+      "loss": 0.1255,
+      "step": 834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029875108804861016,
+      "loss": 0.3114,
+      "step": 835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029874736237246117,
+      "loss": 0.2699,
+      "step": 836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029874363117078385,
+      "loss": 0.4193,
+      "step": 837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987398944437168,
+      "loss": 0.5109,
+      "step": 838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0002987361521913989,
+      "loss": 0.3867,
+      "step": 839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000298732404413969,
+      "loss": 0.2645,
+      "step": 840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987286511115666,
+      "loss": 0.2494,
+      "step": 841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987248922843308,
+      "loss": 0.4383,
+      "step": 842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029872112793240144,
+      "loss": 0.2563,
+      "step": 843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987173580559183,
+      "loss": 0.2954,
+      "step": 844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987135826550214,
+      "loss": 0.2559,
+      "step": 845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000298709801729851,
+      "loss": 0.215,
+      "step": 846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002987060152805476,
+      "loss": 0.0728,
+      "step": 847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029870222330725173,
+      "loss": 0.2341,
+      "step": 848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986984258101044,
+      "loss": 0.4142,
+      "step": 849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986946227892466,
+      "loss": 0.4502,
+      "step": 850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029869081424481963,
+      "loss": 0.3313,
+      "step": 851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986870001769649,
+      "loss": 0.467,
+      "step": 852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986831805858241,
+      "loss": 0.355,
+      "step": 853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029867935547153924,
+      "loss": 0.1509,
+      "step": 854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029867552483425227,
+      "loss": 0.598,
+      "step": 855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029867168867410556,
+      "loss": 0.3371,
+      "step": 856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029866784699124154,
+      "loss": 0.2219,
+      "step": 857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029866399978580304,
+      "loss": 0.1655,
+      "step": 858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986601470579328,
+      "loss": 0.2223,
+      "step": 859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029865628880777415,
+      "loss": 0.1926,
+      "step": 860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986524250354702,
+      "loss": 0.2586,
+      "step": 861
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029864855574116467,
+      "loss": 0.2022,
+      "step": 862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986446809250012,
+      "loss": 0.0957,
+      "step": 863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986408005871237,
+      "loss": 0.2782,
+      "step": 864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986369147276763,
+      "loss": 0.31,
+      "step": 865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029863302334680347,
+      "loss": 0.1533,
+      "step": 866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986291264446497,
+      "loss": 0.34,
+      "step": 867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029862522402135963,
+      "loss": 0.3287,
+      "step": 868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029862131607707836,
+      "loss": 0.373,
+      "step": 869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0002986174026119511,
+      "loss": 0.2444,
+      "step": 870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029861348362612307,
+      "loss": 0.373,
+      "step": 871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00029860955911973995,
+      "loss": 0.0425,
+      "step": 872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002986056290929475,
+      "loss": 0.3146,
+      "step": 873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029860169354589173,
+      "loss": 0.4037,
+      "step": 874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985977524787188,
+      "loss": 0.3334,
+      "step": 875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029859380589157514,
+      "loss": 0.4214,
+      "step": 876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029858985378460734,
+      "loss": 0.3769,
+      "step": 877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985858961579622,
+      "loss": 0.0954,
+      "step": 878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985819330117867,
+      "loss": 0.3082,
+      "step": 879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985779643462282,
+      "loss": 0.5508,
+      "step": 880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985739901614339,
+      "loss": 0.2706,
+      "step": 881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029857001045755166,
+      "loss": 0.1477,
+      "step": 882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985660252347292,
+      "loss": 0.3255,
+      "step": 883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985620344931145,
+      "loss": 0.307,
+      "step": 884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029855803823285594,
+      "loss": 0.3799,
+      "step": 885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985540364541019,
+      "loss": 0.2414,
+      "step": 886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000298550029157001,
+      "loss": 0.194,
+      "step": 887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029854601634170215,
+      "loss": 0.2157,
+      "step": 888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985419980083544,
+      "loss": 0.446,
+      "step": 889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000298537974157107,
+      "loss": 0.3235,
+      "step": 890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985339447881095,
+      "loss": 0.17,
+      "step": 891
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029852990990151146,
+      "loss": 0.4609,
+      "step": 892
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029852586949746286,
+      "loss": 0.2561,
+      "step": 893
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029852182357611377,
+      "loss": 0.2575,
+      "step": 894
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985177721376144,
+      "loss": 0.0305,
+      "step": 895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029851371518211544,
+      "loss": 0.1379,
+      "step": 896
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985096527097674,
+      "loss": 0.3842,
+      "step": 897
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029850558472072124,
+      "loss": 0.3285,
+      "step": 898
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002985015112151281,
+      "loss": 0.3125,
+      "step": 899
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029849743219313935,
+      "loss": 0.2236,
+      "step": 900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984933476549064,
+      "loss": 0.3962,
+      "step": 901
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984892576005811,
+      "loss": 0.3097,
+      "step": 902
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984851620303153,
+      "loss": 0.0667,
+      "step": 903
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002984810609442611,
+      "loss": 0.2275,
+      "step": 904
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000298476954342571,
+      "loss": 0.1367,
+      "step": 905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029847284222539737,
+      "loss": 0.3192,
+      "step": 906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984687245928931,
+      "loss": 0.2483,
+      "step": 907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984646014452111,
+      "loss": 0.1904,
+      "step": 908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984604727825045,
+      "loss": 0.4165,
+      "step": 909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984563386049267,
+      "loss": 0.2536,
+      "step": 910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984521989126312,
+      "loss": 0.3993,
+      "step": 911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984480537057719,
+      "loss": 0.1342,
+      "step": 912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984439029845027,
+      "loss": 0.1271,
+      "step": 913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029843974674897787,
+      "loss": 0.3424,
+      "step": 914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984355849993517,
+      "loss": 0.3562,
+      "step": 915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029843141773577885,
+      "loss": 0.3063,
+      "step": 916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984272449584141,
+      "loss": 0.34,
+      "step": 917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029842306666741245,
+      "loss": 0.3327,
+      "step": 918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984188828629291,
+      "loss": 0.2286,
+      "step": 919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984146935451195,
+      "loss": 0.3671,
+      "step": 920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029841049871413925,
+      "loss": 0.244,
+      "step": 921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029840629837014427,
+      "loss": 0.272,
+      "step": 922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002984020925132904,
+      "loss": 0.2823,
+      "step": 923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000298397881143734,
+      "loss": 0.2622,
+      "step": 924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029839366426163147,
+      "loss": 0.3328,
+      "step": 925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029838944186713954,
+      "loss": 0.3027,
+      "step": 926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029838521396041495,
+      "loss": 0.2234,
+      "step": 927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983809805416148,
+      "loss": 0.0301,
+      "step": 928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029837674161089633,
+      "loss": 0.1884,
+      "step": 929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983724971684171,
+      "loss": 0.4672,
+      "step": 930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983682472143346,
+      "loss": 0.2784,
+      "step": 931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029836399174880683,
+      "loss": 0.2994,
+      "step": 932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029835973077199185,
+      "loss": 0.255,
+      "step": 933
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029835546428404793,
+      "loss": 0.2622,
+      "step": 934
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983511922851336,
+      "loss": 0.3221,
+      "step": 935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029834691477540745,
+      "loss": 0.2978,
+      "step": 936
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029834263175502844,
+      "loss": 0.1968,
+      "step": 937
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0002983383432241557,
+      "loss": 0.4291,
+      "step": 938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029833404918294855,
+      "loss": 0.1741,
+      "step": 939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029832974963156636,
+      "loss": 0.3613,
+      "step": 940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029832544457016904,
+      "loss": 0.1915,
+      "step": 941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983211339989164,
+      "loss": 0.2864,
+      "step": 942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983168179179685,
+      "loss": 0.0486,
+      "step": 943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983124963274858,
+      "loss": 0.3624,
+      "step": 944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983081692276288,
+      "loss": 0.2245,
+      "step": 945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002983038366185582,
+      "loss": 0.3403,
+      "step": 946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000298299498500435,
+      "loss": 0.2571,
+      "step": 947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029829515487342026,
+      "loss": 0.357,
+      "step": 948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029829080573767546,
+      "loss": 0.3482,
+      "step": 949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982864510933621,
+      "loss": 0.2411,
+      "step": 950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982820909406418,
+      "loss": 0.1143,
+      "step": 951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029827772527967676,
+      "loss": 0.4096,
+      "step": 952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000298273354110629,
+      "loss": 0.2987,
+      "step": 953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000298268977433661,
+      "loss": 0.1878,
+      "step": 954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029826459524893523,
+      "loss": 0.4587,
+      "step": 955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029826020755661454,
+      "loss": 0.2755,
+      "step": 956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029825581435686193,
+      "loss": 0.211,
+      "step": 957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029825141564984053,
+      "loss": 0.2029,
+      "step": 958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029824701143571383,
+      "loss": 0.5266,
+      "step": 959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029824260171464534,
+      "loss": 0.2694,
+      "step": 960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029823818648679897,
+      "loss": 0.2594,
+      "step": 961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029823376575233867,
+      "loss": 0.2366,
+      "step": 962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982293395114286,
+      "loss": 0.1215,
+      "step": 963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982249077642333,
+      "loss": 0.3594,
+      "step": 964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982204705109173,
+      "loss": 0.4075,
+      "step": 965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0002982160277516455,
+      "loss": 0.1653,
+      "step": 966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029821157948658295,
+      "loss": 0.2141,
+      "step": 967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029820712571589477,
+      "loss": 0.2751,
+      "step": 968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029820266643974656,
+      "loss": 0.1869,
+      "step": 969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029819820165830383,
+      "loss": 0.3388,
+      "step": 970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029819373137173253,
+      "loss": 0.2138,
+      "step": 971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981892555801987,
+      "loss": 0.1641,
+      "step": 972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981847742838686,
+      "loss": 0.2002,
+      "step": 973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029818028748290857,
+      "loss": 0.27,
+      "step": 974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029817579517748546,
+      "loss": 0.3715,
+      "step": 975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981712973677661,
+      "loss": 0.1403,
+      "step": 976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029816679405391755,
+      "loss": 0.2226,
+      "step": 977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981622852361071,
+      "loss": 0.2808,
+      "step": 978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981577709145022,
+      "loss": 0.3262,
+      "step": 979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029815325108927063,
+      "loss": 0.2426,
+      "step": 980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981487257605802,
+      "loss": 0.2421,
+      "step": 981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029814419492859906,
+      "loss": 0.4084,
+      "step": 982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981396585934955,
+      "loss": 0.3693,
+      "step": 983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029813511675543804,
+      "loss": 0.1708,
+      "step": 984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029813056941459546,
+      "loss": 0.3751,
+      "step": 985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029812601657113656,
+      "loss": 0.2527,
+      "step": 986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029812145822523053,
+      "loss": 0.2965,
+      "step": 987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029811689437704667,
+      "loss": 0.2195,
+      "step": 988
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029811232502675454,
+      "loss": 0.1591,
+      "step": 989
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981077501745239,
+      "loss": 0.2568,
+      "step": 990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002981031698205246,
+      "loss": 0.4367,
+      "step": 991
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029809858396492693,
+      "loss": 0.333,
+      "step": 992
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029809399260790107,
+      "loss": 0.4648,
+      "step": 993
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980893957496177,
+      "loss": 0.271,
+      "step": 994
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980847933902476,
+      "loss": 0.1867,
+      "step": 995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029808018552996167,
+      "loss": 0.3938,
+      "step": 996
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029807557216893107,
+      "loss": 0.2969,
+      "step": 997
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980709533073271,
+      "loss": 0.3577,
+      "step": 998
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029806632894532155,
+      "loss": 0.4089,
+      "step": 999
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029806169908308603,
+      "loss": 0.2866,
+      "step": 1000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980570637207926,
+      "loss": 0.1652,
+      "step": 1001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00029805242285861345,
+      "loss": 0.327,
+      "step": 1002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980477764967209,
+      "loss": 0.1602,
+      "step": 1003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0002980431246352876,
+      "loss": 0.4538,
+      "step": 1004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980384672744864,
+      "loss": 0.2408,
+      "step": 1005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980338044144902,
+      "loss": 0.392,
+      "step": 1006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029802913605547234,
+      "loss": 0.1815,
+      "step": 1007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980244621976061,
+      "loss": 0.3108,
+      "step": 1008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980197828410652,
+      "loss": 0.214,
+      "step": 1009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029801509798602343,
+      "loss": 0.2501,
+      "step": 1010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980104076326548,
+      "loss": 0.2969,
+      "step": 1011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002980057117811336,
+      "loss": 0.3037,
+      "step": 1012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029800101043163423,
+      "loss": 0.2866,
+      "step": 1013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979963035843313,
+      "loss": 0.4018,
+      "step": 1014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979915912393997,
+      "loss": 0.3364,
+      "step": 1015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029798687339701446,
+      "loss": 0.213,
+      "step": 1016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979821500573509,
+      "loss": 0.2939,
+      "step": 1017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029797742122058434,
+      "loss": 0.3376,
+      "step": 1018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029797268688689055,
+      "loss": 0.5265,
+      "step": 1019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029796794705644537,
+      "loss": 0.2292,
+      "step": 1020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979632017294248,
+      "loss": 0.177,
+      "step": 1021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979584509060053,
+      "loss": 0.3058,
+      "step": 1022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979536945863632,
+      "loss": 0.2582,
+      "step": 1023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029794893277067523,
+      "loss": 0.4344,
+      "step": 1024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979441654591182,
+      "loss": 0.3349,
+      "step": 1025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029793939265186935,
+      "loss": 0.3343,
+      "step": 1026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979346143491058,
+      "loss": 0.4951,
+      "step": 1027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029792983055100525,
+      "loss": 0.0001,
+      "step": 1028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979250412577452,
+      "loss": 0.2421,
+      "step": 1029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979202464695037,
+      "loss": 0.2686,
+      "step": 1030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029791544618645876,
+      "loss": 0.2185,
+      "step": 1031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979106404087888,
+      "loss": 0.3837,
+      "step": 1032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002979058291366723,
+      "loss": 0.2687,
+      "step": 1033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029790101237028795,
+      "loss": 0.2029,
+      "step": 1034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029789619010981474,
+      "loss": 0.3187,
+      "step": 1035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00029789136235543174,
+      "loss": 0.3958,
+      "step": 1036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002978865291073183,
+      "loss": 0.2381,
+      "step": 1037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978816903656541,
+      "loss": 0.1465,
+      "step": 1038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978768461306186,
+      "loss": 0.3293,
+      "step": 1039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000297871996402392,
+      "loss": 0.2872,
+      "step": 1040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029786714118115433,
+      "loss": 0.2983,
+      "step": 1041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000297862280467086,
+      "loss": 0.49,
+      "step": 1042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029785741426036755,
+      "loss": 0.1385,
+      "step": 1043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978525425611797,
+      "loss": 0.326,
+      "step": 1044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029784766536970357,
+      "loss": 0.1912,
+      "step": 1045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978427826861202,
+      "loss": 0.3778,
+      "step": 1046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000297837894510611,
+      "loss": 0.2689,
+      "step": 1047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978330008433575,
+      "loss": 0.1733,
+      "step": 1048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978281016845416,
+      "loss": 0.2042,
+      "step": 1049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978231970343451,
+      "loss": 0.2964,
+      "step": 1050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029781828689295046,
+      "loss": 0.1608,
+      "step": 1051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978133712605399,
+      "loss": 0.2945,
+      "step": 1052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029780845013729603,
+      "loss": 0.29,
+      "step": 1053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002978035235234017,
+      "loss": 0.2813,
+      "step": 1054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977985914190398,
+      "loss": 0.4729,
+      "step": 1055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977936538243938,
+      "loss": 0.258,
+      "step": 1056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977887107396469,
+      "loss": 0.2038,
+      "step": 1057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029778376216498274,
+      "loss": 0.2652,
+      "step": 1058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029777880810058526,
+      "loss": 0.1605,
+      "step": 1059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977738485466384,
+      "loss": 0.1611,
+      "step": 1060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977688835033264,
+      "loss": 0.2582,
+      "step": 1061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029776391297083363,
+      "loss": 0.3598,
+      "step": 1062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977589369493449,
+      "loss": 0.2632,
+      "step": 1063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029775395543904493,
+      "loss": 0.3757,
+      "step": 1064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029774896844011887,
+      "loss": 0.3696,
+      "step": 1065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029774397595275177,
+      "loss": 0.1559,
+      "step": 1066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977389779771293,
+      "loss": 0.3168,
+      "step": 1067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029773397451343707,
+      "loss": 0.3899,
+      "step": 1068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029772896556186086,
+      "loss": 0.2815,
+      "step": 1069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002977239511225868,
+      "loss": 0.4508,
+      "step": 1070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029771893119580117,
+      "loss": 0.2459,
+      "step": 1071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002977139057816904,
+      "loss": 0.3609,
+      "step": 1072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002977088748804412,
+      "loss": 0.3532,
+      "step": 1073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002977038384922405,
+      "loss": 0.3497,
+      "step": 1074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976987966172753,
+      "loss": 0.319,
+      "step": 1075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976937492557329,
+      "loss": 0.2172,
+      "step": 1076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029768869640780084,
+      "loss": 0.3636,
+      "step": 1077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976836380736668,
+      "loss": 0.2489,
+      "step": 1078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029767857425351877,
+      "loss": 0.4424,
+      "step": 1079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029767350494754464,
+      "loss": 0.2484,
+      "step": 1080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029766843015593296,
+      "loss": 0.2877,
+      "step": 1081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029766334987887206,
+      "loss": 0.2808,
+      "step": 1082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029765826411655075,
+      "loss": 0.4201,
+      "step": 1083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029765317286915794,
+      "loss": 0.3839,
+      "step": 1084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029764807613688273,
+      "loss": 0.4093,
+      "step": 1085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976429739199145,
+      "loss": 0.3341,
+      "step": 1086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029763786621844274,
+      "loss": 0.3289,
+      "step": 1087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029763275303265716,
+      "loss": 0.3535,
+      "step": 1088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029762763436274773,
+      "loss": 0.2452,
+      "step": 1089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029762251020890463,
+      "loss": 0.091,
+      "step": 1090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976173805713182,
+      "loss": 0.3739,
+      "step": 1091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029761224545017895,
+      "loss": 0.1952,
+      "step": 1092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029760710484567763,
+      "loss": 0.3398,
+      "step": 1093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002976019587580052,
+      "loss": 0.2518,
+      "step": 1094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975968071873529,
+      "loss": 0.1673,
+      "step": 1095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029759165013391203,
+      "loss": 0.3159,
+      "step": 1096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975864875978741,
+      "loss": 0.2788,
+      "step": 1097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000297581319579431,
+      "loss": 0.3042,
+      "step": 1098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029757614607877464,
+      "loss": 0.2777,
+      "step": 1099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029757096709609725,
+      "loss": 0.2962,
+      "step": 1100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029756578263159117,
+      "loss": 0.4973,
+      "step": 1101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00029756059268544894,
+      "loss": 0.3438,
+      "step": 1102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002975553972578635,
+      "loss": 0.3453,
+      "step": 1103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975501963490277,
+      "loss": 0.3626,
+      "step": 1104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029754498995913475,
+      "loss": 0.0823,
+      "step": 1105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029753977808837816,
+      "loss": 0.2012,
+      "step": 1106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029753456073695146,
+      "loss": 0.1351,
+      "step": 1107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975293379050485,
+      "loss": 0.1471,
+      "step": 1108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975241095928632,
+      "loss": 0.3628,
+      "step": 1109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975188758005899,
+      "loss": 0.1623,
+      "step": 1110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975136365284229,
+      "loss": 0.4324,
+      "step": 1111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975083917765569,
+      "loss": 0.3196,
+      "step": 1112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002975031415451867,
+      "loss": 0.2639,
+      "step": 1113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974978858345074,
+      "loss": 0.1717,
+      "step": 1114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029749262464471416,
+      "loss": 0.1432,
+      "step": 1115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974873579760024,
+      "loss": 0.3492,
+      "step": 1116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974820858285678,
+      "loss": 0.2874,
+      "step": 1117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974768082026062,
+      "loss": 0.1644,
+      "step": 1118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974715250983136,
+      "loss": 0.3983,
+      "step": 1119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974662365158864,
+      "loss": 0.1569,
+      "step": 1120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974609424555209,
+      "loss": 0.2908,
+      "step": 1121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974556429174138,
+      "loss": 0.2726,
+      "step": 1122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000297450337901762,
+      "loss": 0.3583,
+      "step": 1123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029744502740876257,
+      "loss": 0.2787,
+      "step": 1124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974397114386127,
+      "loss": 0.3737,
+      "step": 1125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974343899915099,
+      "loss": 0.2393,
+      "step": 1126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974290630676519,
+      "loss": 0.2516,
+      "step": 1127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029742373066723657,
+      "loss": 0.1069,
+      "step": 1128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974183927904619,
+      "loss": 0.3344,
+      "step": 1129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974130494375263,
+      "loss": 0.2736,
+      "step": 1130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974077006086281,
+      "loss": 0.5029,
+      "step": 1131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002974023463039662,
+      "loss": 0.1002,
+      "step": 1132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029739698652373934,
+      "loss": 0.236,
+      "step": 1133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029739162126814667,
+      "loss": 0.2753,
+      "step": 1134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029738625053738746,
+      "loss": 0.1125,
+      "step": 1135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00029738087433166125,
+      "loss": 0.3363,
+      "step": 1136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973754926511678,
+      "loss": 0.452,
+      "step": 1137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000297370105496107,
+      "loss": 0.1976,
+      "step": 1138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973647128666788,
+      "loss": 0.3582,
+      "step": 1139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973593147630838,
+      "loss": 0.173,
+      "step": 1140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973539111855223,
+      "loss": 0.3876,
+      "step": 1141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973485021341952,
+      "loss": 0.1884,
+      "step": 1142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973430876093033,
+      "loss": 0.3877,
+      "step": 1143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973376676110478,
+      "loss": 0.1421,
+      "step": 1144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973322421396299,
+      "loss": 0.3675,
+      "step": 1145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029732681119525137,
+      "loss": 0.398,
+      "step": 1146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029732137477811373,
+      "loss": 0.243,
+      "step": 1147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029731593288841916,
+      "loss": 0.2215,
+      "step": 1148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002973104855263696,
+      "loss": 0.1239,
+      "step": 1149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029730503269216754,
+      "loss": 0.2609,
+      "step": 1150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972995743860154,
+      "loss": 0.3498,
+      "step": 1151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972941106081161,
+      "loss": 0.3149,
+      "step": 1152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972886413586725,
+      "loss": 0.3199,
+      "step": 1153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972831666378878,
+      "loss": 0.328,
+      "step": 1154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029727768644596537,
+      "loss": 0.1542,
+      "step": 1155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029727220078310874,
+      "loss": 0.1548,
+      "step": 1156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972667096495218,
+      "loss": 0.2062,
+      "step": 1157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972612130454084,
+      "loss": 0.504,
+      "step": 1158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029725571097097275,
+      "loss": 0.3046,
+      "step": 1159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972502034264193,
+      "loss": 0.2044,
+      "step": 1160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972446904119526,
+      "loss": 0.2991,
+      "step": 1161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029723917192777744,
+      "loss": 0.3112,
+      "step": 1162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972336479740988,
+      "loss": 0.2388,
+      "step": 1163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972281185511219,
+      "loss": 0.2511,
+      "step": 1164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029722258365905217,
+      "loss": 0.2506,
+      "step": 1165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029721704329809515,
+      "loss": 0.0758,
+      "step": 1166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972114974684567,
+      "loss": 0.2755,
+      "step": 1167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002972059461703428,
+      "loss": 0.2632,
+      "step": 1168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00029720038940395967,
+      "loss": 0.1529,
+      "step": 1169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971948271695138,
+      "loss": 0.3049,
+      "step": 1170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029718925946721166,
+      "loss": 0.2729,
+      "step": 1171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971836862972602,
+      "loss": 0.2231,
+      "step": 1172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971781076598664,
+      "loss": 0.2673,
+      "step": 1173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029717252355523753,
+      "loss": 0.2893,
+      "step": 1174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971669339835809,
+      "loss": 0.1073,
+      "step": 1175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029716133894510435,
+      "loss": 0.0919,
+      "step": 1176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971557384400155,
+      "loss": 0.3493,
+      "step": 1177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971501324685226,
+      "loss": 0.0997,
+      "step": 1178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971445210308337,
+      "loss": 0.2081,
+      "step": 1179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029713890412715743,
+      "loss": 0.3059,
+      "step": 1180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971332817577023,
+      "loss": 0.336,
+      "step": 1181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971276539226772,
+      "loss": 0.3506,
+      "step": 1182
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029712202062229123,
+      "loss": 0.2015,
+      "step": 1183
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002971163818567536,
+      "loss": 0.1809,
+      "step": 1184
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029711073762627387,
+      "loss": 0.0883,
+      "step": 1185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029710508793106156,
+      "loss": 0.2532,
+      "step": 1186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029709943277132664,
+      "loss": 0.3142,
+      "step": 1187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970937721472792,
+      "loss": 0.2882,
+      "step": 1188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970881060591294,
+      "loss": 0.3583,
+      "step": 1189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970824345070878,
+      "loss": 0.2575,
+      "step": 1190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970767574913651,
+      "loss": 0.4113,
+      "step": 1191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970710750121722,
+      "loss": 0.3148,
+      "step": 1192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970653870697201,
+      "loss": 0.2744,
+      "step": 1193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029705969366422016,
+      "loss": 0.3491,
+      "step": 1194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970539947958838,
+      "loss": 0.1166,
+      "step": 1195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0002970482904649228,
+      "loss": 0.3173,
+      "step": 1196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000297042580671549,
+      "loss": 0.3236,
+      "step": 1197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029703686541597454,
+      "loss": 0.3401,
+      "step": 1198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029703114469841174,
+      "loss": 0.3872,
+      "step": 1199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029702541851907305,
+      "loss": 0.2197,
+      "step": 1200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029701968687817124,
+      "loss": 0.2567,
+      "step": 1201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029701394977591915,
+      "loss": 0.2395,
+      "step": 1202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029700820721253,
+      "loss": 0.2249,
+      "step": 1203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000297002459188217,
+      "loss": 0.2791,
+      "step": 1204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029699670570319374,
+      "loss": 0.2251,
+      "step": 1205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029699094675767395,
+      "loss": 0.2566,
+      "step": 1206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969851823518715,
+      "loss": 0.2271,
+      "step": 1207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029697941248600055,
+      "loss": 0.2225,
+      "step": 1208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029697363716027546,
+      "loss": 0.071,
+      "step": 1209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029696785637491074,
+      "loss": 0.1807,
+      "step": 1210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969620701301211,
+      "loss": 0.3086,
+      "step": 1211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969562784261216,
+      "loss": 0.4453,
+      "step": 1212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029695048126312726,
+      "loss": 0.2389,
+      "step": 1213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029694467864135345,
+      "loss": 0.4797,
+      "step": 1214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029693887056101575,
+      "loss": 0.1506,
+      "step": 1215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029693305702232993,
+      "loss": 0.1264,
+      "step": 1216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969272380255119,
+      "loss": 0.2647,
+      "step": 1217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002969214135707778,
+      "loss": 0.1784,
+      "step": 1218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029691558365834407,
+      "loss": 0.3888,
+      "step": 1219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029690974828842725,
+      "loss": 0.4533,
+      "step": 1220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029690390746124406,
+      "loss": 0.1062,
+      "step": 1221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968980611770115,
+      "loss": 0.4002,
+      "step": 1222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029689220943594676,
+      "loss": 0.3731,
+      "step": 1223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968863522382672,
+      "loss": 0.1414,
+      "step": 1224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029688048958419037,
+      "loss": 0.3628,
+      "step": 1225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029687462147393414,
+      "loss": 0.4786,
+      "step": 1226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968687479077163,
+      "loss": 0.2193,
+      "step": 1227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968628688857553,
+      "loss": 0.2831,
+      "step": 1228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029685698440826933,
+      "loss": 0.2398,
+      "step": 1229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000296851094475477,
+      "loss": 0.2529,
+      "step": 1230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968451990875972,
+      "loss": 0.4112,
+      "step": 1231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0002968392982448489,
+      "loss": 0.1895,
+      "step": 1232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029683339194745126,
+      "loss": 0.429,
+      "step": 1233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029682748019562366,
+      "loss": 0.4093,
+      "step": 1234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00029682156298958573,
+      "loss": 0.121,
+      "step": 1235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029681564032955736,
+      "loss": 0.1331,
+      "step": 1236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002968097122157584,
+      "loss": 0.3481,
+      "step": 1237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029680377864840926,
+      "loss": 0.4508,
+      "step": 1238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967978396277301,
+      "loss": 0.3433,
+      "step": 1239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967918951539418,
+      "loss": 0.1888,
+      "step": 1240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029678594522726493,
+      "loss": 0.3134,
+      "step": 1241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029677998984792075,
+      "loss": 0.188,
+      "step": 1242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029677402901613035,
+      "loss": 0.2568,
+      "step": 1243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029676806273211517,
+      "loss": 0.1592,
+      "step": 1244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967620909960969,
+      "loss": 0.4471,
+      "step": 1245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967561138082972,
+      "loss": 0.2207,
+      "step": 1246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967501311689383,
+      "loss": 0.1948,
+      "step": 1247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029674414307824244,
+      "loss": 0.2478,
+      "step": 1248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967381495364319,
+      "loss": 0.126,
+      "step": 1249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029673215054372946,
+      "loss": 0.204,
+      "step": 1250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029672614610035786,
+      "loss": 0.4296,
+      "step": 1251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029672013620654023,
+      "loss": 0.0881,
+      "step": 1252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029671412086249986,
+      "loss": 0.3226,
+      "step": 1253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029670810006846005,
+      "loss": 0.4581,
+      "step": 1254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029670207382464456,
+      "loss": 0.426,
+      "step": 1255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029669604213127725,
+      "loss": 0.2199,
+      "step": 1256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029669000498858215,
+      "loss": 0.3364,
+      "step": 1257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029668396239678353,
+      "loss": 0.0943,
+      "step": 1258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966779143561058,
+      "loss": 0.2012,
+      "step": 1259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966718608667738,
+      "loss": 0.2538,
+      "step": 1260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966658019290122,
+      "loss": 0.2524,
+      "step": 1261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966597375430461,
+      "loss": 0.2027,
+      "step": 1262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966536677091009,
+      "loss": 0.3939,
+      "step": 1263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029664759242740204,
+      "loss": 0.2256,
+      "step": 1264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966415116981751,
+      "loss": 0.2954,
+      "step": 1265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00029663542552164603,
+      "loss": 0.1404,
+      "step": 1266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002966293338980409,
+      "loss": 0.3788,
+      "step": 1267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029662323682758606,
+      "loss": 0.4256,
+      "step": 1268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029661713431050786,
+      "loss": 0.3142,
+      "step": 1269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002966110263470331,
+      "loss": 0.1264,
+      "step": 1270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029660491293738867,
+      "loss": 0.239,
+      "step": 1271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029659879408180164,
+      "loss": 0.2729,
+      "step": 1272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965926697804993,
+      "loss": 0.3574,
+      "step": 1273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965865400337091,
+      "loss": 0.2101,
+      "step": 1274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965804048416589,
+      "loss": 0.2988,
+      "step": 1275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029657426420457637,
+      "loss": 0.1564,
+      "step": 1276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965681181226899,
+      "loss": 0.179,
+      "step": 1277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965619665962276,
+      "loss": 0.2958,
+      "step": 1278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296555809625418,
+      "loss": 0.2905,
+      "step": 1279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965496472104898,
+      "loss": 0.3293,
+      "step": 1280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296543479351672,
+      "loss": 0.3049,
+      "step": 1281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029653730604919373,
+      "loss": 0.4284,
+      "step": 1282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029653112730328423,
+      "loss": 0.4334,
+      "step": 1283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000296524943114173,
+      "loss": 0.4971,
+      "step": 1284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965187534820899,
+      "loss": 0.1467,
+      "step": 1285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965125584072647,
+      "loss": 0.2174,
+      "step": 1286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002965063578899276,
+      "loss": 0.3223,
+      "step": 1287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029650015193030893,
+      "loss": 0.3447,
+      "step": 1288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029649394052863923,
+      "loss": 0.1574,
+      "step": 1289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002964877236851493,
+      "loss": 0.264,
+      "step": 1290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002964815014000699,
+      "loss": 0.452,
+      "step": 1291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029647527367363235,
+      "loss": 0.1628,
+      "step": 1292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002964690405060679,
+      "loss": 0.4697,
+      "step": 1293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002964628018976081,
+      "loss": 0.293,
+      "step": 1294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029645655784848475,
+      "loss": 0.2422,
+      "step": 1295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029645030835892967,
+      "loss": 0.2475,
+      "step": 1296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002964440534291752,
+      "loss": 0.3514,
+      "step": 1297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029643779305945354,
+      "loss": 0.5646,
+      "step": 1298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029643152724999724,
+      "loss": 0.4192,
+      "step": 1299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029642525600103917,
+      "loss": 0.3455,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002964189793128122,
+      "loss": 0.3523,
+      "step": 1301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002964126971855496,
+      "loss": 0.3214,
+      "step": 1302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002964064096194845,
+      "loss": 0.1744,
+      "step": 1303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029640011661485075,
+      "loss": 0.2651,
+      "step": 1304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029639381817188196,
+      "loss": 0.1123,
+      "step": 1305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963875142908121,
+      "loss": 0.2294,
+      "step": 1306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963812049718753,
+      "loss": 0.1689,
+      "step": 1307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963748902153061,
+      "loss": 0.3738,
+      "step": 1308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963685700213389,
+      "loss": 0.1853,
+      "step": 1309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963622443902086,
+      "loss": 0.2779,
+      "step": 1310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963559133221501,
+      "loss": 0.3452,
+      "step": 1311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963495768173986,
+      "loss": 0.1872,
+      "step": 1312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963432348761895,
+      "loss": 0.4343,
+      "step": 1313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963368874987583,
+      "loss": 0.2362,
+      "step": 1314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029633053468534096,
+      "loss": 0.179,
+      "step": 1315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029632417643617333,
+      "loss": 0.274,
+      "step": 1316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963178127514916,
+      "loss": 0.2406,
+      "step": 1317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029631144363153223,
+      "loss": 0.2242,
+      "step": 1318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002963050690765318,
+      "loss": 0.2934,
+      "step": 1319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962986890867271,
+      "loss": 0.2068,
+      "step": 1320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029629230366235506,
+      "loss": 0.2605,
+      "step": 1321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000296285912803653,
+      "loss": 0.2402,
+      "step": 1322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962795165108582,
+      "loss": 0.2616,
+      "step": 1323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962731147842084,
+      "loss": 0.2272,
+      "step": 1324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029626670762394127,
+      "loss": 0.2791,
+      "step": 1325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029626029503029486,
+      "loss": 0.3122,
+      "step": 1326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962538770035074,
+      "loss": 0.2994,
+      "step": 1327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029624745354381733,
+      "loss": 0.2191,
+      "step": 1328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029624102465146326,
+      "loss": 0.2085,
+      "step": 1329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962345903266839,
+      "loss": 0.2751,
+      "step": 1330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029622815056971837,
+      "loss": 0.2281,
+      "step": 1331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00029622170538080584,
+      "loss": 0.2943,
+      "step": 1332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0002962152547601858,
+      "loss": 0.3107,
+      "step": 1333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029620879870809777,
+      "loss": 0.2572,
+      "step": 1334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029620233722478165,
+      "loss": 0.2104,
+      "step": 1335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961958703104774,
+      "loss": 0.3423,
+      "step": 1336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029618939796542524,
+      "loss": 0.1747,
+      "step": 1337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029618292018986574,
+      "loss": 0.1385,
+      "step": 1338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961764369840394,
+      "loss": 0.2965,
+      "step": 1339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961699483481871,
+      "loss": 0.1608,
+      "step": 1340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029616345428254984,
+      "loss": 0.1914,
+      "step": 1341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029615695478736885,
+      "loss": 0.4196,
+      "step": 1342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961504498628856,
+      "loss": 0.3067,
+      "step": 1343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029614393950934176,
+      "loss": 0.1065,
+      "step": 1344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961374237269791,
+      "loss": 0.3306,
+      "step": 1345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961309025160397,
+      "loss": 0.0247,
+      "step": 1346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002961243758767658,
+      "loss": 0.3364,
+      "step": 1347
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029611784380939983,
+      "loss": 0.2108,
+      "step": 1348
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029611130631418446,
+      "loss": 0.2044,
+      "step": 1349
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029610476339136254,
+      "loss": 0.3439,
+      "step": 1350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029609821504117704,
+      "loss": 0.2419,
+      "step": 1351
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029609166126387136,
+      "loss": 0.1019,
+      "step": 1352
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029608510205968885,
+      "loss": 0.332,
+      "step": 1353
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960785374288732,
+      "loss": 0.2942,
+      "step": 1354
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029607196737166825,
+      "loss": 0.2483,
+      "step": 1355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029606539188831805,
+      "loss": 0.1645,
+      "step": 1356
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960588109790669,
+      "loss": 0.2791,
+      "step": 1357
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960522246441592,
+      "loss": 0.2799,
+      "step": 1358
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029604563288383967,
+      "loss": 0.2376,
+      "step": 1359
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029603903569835313,
+      "loss": 0.1061,
+      "step": 1360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029603243308794474,
+      "loss": 0.1928,
+      "step": 1361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029602582505285964,
+      "loss": 0.3585,
+      "step": 1362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960192115933434,
+      "loss": 0.178,
+      "step": 1363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002960125927096416,
+      "loss": 0.4452,
+      "step": 1364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029600596840200015,
+      "loss": 0.3869,
+      "step": 1365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00029599933867066516,
+      "loss": 0.2023,
+      "step": 1366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959927035158829,
+      "loss": 0.3678,
+      "step": 1367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959860629378998,
+      "loss": 0.1359,
+      "step": 1368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029597941693696253,
+      "loss": 0.1157,
+      "step": 1369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000295972765513318,
+      "loss": 0.2145,
+      "step": 1370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029596610866721335,
+      "loss": 0.1947,
+      "step": 1371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959594463988958,
+      "loss": 0.215,
+      "step": 1372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959527787086128,
+      "loss": 0.3155,
+      "step": 1373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029594610559661213,
+      "loss": 0.1856,
+      "step": 1374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959394270631415,
+      "loss": 0.2009,
+      "step": 1375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959327431084493,
+      "loss": 0.2801,
+      "step": 1376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002959260537327835,
+      "loss": 0.4042,
+      "step": 1377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029591935893639277,
+      "loss": 0.378,
+      "step": 1378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029591265871952577,
+      "loss": 0.4197,
+      "step": 1379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029590595308243134,
+      "loss": 0.1953,
+      "step": 1380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029589924202535865,
+      "loss": 0.2923,
+      "step": 1381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000295892525548557,
+      "loss": 0.1972,
+      "step": 1382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029588580365227577,
+      "loss": 0.2984,
+      "step": 1383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958790763367648,
+      "loss": 0.1803,
+      "step": 1384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029587234360227394,
+      "loss": 0.3282,
+      "step": 1385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958656054490533,
+      "loss": 0.296,
+      "step": 1386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958588618773531,
+      "loss": 0.309,
+      "step": 1387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958521128874239,
+      "loss": 0.2411,
+      "step": 1388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029584535847951645,
+      "loss": 0.2656,
+      "step": 1389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958385986538816,
+      "loss": 0.1958,
+      "step": 1390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029583183341077055,
+      "loss": 0.21,
+      "step": 1391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958250627504345,
+      "loss": 0.175,
+      "step": 1392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029581828667312496,
+      "loss": 0.2099,
+      "step": 1393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958115051790937,
+      "loss": 0.0958,
+      "step": 1394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002958047182685926,
+      "loss": 0.3187,
+      "step": 1395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002957979259418738,
+      "loss": 0.1397,
+      "step": 1396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002957911281991896,
+      "loss": 0.2258,
+      "step": 1397
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029578432504079257,
+      "loss": 0.3243,
+      "step": 1398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002957775164669353,
+      "loss": 0.3058,
+      "step": 1399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957707024778708,
+      "loss": 0.3556,
+      "step": 1400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029576388307385215,
+      "loss": 0.1081,
+      "step": 1401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957570582551327,
+      "loss": 0.1427,
+      "step": 1402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000295750228021966,
+      "loss": 0.3631,
+      "step": 1403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957433923746057,
+      "loss": 0.219,
+      "step": 1404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957365513133058,
+      "loss": 0.3122,
+      "step": 1405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029572970483832036,
+      "loss": 0.3059,
+      "step": 1406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029572285294990374,
+      "loss": 0.1331,
+      "step": 1407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957159956483105,
+      "loss": 0.3305,
+      "step": 1408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002957091329337953,
+      "loss": 0.0518,
+      "step": 1409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000295702264806613,
+      "loss": 0.3776,
+      "step": 1410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000295695391267019,
+      "loss": 0.2985,
+      "step": 1411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956885123152684,
+      "loss": 0.2693,
+      "step": 1412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029568162795161676,
+      "loss": 0.1771,
+      "step": 1413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956747381763199,
+      "loss": 0.259,
+      "step": 1414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956678429896336,
+      "loss": 0.392,
+      "step": 1415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956609423918142,
+      "loss": 0.365,
+      "step": 1416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029565403638311794,
+      "loss": 0.3014,
+      "step": 1417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956471249638013,
+      "loss": 0.362,
+      "step": 1418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956402081341211,
+      "loss": 0.2639,
+      "step": 1419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029563328589433426,
+      "loss": 0.1552,
+      "step": 1420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956263582446979,
+      "loss": 0.4239,
+      "step": 1421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956194251854694,
+      "loss": 0.2124,
+      "step": 1422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956124867169063,
+      "loss": 0.3118,
+      "step": 1423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002956055428392663,
+      "loss": 0.3133,
+      "step": 1424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002955985935528073,
+      "loss": 0.3846,
+      "step": 1425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029559163885778763,
+      "loss": 0.3456,
+      "step": 1426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029558467875446544,
+      "loss": 0.3449,
+      "step": 1427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029557771324309945,
+      "loss": 0.3846,
+      "step": 1428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002955707423239482,
+      "loss": 0.4196,
+      "step": 1429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002955637659972708,
+      "loss": 0.297,
+      "step": 1430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029555678426332636,
+      "loss": 0.2481,
+      "step": 1431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0002955497971223742,
+      "loss": 0.2012,
+      "step": 1432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029554280457467396,
+      "loss": 0.2302,
+      "step": 1433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955358066204853,
+      "loss": 0.3291,
+      "step": 1434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955288032600682,
+      "loss": 0.1786,
+      "step": 1435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955217944936828,
+      "loss": 0.2285,
+      "step": 1436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955147803215895,
+      "loss": 0.3216,
+      "step": 1437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955077607440488,
+      "loss": 0.1524,
+      "step": 1438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002955007357613215,
+      "loss": 0.3723,
+      "step": 1439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029549370537366857,
+      "loss": 0.192,
+      "step": 1440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029548666958135116,
+      "loss": 0.2998,
+      "step": 1441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029547962838463054,
+      "loss": 0.1838,
+      "step": 1442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029547258178376837,
+      "loss": 0.191,
+      "step": 1443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029546552977902644,
+      "loss": 0.2441,
+      "step": 1444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954584723706666,
+      "loss": 0.3127,
+      "step": 1445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029545140955895106,
+      "loss": 0.3181,
+      "step": 1446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954443413441422,
+      "loss": 0.2749,
+      "step": 1447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029543726772650255,
+      "loss": 0.3765,
+      "step": 1448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029543018870629485,
+      "loss": 0.4265,
+      "step": 1449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954231042837822,
+      "loss": 0.2235,
+      "step": 1450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029541601445922766,
+      "loss": 0.3498,
+      "step": 1451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954089192328946,
+      "loss": 0.2881,
+      "step": 1452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002954018186050466,
+      "loss": 0.3153,
+      "step": 1453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953947125759474,
+      "loss": 0.2107,
+      "step": 1454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029538760114586105,
+      "loss": 0.4518,
+      "step": 1455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029538048431505164,
+      "loss": 0.1184,
+      "step": 1456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953733620837835,
+      "loss": 0.1483,
+      "step": 1457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029536623445232136,
+      "loss": 0.1925,
+      "step": 1458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953591014209298,
+      "loss": 0.3289,
+      "step": 1459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029535196298987394,
+      "loss": 0.2754,
+      "step": 1460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029534481915941886,
+      "loss": 0.3737,
+      "step": 1461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029533766992983,
+      "loss": 0.2797,
+      "step": 1462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953305153013729,
+      "loss": 0.3239,
+      "step": 1463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00029532335527431334,
+      "loss": 0.4888,
+      "step": 1464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0002953161898489173,
+      "loss": 0.2427,
+      "step": 1465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002953090190254509,
+      "loss": 0.3419,
+      "step": 1466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002953018428041806,
+      "loss": 0.4325,
+      "step": 1467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952946611853729,
+      "loss": 0.2913,
+      "step": 1468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029528747416929463,
+      "loss": 0.332,
+      "step": 1469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029528028175621274,
+      "loss": 0.2977,
+      "step": 1470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952730839463944,
+      "loss": 0.2402,
+      "step": 1471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029526588074010705,
+      "loss": 0.2589,
+      "step": 1472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952586721376182,
+      "loss": 0.2131,
+      "step": 1473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029525145813919564,
+      "loss": 0.297,
+      "step": 1474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029524423874510735,
+      "loss": 0.4237,
+      "step": 1475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952370139556215,
+      "loss": 0.2751,
+      "step": 1476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029522978377100647,
+      "loss": 0.0939,
+      "step": 1477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952225481915309,
+      "loss": 0.1014,
+      "step": 1478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029521530721746353,
+      "loss": 0.1495,
+      "step": 1479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952080608490733,
+      "loss": 0.1863,
+      "step": 1480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002952008090866295,
+      "loss": 0.432,
+      "step": 1481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029519355193040136,
+      "loss": 0.284,
+      "step": 1482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029518628938065856,
+      "loss": 0.2404,
+      "step": 1483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951790214376709,
+      "loss": 0.2782,
+      "step": 1484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951717481017083,
+      "loss": 0.4125,
+      "step": 1485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029516446937304097,
+      "loss": 0.2794,
+      "step": 1486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029515718525193927,
+      "loss": 0.1577,
+      "step": 1487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029514989573867386,
+      "loss": 0.2785,
+      "step": 1488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029514260083351546,
+      "loss": 0.2661,
+      "step": 1489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029513530053673506,
+      "loss": 0.3852,
+      "step": 1490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029512799484860384,
+      "loss": 0.164,
+      "step": 1491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951206837693932,
+      "loss": 0.363,
+      "step": 1492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029511336729937473,
+      "loss": 0.2935,
+      "step": 1493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002951060454388202,
+      "loss": 0.4491,
+      "step": 1494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029509871818800157,
+      "loss": 0.2144,
+      "step": 1495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002950913855471911,
+      "loss": 0.2352,
+      "step": 1496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029508404751666116,
+      "loss": 0.4346,
+      "step": 1497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00029507670409668424,
+      "loss": 0.359,
+      "step": 1498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029506935528753324,
+      "loss": 0.3834,
+      "step": 1499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950620010894811,
+      "loss": 0.35,
+      "step": 1500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000295054641502801,
+      "loss": 0.384,
+      "step": 1501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029504727652776634,
+      "loss": 0.4987,
+      "step": 1502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029503990616465067,
+      "loss": 0.0797,
+      "step": 1503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950325304137279,
+      "loss": 0.174,
+      "step": 1504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950251492752718,
+      "loss": 0.0468,
+      "step": 1505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029501776274955683,
+      "loss": 0.2689,
+      "step": 1506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029501037083685716,
+      "loss": 0.206,
+      "step": 1507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002950029735374475,
+      "loss": 0.2429,
+      "step": 1508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029499557085160247,
+      "loss": 0.2554,
+      "step": 1509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949881627795973,
+      "loss": 0.2931,
+      "step": 1510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029498074932170703,
+      "loss": 0.2803,
+      "step": 1511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029497333047820705,
+      "loss": 0.2717,
+      "step": 1512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029496590624937297,
+      "loss": 0.2346,
+      "step": 1513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949584766354806,
+      "loss": 0.2544,
+      "step": 1514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029495104163680593,
+      "loss": 0.202,
+      "step": 1515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949436012536251,
+      "loss": 0.3041,
+      "step": 1516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029493615548621454,
+      "loss": 0.1258,
+      "step": 1517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949287043348508,
+      "loss": 0.3768,
+      "step": 1518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029492124779981075,
+      "loss": 0.4258,
+      "step": 1519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029491378588137126,
+      "loss": 0.2839,
+      "step": 1520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002949063185798096,
+      "loss": 0.1417,
+      "step": 1521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029489884589540313,
+      "loss": 0.1888,
+      "step": 1522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948913678284295,
+      "loss": 0.0993,
+      "step": 1523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948838843791664,
+      "loss": 0.2891,
+      "step": 1524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029487639554789187,
+      "loss": 0.1256,
+      "step": 1525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948689013348841,
+      "loss": 0.361,
+      "step": 1526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029486140174042147,
+      "loss": 0.2919,
+      "step": 1527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948538967647825,
+      "loss": 0.2055,
+      "step": 1528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029484638640824614,
+      "loss": 0.2445,
+      "step": 1529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0002948388706710913,
+      "loss": 0.2786,
+      "step": 1530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029483134955359707,
+      "loss": 0.1127,
+      "step": 1531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029482382305604296,
+      "loss": 0.4242,
+      "step": 1532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002948162911787085,
+      "loss": 0.3489,
+      "step": 1533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029480875392187347,
+      "loss": 0.1562,
+      "step": 1534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002948012112858179,
+      "loss": 0.3839,
+      "step": 1535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000294793663270822,
+      "loss": 0.2368,
+      "step": 1536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947861098771661,
+      "loss": 0.3223,
+      "step": 1537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029477855110513076,
+      "loss": 0.246,
+      "step": 1538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947709869549968,
+      "loss": 0.1221,
+      "step": 1539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947634174270452,
+      "loss": 0.2034,
+      "step": 1540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029475584252155725,
+      "loss": 0.2863,
+      "step": 1541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947482622388142,
+      "loss": 0.4638,
+      "step": 1542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029474067657909764,
+      "loss": 0.269,
+      "step": 1543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029473308554268945,
+      "loss": 0.3791,
+      "step": 1544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947254891298715,
+      "loss": 0.1553,
+      "step": 1545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029471788734092606,
+      "loss": 0.2044,
+      "step": 1546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002947102801761355,
+      "loss": 0.3578,
+      "step": 1547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029470266763578236,
+      "loss": 0.2478,
+      "step": 1548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946950497201495,
+      "loss": 0.2923,
+      "step": 1549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029468742642951986,
+      "loss": 0.3726,
+      "step": 1550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029467979776417655,
+      "loss": 0.5185,
+      "step": 1551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029467216372440307,
+      "loss": 0.3456,
+      "step": 1552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000294664524310483,
+      "loss": 0.328,
+      "step": 1553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946568795227,
+      "loss": 0.3352,
+      "step": 1554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029464922936133815,
+      "loss": 0.388,
+      "step": 1555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946415738266816,
+      "loss": 0.2405,
+      "step": 1556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946339129190148,
+      "loss": 0.3925,
+      "step": 1557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946262466386222,
+      "loss": 0.2795,
+      "step": 1558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029461857498578865,
+      "loss": 0.2927,
+      "step": 1559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029461089796079914,
+      "loss": 0.2759,
+      "step": 1560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002946032155639389,
+      "loss": 0.4036,
+      "step": 1561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029459552779549316,
+      "loss": 0.103,
+      "step": 1562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0002945878346557477,
+      "loss": 0.3789,
+      "step": 1563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00029458013614498804,
+      "loss": 0.3293,
+      "step": 1564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945724322635004,
+      "loss": 0.2893,
+      "step": 1565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945647230115708,
+      "loss": 0.3492,
+      "step": 1566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029455700838948575,
+      "loss": 0.0982,
+      "step": 1567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945492883975317,
+      "loss": 0.4754,
+      "step": 1568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029454156303599543,
+      "loss": 0.2101,
+      "step": 1569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000294533832305164,
+      "loss": 0.2007,
+      "step": 1570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945260962053245,
+      "loss": 0.302,
+      "step": 1571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029451835473676445,
+      "loss": 0.2931,
+      "step": 1572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002945106078997712,
+      "loss": 0.1154,
+      "step": 1573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029450285569463275,
+      "loss": 0.3683,
+      "step": 1574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944950981216369,
+      "loss": 0.2305,
+      "step": 1575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944873351810719,
+      "loss": 0.1063,
+      "step": 1576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944795668732261,
+      "loss": 0.0388,
+      "step": 1577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029447179319838803,
+      "loss": 0.2466,
+      "step": 1578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944640141568465,
+      "loss": 0.1594,
+      "step": 1579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029445622974889055,
+      "loss": 0.3182,
+      "step": 1580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944484399748093,
+      "loss": 0.3085,
+      "step": 1581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000294440644834892,
+      "loss": 0.3983,
+      "step": 1582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944328443294284,
+      "loss": 0.3938,
+      "step": 1583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029442503845870813,
+      "loss": 0.1276,
+      "step": 1584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944172272230212,
+      "loss": 0.2155,
+      "step": 1585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002944094106226578,
+      "loss": 0.4954,
+      "step": 1586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029440158865790824,
+      "loss": 0.2417,
+      "step": 1587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029439376132906316,
+      "loss": 0.3916,
+      "step": 1588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029438592863641324,
+      "loss": 0.3094,
+      "step": 1589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943780905802495,
+      "loss": 0.3086,
+      "step": 1590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943702471608631,
+      "loss": 0.2039,
+      "step": 1591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943623983785453,
+      "loss": 0.3783,
+      "step": 1592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943545442335878,
+      "loss": 0.18,
+      "step": 1593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943466847262823,
+      "loss": 0.2965,
+      "step": 1594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943388198569208,
+      "loss": 0.3282,
+      "step": 1595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029433094962579535,
+      "loss": 0.2572,
+      "step": 1596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0002943230740331984,
+      "loss": 0.2609,
+      "step": 1597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002943151930794224,
+      "loss": 0.2669,
+      "step": 1598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029430730676476024,
+      "loss": 0.2585,
+      "step": 1599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029429941508950485,
+      "loss": 0.18,
+      "step": 1600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029429151805394926,
+      "loss": 0.2206,
+      "step": 1601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029428361565838696,
+      "loss": 0.3159,
+      "step": 1602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029427570790311144,
+      "loss": 0.341,
+      "step": 1603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942677947884164,
+      "loss": 0.2141,
+      "step": 1604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029425987631459587,
+      "loss": 0.2033,
+      "step": 1605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029425195248194405,
+      "loss": 0.2822,
+      "step": 1606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029424402329075516,
+      "loss": 0.1612,
+      "step": 1607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029423608874132377,
+      "loss": 0.3472,
+      "step": 1608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942281488339447,
+      "loss": 0.1743,
+      "step": 1609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942202035689128,
+      "loss": 0.2686,
+      "step": 1610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002942122529465233,
+      "loss": 0.1995,
+      "step": 1611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029420429696707143,
+      "loss": 0.446,
+      "step": 1612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941963356308529,
+      "loss": 0.1584,
+      "step": 1613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941883689381633,
+      "loss": 0.2495,
+      "step": 1614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029418039688929867,
+      "loss": 0.1637,
+      "step": 1615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029417241948455504,
+      "loss": 0.1845,
+      "step": 1616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941644367242288,
+      "loss": 0.2457,
+      "step": 1617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941564486086165,
+      "loss": 0.2406,
+      "step": 1618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029414845513801494,
+      "loss": 0.3969,
+      "step": 1619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941404563127209,
+      "loss": 0.4207,
+      "step": 1620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029413245213303165,
+      "loss": 0.3595,
+      "step": 1621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029412444259924446,
+      "loss": 0.1899,
+      "step": 1622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029411642771165686,
+      "loss": 0.374,
+      "step": 1623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941084074705666,
+      "loss": 0.2549,
+      "step": 1624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002941003818762716,
+      "loss": 0.3674,
+      "step": 1625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029409235092906996,
+      "loss": 0.1308,
+      "step": 1626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029408431462926005,
+      "loss": 0.3371,
+      "step": 1627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029407627297714034,
+      "loss": 0.1599,
+      "step": 1628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029406822597300965,
+      "loss": 0.3132,
+      "step": 1629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002940601736171668,
+      "loss": 0.3197,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029405211590991095,
+      "loss": 0.1463,
+      "step": 1631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002940440528515414,
+      "loss": 0.1995,
+      "step": 1632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029403598444235776,
+      "loss": 0.1711,
+      "step": 1633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029402791068265964,
+      "loss": 0.2138,
+      "step": 1634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029401983157274704,
+      "loss": 0.3376,
+      "step": 1635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029401174711292,
+      "loss": 0.1361,
+      "step": 1636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029400365730347883,
+      "loss": 0.1845,
+      "step": 1637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029399556214472413,
+      "loss": 0.3565,
+      "step": 1638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029398746163695653,
+      "loss": 0.4043,
+      "step": 1639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029397935578047695,
+      "loss": 0.3053,
+      "step": 1640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939712445755866,
+      "loss": 0.3048,
+      "step": 1641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029396312802258657,
+      "loss": 0.2179,
+      "step": 1642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029395500612177863,
+      "loss": 0.3258,
+      "step": 1643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939468788734643,
+      "loss": 0.1433,
+      "step": 1644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939387462779455,
+      "loss": 0.2505,
+      "step": 1645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029393060833552445,
+      "loss": 0.3189,
+      "step": 1646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939224650465034,
+      "loss": 0.2461,
+      "step": 1647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029391431641118473,
+      "loss": 0.3148,
+      "step": 1648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002939061624298713,
+      "loss": 0.3786,
+      "step": 1649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029389800310286596,
+      "loss": 0.2087,
+      "step": 1650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938898384304717,
+      "loss": 0.2936,
+      "step": 1651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938816684129919,
+      "loss": 0.1656,
+      "step": 1652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029387349305073007,
+      "loss": 0.0771,
+      "step": 1653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938653123439899,
+      "loss": 0.134,
+      "step": 1654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029385712629307524,
+      "loss": 0.3098,
+      "step": 1655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938489348982902,
+      "loss": 0.2677,
+      "step": 1656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029384073815993906,
+      "loss": 0.1637,
+      "step": 1657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938325360783263,
+      "loss": 0.348,
+      "step": 1658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029382432865375665,
+      "loss": 0.2089,
+      "step": 1659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029381611588653487,
+      "loss": 0.1811,
+      "step": 1660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0002938078977769662,
+      "loss": 0.3811,
+      "step": 1661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029379967432535583,
+      "loss": 0.4252,
+      "step": 1662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00029379144553200924,
+      "loss": 0.2892,
+      "step": 1663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029378321139723207,
+      "loss": 0.2829,
+      "step": 1664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937749719213302,
+      "loss": 0.4034,
+      "step": 1665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937667271046099,
+      "loss": 0.1469,
+      "step": 1666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937584769473772,
+      "loss": 0.1851,
+      "step": 1667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937502214499386,
+      "loss": 0.2249,
+      "step": 1668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937419606126009,
+      "loss": 0.3667,
+      "step": 1669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029373369443567077,
+      "loss": 0.2854,
+      "step": 1670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937254229194555,
+      "loss": 0.2905,
+      "step": 1671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002937171460642622,
+      "loss": 0.2118,
+      "step": 1672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029370886387039834,
+      "loss": 0.1213,
+      "step": 1673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029370057633817167,
+      "loss": 0.3235,
+      "step": 1674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936922834678899,
+      "loss": 0.1727,
+      "step": 1675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029368398525986126,
+      "loss": 0.1682,
+      "step": 1676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936756817143939,
+      "loss": 0.4145,
+      "step": 1677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936673728317963,
+      "loss": 0.3664,
+      "step": 1678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029365905861237704,
+      "loss": 0.3116,
+      "step": 1679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029365073905644507,
+      "loss": 0.2281,
+      "step": 1680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002936424141643094,
+      "loss": 0.2421,
+      "step": 1681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029363408393627923,
+      "loss": 0.1655,
+      "step": 1682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029362574837266407,
+      "loss": 0.0844,
+      "step": 1683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029361740747377356,
+      "loss": 0.2977,
+      "step": 1684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029360906123991756,
+      "loss": 0.2071,
+      "step": 1685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000293600709671406,
+      "loss": 0.3733,
+      "step": 1686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029359235276854914,
+      "loss": 0.2254,
+      "step": 1687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029358399053165755,
+      "loss": 0.3213,
+      "step": 1688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029357562296104166,
+      "loss": 0.2065,
+      "step": 1689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935672500570125,
+      "loss": 0.3372,
+      "step": 1690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029355887181988096,
+      "loss": 0.2746,
+      "step": 1691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935504882499584,
+      "loss": 0.2462,
+      "step": 1692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029354209934755605,
+      "loss": 0.3108,
+      "step": 1693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935337051129857,
+      "loss": 0.2581,
+      "step": 1694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029352530554655913,
+      "loss": 0.1682,
+      "step": 1695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002935169006485883,
+      "loss": 0.2696,
+      "step": 1696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029350849041938544,
+      "loss": 0.1737,
+      "step": 1697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002935000748592631,
+      "loss": 0.264,
+      "step": 1698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934916539685337,
+      "loss": 0.1505,
+      "step": 1699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934832277475102,
+      "loss": 0.2411,
+      "step": 1700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029347479619650554,
+      "loss": 0.2302,
+      "step": 1701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934663593158329,
+      "loss": 0.2335,
+      "step": 1702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934579171058057,
+      "loss": 0.043,
+      "step": 1703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029344946956673764,
+      "loss": 0.4307,
+      "step": 1704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934410166989424,
+      "loss": 0.3111,
+      "step": 1705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000293432558502734,
+      "loss": 0.3435,
+      "step": 1706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029342409497842674,
+      "loss": 0.2399,
+      "step": 1707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934156261263349,
+      "loss": 0.2959,
+      "step": 1708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002934071519467731,
+      "loss": 0.2863,
+      "step": 1709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933986724400562,
+      "loss": 0.354,
+      "step": 1710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933901876064991,
+      "loss": 0.2712,
+      "step": 1711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000293381697446417,
+      "loss": 0.2795,
+      "step": 1712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933732019601253,
+      "loss": 0.4045,
+      "step": 1713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933647011479396,
+      "loss": 0.1364,
+      "step": 1714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933561950101757,
+      "loss": 0.3542,
+      "step": 1715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933476835471495,
+      "loss": 0.1299,
+      "step": 1716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029333916675917726,
+      "loss": 0.193,
+      "step": 1717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933306446465753,
+      "loss": 0.2628,
+      "step": 1718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029332211720966016,
+      "loss": 0.2151,
+      "step": 1719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029331358444874864,
+      "loss": 0.2057,
+      "step": 1720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002933050463641578,
+      "loss": 0.295,
+      "step": 1721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002932965029562047,
+      "loss": 0.2486,
+      "step": 1722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002932879542252067,
+      "loss": 0.2549,
+      "step": 1723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002932794001714814,
+      "loss": 0.1619,
+      "step": 1724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002932708407953466,
+      "loss": 0.289,
+      "step": 1725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002932622760971202,
+      "loss": 0.256,
+      "step": 1726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029325370607712027,
+      "loss": 0.3175,
+      "step": 1727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029324513073566527,
+      "loss": 0.1346,
+      "step": 1728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00029323655007307373,
+      "loss": 0.1361,
+      "step": 1729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029322796408966445,
+      "loss": 0.2314,
+      "step": 1730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029321937278575626,
+      "loss": 0.2486,
+      "step": 1731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002932107761616684,
+      "loss": 0.3394,
+      "step": 1732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002932021742177201,
+      "loss": 0.4331,
+      "step": 1733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029319356695423104,
+      "loss": 0.4352,
+      "step": 1734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931849543715208,
+      "loss": 0.3157,
+      "step": 1735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931763364699095,
+      "loss": 0.373,
+      "step": 1736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029316771324971704,
+      "loss": 0.1119,
+      "step": 1737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029315908471126397,
+      "loss": 0.2445,
+      "step": 1738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931504508548707,
+      "loss": 0.3022,
+      "step": 1739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931418116808579,
+      "loss": 0.3943,
+      "step": 1740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931331671895466,
+      "loss": 0.1757,
+      "step": 1741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029312451738125787,
+      "loss": 0.222,
+      "step": 1742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029311586225631305,
+      "loss": 0.1426,
+      "step": 1743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002931072018150336,
+      "loss": 0.2361,
+      "step": 1744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029309853605774126,
+      "loss": 0.2634,
+      "step": 1745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029308986498475793,
+      "loss": 0.4707,
+      "step": 1746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930811885964058,
+      "loss": 0.0675,
+      "step": 1747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029307250689300697,
+      "loss": 0.3394,
+      "step": 1748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029306381987488417,
+      "loss": 0.361,
+      "step": 1749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029305512754235994,
+      "loss": 0.1413,
+      "step": 1750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930464298957572,
+      "loss": 0.3305,
+      "step": 1751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029303772693539914,
+      "loss": 0.1752,
+      "step": 1752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000293029018661609,
+      "loss": 0.2549,
+      "step": 1753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930203050747102,
+      "loss": 0.1887,
+      "step": 1754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002930115861750264,
+      "loss": 0.1983,
+      "step": 1755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029300286196288166,
+      "loss": 0.2689,
+      "step": 1756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002929941324385999,
+      "loss": 0.2104,
+      "step": 1757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029298539760250545,
+      "loss": 0.3662,
+      "step": 1758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002929766574549228,
+      "loss": 0.5166,
+      "step": 1759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002929679119961766,
+      "loss": 0.268,
+      "step": 1760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002929591612265917,
+      "loss": 0.1503,
+      "step": 1761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00029295040514649317,
+      "loss": 0.4117,
+      "step": 1762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002929416437562063,
+      "loss": 0.2704,
+      "step": 1763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029293287705605656,
+      "loss": 0.3453,
+      "step": 1764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029292410504636957,
+      "loss": 0.3973,
+      "step": 1765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002929153277274712,
+      "loss": 0.2841,
+      "step": 1766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002929065450996875,
+      "loss": 0.2332,
+      "step": 1767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928977571633447,
+      "loss": 0.1952,
+      "step": 1768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928889639187693,
+      "loss": 0.2351,
+      "step": 1769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928801653662879,
+      "loss": 0.267,
+      "step": 1770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029287136150622735,
+      "loss": 0.2546,
+      "step": 1771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029286255233891464,
+      "loss": 0.2763,
+      "step": 1772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928537378646771,
+      "loss": 0.3231,
+      "step": 1773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928449180838421,
+      "loss": 0.0946,
+      "step": 1774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928360929967373,
+      "loss": 0.2822,
+      "step": 1775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928272626036905,
+      "loss": 0.3929,
+      "step": 1776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002928184269050297,
+      "loss": 0.1814,
+      "step": 1777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029280958590108323,
+      "loss": 0.3385,
+      "step": 1778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029280073959217934,
+      "loss": 0.066,
+      "step": 1779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927918879786468,
+      "loss": 0.2809,
+      "step": 1780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927830310608143,
+      "loss": 0.0748,
+      "step": 1781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029277416883901095,
+      "loss": 0.1194,
+      "step": 1782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029276530131356585,
+      "loss": 0.179,
+      "step": 1783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029275642848480853,
+      "loss": 0.2202,
+      "step": 1784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029274755035306845,
+      "loss": 0.1528,
+      "step": 1785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927386669186756,
+      "loss": 0.2398,
+      "step": 1786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927297781819597,
+      "loss": 0.1214,
+      "step": 1787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927208841432512,
+      "loss": 0.2905,
+      "step": 1788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927119848028804,
+      "loss": 0.2248,
+      "step": 1789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002927030801611778,
+      "loss": 0.2592,
+      "step": 1790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002926941702184742,
+      "loss": 0.3307,
+      "step": 1791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002926852549751007,
+      "loss": 0.4024,
+      "step": 1792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00029267633443138836,
+      "loss": 0.2335,
+      "step": 1793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0002926674085876686,
+      "loss": 0.3413,
+      "step": 1794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029265847744427303,
+      "loss": 0.3006,
+      "step": 1795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002926495410015333,
+      "loss": 0.3663,
+      "step": 1796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029264059925978144,
+      "loss": 0.1595,
+      "step": 1797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029263165221934967,
+      "loss": 0.2729,
+      "step": 1798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029262269988057025,
+      "loss": 0.2582,
+      "step": 1799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002926137422437757,
+      "loss": 0.2758,
+      "step": 1800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002926047793092989,
+      "loss": 0.285,
+      "step": 1801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029259581107747275,
+      "loss": 0.2293,
+      "step": 1802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925868375486304,
+      "loss": 0.2636,
+      "step": 1803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925778587231051,
+      "loss": 0.2176,
+      "step": 1804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029256887460123046,
+      "loss": 0.2073,
+      "step": 1805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029255988518334027,
+      "loss": 0.3329,
+      "step": 1806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925508904697683,
+      "loss": 0.3212,
+      "step": 1807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925418904608489,
+      "loss": 0.1044,
+      "step": 1808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029253288515691616,
+      "loss": 0.3573,
+      "step": 1809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925238745583048,
+      "loss": 0.0903,
+      "step": 1810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002925148586653494,
+      "loss": 0.187,
+      "step": 1811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029250583747838485,
+      "loss": 0.261,
+      "step": 1812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029249681099774644,
+      "loss": 0.2489,
+      "step": 1813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029248777922376935,
+      "loss": 0.4501,
+      "step": 1814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924787421567891,
+      "loss": 0.2202,
+      "step": 1815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029246969979714136,
+      "loss": 0.2226,
+      "step": 1816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029246065214516207,
+      "loss": 0.2908,
+      "step": 1817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029245159920118736,
+      "loss": 0.2637,
+      "step": 1818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029244254096555346,
+      "loss": 0.3328,
+      "step": 1819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029243347743859683,
+      "loss": 0.2955,
+      "step": 1820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924244086206542,
+      "loss": 0.2782,
+      "step": 1821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924153345120625,
+      "loss": 0.3605,
+      "step": 1822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002924062551131587,
+      "loss": 0.3518,
+      "step": 1823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029239717042428017,
+      "loss": 0.0498,
+      "step": 1824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029238808044576434,
+      "loss": 0.368,
+      "step": 1825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002923789851779488,
+      "loss": 0.1863,
+      "step": 1826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029236988462117157,
+      "loss": 0.1469,
+      "step": 1827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029236077877577057,
+      "loss": 0.2805,
+      "step": 1828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029235166764208413,
+      "loss": 0.1966,
+      "step": 1829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029234255122045065,
+      "loss": 0.3393,
+      "step": 1830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029233342951120886,
+      "loss": 0.47,
+      "step": 1831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002923243025146975,
+      "loss": 0.1976,
+      "step": 1832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029231517023125566,
+      "loss": 0.312,
+      "step": 1833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002923060326612226,
+      "loss": 0.2675,
+      "step": 1834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922968898049378,
+      "loss": 0.1861,
+      "step": 1835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029228774166274073,
+      "loss": 0.246,
+      "step": 1836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029227858823497135,
+      "loss": 0.287,
+      "step": 1837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029226942952196964,
+      "loss": 0.2968,
+      "step": 1838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029226026552407584,
+      "loss": 0.3254,
+      "step": 1839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922510962416303,
+      "loss": 0.2417,
+      "step": 1840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029224192167497375,
+      "loss": 0.216,
+      "step": 1841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029223274182444694,
+      "loss": 0.2091,
+      "step": 1842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922235566903908,
+      "loss": 0.2453,
+      "step": 1843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002922143662731466,
+      "loss": 0.438,
+      "step": 1844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029220517057305576,
+      "loss": 0.1645,
+      "step": 1845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029219596959045987,
+      "loss": 0.2165,
+      "step": 1846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029218676332570066,
+      "loss": 0.3568,
+      "step": 1847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921775517791202,
+      "loss": 0.1404,
+      "step": 1848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921683349510606,
+      "loss": 0.151,
+      "step": 1849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921591128418642,
+      "loss": 0.2076,
+      "step": 1850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029214988545187367,
+      "loss": 0.2121,
+      "step": 1851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002921406527814318,
+      "loss": 0.2642,
+      "step": 1852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029213141483088145,
+      "loss": 0.1925,
+      "step": 1853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029212217160056586,
+      "loss": 0.1627,
+      "step": 1854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029211292309082836,
+      "loss": 0.1025,
+      "step": 1855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029210366930201246,
+      "loss": 0.2079,
+      "step": 1856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029209441023446203,
+      "loss": 0.3006,
+      "step": 1857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029208514588852097,
+      "loss": 0.2754,
+      "step": 1858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002920758762645333,
+      "loss": 0.0574,
+      "step": 1859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00029206660136284353,
+      "loss": 0.2271,
+      "step": 1860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002920573211837961,
+      "loss": 0.119,
+      "step": 1861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029204803572773576,
+      "loss": 0.2214,
+      "step": 1862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029203874499500747,
+      "loss": 0.3398,
+      "step": 1863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029202944898595634,
+      "loss": 0.1581,
+      "step": 1864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002920201477009277,
+      "loss": 0.3694,
+      "step": 1865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029201084114026697,
+      "loss": 0.3016,
+      "step": 1866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029200152930431995,
+      "loss": 0.4154,
+      "step": 1867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029199221219343256,
+      "loss": 0.2293,
+      "step": 1868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919828898079509,
+      "loss": 0.1542,
+      "step": 1869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919735621482212,
+      "loss": 0.3505,
+      "step": 1870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029196422921459,
+      "loss": 0.2958,
+      "step": 1871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000291954891007404,
+      "loss": 0.2408,
+      "step": 1872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919455475270101,
+      "loss": 0.2094,
+      "step": 1873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029193619877375534,
+      "loss": 0.181,
+      "step": 1874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000291926844747987,
+      "loss": 0.3006,
+      "step": 1875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919174854500526,
+      "loss": 0.1731,
+      "step": 1876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002919081208802998,
+      "loss": 0.3105,
+      "step": 1877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029189875103907643,
+      "loss": 0.1065,
+      "step": 1878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918893759267306,
+      "loss": 0.127,
+      "step": 1879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029187999554361054,
+      "loss": 0.1069,
+      "step": 1880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918706098900647,
+      "loss": 0.1772,
+      "step": 1881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029186121896644163,
+      "loss": 0.1517,
+      "step": 1882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918518227730904,
+      "loss": 0.2422,
+      "step": 1883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918424213103599,
+      "loss": 0.1662,
+      "step": 1884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918330145785994,
+      "loss": 0.1921,
+      "step": 1885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029182360257815835,
+      "loss": 0.1529,
+      "step": 1886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918141853093863,
+      "loss": 0.253,
+      "step": 1887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002918047627726332,
+      "loss": 0.1485,
+      "step": 1888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029179533496824896,
+      "loss": 0.041,
+      "step": 1889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029178590189658386,
+      "loss": 0.1449,
+      "step": 1890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029177646355798827,
+      "loss": 0.3484,
+      "step": 1891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00029176701995281285,
+      "loss": 0.2447,
+      "step": 1892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0002917575710814083,
+      "loss": 0.1739,
+      "step": 1893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002917481169441257,
+      "loss": 0.3498,
+      "step": 1894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002917386575413163,
+      "loss": 0.2539,
+      "step": 1895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029172919287333136,
+      "loss": 0.3209,
+      "step": 1896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002917197229405225,
+      "loss": 0.4159,
+      "step": 1897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002917102477432416,
+      "loss": 0.3969,
+      "step": 1898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002917007672818405,
+      "loss": 0.3017,
+      "step": 1899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916912815566715,
+      "loss": 0.1974,
+      "step": 1900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916817905680868,
+      "loss": 0.1618,
+      "step": 1901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916722943164391,
+      "loss": 0.0767,
+      "step": 1902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916627928020811,
+      "loss": 0.3935,
+      "step": 1903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916532860253658,
+      "loss": 0.188,
+      "step": 1904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029164377398664627,
+      "loss": 0.1811,
+      "step": 1905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029163425668627595,
+      "loss": 0.1736,
+      "step": 1906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916247341246083,
+      "loss": 0.2487,
+      "step": 1907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029161520630199714,
+      "loss": 0.115,
+      "step": 1908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002916056732187963,
+      "loss": 0.2255,
+      "step": 1909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029159613487536,
+      "loss": 0.2397,
+      "step": 1910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915865912720425,
+      "loss": 0.31,
+      "step": 1911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915770424091983,
+      "loss": 0.1988,
+      "step": 1912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915674882871822,
+      "loss": 0.4858,
+      "step": 1913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000291557928906349,
+      "loss": 0.4887,
+      "step": 1914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915483642670539,
+      "loss": 0.2873,
+      "step": 1915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915387943696521,
+      "loss": 0.0864,
+      "step": 1916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915292192144992,
+      "loss": 0.1039,
+      "step": 1917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915196388019508,
+      "loss": 0.4565,
+      "step": 1918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002915100531323628,
+      "loss": 0.1106,
+      "step": 1919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029150046220609136,
+      "loss": 0.1282,
+      "step": 1920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002914908660234927,
+      "loss": 0.194,
+      "step": 1921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029148126458492326,
+      "loss": 0.3163,
+      "step": 1922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029147165789073973,
+      "loss": 0.3475,
+      "step": 1923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000291462045941299,
+      "loss": 0.2512,
+      "step": 1924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002914524287369581,
+      "loss": 0.2025,
+      "step": 1925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002914428062780742,
+      "loss": 0.111,
+      "step": 1926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029143317856500486,
+      "loss": 0.1849,
+      "step": 1927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029142354559810767,
+      "loss": 0.334,
+      "step": 1928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029141390737774057,
+      "loss": 0.3098,
+      "step": 1929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029140426390426144,
+      "loss": 0.261,
+      "step": 1930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913946151780286,
+      "loss": 0.2775,
+      "step": 1931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913849611994004,
+      "loss": 0.2843,
+      "step": 1932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913753019687355,
+      "loss": 0.3453,
+      "step": 1933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029136563748639275,
+      "loss": 0.2209,
+      "step": 1934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029135596775273106,
+      "loss": 0.2948,
+      "step": 1935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913462927681098,
+      "loss": 0.1759,
+      "step": 1936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029133661253288813,
+      "loss": 0.2421,
+      "step": 1937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029132692704742587,
+      "loss": 0.4065,
+      "step": 1938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913172363120826,
+      "loss": 0.2618,
+      "step": 1939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002913075403272185,
+      "loss": 0.233,
+      "step": 1940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912978390931936,
+      "loss": 0.311,
+      "step": 1941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912881326103684,
+      "loss": 0.1311,
+      "step": 1942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029127842087910334,
+      "loss": 0.3583,
+      "step": 1943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029126870389975924,
+      "loss": 0.1296,
+      "step": 1944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029125898167269715,
+      "loss": 0.1133,
+      "step": 1945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029124925419827803,
+      "loss": 0.3254,
+      "step": 1946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029123952147686337,
+      "loss": 0.4281,
+      "step": 1947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029122978350881463,
+      "loss": 0.1678,
+      "step": 1948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002912200402944936,
+      "loss": 0.0841,
+      "step": 1949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029121029183426224,
+      "loss": 0.1999,
+      "step": 1950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029120053812848256,
+      "loss": 0.1661,
+      "step": 1951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029119077917751703,
+      "loss": 0.1401,
+      "step": 1952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911810149817281,
+      "loss": 0.2997,
+      "step": 1953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911712455414784,
+      "loss": 0.1919,
+      "step": 1954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911614708571309,
+      "loss": 0.0866,
+      "step": 1955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911516909290488,
+      "loss": 0.266,
+      "step": 1956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911419057575952,
+      "loss": 0.2658,
+      "step": 1957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002911321153431338,
+      "loss": 0.268,
+      "step": 1958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029112231968602813,
+      "loss": 0.2653,
+      "step": 1959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029111251878664217,
+      "loss": 0.3122,
+      "step": 1960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002911027126453398,
+      "loss": 0.2765,
+      "step": 1961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910929012624856,
+      "loss": 0.2144,
+      "step": 1962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910830846384438,
+      "loss": 0.3085,
+      "step": 1963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029107326277357914,
+      "loss": 0.3016,
+      "step": 1964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029106343566825645,
+      "loss": 0.3617,
+      "step": 1965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910536033228408,
+      "loss": 0.3215,
+      "step": 1966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910437657376974,
+      "loss": 0.2536,
+      "step": 1967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029103392291319173,
+      "loss": 0.3661,
+      "step": 1968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029102407484968946,
+      "loss": 0.2418,
+      "step": 1969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002910142215475563,
+      "loss": 0.2747,
+      "step": 1970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029100436300715833,
+      "loss": 0.2469,
+      "step": 1971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029099449922886174,
+      "loss": 0.2915,
+      "step": 1972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000290984630213033,
+      "loss": 0.3442,
+      "step": 1973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909747559600387,
+      "loss": 0.3356,
+      "step": 1974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029096487647024556,
+      "loss": 0.1776,
+      "step": 1975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909549917440207,
+      "loss": 0.2703,
+      "step": 1976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909451017817312,
+      "loss": 0.3472,
+      "step": 1977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909352065837445,
+      "loss": 0.3789,
+      "step": 1978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909253061504282,
+      "loss": 0.4946,
+      "step": 1979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029091540048215,
+      "loss": 0.0983,
+      "step": 1980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002909054895792779,
+      "loss": 0.3091,
+      "step": 1981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908955734421801,
+      "loss": 0.2296,
+      "step": 1982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908856520712249,
+      "loss": 0.2374,
+      "step": 1983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908757254667808,
+      "loss": 0.2646,
+      "step": 1984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029086579362921667,
+      "loss": 0.4312,
+      "step": 1985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908558565589014,
+      "loss": 0.2981,
+      "step": 1986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029084591425620413,
+      "loss": 0.3897,
+      "step": 1987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908359667214941,
+      "loss": 0.3564,
+      "step": 1988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000290826013955141,
+      "loss": 0.3288,
+      "step": 1989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002908160559575144,
+      "loss": 0.1348,
+      "step": 1990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029080609272898433,
+      "loss": 0.3561,
+      "step": 1991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002907961242699207,
+      "loss": 0.0464,
+      "step": 1992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029078615058069404,
+      "loss": 0.36,
+      "step": 1993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907761716616747,
+      "loss": 0.2961,
+      "step": 1994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907661875132334,
+      "loss": 0.3437,
+      "step": 1995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029075619813574103,
+      "loss": 0.297,
+      "step": 1996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907462035295686,
+      "loss": 0.1672,
+      "step": 1997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907362036950876,
+      "loss": 0.2305,
+      "step": 1998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002907261986326692,
+      "loss": 0.185,
+      "step": 1999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029071618834268525,
+      "loss": 0.135,
+      "step": 2000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029070617282550754,
+      "loss": 0.3249,
+      "step": 2001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029069615208150815,
+      "loss": 0.2952,
+      "step": 2002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906861261110593,
+      "loss": 0.2556,
+      "step": 2003
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029067609491453334,
+      "loss": 0.3139,
+      "step": 2004
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906660584923031,
+      "loss": 0.1327,
+      "step": 2005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906560168447412,
+      "loss": 0.4226,
+      "step": 2006
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906459699722208,
+      "loss": 0.1889,
+      "step": 2007
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000290635917875115,
+      "loss": 0.0995,
+      "step": 2008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906258605537973,
+      "loss": 0.2354,
+      "step": 2009
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906157980086413,
+      "loss": 0.3073,
+      "step": 2010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002906057302400207,
+      "loss": 0.5341,
+      "step": 2011
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029059565724830953,
+      "loss": 0.1667,
+      "step": 2012
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029058557903388204,
+      "loss": 0.4069,
+      "step": 2013
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905754955971125,
+      "loss": 0.1942,
+      "step": 2014
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905654069383756,
+      "loss": 0.2601,
+      "step": 2015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029055531305804597,
+      "loss": 0.3224,
+      "step": 2016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029054521395649864,
+      "loss": 0.2658,
+      "step": 2017
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905351096341088,
+      "loss": 0.2879,
+      "step": 2018
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029052500009125174,
+      "loss": 0.2601,
+      "step": 2019
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000290514885328303,
+      "loss": 0.3574,
+      "step": 2020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002905047653456383,
+      "loss": 0.2322,
+      "step": 2021
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029049464014363366,
+      "loss": 0.4208,
+      "step": 2022
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0002904845097226651,
+      "loss": 0.2584,
+      "step": 2023
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029047437408310893,
+      "loss": 0.2094,
+      "step": 2024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00029046423322534175,
+      "loss": 0.1953,
+      "step": 2025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002904540871497402,
+      "loss": 0.1669,
+      "step": 2026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029044393585668125,
+      "loss": 0.2679,
+      "step": 2027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002904337793465419,
+      "loss": 0.2149,
+      "step": 2028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002904236176196994,
+      "loss": 0.2087,
+      "step": 2029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029041345067653136,
+      "loss": 0.279,
+      "step": 2030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029040327851741535,
+      "loss": 0.2369,
+      "step": 2031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903931011427293,
+      "loss": 0.2415,
+      "step": 2032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029038291855285124,
+      "loss": 0.2846,
+      "step": 2033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029037273074815936,
+      "loss": 0.1907,
+      "step": 2034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029036253772903226,
+      "loss": 0.2177,
+      "step": 2035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029035233949584844,
+      "loss": 0.234,
+      "step": 2036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903421360489868,
+      "loss": 0.2223,
+      "step": 2037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903319273888264,
+      "loss": 0.1903,
+      "step": 2038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002903217135157464,
+      "loss": 0.1627,
+      "step": 2039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029031149443012615,
+      "loss": 0.3138,
+      "step": 2040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029030127013234545,
+      "loss": 0.1485,
+      "step": 2041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002902910406227839,
+      "loss": 0.3189,
+      "step": 2042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029028080590182164,
+      "loss": 0.3512,
+      "step": 2043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029027056596983884,
+      "loss": 0.407,
+      "step": 2044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029026032082721584,
+      "loss": 0.3548,
+      "step": 2045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002902500704743332,
+      "loss": 0.2699,
+      "step": 2046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029023981491157173,
+      "loss": 0.2564,
+      "step": 2047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029022955413931243,
+      "loss": 0.3234,
+      "step": 2048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029021928815793635,
+      "loss": 0.3827,
+      "step": 2049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002902090169678249,
+      "loss": 0.2275,
+      "step": 2050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901987405693597,
+      "loss": 0.4564,
+      "step": 2051
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901884589629224,
+      "loss": 0.2218,
+      "step": 2052
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901781721488949,
+      "loss": 0.2228,
+      "step": 2053
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901678801276594,
+      "loss": 0.1578,
+      "step": 2054
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029015758289959825,
+      "loss": 0.1601,
+      "step": 2055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901472804650938,
+      "loss": 0.2123,
+      "step": 2056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0002901369728245289,
+      "loss": 0.1674,
+      "step": 2057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029012665997828647,
+      "loss": 0.1568,
+      "step": 2058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029011634192674943,
+      "loss": 0.2364,
+      "step": 2059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002901060186703013,
+      "loss": 0.3734,
+      "step": 2060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029009569020932533,
+      "loss": 0.152,
+      "step": 2061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029008535654420533,
+      "loss": 0.2337,
+      "step": 2062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900750176753251,
+      "loss": 0.2528,
+      "step": 2063
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900646736030688,
+      "loss": 0.1789,
+      "step": 2064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029005432432782055,
+      "loss": 0.3341,
+      "step": 2065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900439698499649,
+      "loss": 0.2544,
+      "step": 2066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900336101698864,
+      "loss": 0.2642,
+      "step": 2067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029002324528797,
+      "loss": 0.196,
+      "step": 2068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029001287520460055,
+      "loss": 0.2847,
+      "step": 2069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002900024999201634,
+      "loss": 0.2101,
+      "step": 2070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028999211943504403,
+      "loss": 0.156,
+      "step": 2071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899817337496278,
+      "loss": 0.3996,
+      "step": 2072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899713428643008,
+      "loss": 0.3294,
+      "step": 2073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899609467794488,
+      "loss": 0.2909,
+      "step": 2074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000289950545495458,
+      "loss": 0.3412,
+      "step": 2075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028994013901271495,
+      "loss": 0.3636,
+      "step": 2076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028992972733160604,
+      "loss": 0.2887,
+      "step": 2077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002899193104525181,
+      "loss": 0.1099,
+      "step": 2078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028990888837583813,
+      "loss": 0.2923,
+      "step": 2079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028989846110195325,
+      "loss": 0.3496,
+      "step": 2080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002898880286312507,
+      "loss": 0.2362,
+      "step": 2081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028987759096411815,
+      "loss": 0.2818,
+      "step": 2082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002898671481009433,
+      "loss": 0.2228,
+      "step": 2083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000289856700042114,
+      "loss": 0.2404,
+      "step": 2084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002898462467880185,
+      "loss": 0.249,
+      "step": 2085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000289835788339045,
+      "loss": 0.3787,
+      "step": 2086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000289825324695582,
+      "loss": 0.3229,
+      "step": 2087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028981485585801825,
+      "loss": 0.2444,
+      "step": 2088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002898043818267426,
+      "loss": 0.3079,
+      "step": 2089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00028979390260214417,
+      "loss": 0.2206,
+      "step": 2090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002897834181846121,
+      "loss": 0.4328,
+      "step": 2091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028977292857453605,
+      "loss": 0.2989,
+      "step": 2092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002897624337723055,
+      "loss": 0.2382,
+      "step": 2093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028975193377831046,
+      "loss": 0.1452,
+      "step": 2094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002897414285929409,
+      "loss": 0.3865,
+      "step": 2095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028973091821658705,
+      "loss": 0.2085,
+      "step": 2096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028972040264963936,
+      "loss": 0.2957,
+      "step": 2097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028970988189248837,
+      "loss": 0.3706,
+      "step": 2098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028969935594552504,
+      "loss": 0.0666,
+      "step": 2099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002896888248091402,
+      "loss": 0.1943,
+      "step": 2100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028967828848372527,
+      "loss": 0.3739,
+      "step": 2101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028966774696967146,
+      "loss": 0.2828,
+      "step": 2102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028965720026737047,
+      "loss": 0.2575,
+      "step": 2103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028964664837721396,
+      "loss": 0.233,
+      "step": 2104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000289636091299594,
+      "loss": 0.2298,
+      "step": 2105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028962552903490275,
+      "loss": 0.2106,
+      "step": 2106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002896149615835325,
+      "loss": 0.2942,
+      "step": 2107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028960438894587594,
+      "loss": 0.241,
+      "step": 2108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028959381112232565,
+      "loss": 0.2932,
+      "step": 2109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028958322811327463,
+      "loss": 0.1218,
+      "step": 2110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000289572639919116,
+      "loss": 0.2971,
+      "step": 2111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002895620465402431,
+      "loss": 0.3583,
+      "step": 2112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002895514479770495,
+      "loss": 0.2912,
+      "step": 2113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028954084422992877,
+      "loss": 0.2658,
+      "step": 2114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002895302352992749,
+      "loss": 0.2315,
+      "step": 2115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002895196211854819,
+      "loss": 0.2584,
+      "step": 2116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028950900188894416,
+      "loss": 0.1992,
+      "step": 2117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002894983774100561,
+      "loss": 0.3014,
+      "step": 2118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002894877477492124,
+      "loss": 0.3449,
+      "step": 2119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002894771129068079,
+      "loss": 0.1688,
+      "step": 2120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028946647288323766,
+      "loss": 0.3424,
+      "step": 2121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000289455827678897,
+      "loss": 0.0806,
+      "step": 2122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002894451772941812,
+      "loss": 0.2518,
+      "step": 2123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00028943452172948604,
+      "loss": 0.2594,
+      "step": 2124
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002894238609852073,
+      "loss": 0.2004,
+      "step": 2125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028941319506174093,
+      "loss": 0.1579,
+      "step": 2126
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028940252395948324,
+      "loss": 0.2112,
+      "step": 2127
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028939184767883057,
+      "loss": 0.4063,
+      "step": 2128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893811662201795,
+      "loss": 0.1858,
+      "step": 2129
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028937047958392683,
+      "loss": 0.196,
+      "step": 2130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893597877704696,
+      "loss": 0.0839,
+      "step": 2131
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028934909078020486,
+      "loss": 0.222,
+      "step": 2132
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028933838861353007,
+      "loss": 0.1792,
+      "step": 2133
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893276812708427,
+      "loss": 0.2623,
+      "step": 2134
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893169687525406,
+      "loss": 0.2668,
+      "step": 2135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893062510590217,
+      "loss": 0.3443,
+      "step": 2136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028929552819068397,
+      "loss": 0.1931,
+      "step": 2137
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892848001479259,
+      "loss": 0.2402,
+      "step": 2138
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028927406693114596,
+      "loss": 0.2735,
+      "step": 2139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028926332854074284,
+      "loss": 0.2428,
+      "step": 2140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028925258497711546,
+      "loss": 0.1069,
+      "step": 2141
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892418362406629,
+      "loss": 0.1721,
+      "step": 2142
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892310823317845,
+      "loss": 0.1699,
+      "step": 2143
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028922032325087965,
+      "loss": 0.2209,
+      "step": 2144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028920955899834795,
+      "loss": 0.1692,
+      "step": 2145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028919878957458947,
+      "loss": 0.0921,
+      "step": 2146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891880149800041,
+      "loss": 0.2417,
+      "step": 2147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028917723521499223,
+      "loss": 0.188,
+      "step": 2148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028916645027995415,
+      "loss": 0.2479,
+      "step": 2149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891556601752905,
+      "loss": 0.185,
+      "step": 2150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891448649014022,
+      "loss": 0.2642,
+      "step": 2151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891340644586902,
+      "loss": 0.0475,
+      "step": 2152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028912325884755566,
+      "loss": 0.2792,
+      "step": 2153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002891124480684001,
+      "loss": 0.2301,
+      "step": 2154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000289101632121625,
+      "loss": 0.1342,
+      "step": 2155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028909081100763224,
+      "loss": 0.3833,
+      "step": 2156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028907998472682366,
+      "loss": 0.2393,
+      "step": 2157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002890691532796015,
+      "loss": 0.2558,
+      "step": 2158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028905831666636817,
+      "loss": 0.2953,
+      "step": 2159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028904747488752607,
+      "loss": 0.1786,
+      "step": 2160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002890366279434781,
+      "loss": 0.2859,
+      "step": 2161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002890257758346271,
+      "loss": 0.4017,
+      "step": 2162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028901491856137625,
+      "loss": 0.2644,
+      "step": 2163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028900405612412877,
+      "loss": 0.2269,
+      "step": 2164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028899318852328826,
+      "loss": 0.189,
+      "step": 2165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889823157592584,
+      "loss": 0.3139,
+      "step": 2166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889714378324431,
+      "loss": 0.2782,
+      "step": 2167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889605547432464,
+      "loss": 0.2221,
+      "step": 2168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028894966649207257,
+      "loss": 0.286,
+      "step": 2169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028893877307932606,
+      "loss": 0.234,
+      "step": 2170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889278745054117,
+      "loss": 0.3447,
+      "step": 2171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028891697077073405,
+      "loss": 0.3069,
+      "step": 2172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002889060618756984,
+      "loss": 0.4429,
+      "step": 2173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028889514782070983,
+      "loss": 0.1042,
+      "step": 2174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028888422860617386,
+      "loss": 0.3361,
+      "step": 2175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028887330423249606,
+      "loss": 0.023,
+      "step": 2176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028886237470008224,
+      "loss": 0.1892,
+      "step": 2177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028885144000933844,
+      "loss": 0.207,
+      "step": 2178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002888405001606708,
+      "loss": 0.2621,
+      "step": 2179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002888295551544857,
+      "loss": 0.28,
+      "step": 2180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002888186049911898,
+      "loss": 0.2465,
+      "step": 2181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028880764967118976,
+      "loss": 0.4876,
+      "step": 2182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887966891948926,
+      "loss": 0.2097,
+      "step": 2183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028878572356270547,
+      "loss": 0.3362,
+      "step": 2184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028877475277503566,
+      "loss": 0.2765,
+      "step": 2185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887637768322908,
+      "loss": 0.1307,
+      "step": 2186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887527957348785,
+      "loss": 0.3246,
+      "step": 2187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887418094832068,
+      "loss": 0.2496,
+      "step": 2188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002887308180776836,
+      "loss": 0.1594,
+      "step": 2189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028871982151871747,
+      "loss": 0.3247,
+      "step": 2190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028870881980671664,
+      "loss": 0.5083,
+      "step": 2191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028869781294209003,
+      "loss": 0.2161,
+      "step": 2192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028868680092524634,
+      "loss": 0.1117,
+      "step": 2193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886757837565947,
+      "loss": 0.2273,
+      "step": 2194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028866476143654435,
+      "loss": 0.4372,
+      "step": 2195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028865373396550474,
+      "loss": 0.3961,
+      "step": 2196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886427013438855,
+      "loss": 0.2009,
+      "step": 2197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886316635720965,
+      "loss": 0.3717,
+      "step": 2198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028862062065054773,
+      "loss": 0.3502,
+      "step": 2199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002886095725796494,
+      "loss": 0.1152,
+      "step": 2200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028859851935981193,
+      "loss": 0.2331,
+      "step": 2201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028858746099144584,
+      "loss": 0.1267,
+      "step": 2202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028857639747496204,
+      "loss": 0.2809,
+      "step": 2203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028856532881077143,
+      "loss": 0.5856,
+      "step": 2204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885542549992852,
+      "loss": 0.3441,
+      "step": 2205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885431760409147,
+      "loss": 0.2123,
+      "step": 2206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885320919360715,
+      "loss": 0.3094,
+      "step": 2207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028852100268516727,
+      "loss": 0.2092,
+      "step": 2208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002885099082886141,
+      "loss": 0.1889,
+      "step": 2209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884988087468239,
+      "loss": 0.2273,
+      "step": 2210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028848770406020915,
+      "loss": 0.0473,
+      "step": 2211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884765942291823,
+      "loss": 0.2593,
+      "step": 2212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884654792541561,
+      "loss": 0.3744,
+      "step": 2213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028845435913554335,
+      "loss": 0.2797,
+      "step": 2214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028844323387375716,
+      "loss": 0.3436,
+      "step": 2215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884321034692109,
+      "loss": 0.2315,
+      "step": 2216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002884209679223178,
+      "loss": 0.2713,
+      "step": 2217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028840982723349177,
+      "loss": 0.3557,
+      "step": 2218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028839868140314646,
+      "loss": 0.1741,
+      "step": 2219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00028838753043169605,
+      "loss": 0.1037,
+      "step": 2220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002883763743195547,
+      "loss": 0.3386,
+      "step": 2221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002883652130671368,
+      "loss": 0.122,
+      "step": 2222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000288354046674857,
+      "loss": 0.0992,
+      "step": 2223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002883428751431301,
+      "loss": 0.1162,
+      "step": 2224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002883316984723711,
+      "loss": 0.393,
+      "step": 2225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028832051666299505,
+      "loss": 0.1881,
+      "step": 2226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002883093297154175,
+      "loss": 0.4579,
+      "step": 2227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028829813763005393,
+      "loss": 0.3657,
+      "step": 2228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882869404073201,
+      "loss": 0.1946,
+      "step": 2229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028827573804763196,
+      "loss": 0.2025,
+      "step": 2230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882645305514057,
+      "loss": 0.252,
+      "step": 2231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882533179190575,
+      "loss": 0.2377,
+      "step": 2232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000288242100151004,
+      "loss": 0.1884,
+      "step": 2233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002882308772476619,
+      "loss": 0.3037,
+      "step": 2234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000288219649209448,
+      "loss": 0.2095,
+      "step": 2235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028820841603677955,
+      "loss": 0.2796,
+      "step": 2236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881971777300736,
+      "loss": 0.4928,
+      "step": 2237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881859342897479,
+      "loss": 0.2571,
+      "step": 2238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002881746857162199,
+      "loss": 0.2727,
+      "step": 2239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028816343200990753,
+      "loss": 0.1297,
+      "step": 2240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028815217317122884,
+      "loss": 0.1976,
+      "step": 2241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000288140909200602,
+      "loss": 0.3672,
+      "step": 2242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028812964009844547,
+      "loss": 0.2956,
+      "step": 2243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028811836586517797,
+      "loss": 0.0831,
+      "step": 2244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028810708650121815,
+      "loss": 0.222,
+      "step": 2245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000288095802006985,
+      "loss": 0.3752,
+      "step": 2246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880845123828978,
+      "loss": 0.1161,
+      "step": 2247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028807321762937596,
+      "loss": 0.3533,
+      "step": 2248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880619177468389,
+      "loss": 0.1835,
+      "step": 2249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028805061273570656,
+      "loss": 0.3007,
+      "step": 2250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880393025963987,
+      "loss": 0.2778,
+      "step": 2251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880279873293356,
+      "loss": 0.3493,
+      "step": 2252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028801666693493744,
+      "loss": 0.352,
+      "step": 2253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002880053414136249,
+      "loss": 0.1785,
+      "step": 2254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00028799401076581865,
+      "loss": 0.3449,
+      "step": 2255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002879826749919395,
+      "loss": 0.2181,
+      "step": 2256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028797133409240867,
+      "loss": 0.3418,
+      "step": 2257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028795998806764735,
+      "loss": 0.1986,
+      "step": 2258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000287948636918077,
+      "loss": 0.3454,
+      "step": 2259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028793728064411937,
+      "loss": 0.1458,
+      "step": 2260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002879259192461962,
+      "loss": 0.2994,
+      "step": 2261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028791455272472965,
+      "loss": 0.1364,
+      "step": 2262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028790318108014193,
+      "loss": 0.2571,
+      "step": 2263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878918043128554,
+      "loss": 0.39,
+      "step": 2264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028788042242329265,
+      "loss": 0.4226,
+      "step": 2265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878690354118766,
+      "loss": 0.2179,
+      "step": 2266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878576432790301,
+      "loss": 0.3487,
+      "step": 2267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028784624602517654,
+      "loss": 0.0784,
+      "step": 2268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028783484365073905,
+      "loss": 0.2921,
+      "step": 2269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002878234361561414,
+      "loss": 0.2718,
+      "step": 2270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028781202354180726,
+      "loss": 0.0946,
+      "step": 2271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028780060580816053,
+      "loss": 0.2097,
+      "step": 2272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002877891829556254,
+      "loss": 0.1387,
+      "step": 2273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002877777549846262,
+      "loss": 0.1482,
+      "step": 2274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028776632189558745,
+      "loss": 0.3785,
+      "step": 2275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002877548836889338,
+      "loss": 0.2298,
+      "step": 2276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028774344036509023,
+      "loss": 0.4382,
+      "step": 2277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028773199192448173,
+      "loss": 0.3168,
+      "step": 2278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002877205383675337,
+      "loss": 0.1655,
+      "step": 2279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028770907969467147,
+      "loss": 0.2669,
+      "step": 2280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028769761590632074,
+      "loss": 0.1088,
+      "step": 2281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876861470029074,
+      "loss": 0.2163,
+      "step": 2282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876746729848575,
+      "loss": 0.2921,
+      "step": 2283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028766319385259713,
+      "loss": 0.268,
+      "step": 2284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028765170960655287,
+      "loss": 0.1042,
+      "step": 2285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876402202471513,
+      "loss": 0.4105,
+      "step": 2286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002876287257748191,
+      "loss": 0.286,
+      "step": 2287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00028761722618998335,
+      "loss": 0.2007,
+      "step": 2288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002876057214930712,
+      "loss": 0.1585,
+      "step": 2289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028759421168451005,
+      "loss": 0.2759,
+      "step": 2290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002875826967647274,
+      "loss": 0.1763,
+      "step": 2291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287571176734151,
+      "loss": 0.3029,
+      "step": 2292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028755965159320886,
+      "loss": 0.1643,
+      "step": 2293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287548121342329,
+      "loss": 0.2311,
+      "step": 2294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028753658598193986,
+      "loss": 0.2428,
+      "step": 2295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028752504551246983,
+      "loss": 0.3139,
+      "step": 2296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028751349993434765,
+      "loss": 0.3878,
+      "step": 2297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002875019492480022,
+      "loss": 0.1835,
+      "step": 2298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874903934538626,
+      "loss": 0.3283,
+      "step": 2299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287478832552358,
+      "loss": 0.2655,
+      "step": 2300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874672665439179,
+      "loss": 0.1229,
+      "step": 2301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028745569542897205,
+      "loss": 0.321,
+      "step": 2302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028744411920795024,
+      "loss": 0.2315,
+      "step": 2303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028743253788128233,
+      "loss": 0.3216,
+      "step": 2304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028742095144939874,
+      "loss": 0.2566,
+      "step": 2305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002874093599127298,
+      "loss": 0.1839,
+      "step": 2306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028739776327170604,
+      "loss": 0.3802,
+      "step": 2307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028738616152675827,
+      "loss": 0.4218,
+      "step": 2308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028737455467831756,
+      "loss": 0.2013,
+      "step": 2309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873629427268149,
+      "loss": 0.128,
+      "step": 2310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873513256726818,
+      "loss": 0.1239,
+      "step": 2311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873397035163497,
+      "loss": 0.4136,
+      "step": 2312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028732807625825036,
+      "loss": 0.3143,
+      "step": 2313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873164438988157,
+      "loss": 0.2253,
+      "step": 2314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002873048064384778,
+      "loss": 0.6165,
+      "step": 2315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287293163877669,
+      "loss": 0.2949,
+      "step": 2316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002872815162168218,
+      "loss": 0.2973,
+      "step": 2317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00028726986345636884,
+      "loss": 0.4118,
+      "step": 2318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287258205596743,
+      "loss": 0.3028,
+      "step": 2319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002872465426383773,
+      "loss": 0.3699,
+      "step": 2320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000287234874581705,
+      "loss": 0.2402,
+      "step": 2321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028722320142715954,
+      "loss": 0.1611,
+      "step": 2322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002872115231751746,
+      "loss": 0.3623,
+      "step": 2323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028719983982618394,
+      "loss": 0.4318,
+      "step": 2324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028718815138062154,
+      "loss": 0.411,
+      "step": 2325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028717645783892166,
+      "loss": 0.2506,
+      "step": 2326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871647592015186,
+      "loss": 0.1508,
+      "step": 2327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028715305546884697,
+      "loss": 0.26,
+      "step": 2328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028714134664134153,
+      "loss": 0.2502,
+      "step": 2329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028712963271943724,
+      "loss": 0.3457,
+      "step": 2330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028711791370356917,
+      "loss": 0.3628,
+      "step": 2331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002871061895941728,
+      "loss": 0.2857,
+      "step": 2332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870944603916835,
+      "loss": 0.2297,
+      "step": 2333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000287082726096537,
+      "loss": 0.2671,
+      "step": 2334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870709867091692,
+      "loss": 0.2219,
+      "step": 2335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028705924223001625,
+      "loss": 0.4238,
+      "step": 2336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028704749265951433,
+      "loss": 0.2338,
+      "step": 2337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870357379981,
+      "loss": 0.2755,
+      "step": 2338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870239782462098,
+      "loss": 0.166,
+      "step": 2339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870122134042807,
+      "loss": 0.3466,
+      "step": 2340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002870004434727496,
+      "loss": 0.1339,
+      "step": 2341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028698866845205376,
+      "loss": 0.3264,
+      "step": 2342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002869768883426306,
+      "loss": 0.1759,
+      "step": 2343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028696510314491777,
+      "loss": 0.1876,
+      "step": 2344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000286953312859353,
+      "loss": 0.2871,
+      "step": 2345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028694151748637426,
+      "loss": 0.1806,
+      "step": 2346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002869297170264197,
+      "loss": 0.2895,
+      "step": 2347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028691791147992775,
+      "loss": 0.3637,
+      "step": 2348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028690610084733685,
+      "loss": 0.2857,
+      "step": 2349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002868942851290858,
+      "loss": 0.0464,
+      "step": 2350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00028688246432561346,
+      "loss": 0.4376,
+      "step": 2351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000286870638437359,
+      "loss": 0.4993,
+      "step": 2352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002868588074647617,
+      "loss": 0.1392,
+      "step": 2353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000286846971408261,
+      "loss": 0.1963,
+      "step": 2354
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002868351302682967,
+      "loss": 0.3059,
+      "step": 2355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002868232840453085,
+      "loss": 0.2506,
+      "step": 2356
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028681143273973656,
+      "loss": 0.2321,
+      "step": 2357
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867995763520211,
+      "loss": 0.2557,
+      "step": 2358
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028678771488260243,
+      "loss": 0.5061,
+      "step": 2359
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867758483319214,
+      "loss": 0.3687,
+      "step": 2360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028676397670041867,
+      "loss": 0.1397,
+      "step": 2361
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028675209998853523,
+      "loss": 0.2158,
+      "step": 2362
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028674021819671234,
+      "loss": 0.3773,
+      "step": 2363
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867283313253913,
+      "loss": 0.1872,
+      "step": 2364
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867164393750137,
+      "loss": 0.1862,
+      "step": 2365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002867045423460214,
+      "loss": 0.2293,
+      "step": 2366
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866926402388561,
+      "loss": 0.3582,
+      "step": 2367
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028668073305396013,
+      "loss": 0.2238,
+      "step": 2368
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028666882079177565,
+      "loss": 0.1499,
+      "step": 2369
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866569034527454,
+      "loss": 0.2204,
+      "step": 2370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866449810373118,
+      "loss": 0.3417,
+      "step": 2371
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866330535459179,
+      "loss": 0.2349,
+      "step": 2372
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866211209790068,
+      "loss": 0.1945,
+      "step": 2373
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002866091833370216,
+      "loss": 0.2949,
+      "step": 2374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865972406204059,
+      "loss": 0.408,
+      "step": 2375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865852928296033,
+      "loss": 0.3832,
+      "step": 2376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028657333996505755,
+      "loss": 0.1489,
+      "step": 2377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865613820272127,
+      "loss": 0.0311,
+      "step": 2378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865494190165131,
+      "loss": 0.2701,
+      "step": 2379
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865374509334029,
+      "loss": 0.2397,
+      "step": 2380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865254777783268,
+      "loss": 0.1889,
+      "step": 2381
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028651349955172953,
+      "loss": 0.4901,
+      "step": 2382
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002865015162540561,
+      "loss": 0.417,
+      "step": 2383
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00028648952788575167,
+      "loss": 0.3234,
+      "step": 2384
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002864775344472615,
+      "loss": 0.3799,
+      "step": 2385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002864655359390312,
+      "loss": 0.0394,
+      "step": 2386
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002864535323615063,
+      "loss": 0.0823,
+      "step": 2387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002864415237151329,
+      "loss": 0.1259,
+      "step": 2388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000286429510000357,
+      "loss": 0.3381,
+      "step": 2389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002864174912176249,
+      "loss": 0.2694,
+      "step": 2390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028640546736738296,
+      "loss": 0.1879,
+      "step": 2391
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000286393438450078,
+      "loss": 0.2285,
+      "step": 2392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028638140446615674,
+      "loss": 0.3317,
+      "step": 2393
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863693654160662,
+      "loss": 0.3492,
+      "step": 2394
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863573213002537,
+      "loss": 0.2729,
+      "step": 2395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002863452721191665,
+      "loss": 0.2763,
+      "step": 2396
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028633321787325237,
+      "loss": 0.2695,
+      "step": 2397
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028632115856295896,
+      "loss": 0.2273,
+      "step": 2398
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028630909418873423,
+      "loss": 0.082,
+      "step": 2399
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862970247510264,
+      "loss": 0.2405,
+      "step": 2400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028628495025028377,
+      "loss": 0.271,
+      "step": 2401
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862728706869549,
+      "loss": 0.2416,
+      "step": 2402
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862607860614885,
+      "loss": 0.1304,
+      "step": 2403
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862486963743335,
+      "loss": 0.1423,
+      "step": 2404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028623660162593894,
+      "loss": 0.1282,
+      "step": 2405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862245018167542,
+      "loss": 0.1624,
+      "step": 2406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002862123969472286,
+      "loss": 0.1764,
+      "step": 2407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028620028701781193,
+      "loss": 0.2475,
+      "step": 2408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000286188172028954,
+      "loss": 0.3073,
+      "step": 2409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002861760519811049,
+      "loss": 0.3634,
+      "step": 2410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028616392687471476,
+      "loss": 0.2312,
+      "step": 2411
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002861517967102341,
+      "loss": 0.1431,
+      "step": 2412
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028613966148811334,
+      "loss": 0.1989,
+      "step": 2413
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028612752120880345,
+      "loss": 0.2999,
+      "step": 2414
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002861153758727553,
+      "loss": 0.0608,
+      "step": 2415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028610322548042015,
+      "loss": 0.327,
+      "step": 2416
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028609107003224924,
+      "loss": 0.3088,
+      "step": 2417
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002860789095286942,
+      "loss": 0.1438,
+      "step": 2418
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028606674397020664,
+      "loss": 0.1453,
+      "step": 2419
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00028605457335723864,
+      "loss": 0.3434,
+      "step": 2420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002860423976902422,
+      "loss": 0.2146,
+      "step": 2421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002860302169696696,
+      "loss": 0.3015,
+      "step": 2422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028601803119597335,
+      "loss": 0.2598,
+      "step": 2423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028600584036960613,
+      "loss": 0.1453,
+      "step": 2424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859936444910208,
+      "loss": 0.1685,
+      "step": 2425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028598144356067033,
+      "loss": 0.209,
+      "step": 2426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028596923757900797,
+      "loss": 0.3207,
+      "step": 2427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028595702654648716,
+      "loss": 0.1782,
+      "step": 2428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859448104635616,
+      "loss": 0.1517,
+      "step": 2429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859325893306849,
+      "loss": 0.1806,
+      "step": 2430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028592036314831117,
+      "loss": 0.2877,
+      "step": 2431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002859081319168945,
+      "loss": 0.1717,
+      "step": 2432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002858958956368893,
+      "loss": 0.1475,
+      "step": 2433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028588365430875004,
+      "loss": 0.3255,
+      "step": 2434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028587140793293156,
+      "loss": 0.2828,
+      "step": 2435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002858591565098887,
+      "loss": 0.217,
+      "step": 2436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028584690004007657,
+      "loss": 0.2412,
+      "step": 2437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028583463852395046,
+      "loss": 0.2234,
+      "step": 2438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028582237196196587,
+      "loss": 0.1458,
+      "step": 2439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028581010035457844,
+      "loss": 0.2847,
+      "step": 2440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028579782370224407,
+      "loss": 0.1574,
+      "step": 2441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857855420054188,
+      "loss": 0.2458,
+      "step": 2442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028577325526455875,
+      "loss": 0.1318,
+      "step": 2443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857609634801205,
+      "loss": 0.3089,
+      "step": 2444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857486666525605,
+      "loss": 0.2163,
+      "step": 2445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028573636478233566,
+      "loss": 0.2351,
+      "step": 2446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028572405786990294,
+      "loss": 0.2186,
+      "step": 2447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002857117459157194,
+      "loss": 0.2132,
+      "step": 2448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002856994289202425,
+      "loss": 0.1743,
+      "step": 2449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002856871068839298,
+      "loss": 0.313,
+      "step": 2450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002856747798072389,
+      "loss": 0.1392,
+      "step": 2451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00028566244769062785,
+      "loss": 0.1828,
+      "step": 2452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0002856501105345547,
+      "loss": 0.305,
+      "step": 2453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002856377683394777,
+      "loss": 0.2123,
+      "step": 2454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002856254211058554,
+      "loss": 0.2277,
+      "step": 2455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002856130688341464,
+      "loss": 0.2361,
+      "step": 2456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028560071152480956,
+      "loss": 0.3193,
+      "step": 2457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002855883491783039,
+      "loss": 0.2176,
+      "step": 2458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028557598179508874,
+      "loss": 0.2991,
+      "step": 2459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028556360937562344,
+      "loss": 0.2926,
+      "step": 2460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028555123192036756,
+      "loss": 0.2226,
+      "step": 2461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028553884942978086,
+      "loss": 0.4158,
+      "step": 2462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028552646190432347,
+      "loss": 0.2615,
+      "step": 2463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028551406934445537,
+      "loss": 0.3113,
+      "step": 2464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000285501671750637,
+      "loss": 0.3156,
+      "step": 2465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002854892691233289,
+      "loss": 0.0859,
+      "step": 2466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028547686146299176,
+      "loss": 0.3284,
+      "step": 2467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028546444877008655,
+      "loss": 0.2538,
+      "step": 2468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028545203104507423,
+      "loss": 0.175,
+      "step": 2469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028543960828841624,
+      "loss": 0.1089,
+      "step": 2470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028542718050057395,
+      "loss": 0.2567,
+      "step": 2471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000285414747682009,
+      "loss": 0.2362,
+      "step": 2472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028540230983318334,
+      "loss": 0.4656,
+      "step": 2473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002853898669545589,
+      "loss": 0.4329,
+      "step": 2474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028537741904659795,
+      "loss": 0.3118,
+      "step": 2475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028536496610976286,
+      "loss": 0.3089,
+      "step": 2476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002853525081445162,
+      "loss": 0.1194,
+      "step": 2477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002853400451513208,
+      "loss": 0.1869,
+      "step": 2478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028532757713063965,
+      "loss": 0.2219,
+      "step": 2479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002853151040829358,
+      "loss": 0.1605,
+      "step": 2480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028530262600867266,
+      "loss": 0.3385,
+      "step": 2481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002852901429083137,
+      "loss": 0.1001,
+      "step": 2482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002852776547823227,
+      "loss": 0.1294,
+      "step": 2483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028526516163116357,
+      "loss": 0.5129,
+      "step": 2484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0002852526634553003,
+      "loss": 0.3074,
+      "step": 2485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00028524016025519716,
+      "loss": 0.2502,
+      "step": 2486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028522765203131873,
+      "loss": 0.2607,
+      "step": 2487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028521513878412955,
+      "loss": 0.3722,
+      "step": 2488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002852026205140945,
+      "loss": 0.3353,
+      "step": 2489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028519009722167857,
+      "loss": 0.3443,
+      "step": 2490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028517756890734697,
+      "loss": 0.2284,
+      "step": 2491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028516503557156505,
+      "loss": 0.2451,
+      "step": 2492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851524972147985,
+      "loss": 0.3064,
+      "step": 2493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851399538375129,
+      "loss": 0.3499,
+      "step": 2494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002851274054401744,
+      "loss": 0.3286,
+      "step": 2495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028511485202324904,
+      "loss": 0.3517,
+      "step": 2496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028510229358720313,
+      "loss": 0.1303,
+      "step": 2497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002850897301325032,
+      "loss": 0.0878,
+      "step": 2498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028507716165961593,
+      "loss": 0.2626,
+      "step": 2499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002850645881690082,
+      "loss": 0.3046,
+      "step": 2500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002850520096611471,
+      "loss": 0.3793,
+      "step": 2501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028503942613649995,
+      "loss": 0.1084,
+      "step": 2502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002850268375955341,
+      "loss": 0.2113,
+      "step": 2503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028501424403871707,
+      "loss": 0.1279,
+      "step": 2504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028500164546651694,
+      "loss": 0.1393,
+      "step": 2505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849890418794015,
+      "loss": 0.3027,
+      "step": 2506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028497643327783905,
+      "loss": 0.27,
+      "step": 2507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849638196622979,
+      "loss": 0.1822,
+      "step": 2508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849512010332466,
+      "loss": 0.1452,
+      "step": 2509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002849385773911539,
+      "loss": 0.21,
+      "step": 2510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028492594873648884,
+      "loss": 0.438,
+      "step": 2511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028491331506972045,
+      "loss": 0.1076,
+      "step": 2512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028490067639131794,
+      "loss": 0.4603,
+      "step": 2513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000284888032701751,
+      "loss": 0.129,
+      "step": 2514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00028487538400148916,
+      "loss": 0.1634,
+      "step": 2515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848627302910023,
+      "loss": 0.0546,
+      "step": 2516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848500715707605,
+      "loss": 0.292,
+      "step": 2517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000284837407841234,
+      "loss": 0.1587,
+      "step": 2518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002848247391028932,
+      "loss": 0.2358,
+      "step": 2519
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002848120653562087,
+      "loss": 0.3936,
+      "step": 2520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002847993866016513,
+      "loss": 0.1853,
+      "step": 2521
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000284786702839692,
+      "loss": 0.2624,
+      "step": 2522
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000284774014070802,
+      "loss": 0.3168,
+      "step": 2523
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002847613202954525,
+      "loss": 0.3174,
+      "step": 2524
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002847486215141152,
+      "loss": 0.3391,
+      "step": 2525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028473591772726174,
+      "loss": 0.2767,
+      "step": 2526
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028472320893536404,
+      "loss": 0.2851,
+      "step": 2527
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002847104951388942,
+      "loss": 0.2464,
+      "step": 2528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028469777633832456,
+      "loss": 0.1894,
+      "step": 2529
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002846850525341274,
+      "loss": 0.1279,
+      "step": 2530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002846723237267756,
+      "loss": 0.3391,
+      "step": 2531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002846595899167419,
+      "loss": 0.4101,
+      "step": 2532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028464685110449927,
+      "loss": 0.3756,
+      "step": 2533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000284634107290521,
+      "loss": 0.2937,
+      "step": 2534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002846213584752804,
+      "loss": 0.1586,
+      "step": 2535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002846086046592512,
+      "loss": 0.1591,
+      "step": 2536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000284595845842907,
+      "loss": 0.2577,
+      "step": 2537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845830820267219,
+      "loss": 0.2831,
+      "step": 2538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845703132111699,
+      "loss": 0.2121,
+      "step": 2539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845575393967254,
+      "loss": 0.1888,
+      "step": 2540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845447605838629,
+      "loss": 0.1766,
+      "step": 2541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845319767730571,
+      "loss": 0.1523,
+      "step": 2542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002845191879647829,
+      "loss": 0.3325,
+      "step": 2543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028450639415951526,
+      "loss": 0.15,
+      "step": 2544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028449359535772956,
+      "loss": 0.1572,
+      "step": 2545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028448079155990127,
+      "loss": 0.2961,
+      "step": 2546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028446798276650587,
+      "loss": 0.2128,
+      "step": 2547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028445516897801916,
+      "loss": 0.2533,
+      "step": 2548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002844423501949173,
+      "loss": 0.3955,
+      "step": 2549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002844295264176763,
+      "loss": 0.1414,
+      "step": 2550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00028441669764677266,
+      "loss": 0.2914,
+      "step": 2551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0002844038638826829,
+      "loss": 0.3946,
+      "step": 2552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028439102512588374,
+      "loss": 0.3383,
+      "step": 2553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843781813768521,
+      "loss": 0.2657,
+      "step": 2554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028436533263606503,
+      "loss": 0.1603,
+      "step": 2555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843524789039999,
+      "loss": 0.2567,
+      "step": 2556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028433962018113414,
+      "loss": 0.2055,
+      "step": 2557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843267564679454,
+      "loss": 0.3409,
+      "step": 2558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002843138877649117,
+      "loss": 0.2079,
+      "step": 2559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028430101407251083,
+      "loss": 0.2987,
+      "step": 2560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002842881353912212,
+      "loss": 0.2349,
+      "step": 2561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000284275251721521,
+      "loss": 0.0914,
+      "step": 2562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002842623630638891,
+      "loss": 0.4576,
+      "step": 2563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002842494694188041,
+      "loss": 0.2337,
+      "step": 2564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028423657078674496,
+      "loss": 0.266,
+      "step": 2565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028422366716819087,
+      "loss": 0.3541,
+      "step": 2566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028421075856362117,
+      "loss": 0.1284,
+      "step": 2567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028419784497351534,
+      "loss": 0.2697,
+      "step": 2568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841849263983531,
+      "loss": 0.201,
+      "step": 2569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841720028386144,
+      "loss": 0.2682,
+      "step": 2570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841590742947792,
+      "loss": 0.2495,
+      "step": 2571
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841461407673278,
+      "loss": 0.3568,
+      "step": 2572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028413320225674065,
+      "loss": 0.3132,
+      "step": 2573
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841202587634984,
+      "loss": 0.1652,
+      "step": 2574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002841073102880818,
+      "loss": 0.2361,
+      "step": 2575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840943568309719,
+      "loss": 0.0557,
+      "step": 2576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840813983926499,
+      "loss": 0.2375,
+      "step": 2577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840684349735972,
+      "loss": 0.2281,
+      "step": 2578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028405546657429517,
+      "loss": 0.2889,
+      "step": 2579
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028404249319522574,
+      "loss": 0.2733,
+      "step": 2580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840295148368707,
+      "loss": 0.3435,
+      "step": 2581
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002840165314997123,
+      "loss": 0.0879,
+      "step": 2582
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028400354318423274,
+      "loss": 0.1987,
+      "step": 2583
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00028399054989091446,
+      "loss": 0.2304,
+      "step": 2584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0002839775516202403,
+      "loss": 0.2919,
+      "step": 2585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028396454837269284,
+      "loss": 0.2287,
+      "step": 2586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002839515401487553,
+      "loss": 0.3576,
+      "step": 2587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002839385269489109,
+      "loss": 0.3834,
+      "step": 2588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028392550877364293,
+      "loss": 0.2523,
+      "step": 2589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002839124856234351,
+      "loss": 0.1565,
+      "step": 2590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028389945749877107,
+      "loss": 0.2351,
+      "step": 2591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002838864244001349,
+      "loss": 0.303,
+      "step": 2592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002838733863280107,
+      "loss": 0.1637,
+      "step": 2593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028386034328288273,
+      "loss": 0.2744,
+      "step": 2594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028384729526523554,
+      "loss": 0.3423,
+      "step": 2595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028383424227555385,
+      "loss": 0.1597,
+      "step": 2596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028382118431432256,
+      "loss": 0.2284,
+      "step": 2597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028380812138202664,
+      "loss": 0.3278,
+      "step": 2598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028379505347915143,
+      "loss": 0.3565,
+      "step": 2599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028378198060618234,
+      "loss": 0.2924,
+      "step": 2600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028376890276360497,
+      "loss": 0.1676,
+      "step": 2601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028375581995190516,
+      "loss": 0.3149,
+      "step": 2602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028374273217156884,
+      "loss": 0.2699,
+      "step": 2603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002837296394230822,
+      "loss": 0.1266,
+      "step": 2604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002837165417069317,
+      "loss": 0.4045,
+      "step": 2605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002837034390236037,
+      "loss": 0.258,
+      "step": 2606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028369033137358506,
+      "loss": 0.2908,
+      "step": 2607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028367721875736273,
+      "loss": 0.2548,
+      "step": 2608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028366410117542366,
+      "loss": 0.0907,
+      "step": 2609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028365097862825513,
+      "loss": 0.1327,
+      "step": 2610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028363785111634477,
+      "loss": 0.2357,
+      "step": 2611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002836247186401801,
+      "loss": 0.2082,
+      "step": 2612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028361158120024897,
+      "loss": 0.2264,
+      "step": 2613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002835984387970394,
+      "loss": 0.2228,
+      "step": 2614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028358529143103966,
+      "loss": 0.2715,
+      "step": 2615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028357213910273806,
+      "loss": 0.1989,
+      "step": 2616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028355898181262317,
+      "loss": 0.2547,
+      "step": 2617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00028354581956118374,
+      "loss": 0.2238,
+      "step": 2618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028353265234890875,
+      "loss": 0.1643,
+      "step": 2619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028351948017628733,
+      "loss": 0.2952,
+      "step": 2620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028350630304380874,
+      "loss": 0.1857,
+      "step": 2621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028349312095196253,
+      "loss": 0.1306,
+      "step": 2622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028347993390123835,
+      "loss": 0.259,
+      "step": 2623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000283466741892126,
+      "loss": 0.3032,
+      "step": 2624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028345354492511564,
+      "loss": 0.3668,
+      "step": 2625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834403430006974,
+      "loss": 0.3183,
+      "step": 2626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834271361193617,
+      "loss": 0.3194,
+      "step": 2627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028341392428159917,
+      "loss": 0.2337,
+      "step": 2628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002834007074879006,
+      "loss": 0.349,
+      "step": 2629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028338748573875696,
+      "loss": 0.4203,
+      "step": 2630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028337425903465936,
+      "loss": 0.2658,
+      "step": 2631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028336102737609915,
+      "loss": 0.0937,
+      "step": 2632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833477907635679,
+      "loss": 0.0448,
+      "step": 2633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833345491975572,
+      "loss": 0.2447,
+      "step": 2634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000283321302678559,
+      "loss": 0.2931,
+      "step": 2635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002833080512070654,
+      "loss": 0.1493,
+      "step": 2636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832947947835686,
+      "loss": 0.1988,
+      "step": 2637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028328153340856106,
+      "loss": 0.2227,
+      "step": 2638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028326826708253544,
+      "loss": 0.3022,
+      "step": 2639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832549958059845,
+      "loss": 0.3969,
+      "step": 2640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028324171957940116,
+      "loss": 0.247,
+      "step": 2641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002832284384032787,
+      "loss": 0.2796,
+      "step": 2642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028321515227811044,
+      "loss": 0.079,
+      "step": 2643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028320186120438993,
+      "loss": 0.1519,
+      "step": 2644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002831885651826109,
+      "loss": 0.2059,
+      "step": 2645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028317526421326724,
+      "loss": 0.2823,
+      "step": 2646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028316195829685304,
+      "loss": 0.2665,
+      "step": 2647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002831486474338626,
+      "loss": 0.2914,
+      "step": 2648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028313533162479035,
+      "loss": 0.288,
+      "step": 2649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00028312201087013093,
+      "loss": 0.2685,
+      "step": 2650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0002831086851703792,
+      "loss": 0.3462,
+      "step": 2651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028309535452603014,
+      "loss": 0.1232,
+      "step": 2652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000283082018937579,
+      "loss": 0.2011,
+      "step": 2653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028306867840552103,
+      "loss": 0.2888,
+      "step": 2654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028305533293035195,
+      "loss": 0.4717,
+      "step": 2655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028304198251256744,
+      "loss": 0.1238,
+      "step": 2656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002830286271526634,
+      "loss": 0.2843,
+      "step": 2657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000283015266851136,
+      "loss": 0.1643,
+      "step": 2658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002830019016084814,
+      "loss": 0.2778,
+      "step": 2659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002829885314251963,
+      "loss": 0.3133,
+      "step": 2660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028297515630177717,
+      "loss": 0.2423,
+      "step": 2661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028296177623872096,
+      "loss": 0.2861,
+      "step": 2662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002829483912365246,
+      "loss": 0.166,
+      "step": 2663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028293500129568546,
+      "loss": 0.1727,
+      "step": 2664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028292160641670087,
+      "loss": 0.4252,
+      "step": 2665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002829082066000683,
+      "loss": 0.1348,
+      "step": 2666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028289480184628566,
+      "loss": 0.2926,
+      "step": 2667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828813921558508,
+      "loss": 0.2686,
+      "step": 2668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828679775292619,
+      "loss": 0.1731,
+      "step": 2669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028285455796701734,
+      "loss": 0.285,
+      "step": 2670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828411334696155,
+      "loss": 0.2841,
+      "step": 2671
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002828277040375551,
+      "loss": 0.3598,
+      "step": 2672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000282814269671335,
+      "loss": 0.346,
+      "step": 2673
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028280083037145426,
+      "loss": 0.2712,
+      "step": 2674
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002827873861384121,
+      "loss": 0.3283,
+      "step": 2675
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028277393697270797,
+      "loss": 0.1587,
+      "step": 2676
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002827604828748414,
+      "loss": 0.1677,
+      "step": 2677
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028274702384531225,
+      "loss": 0.2849,
+      "step": 2678
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028273355988462046,
+      "loss": 0.4161,
+      "step": 2679
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028272009099326607,
+      "loss": 0.1037,
+      "step": 2680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028270661717174955,
+      "loss": 0.3018,
+      "step": 2681
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002826931384205714,
+      "loss": 0.3055,
+      "step": 2682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002826796547402322,
+      "loss": 0.1702,
+      "step": 2683
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00028266616613123295,
+      "loss": 0.3672,
+      "step": 2684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002826526725940746,
+      "loss": 0.2113,
+      "step": 2685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028263917412925854,
+      "loss": 0.3316,
+      "step": 2686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002826256707372861,
+      "loss": 0.1709,
+      "step": 2687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028261216241865894,
+      "loss": 0.1832,
+      "step": 2688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825986491738788,
+      "loss": 0.2113,
+      "step": 2689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825851310034477,
+      "loss": 0.263,
+      "step": 2690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825716079078678,
+      "loss": 0.2749,
+      "step": 2691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825580798876414,
+      "loss": 0.2663,
+      "step": 2692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825445469432711,
+      "loss": 0.3025,
+      "step": 2693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028253100907525946,
+      "loss": 0.1546,
+      "step": 2694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028251746628410956,
+      "loss": 0.2793,
+      "step": 2695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002825039185703244,
+      "loss": 0.2866,
+      "step": 2696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028249036593440717,
+      "loss": 0.249,
+      "step": 2697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002824768083768614,
+      "loss": 0.3001,
+      "step": 2698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028246324589819075,
+      "loss": 0.1807,
+      "step": 2699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002824496784988989,
+      "loss": 0.2062,
+      "step": 2700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028243610617948985,
+      "loss": 0.096,
+      "step": 2701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028242252894046787,
+      "loss": 0.1426,
+      "step": 2702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002824089467823373,
+      "loss": 0.1888,
+      "step": 2703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823953597056026,
+      "loss": 0.1842,
+      "step": 2704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028238176771076855,
+      "loss": 0.3438,
+      "step": 2705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028236817079834,
+      "loss": 0.3248,
+      "step": 2706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823545689688221,
+      "loss": 0.1331,
+      "step": 2707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823409622227201,
+      "loss": 0.1972,
+      "step": 2708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028232735056053944,
+      "loss": 0.0681,
+      "step": 2709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823137339827857,
+      "loss": 0.4132,
+      "step": 2710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002823001124899648,
+      "loss": 0.3165,
+      "step": 2711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822864860825827,
+      "loss": 0.1851,
+      "step": 2712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822728547611455,
+      "loss": 0.2708,
+      "step": 2713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822592185261597,
+      "loss": 0.1629,
+      "step": 2714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028224557737813183,
+      "loss": 0.19,
+      "step": 2715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00028223193131756845,
+      "loss": 0.1157,
+      "step": 2716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0002822182803449766,
+      "loss": 0.3271,
+      "step": 2717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028220462446086343,
+      "loss": 0.2251,
+      "step": 2718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028219096366573616,
+      "loss": 0.2614,
+      "step": 2719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028217729796010217,
+      "loss": 0.3591,
+      "step": 2720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028216362734446917,
+      "loss": 0.3745,
+      "step": 2721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000282149951819345,
+      "loss": 0.2142,
+      "step": 2722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821362713852377,
+      "loss": 0.148,
+      "step": 2723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002821225860426553,
+      "loss": 0.272,
+      "step": 2724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028210889579210634,
+      "loss": 0.2098,
+      "step": 2725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002820952006340993,
+      "loss": 0.231,
+      "step": 2726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002820815005691429,
+      "loss": 0.3557,
+      "step": 2727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002820677955977461,
+      "loss": 0.3056,
+      "step": 2728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028205408572041797,
+      "loss": 0.4672,
+      "step": 2729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002820403709376678,
+      "loss": 0.3447,
+      "step": 2730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028202665125000504,
+      "loss": 0.2766,
+      "step": 2731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002820129266579394,
+      "loss": 0.1602,
+      "step": 2732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028199919716198065,
+      "loss": 0.133,
+      "step": 2733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028198546276263876,
+      "loss": 0.2877,
+      "step": 2734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000281971723460424,
+      "loss": 0.3307,
+      "step": 2735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028195797925584676,
+      "loss": 0.3042,
+      "step": 2736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002819442301494175,
+      "loss": 0.3478,
+      "step": 2737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000281930476141647,
+      "loss": 0.3621,
+      "step": 2738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028191671723304623,
+      "loss": 0.2547,
+      "step": 2739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028190295342412627,
+      "loss": 0.1766,
+      "step": 2740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002818891847153983,
+      "loss": 0.3334,
+      "step": 2741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000281875411107374,
+      "loss": 0.1989,
+      "step": 2742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028186163260056487,
+      "loss": 0.2245,
+      "step": 2743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028184784919548276,
+      "loss": 0.2581,
+      "step": 2744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002818340608926397,
+      "loss": 0.2093,
+      "step": 2745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002818202676925478,
+      "loss": 0.1968,
+      "step": 2746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002818064695957196,
+      "loss": 0.2583,
+      "step": 2747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028179266660266754,
+      "loss": 0.0873,
+      "step": 2748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002817788587139044,
+      "loss": 0.3419,
+      "step": 2749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00028176504592994313,
+      "loss": 0.1565,
+      "step": 2750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002817512282512968,
+      "loss": 0.1243,
+      "step": 2751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028173740567847866,
+      "loss": 0.1802,
+      "step": 2752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002817235782120022,
+      "loss": 0.2511,
+      "step": 2753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002817097458523811,
+      "loss": 0.2849,
+      "step": 2754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816959086001292,
+      "loss": 0.1832,
+      "step": 2755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816820664557605,
+      "loss": 0.171,
+      "step": 2756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028166821941978914,
+      "loss": 0.3566,
+      "step": 2757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816543674927296,
+      "loss": 0.4093,
+      "step": 2758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028164051067509633,
+      "loss": 0.132,
+      "step": 2759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002816266489674041,
+      "loss": 0.1953,
+      "step": 2760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028161278237016793,
+      "loss": 0.1812,
+      "step": 2761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028159891088390285,
+      "loss": 0.2771,
+      "step": 2762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028158503450912407,
+      "loss": 0.1141,
+      "step": 2763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028157115324634714,
+      "loss": 0.2457,
+      "step": 2764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028155726709608775,
+      "loss": 0.3757,
+      "step": 2765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002815433760588616,
+      "loss": 0.2743,
+      "step": 2766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002815294801351848,
+      "loss": 0.1273,
+      "step": 2767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028151557932557355,
+      "loss": 0.1151,
+      "step": 2768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002815016736305442,
+      "loss": 0.109,
+      "step": 2769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814877630506133,
+      "loss": 0.2279,
+      "step": 2770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814738475862976,
+      "loss": 0.1263,
+      "step": 2771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028145992723811394,
+      "loss": 0.0822,
+      "step": 2772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814460020065795,
+      "loss": 0.3049,
+      "step": 2773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814320718922116,
+      "loss": 0.3453,
+      "step": 2774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814181368955276,
+      "loss": 0.2781,
+      "step": 2775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002814041970170452,
+      "loss": 0.2899,
+      "step": 2776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028139025225728224,
+      "loss": 0.1927,
+      "step": 2777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813763026167567,
+      "loss": 0.1704,
+      "step": 2778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028136234809598683,
+      "loss": 0.2539,
+      "step": 2779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813483886954909,
+      "loss": 0.247,
+      "step": 2780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00028133442441578745,
+      "loss": 0.3486,
+      "step": 2781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813204552573953,
+      "loss": 0.358,
+      "step": 2782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002813064812208334,
+      "loss": 0.2872,
+      "step": 2783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028129250230662066,
+      "loss": 0.1107,
+      "step": 2784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002812785185152766,
+      "loss": 0.2275,
+      "step": 2785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028126452984732046,
+      "loss": 0.2844,
+      "step": 2786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028125053630327195,
+      "loss": 0.4154,
+      "step": 2787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028123653788365095,
+      "loss": 0.1475,
+      "step": 2788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028122253458897744,
+      "loss": 0.1892,
+      "step": 2789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028120852641977157,
+      "loss": 0.2618,
+      "step": 2790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002811945133765537,
+      "loss": 0.3244,
+      "step": 2791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028118049545984434,
+      "loss": 0.2594,
+      "step": 2792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002811664726701643,
+      "loss": 0.1388,
+      "step": 2793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028115244500803444,
+      "loss": 0.2609,
+      "step": 2794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002811384124739758,
+      "loss": 0.1731,
+      "step": 2795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002811243750685097,
+      "loss": 0.2302,
+      "step": 2796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028111033279215765,
+      "loss": 0.1927,
+      "step": 2797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810962856454412,
+      "loss": 0.2406,
+      "step": 2798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810822336288822,
+      "loss": 0.1888,
+      "step": 2799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028106817674300254,
+      "loss": 0.2021,
+      "step": 2800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810541149883245,
+      "loss": 0.1487,
+      "step": 2801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002810400483653704,
+      "loss": 0.2595,
+      "step": 2802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028102597687466277,
+      "loss": 0.2433,
+      "step": 2803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028101190051672435,
+      "loss": 0.2186,
+      "step": 2804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028099781929207803,
+      "loss": 0.1346,
+      "step": 2805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002809837332012468,
+      "loss": 0.3087,
+      "step": 2806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028096964224475407,
+      "loss": 0.4218,
+      "step": 2807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028095554642312316,
+      "loss": 0.0904,
+      "step": 2808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028094144573687774,
+      "loss": 0.1866,
+      "step": 2809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002809273401865416,
+      "loss": 0.2224,
+      "step": 2810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028091322977263867,
+      "loss": 0.3812,
+      "step": 2811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002808991144956932,
+      "loss": 0.3513,
+      "step": 2812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002808849943562294,
+      "loss": 0.2629,
+      "step": 2813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00028087086935477194,
+      "loss": 0.1646,
+      "step": 2814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0002808567394918455,
+      "loss": 0.2088,
+      "step": 2815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028084260476797485,
+      "loss": 0.1178,
+      "step": 2816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002808284651836852,
+      "loss": 0.2981,
+      "step": 2817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028081432073950163,
+      "loss": 0.2812,
+      "step": 2818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002808001714359497,
+      "loss": 0.2207,
+      "step": 2819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000280786017273555,
+      "loss": 0.4173,
+      "step": 2820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002807718582528432,
+      "loss": 0.3902,
+      "step": 2821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028075769437434044,
+      "loss": 0.1241,
+      "step": 2822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002807435256385727,
+      "loss": 0.1963,
+      "step": 2823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028072935204606647,
+      "loss": 0.3101,
+      "step": 2824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002807151735973481,
+      "loss": 0.4083,
+      "step": 2825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028070099029294443,
+      "loss": 0.3701,
+      "step": 2826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028068680213338217,
+      "loss": 0.1228,
+      "step": 2827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028067260911918845,
+      "loss": 0.1671,
+      "step": 2828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028065841125089053,
+      "loss": 0.28,
+      "step": 2829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002806442085290158,
+      "loss": 0.2628,
+      "step": 2830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028063000095409176,
+      "loss": 0.2091,
+      "step": 2831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028061578852664633,
+      "loss": 0.4458,
+      "step": 2832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028060157124720737,
+      "loss": 0.3483,
+      "step": 2833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000280587349116303,
+      "loss": 0.2457,
+      "step": 2834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028057312213446156,
+      "loss": 0.1704,
+      "step": 2835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028055889030221154,
+      "loss": 0.2389,
+      "step": 2836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002805446536200816,
+      "loss": 0.3695,
+      "step": 2837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002805304120886006,
+      "loss": 0.0928,
+      "step": 2838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002805161657082976,
+      "loss": 0.4602,
+      "step": 2839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002805019144797017,
+      "loss": 0.3345,
+      "step": 2840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804876584033424,
+      "loss": 0.15,
+      "step": 2841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804733974797493,
+      "loss": 0.1458,
+      "step": 2842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804591317094521,
+      "loss": 0.1689,
+      "step": 2843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028044486109298064,
+      "loss": 0.223,
+      "step": 2844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028043058563086517,
+      "loss": 0.2854,
+      "step": 2845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804163053236359,
+      "loss": 0.342,
+      "step": 2846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002804020201718234,
+      "loss": 0.2238,
+      "step": 2847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00028038773017595816,
+      "loss": 0.2057,
+      "step": 2848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028037343533657115,
+      "loss": 0.3458,
+      "step": 2849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002803591356541933,
+      "loss": 0.2681,
+      "step": 2850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028034483112935585,
+      "loss": 0.1759,
+      "step": 2851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002803305217625902,
+      "loss": 0.1148,
+      "step": 2852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002803162075544278,
+      "loss": 0.3498,
+      "step": 2853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002803018885054005,
+      "loss": 0.3244,
+      "step": 2854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028028756461604006,
+      "loss": 0.2269,
+      "step": 2855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002802732358868787,
+      "loss": 0.171,
+      "step": 2856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002802589023184487,
+      "loss": 0.2071,
+      "step": 2857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028024456391128244,
+      "loss": 0.2708,
+      "step": 2858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028023022066591254,
+      "loss": 0.2348,
+      "step": 2859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028021587258287184,
+      "loss": 0.1686,
+      "step": 2860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028020151966269335,
+      "loss": 0.4509,
+      "step": 2861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801871619059102,
+      "loss": 0.1086,
+      "step": 2862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028017279931305577,
+      "loss": 0.2836,
+      "step": 2863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801584318846636,
+      "loss": 0.2156,
+      "step": 2864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801440596212673,
+      "loss": 0.2417,
+      "step": 2865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801296825234009,
+      "loss": 0.34,
+      "step": 2866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002801153005915983,
+      "loss": 0.2535,
+      "step": 2867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028010091382639393,
+      "loss": 0.2968,
+      "step": 2868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028008652222832214,
+      "loss": 0.0441,
+      "step": 2869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00028007212579791746,
+      "loss": 0.1775,
+      "step": 2870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800577245357148,
+      "loss": 0.2941,
+      "step": 2871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000280043318442249,
+      "loss": 0.2132,
+      "step": 2872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800289075180553,
+      "loss": 0.3304,
+      "step": 2873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000280014491763669,
+      "loss": 0.3152,
+      "step": 2874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002800000711796256,
+      "loss": 0.3198,
+      "step": 2875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027998564576646077,
+      "loss": 0.2334,
+      "step": 2876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002799712155247104,
+      "loss": 0.2312,
+      "step": 2877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002799567804549105,
+      "loss": 0.4106,
+      "step": 2878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00027994234055759724,
+      "loss": 0.3122,
+      "step": 2879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002799278958333072,
+      "loss": 0.285,
+      "step": 2880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002799134462825767,
+      "loss": 0.331,
+      "step": 2881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027989899190594274,
+      "loss": 0.3346,
+      "step": 2882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002798845327039421,
+      "loss": 0.425,
+      "step": 2883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000279870068677112,
+      "loss": 0.2932,
+      "step": 2884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002798555998259897,
+      "loss": 0.2167,
+      "step": 2885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027984112615111263,
+      "loss": 0.2228,
+      "step": 2886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027982664765301846,
+      "loss": 0.2081,
+      "step": 2887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002798121643322451,
+      "loss": 0.3677,
+      "step": 2888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002797976761893304,
+      "loss": 0.1987,
+      "step": 2889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002797831832248128,
+      "loss": 0.2948,
+      "step": 2890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027976868543923045,
+      "loss": 0.254,
+      "step": 2891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000279754182833122,
+      "loss": 0.3475,
+      "step": 2892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027973967540702614,
+      "loss": 0.271,
+      "step": 2893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002797251631614818,
+      "loss": 0.2212,
+      "step": 2894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002797106460970281,
+      "loss": 0.1665,
+      "step": 2895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002796961242142042,
+      "loss": 0.1631,
+      "step": 2896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027968159751354965,
+      "loss": 0.2122,
+      "step": 2897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002796670659956041,
+      "loss": 0.232,
+      "step": 2898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002796525296609072,
+      "loss": 0.1682,
+      "step": 2899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027963798850999907,
+      "loss": 0.3035,
+      "step": 2900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002796234425434198,
+      "loss": 0.3432,
+      "step": 2901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002796088917617098,
+      "loss": 0.1065,
+      "step": 2902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027959433616540956,
+      "loss": 0.3399,
+      "step": 2903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002795797757550597,
+      "loss": 0.2427,
+      "step": 2904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002795652105312012,
+      "loss": 0.2693,
+      "step": 2905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002795506404943751,
+      "loss": 0.25,
+      "step": 2906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002795360656451225,
+      "loss": 0.3361,
+      "step": 2907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000279521485983985,
+      "loss": 0.3724,
+      "step": 2908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002795069015115042,
+      "loss": 0.2386,
+      "step": 2909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002794923122282217,
+      "loss": 0.1653,
+      "step": 2910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00027947771813467947,
+      "loss": 0.3096,
+      "step": 2911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002794631192314198,
+      "loss": 0.3544,
+      "step": 2912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002794485155189849,
+      "loss": 0.3441,
+      "step": 2913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002794339069979172,
+      "loss": 0.2595,
+      "step": 2914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027941929366875944,
+      "loss": 0.2483,
+      "step": 2915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002794046755320544,
+      "loss": 0.2754,
+      "step": 2916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027939005258834516,
+      "loss": 0.108,
+      "step": 2917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027937542483817495,
+      "loss": 0.3532,
+      "step": 2918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027936079228208703,
+      "loss": 0.2185,
+      "step": 2919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793461549206251,
+      "loss": 0.2841,
+      "step": 2920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002793315127543327,
+      "loss": 0.1025,
+      "step": 2921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027931686578375393,
+      "loss": 0.2887,
+      "step": 2922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027930221400943283,
+      "loss": 0.4155,
+      "step": 2923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027928755743191364,
+      "loss": 0.3773,
+      "step": 2924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027927289605174084,
+      "loss": 0.3285,
+      "step": 2925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000279258229869459,
+      "loss": 0.2393,
+      "step": 2926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027924355888561306,
+      "loss": 0.4374,
+      "step": 2927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002792288831007478,
+      "loss": 0.1063,
+      "step": 2928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027921420251540854,
+      "loss": 0.4349,
+      "step": 2929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002791995171301406,
+      "loss": 0.1938,
+      "step": 2930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027918482694548944,
+      "loss": 0.2145,
+      "step": 2931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002791701319620008,
+      "loss": 0.229,
+      "step": 2932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027915543218022055,
+      "loss": 0.2036,
+      "step": 2933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027914072760069474,
+      "loss": 0.1923,
+      "step": 2934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027912601822396957,
+      "loss": 0.2781,
+      "step": 2935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002791113040505915,
+      "loss": 0.2242,
+      "step": 2936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027909658508110715,
+      "loss": 0.1302,
+      "step": 2937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027908186131606316,
+      "loss": 0.3349,
+      "step": 2938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002790671327560066,
+      "loss": 0.2332,
+      "step": 2939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002790523994014845,
+      "loss": 0.3597,
+      "step": 2940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027903766125304423,
+      "loss": 0.2724,
+      "step": 2941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002790229183112333,
+      "loss": 0.2508,
+      "step": 2942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027900817057659926,
+      "loss": 0.1282,
+      "step": 2943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00027899341804969,
+      "loss": 0.2823,
+      "step": 2944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002789786607310535,
+      "loss": 0.3713,
+      "step": 2945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000278963898621238,
+      "loss": 0.2512,
+      "step": 2946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0002789491317207919,
+      "loss": 0.1621,
+      "step": 2947
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027893436003026363,
+      "loss": 0.1696,
+      "step": 2948
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000278919583550202,
+      "loss": 0.0798,
+      "step": 2949
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002789048022811559,
+      "loss": 0.4016,
+      "step": 2950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027889001622367443,
+      "loss": 0.0911,
+      "step": 2951
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027887522537830677,
+      "loss": 0.2245,
+      "step": 2952
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027886042974560246,
+      "loss": 0.1771,
+      "step": 2953
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027884562932611103,
+      "loss": 0.2591,
+      "step": 2954
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027883082412038233,
+      "loss": 0.2654,
+      "step": 2955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002788160141289663,
+      "loss": 0.2631,
+      "step": 2956
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002788011993524131,
+      "loss": 0.338,
+      "step": 2957
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000278786379791273,
+      "loss": 0.0909,
+      "step": 2958
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027877155544609665,
+      "loss": 0.1327,
+      "step": 2959
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002787567263174346,
+      "loss": 0.1947,
+      "step": 2960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002787418924058377,
+      "loss": 0.302,
+      "step": 2961
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027872705371185705,
+      "loss": 0.348,
+      "step": 2962
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002787122102360439,
+      "loss": 0.2807,
+      "step": 2963
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027869736197894953,
+      "loss": 0.2256,
+      "step": 2964
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002786825089411255,
+      "loss": 0.4669,
+      "step": 2965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002786676511231237,
+      "loss": 0.2389,
+      "step": 2966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000278652788525496,
+      "loss": 0.2763,
+      "step": 2967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027863792114879443,
+      "loss": 0.4711,
+      "step": 2968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027862304899357133,
+      "loss": 0.2104,
+      "step": 2969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027860817206037914,
+      "loss": 0.2793,
+      "step": 2970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027859329034977053,
+      "loss": 0.2689,
+      "step": 2971
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002785784038622983,
+      "loss": 0.4088,
+      "step": 2972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027856351259851535,
+      "loss": 0.0908,
+      "step": 2973
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002785486165589749,
+      "loss": 0.1338,
+      "step": 2974
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002785337157442304,
+      "loss": 0.271,
+      "step": 2975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002785188101548352,
+      "loss": 0.4213,
+      "step": 2976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002785038997913431,
+      "loss": 0.2711,
+      "step": 2977
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000278488984654308,
+      "loss": 0.0954,
+      "step": 2978
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00027847406474428384,
+      "loss": 0.0943,
+      "step": 2979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000278459140061825,
+      "loss": 0.2283,
+      "step": 2980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027844421060748575,
+      "loss": 0.1015,
+      "step": 2981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027842927638182076,
+      "loss": 0.1019,
+      "step": 2982
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027841433738538476,
+      "loss": 0.1867,
+      "step": 2983
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783993936187327,
+      "loss": 0.2826,
+      "step": 2984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783844450824197,
+      "loss": 0.2547,
+      "step": 2985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027836949177700106,
+      "loss": 0.2411,
+      "step": 2986
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783545337030322,
+      "loss": 0.3398,
+      "step": 2987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027833957086106887,
+      "loss": 0.2708,
+      "step": 2988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002783246032516668,
+      "loss": 0.2745,
+      "step": 2989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000278309630875382,
+      "loss": 0.1331,
+      "step": 2990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002782946537327707,
+      "loss": 0.1087,
+      "step": 2991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002782796718243893,
+      "loss": 0.102,
+      "step": 2992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002782646851507942,
+      "loss": 0.2893,
+      "step": 2993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027824969371254216,
+      "loss": 0.1501,
+      "step": 2994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027823469751019016,
+      "loss": 0.2838,
+      "step": 2995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027821969654429513,
+      "loss": 0.2471,
+      "step": 2996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002782046908154144,
+      "loss": 0.1883,
+      "step": 2997
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002781896803241054,
+      "loss": 0.332,
+      "step": 2998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027817466507092564,
+      "loss": 0.0579,
+      "step": 2999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027815964505643296,
+      "loss": 0.2257,
+      "step": 3000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002781446202811853,
+      "loss": 0.3317,
+      "step": 3001
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027812959074574074,
+      "loss": 0.1632,
+      "step": 3002
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002781145564506576,
+      "loss": 0.4839,
+      "step": 3003
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002780995173964945,
+      "loss": 0.3565,
+      "step": 3004
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027808447358380987,
+      "loss": 0.2017,
+      "step": 3005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027806942501316267,
+      "loss": 0.1961,
+      "step": 3006
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002780543716851119,
+      "loss": 0.1499,
+      "step": 3007
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027803931360021676,
+      "loss": 0.2741,
+      "step": 3008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027802425075903656,
+      "loss": 0.176,
+      "step": 3009
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002780091831621309,
+      "loss": 0.0317,
+      "step": 3010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002779941108100594,
+      "loss": 0.202,
+      "step": 3011
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00027797903370338213,
+      "loss": 0.1281,
+      "step": 3012
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000277963951842659,
+      "loss": 0.2853,
+      "step": 3013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002779488652284503,
+      "loss": 0.2241,
+      "step": 3014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027793377386131646,
+      "loss": 0.0949,
+      "step": 3015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002779186777418181,
+      "loss": 0.0699,
+      "step": 3016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000277903576870516,
+      "loss": 0.2135,
+      "step": 3017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027788847124797104,
+      "loss": 0.1131,
+      "step": 3018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027787336087474446,
+      "loss": 0.2153,
+      "step": 3019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002778582457513975,
+      "loss": 0.2979,
+      "step": 3020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002778431258784916,
+      "loss": 0.1881,
+      "step": 3021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002778280012565885,
+      "loss": 0.2618,
+      "step": 3022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027781287188625005,
+      "loss": 0.3402,
+      "step": 3023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027779773776803816,
+      "loss": 0.2915,
+      "step": 3024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002777825989025151,
+      "loss": 0.1825,
+      "step": 3025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027776745529024317,
+      "loss": 0.2974,
+      "step": 3026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027775230693178503,
+      "loss": 0.2263,
+      "step": 3027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027773715382770325,
+      "loss": 0.3638,
+      "step": 3028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027772199597856083,
+      "loss": 0.2331,
+      "step": 3029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027770683338492077,
+      "loss": 0.3896,
+      "step": 3030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002776916660473464,
+      "loss": 0.185,
+      "step": 3031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027767649396640107,
+      "loss": 0.2154,
+      "step": 3032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027766131714264836,
+      "loss": 0.1165,
+      "step": 3033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002776461355766521,
+      "loss": 0.3209,
+      "step": 3034
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002776309492689762,
+      "loss": 0.2406,
+      "step": 3035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027761575822018487,
+      "loss": 0.1598,
+      "step": 3036
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027760056243084227,
+      "loss": 0.4342,
+      "step": 3037
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027758536190151305,
+      "loss": 0.3291,
+      "step": 3038
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775701566327617,
+      "loss": 0.1664,
+      "step": 3039
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027755494662515316,
+      "loss": 0.1478,
+      "step": 3040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775397318792524,
+      "loss": 0.2573,
+      "step": 3041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775245123956246,
+      "loss": 0.1603,
+      "step": 3042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002775092881748352,
+      "loss": 0.2765,
+      "step": 3043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027749405921744956,
+      "loss": 0.1426,
+      "step": 3044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0002774788255240335,
+      "loss": 0.2049,
+      "step": 3045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00027746358709515297,
+      "loss": 0.2093,
+      "step": 3046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027744834393137396,
+      "loss": 0.0998,
+      "step": 3047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027743309603326267,
+      "loss": 0.4802,
+      "step": 3048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002774178434013856,
+      "loss": 0.3505,
+      "step": 3049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002774025860363092,
+      "loss": 0.2626,
+      "step": 3050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002773873239386004,
+      "loss": 0.3235,
+      "step": 3051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002773720571088261,
+      "loss": 0.1511,
+      "step": 3052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027735678554755336,
+      "loss": 0.1635,
+      "step": 3053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027734150925534955,
+      "loss": 0.2067,
+      "step": 3054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027732622823278206,
+      "loss": 0.2992,
+      "step": 3055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002773109424804186,
+      "loss": 0.3273,
+      "step": 3056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002772956519988269,
+      "loss": 0.3501,
+      "step": 3057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027728035678857506,
+      "loss": 0.1885,
+      "step": 3058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002772650568502312,
+      "loss": 0.3637,
+      "step": 3059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002772497521843637,
+      "loss": 0.0859,
+      "step": 3060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027723444279154107,
+      "loss": 0.1379,
+      "step": 3061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027721912867233195,
+      "loss": 0.4228,
+      "step": 3062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002772038098273053,
+      "loss": 0.1792,
+      "step": 3063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002771884862570302,
+      "loss": 0.1487,
+      "step": 3064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027717315796207574,
+      "loss": 0.1903,
+      "step": 3065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027715782494301135,
+      "loss": 0.2113,
+      "step": 3066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002771424872004067,
+      "loss": 0.3263,
+      "step": 3067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027712714473483155,
+      "loss": 0.1446,
+      "step": 3068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027711179754685573,
+      "loss": 0.1741,
+      "step": 3069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027709644563704935,
+      "loss": 0.3087,
+      "step": 3070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027708108900598275,
+      "loss": 0.2163,
+      "step": 3071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002770657276542263,
+      "loss": 0.2652,
+      "step": 3072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027705036158235074,
+      "loss": 0.0829,
+      "step": 3073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027703499079092685,
+      "loss": 0.4331,
+      "step": 3074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027701961528052557,
+      "loss": 0.2564,
+      "step": 3075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000277004235051718,
+      "loss": 0.2909,
+      "step": 3076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002769888501050756,
+      "loss": 0.3197,
+      "step": 3077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002769734604411698,
+      "loss": 0.2132,
+      "step": 3078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00027695806606057227,
+      "loss": 0.2108,
+      "step": 3079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002769426669638549,
+      "loss": 0.2105,
+      "step": 3080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027692726315158966,
+      "loss": 0.3406,
+      "step": 3081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002769118546243489,
+      "loss": 0.2024,
+      "step": 3082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768964413827048,
+      "loss": 0.1227,
+      "step": 3083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768810234272301,
+      "loss": 0.0645,
+      "step": 3084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768656007584974,
+      "loss": 0.1705,
+      "step": 3085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768501733770797,
+      "loss": 0.0636,
+      "step": 3086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027683474128355006,
+      "loss": 0.2961,
+      "step": 3087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002768193044784817,
+      "loss": 0.1309,
+      "step": 3088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027680386296244803,
+      "loss": 0.1639,
+      "step": 3089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002767884167360228,
+      "loss": 0.2948,
+      "step": 3090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002767729657997796,
+      "loss": 0.3198,
+      "step": 3091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027675751015429254,
+      "loss": 0.2826,
+      "step": 3092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027674204980013566,
+      "loss": 0.1946,
+      "step": 3093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002767265847378833,
+      "loss": 0.2728,
+      "step": 3094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002767111149681099,
+      "loss": 0.2506,
+      "step": 3095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002766956404913902,
+      "loss": 0.2833,
+      "step": 3096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000276680161308299,
+      "loss": 0.1963,
+      "step": 3097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002766646774194113,
+      "loss": 0.2564,
+      "step": 3098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027664918882530225,
+      "loss": 0.4011,
+      "step": 3099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027663369552654725,
+      "loss": 0.2793,
+      "step": 3100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002766181975237218,
+      "loss": 0.2063,
+      "step": 3101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027660269481740156,
+      "loss": 0.1428,
+      "step": 3102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765871874081625,
+      "loss": 0.1065,
+      "step": 3103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765716752965807,
+      "loss": 0.1902,
+      "step": 3104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027655615848323225,
+      "loss": 0.1396,
+      "step": 3105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002765406369686937,
+      "loss": 0.1874,
+      "step": 3106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027652511075354155,
+      "loss": 0.1404,
+      "step": 3107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027650957983835255,
+      "loss": 0.2913,
+      "step": 3108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002764940442237037,
+      "loss": 0.1374,
+      "step": 3109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000276478503910172,
+      "loss": 0.3058,
+      "step": 3110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002764629588983348,
+      "loss": 0.1291,
+      "step": 3111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00027644740918876943,
+      "loss": 0.0493,
+      "step": 3112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027643185478205373,
+      "loss": 0.1412,
+      "step": 3113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027641629567876536,
+      "loss": 0.2605,
+      "step": 3114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027640073187948225,
+      "loss": 0.2372,
+      "step": 3115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002763851633847827,
+      "loss": 0.2423,
+      "step": 3116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027636959019524494,
+      "loss": 0.2252,
+      "step": 3117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002763540123114474,
+      "loss": 0.3157,
+      "step": 3118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002763384297339689,
+      "loss": 0.4746,
+      "step": 3119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002763228424633882,
+      "loss": 0.2849,
+      "step": 3120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027630725050028436,
+      "loss": 0.112,
+      "step": 3121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027629165384523654,
+      "loss": 0.2601,
+      "step": 3122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027627605249882413,
+      "loss": 0.2981,
+      "step": 3123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027626044646162664,
+      "loss": 0.0742,
+      "step": 3124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027624483573422394,
+      "loss": 0.3298,
+      "step": 3125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027622922031719565,
+      "loss": 0.1262,
+      "step": 3126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002762136002111221,
+      "loss": 0.2643,
+      "step": 3127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027619797541658333,
+      "loss": 0.4341,
+      "step": 3128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027618234593415993,
+      "loss": 0.0843,
+      "step": 3129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761667117644324,
+      "loss": 0.4072,
+      "step": 3130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027615107290798144,
+      "loss": 0.2317,
+      "step": 3131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761354293653881,
+      "loss": 0.2567,
+      "step": 3132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002761197811372334,
+      "loss": 0.2248,
+      "step": 3133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027610412822409876,
+      "loss": 0.1167,
+      "step": 3134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002760884706265655,
+      "loss": 0.3215,
+      "step": 3135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027607280834521534,
+      "loss": 0.1403,
+      "step": 3136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027605714138063,
+      "loss": 0.4083,
+      "step": 3137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027604146973339154,
+      "loss": 0.2584,
+      "step": 3138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002760257934040821,
+      "loss": 0.1251,
+      "step": 3139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00027601011239328406,
+      "loss": 0.2409,
+      "step": 3140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002759944267015798,
+      "loss": 0.2455,
+      "step": 3141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002759787363295521,
+      "loss": 0.1347,
+      "step": 3142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002759630412777837,
+      "loss": 0.2037,
+      "step": 3143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002759473415468577,
+      "loss": 0.3219,
+      "step": 3144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002759316371373574,
+      "loss": 0.217,
+      "step": 3145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000275915928049866,
+      "loss": 0.2135,
+      "step": 3146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002759002142849672,
+      "loss": 0.3352,
+      "step": 3147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002758844958432445,
+      "loss": 0.2236,
+      "step": 3148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027586877272528206,
+      "loss": 0.0996,
+      "step": 3149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002758530449316638,
+      "loss": 0.2285,
+      "step": 3150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002758373124629739,
+      "loss": 0.2598,
+      "step": 3151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027582157531979694,
+      "loss": 0.2387,
+      "step": 3152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027580583350271737,
+      "loss": 0.2015,
+      "step": 3153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757900870123201,
+      "loss": 0.1831,
+      "step": 3154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757743358491899,
+      "loss": 0.2014,
+      "step": 3155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757585800139119,
+      "loss": 0.1008,
+      "step": 3156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757428195070715,
+      "loss": 0.132,
+      "step": 3157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757270543292541,
+      "loss": 0.3459,
+      "step": 3158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002757112844810453,
+      "loss": 0.1536,
+      "step": 3159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027569550996303096,
+      "loss": 0.1862,
+      "step": 3160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000275679730775797,
+      "loss": 0.298,
+      "step": 3161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027566394691992956,
+      "loss": 0.2687,
+      "step": 3162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002756481583960151,
+      "loss": 0.3277,
+      "step": 3163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027563236520463996,
+      "loss": 0.2488,
+      "step": 3164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027561656734639085,
+      "loss": 0.1432,
+      "step": 3165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002756007648218546,
+      "loss": 0.2672,
+      "step": 3166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755849576316183,
+      "loss": 0.1289,
+      "step": 3167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755691457762691,
+      "loss": 0.2648,
+      "step": 3168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755533292563944,
+      "loss": 0.202,
+      "step": 3169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002755375080725816,
+      "loss": 0.1639,
+      "step": 3170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027552168222541855,
+      "loss": 0.2008,
+      "step": 3171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027550585171549316,
+      "loss": 0.3806,
+      "step": 3172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027549001654339337,
+      "loss": 0.3185,
+      "step": 3173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0002754741767097074,
+      "loss": 0.3633,
+      "step": 3174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027545833221502385,
+      "loss": 0.3652,
+      "step": 3175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027544248305993107,
+      "loss": 0.3253,
+      "step": 3176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027542662924501796,
+      "loss": 0.1065,
+      "step": 3177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00027541077077087336,
+      "loss": 0.3297,
+      "step": 3178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002753949076380863,
+      "loss": 0.284,
+      "step": 3179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002753790398472463,
+      "loss": 0.2489,
+      "step": 3180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027536316739894257,
+      "loss": 0.3624,
+      "step": 3181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002753472902937648,
+      "loss": 0.1805,
+      "step": 3182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002753314085323027,
+      "loss": 0.1206,
+      "step": 3183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027531552211514635,
+      "loss": 0.2407,
+      "step": 3184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027529963104288587,
+      "loss": 0.1456,
+      "step": 3185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002752837353161115,
+      "loss": 0.2735,
+      "step": 3186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027526783493541377,
+      "loss": 0.277,
+      "step": 3187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027525192990138334,
+      "loss": 0.1549,
+      "step": 3188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000275236020214611,
+      "loss": 0.0884,
+      "step": 3189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027522010587568773,
+      "loss": 0.2218,
+      "step": 3190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027520418688520477,
+      "loss": 0.2646,
+      "step": 3191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027518826324375345,
+      "loss": 0.2875,
+      "step": 3192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002751723349519252,
+      "loss": 0.1993,
+      "step": 3193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027515640201031174,
+      "loss": 0.437,
+      "step": 3194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000275140464419505,
+      "loss": 0.2915,
+      "step": 3195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027512452218009703,
+      "loss": 0.2211,
+      "step": 3196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027510857529267996,
+      "loss": 0.286,
+      "step": 3197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027509262375784616,
+      "loss": 0.1374,
+      "step": 3198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027507666757618825,
+      "loss": 0.3067,
+      "step": 3199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002750607067482989,
+      "loss": 0.282,
+      "step": 3200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027504474127477106,
+      "loss": 0.185,
+      "step": 3201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027502877115619773,
+      "loss": 0.2296,
+      "step": 3202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027501279639317226,
+      "loss": 0.2104,
+      "step": 3203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000274996816986288,
+      "loss": 0.2042,
+      "step": 3204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027498083293613844,
+      "loss": 0.1803,
+      "step": 3205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027496484424331755,
+      "loss": 0.287,
+      "step": 3206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002749488509084191,
+      "loss": 0.258,
+      "step": 3207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027493285293203726,
+      "loss": 0.0719,
+      "step": 3208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0002749168503147663,
+      "loss": 0.1313,
+      "step": 3209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027490084305720067,
+      "loss": 0.233,
+      "step": 3210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00027488483115993496,
+      "loss": 0.2434,
+      "step": 3211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000274868814623564,
+      "loss": 0.1712,
+      "step": 3212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002748527934486828,
+      "loss": 0.2347,
+      "step": 3213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027483676763588637,
+      "loss": 0.1736,
+      "step": 3214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027482073718577017,
+      "loss": 0.1758,
+      "step": 3215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002748047020989296,
+      "loss": 0.139,
+      "step": 3216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002747886623759603,
+      "loss": 0.3934,
+      "step": 3217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002747726180174582,
+      "loss": 0.2685,
+      "step": 3218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027475656902401917,
+      "loss": 0.2739,
+      "step": 3219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027474051539623946,
+      "loss": 0.4702,
+      "step": 3220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027472445713471545,
+      "loss": 0.1935,
+      "step": 3221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002747083942400436,
+      "loss": 0.288,
+      "step": 3222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002746923267128205,
+      "loss": 0.078,
+      "step": 3223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027467625455364326,
+      "loss": 0.3898,
+      "step": 3224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002746601777631087,
+      "loss": 0.0758,
+      "step": 3225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002746440963418141,
+      "loss": 0.3802,
+      "step": 3226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027462801029035685,
+      "loss": 0.1821,
+      "step": 3227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027461191960933445,
+      "loss": 0.1343,
+      "step": 3228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027459582429934475,
+      "loss": 0.1599,
+      "step": 3229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002745797243609855,
+      "loss": 0.3983,
+      "step": 3230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002745636197948548,
+      "loss": 0.1683,
+      "step": 3231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002745475106015509,
+      "loss": 0.2656,
+      "step": 3232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027453139678167225,
+      "loss": 0.2235,
+      "step": 3233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027451527833581736,
+      "loss": 0.3542,
+      "step": 3234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027449915526458503,
+      "loss": 0.1997,
+      "step": 3235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027448302756857417,
+      "loss": 0.2391,
+      "step": 3236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002744668952483839,
+      "loss": 0.2282,
+      "step": 3237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002744507583046134,
+      "loss": 0.2222,
+      "step": 3238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027443461673786227,
+      "loss": 0.2152,
+      "step": 3239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00027441847054873,
+      "loss": 0.3113,
+      "step": 3240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002744023197378164,
+      "loss": 0.2279,
+      "step": 3241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002743861643057214,
+      "loss": 0.1478,
+      "step": 3242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0002743700042530452,
+      "loss": 0.1816,
+      "step": 3243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000274353839580388,
+      "loss": 0.2872,
+      "step": 3244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027433767028835036,
+      "loss": 0.2878,
+      "step": 3245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002743214963775329,
+      "loss": 0.2918,
+      "step": 3246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002743053178485364,
+      "loss": 0.1381,
+      "step": 3247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027428913470196187,
+      "loss": 0.0711,
+      "step": 3248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027427294693841043,
+      "loss": 0.2049,
+      "step": 3249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027425675455848347,
+      "loss": 0.2856,
+      "step": 3250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027424055756278243,
+      "loss": 0.1846,
+      "step": 3251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027422435595190906,
+      "loss": 0.4586,
+      "step": 3252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002742081497264651,
+      "loss": 0.3633,
+      "step": 3253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027419193888705265,
+      "loss": 0.2142,
+      "step": 3254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002741757234342739,
+      "loss": 0.2346,
+      "step": 3255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027415950336873103,
+      "loss": 0.2727,
+      "step": 3256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002741432786910268,
+      "loss": 0.2768,
+      "step": 3257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002741270494017638,
+      "loss": 0.0553,
+      "step": 3258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002741108155015449,
+      "loss": 0.2747,
+      "step": 3259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027409457699097324,
+      "loss": 0.1087,
+      "step": 3260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027407833387065186,
+      "loss": 0.2803,
+      "step": 3261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027406208614118424,
+      "loss": 0.2811,
+      "step": 3262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027404583380317396,
+      "loss": 0.297,
+      "step": 3263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002740295768572247,
+      "loss": 0.242,
+      "step": 3264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002740133153039403,
+      "loss": 0.2259,
+      "step": 3265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027399704914392504,
+      "loss": 0.3109,
+      "step": 3266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002739807783777829,
+      "loss": 0.2662,
+      "step": 3267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027396450300611855,
+      "loss": 0.2297,
+      "step": 3268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027394822302953634,
+      "loss": 0.2364,
+      "step": 3269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002739319384486411,
+      "loss": 0.2125,
+      "step": 3270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027391564926403783,
+      "loss": 0.294,
+      "step": 3271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027389935547633156,
+      "loss": 0.2445,
+      "step": 3272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002738830570861276,
+      "loss": 0.1727,
+      "step": 3273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002738667540940313,
+      "loss": 0.1416,
+      "step": 3274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002738504465006483,
+      "loss": 0.302,
+      "step": 3275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0002738341343065845,
+      "loss": 0.3265,
+      "step": 3276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00027381781751244564,
+      "loss": 0.0901,
+      "step": 3277
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000273801496118838,
+      "loss": 0.2115,
+      "step": 3278
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002737851701263678,
+      "loss": 0.2371,
+      "step": 3279
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027376883953564155,
+      "loss": 0.2913,
+      "step": 3280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027375250434726587,
+      "loss": 0.1525,
+      "step": 3281
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002737361645618475,
+      "loss": 0.3679,
+      "step": 3282
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027371982017999354,
+      "loss": 0.1511,
+      "step": 3283
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002737034712023111,
+      "loss": 0.2491,
+      "step": 3284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027368711762940735,
+      "loss": 0.2417,
+      "step": 3285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027367075946188997,
+      "loss": 0.0439,
+      "step": 3286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002736543967003665,
+      "loss": 0.3527,
+      "step": 3287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002736380293454448,
+      "loss": 0.1656,
+      "step": 3288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027362165739773286,
+      "loss": 0.2286,
+      "step": 3289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002736052808578389,
+      "loss": 0.1586,
+      "step": 3290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002735888997263712,
+      "loss": 0.29,
+      "step": 3291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027357251400393835,
+      "loss": 0.0726,
+      "step": 3292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002735561236911489,
+      "loss": 0.235,
+      "step": 3293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027353972878861187,
+      "loss": 0.12,
+      "step": 3294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027352332929693614,
+      "loss": 0.1328,
+      "step": 3295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027350692521673094,
+      "loss": 0.168,
+      "step": 3296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027349051654860574,
+      "loss": 0.2157,
+      "step": 3297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027347410329316987,
+      "loss": 0.0155,
+      "step": 3298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027345768545103323,
+      "loss": 0.1662,
+      "step": 3299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027344126302280555,
+      "loss": 0.1112,
+      "step": 3300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027342483600909694,
+      "loss": 0.1342,
+      "step": 3301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027340840441051766,
+      "loss": 0.1215,
+      "step": 3302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000273391968227678,
+      "loss": 0.0831,
+      "step": 3303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002733755274611886,
+      "loss": 0.3257,
+      "step": 3304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002733590821116601,
+      "loss": 0.1576,
+      "step": 3305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027334263217970354,
+      "loss": 0.2056,
+      "step": 3306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002733261776659298,
+      "loss": 0.12,
+      "step": 3307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027330971857095026,
+      "loss": 0.2468,
+      "step": 3308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00027329325489537626,
+      "loss": 0.1675,
+      "step": 3309
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027327678663981945,
+      "loss": 0.0973,
+      "step": 3310
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027326031380489146,
+      "loss": 0.1616,
+      "step": 3311
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002732438363912043,
+      "loss": 0.2437,
+      "step": 3312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027322735439937,
+      "loss": 0.0661,
+      "step": 3313
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027321086783000086,
+      "loss": 0.118,
+      "step": 3314
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002731943766837093,
+      "loss": 0.0296,
+      "step": 3315
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002731778809611079,
+      "loss": 0.1171,
+      "step": 3316
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002731613806628094,
+      "loss": 0.3711,
+      "step": 3317
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027314487578942683,
+      "loss": 0.2328,
+      "step": 3318
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027312836634157327,
+      "loss": 0.1042,
+      "step": 3319
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002731118523198619,
+      "loss": 0.1437,
+      "step": 3320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027309533372490627,
+      "loss": 0.2858,
+      "step": 3321
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027307881055731994,
+      "loss": 0.3004,
+      "step": 3322
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002730622828177168,
+      "loss": 0.2877,
+      "step": 3323
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002730457505067107,
+      "loss": 0.2514,
+      "step": 3324
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002730292136249157,
+      "loss": 0.158,
+      "step": 3325
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002730126721729463,
+      "loss": 0.154,
+      "step": 3326
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027299612615141685,
+      "loss": 0.0426,
+      "step": 3327
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027297957556094197,
+      "loss": 0.1141,
+      "step": 3328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002729630204021365,
+      "loss": 0.1253,
+      "step": 3329
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002729464606756154,
+      "loss": 0.1381,
+      "step": 3330
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027292989638199385,
+      "loss": 0.2154,
+      "step": 3331
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002729133275218871,
+      "loss": 0.3223,
+      "step": 3332
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002728967540959107,
+      "loss": 0.1823,
+      "step": 3333
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002728801761046803,
+      "loss": 0.1217,
+      "step": 3334
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027286359354881164,
+      "loss": 0.2968,
+      "step": 3335
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002728470064289208,
+      "loss": 0.1094,
+      "step": 3336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027283041474562397,
+      "loss": 0.255,
+      "step": 3337
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027281381849953736,
+      "loss": 0.2227,
+      "step": 3338
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027279721769127757,
+      "loss": 0.1967,
+      "step": 3339
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00027278061232146125,
+      "loss": 0.0713,
+      "step": 3340
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002727640023907052,
+      "loss": 0.0947,
+      "step": 3341
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002727473878996265,
+      "loss": 0.1423,
+      "step": 3342
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027273076884884223,
+      "loss": 0.1496,
+      "step": 3343
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027271414523896985,
+      "loss": 0.1354,
+      "step": 3344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027269751707062684,
+      "loss": 0.0356,
+      "step": 3345
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002726808843444309,
+      "loss": 0.1584,
+      "step": 3346
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002726642470609998,
+      "loss": 0.0764,
+      "step": 3347
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027264760522095164,
+      "loss": 0.1286,
+      "step": 3348
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027263095882490467,
+      "loss": 0.2448,
+      "step": 3349
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002726143078734771,
+      "loss": 0.1866,
+      "step": 3350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027259765236728765,
+      "loss": 0.2669,
+      "step": 3351
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002725809923069549,
+      "loss": 0.1869,
+      "step": 3352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027256432769309774,
+      "loss": 0.2645,
+      "step": 3353
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027254765852633523,
+      "loss": 0.0488,
+      "step": 3354
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002725309848072866,
+      "loss": 0.0975,
+      "step": 3355
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027251430653657115,
+      "loss": 0.1217,
+      "step": 3356
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002724976237148085,
+      "loss": 0.1601,
+      "step": 3357
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027248093634261837,
+      "loss": 0.158,
+      "step": 3358
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027246424442062056,
+      "loss": 0.1234,
+      "step": 3359
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027244754794943527,
+      "loss": 0.085,
+      "step": 3360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027243084692968263,
+      "loss": 0.1505,
+      "step": 3361
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000272414141361983,
+      "loss": 0.1158,
+      "step": 3362
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027239743124695705,
+      "loss": 0.1963,
+      "step": 3363
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723807165852255,
+      "loss": 0.172,
+      "step": 3364
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723639973774091,
+      "loss": 0.2331,
+      "step": 3365
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723472736241291,
+      "loss": 0.2871,
+      "step": 3366
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723305453260066,
+      "loss": 0.2539,
+      "step": 3367
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002723138124836632,
+      "loss": 0.1854,
+      "step": 3368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027229707509772026,
+      "loss": 0.1978,
+      "step": 3369
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002722803331687996,
+      "loss": 0.1874,
+      "step": 3370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027226358669752316,
+      "loss": 0.2274,
+      "step": 3371
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027224683568451304,
+      "loss": 0.2186,
+      "step": 3372
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002722300801303914,
+      "loss": 0.1261,
+      "step": 3373
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00027221332003578074,
+      "loss": 0.186,
+      "step": 3374
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0002721965554013037,
+      "loss": 0.0777,
+      "step": 3375
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002721797862275829,
+      "loss": 0.2326,
+      "step": 3376
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002721630125152413,
+      "loss": 0.2971,
+      "step": 3377
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002721462342649021,
+      "loss": 0.2964,
+      "step": 3378
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027212945147718845,
+      "loss": 0.1038,
+      "step": 3379
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027211266415272384,
+      "loss": 0.2724,
+      "step": 3380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720958722921318,
+      "loss": 0.0877,
+      "step": 3381
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720790758960362,
+      "loss": 0.1163,
+      "step": 3382
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027206227496506086,
+      "loss": 0.2903,
+      "step": 3383
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027204546949983,
+      "loss": 0.1815,
+      "step": 3384
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720286595009679,
+      "loss": 0.0009,
+      "step": 3385
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002720118449690988,
+      "loss": 0.2752,
+      "step": 3386
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002719950259048476,
+      "loss": 0.2314,
+      "step": 3387
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002719782023088389,
+      "loss": 0.0968,
+      "step": 3388
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027196137418169765,
+      "loss": 0.1484,
+      "step": 3389
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027194454152404895,
+      "loss": 0.1342,
+      "step": 3390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027192770433651824,
+      "loss": 0.1346,
+      "step": 3391
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002719108626197308,
+      "loss": 0.1536,
+      "step": 3392
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002718940163743123,
+      "loss": 0.2062,
+      "step": 3393
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027187716560088854,
+      "loss": 0.3212,
+      "step": 3394
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027186031030008556,
+      "loss": 0.1734,
+      "step": 3395
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002718434504725293,
+      "loss": 0.2323,
+      "step": 3396
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027182658611884624,
+      "loss": 0.1788,
+      "step": 3397
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027180971723966276,
+      "loss": 0.2083,
+      "step": 3398
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002717928438356054,
+      "loss": 0.0924,
+      "step": 3399
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002717759659073011,
+      "loss": 0.15,
+      "step": 3400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002717590834553768,
+      "loss": 0.1883,
+      "step": 3401
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027174219648045956,
+      "loss": 0.2759,
+      "step": 3402
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002717253049831768,
+      "loss": 0.1687,
+      "step": 3403
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027170840896415584,
+      "loss": 0.2356,
+      "step": 3404
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002716915084240244,
+      "loss": 0.1936,
+      "step": 3405
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002716746033634103,
+      "loss": 0.0651,
+      "step": 3406
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0002716576937829415,
+      "loss": 0.2496,
+      "step": 3407
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00027164077968324616,
+      "loss": 0.1205,
+      "step": 3408
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027162386106495244,
+      "loss": 0.2588,
+      "step": 3409
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027160693792868906,
+      "loss": 0.244,
+      "step": 3410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027159001027508454,
+      "loss": 0.2204,
+      "step": 3411
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027157307810476766,
+      "loss": 0.1623,
+      "step": 3412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002715561414183674,
+      "loss": 0.2434,
+      "step": 3413
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000271539200216513,
+      "loss": 0.2131,
+      "step": 3414
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027152225449983376,
+      "loss": 0.1353,
+      "step": 3415
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002715053042689591,
+      "loss": 0.1493,
+      "step": 3416
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027148834952451866,
+      "loss": 0.18,
+      "step": 3417
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027147139026714235,
+      "loss": 0.1453,
+      "step": 3418
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027145442649746004,
+      "loss": 0.2393,
+      "step": 3419
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027143745821610197,
+      "loss": 0.1629,
+      "step": 3420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027142048542369844,
+      "loss": 0.2177,
+      "step": 3421
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027140350812088,
+      "loss": 0.1553,
+      "step": 3422
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002713865263082772,
+      "loss": 0.2831,
+      "step": 3423
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027136953998652087,
+      "loss": 0.1238,
+      "step": 3424
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002713525491562421,
+      "loss": 0.0598,
+      "step": 3425
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027133555381807196,
+      "loss": 0.1356,
+      "step": 3426
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027131855397264184,
+      "loss": 0.1114,
+      "step": 3427
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002713015496205832,
+      "loss": 0.1725,
+      "step": 3428
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027128454076252764,
+      "loss": 0.1527,
+      "step": 3429
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002712675273991071,
+      "loss": 0.1478,
+      "step": 3430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027125050953095357,
+      "loss": 0.1048,
+      "step": 3431
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027123348715869913,
+      "loss": 0.1467,
+      "step": 3432
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002712164602829762,
+      "loss": 0.1195,
+      "step": 3433
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027119942890441715,
+      "loss": 0.2043,
+      "step": 3434
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027118239302365477,
+      "loss": 0.1906,
+      "step": 3435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002711653526413219,
+      "loss": 0.0735,
+      "step": 3436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027114830775805144,
+      "loss": 0.0968,
+      "step": 3437
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00027113125837447663,
+      "loss": 0.2562,
+      "step": 3438
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002711142044912308,
+      "loss": 0.1465,
+      "step": 3439
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002710971461089474,
+      "loss": 0.1476,
+      "step": 3440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002710800832282602,
+      "loss": 0.0644,
+      "step": 3441
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002710630158498029,
+      "loss": 0.1899,
+      "step": 3442
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027104594397420965,
+      "loss": 0.2403,
+      "step": 3443
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002710288676021145,
+      "loss": 0.1924,
+      "step": 3444
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002710117867341518,
+      "loss": 0.0795,
+      "step": 3445
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027099470137095615,
+      "loss": 0.1538,
+      "step": 3446
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027097761151316215,
+      "loss": 0.0977,
+      "step": 3447
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002709605171614046,
+      "loss": 0.1705,
+      "step": 3448
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002709434183163186,
+      "loss": 0.4089,
+      "step": 3449
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027092631497853924,
+      "loss": 0.1601,
+      "step": 3450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002709092071487019,
+      "loss": 0.2171,
+      "step": 3451
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002708920948274421,
+      "loss": 0.2063,
+      "step": 3452
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002708749780153955,
+      "loss": 0.1823,
+      "step": 3453
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002708578567131979,
+      "loss": 0.2004,
+      "step": 3454
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027084073092148534,
+      "loss": 0.1014,
+      "step": 3455
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027082360064089396,
+      "loss": 0.1065,
+      "step": 3456
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002708064658720602,
+      "loss": 0.1989,
+      "step": 3457
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002707893266156204,
+      "loss": 0.196,
+      "step": 3458
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027077218287221137,
+      "loss": 0.1608,
+      "step": 3459
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002707550346424699,
+      "loss": 0.0986,
+      "step": 3460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027073788192703304,
+      "loss": 0.2536,
+      "step": 3461
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002707207247265379,
+      "loss": 0.215,
+      "step": 3462
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002707035630416219,
+      "loss": 0.2487,
+      "step": 3463
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027068639687292245,
+      "loss": 0.1762,
+      "step": 3464
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027066922622107726,
+      "loss": 0.1521,
+      "step": 3465
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002706520510867241,
+      "loss": 0.2344,
+      "step": 3466
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002706348714705012,
+      "loss": 0.1818,
+      "step": 3467
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027061768737304646,
+      "loss": 0.1322,
+      "step": 3468
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027060049879499843,
+      "loss": 0.1189,
+      "step": 3469
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002705833057369955,
+      "loss": 0.2243,
+      "step": 3470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027056610819967635,
+      "loss": 0.1345,
+      "step": 3471
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027054890618367986,
+      "loss": 0.1527,
+      "step": 3472
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000270531699689645,
+      "loss": 0.2254,
+      "step": 3473
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00027051448871821094,
+      "loss": 0.1804,
+      "step": 3474
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002704972732700171,
+      "loss": 0.3136,
+      "step": 3475
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002704800533457028,
+      "loss": 0.1994,
+      "step": 3476
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002704628289459079,
+      "loss": 0.2257,
+      "step": 3477
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027044560007127215,
+      "loss": 0.1882,
+      "step": 3478
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002704283667224356,
+      "loss": 0.2816,
+      "step": 3479
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002704111289000383,
+      "loss": 0.1083,
+      "step": 3480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002703938866047207,
+      "loss": 0.2249,
+      "step": 3481
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027037663983712324,
+      "loss": 0.1456,
+      "step": 3482
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027035938859788663,
+      "loss": 0.1586,
+      "step": 3483
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002703421328876517,
+      "loss": 0.2103,
+      "step": 3484
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002703248727070594,
+      "loss": 0.1177,
+      "step": 3485
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027030760805675096,
+      "loss": 0.113,
+      "step": 3486
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027029033893736766,
+      "loss": 0.1922,
+      "step": 3487
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027027306534955104,
+      "loss": 0.1268,
+      "step": 3488
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002702557872939428,
+      "loss": 0.2549,
+      "step": 3489
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002702385047711846,
+      "loss": 0.07,
+      "step": 3490
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027022121778191854,
+      "loss": 0.1017,
+      "step": 3491
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027020392632678687,
+      "loss": 0.1589,
+      "step": 3492
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027018663040643176,
+      "loss": 0.2737,
+      "step": 3493
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701693300214958,
+      "loss": 0.1311,
+      "step": 3494
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027015202517262165,
+      "loss": 0.1355,
+      "step": 3495
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701347158604521,
+      "loss": 0.294,
+      "step": 3496
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701174020856301,
+      "loss": 0.3069,
+      "step": 3497
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002701000838487989,
+      "loss": 0.2206,
+      "step": 3498
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027008276115060174,
+      "loss": 0.1438,
+      "step": 3499
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00027006543399168216,
+      "loss": 0.327,
+      "step": 3500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002700481023726838,
+      "loss": 0.1438,
+      "step": 3501
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002700307662942505,
+      "loss": 0.2474,
+      "step": 3502
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002700134257570262,
+      "loss": 0.1585,
+      "step": 3503
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00026999608076165506,
+      "loss": 0.2062,
+      "step": 3504
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002699787313087814,
+      "loss": 0.2665,
+      "step": 3505
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0002699613773990497,
+      "loss": 0.1386,
+      "step": 3506
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00026994401903310467,
+      "loss": 0.1833,
+      "step": 3507
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000269926656211591,
+      "loss": 0.2064,
+      "step": 3508
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002699092889351538,
+      "loss": 0.1216,
+      "step": 3509
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002698919172044381,
+      "loss": 0.1928,
+      "step": 3510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026987454102008927,
+      "loss": 0.1997,
+      "step": 3511
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026985716038275273,
+      "loss": 0.0409,
+      "step": 3512
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026983977529307417,
+      "loss": 0.1358,
+      "step": 3513
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026982238575169943,
+      "loss": 0.2152,
+      "step": 3514
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002698049917592744,
+      "loss": 0.1408,
+      "step": 3515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026978759331644525,
+      "loss": 0.1888,
+      "step": 3516
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026977019042385834,
+      "loss": 0.2008,
+      "step": 3517
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026975278308216004,
+      "loss": 0.1871,
+      "step": 3518
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000269735371291997,
+      "loss": 0.0356,
+      "step": 3519
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000269717955054016,
+      "loss": 0.1143,
+      "step": 3520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026970053436886415,
+      "loss": 0.1413,
+      "step": 3521
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002696831092371884,
+      "loss": 0.2605,
+      "step": 3522
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002696656796596361,
+      "loss": 0.1637,
+      "step": 3523
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026964824563685476,
+      "loss": 0.1947,
+      "step": 3524
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002696308071694919,
+      "loss": 0.0882,
+      "step": 3525
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002696133642581955,
+      "loss": 0.2384,
+      "step": 3526
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002695959169036133,
+      "loss": 0.2175,
+      "step": 3527
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002695784651063934,
+      "loss": 0.2742,
+      "step": 3528
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002695610088671843,
+      "loss": 0.2204,
+      "step": 3529
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002695435481866343,
+      "loss": 0.1551,
+      "step": 3530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026952608306539204,
+      "loss": 0.1122,
+      "step": 3531
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002695086135041063,
+      "loss": 0.124,
+      "step": 3532
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000269491139503426,
+      "loss": 0.1622,
+      "step": 3533
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002694736610640003,
+      "loss": 0.0543,
+      "step": 3534
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002694561781864785,
+      "loss": 0.1253,
+      "step": 3535
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002694386908715099,
+      "loss": 0.1036,
+      "step": 3536
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026942119911974423,
+      "loss": 0.267,
+      "step": 3537
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026940370293183116,
+      "loss": 0.2265,
+      "step": 3538
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026938620230842076,
+      "loss": 0.2456,
+      "step": 3539
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00026936869725016296,
+      "loss": 0.3714,
+      "step": 3540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002693511877577081,
+      "loss": 0.2474,
+      "step": 3541
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026933367383170665,
+      "loss": 0.1709,
+      "step": 3542
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002693161554728092,
+      "loss": 0.1516,
+      "step": 3543
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002692986326816664,
+      "loss": 0.1371,
+      "step": 3544
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026928110545892925,
+      "loss": 0.0294,
+      "step": 3545
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002692635738052488,
+      "loss": 0.1036,
+      "step": 3546
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026924603772127636,
+      "loss": 0.1972,
+      "step": 3547
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002692284972076633,
+      "loss": 0.107,
+      "step": 3548
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002692109522650612,
+      "loss": 0.1463,
+      "step": 3549
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002691934028941218,
+      "loss": 0.1722,
+      "step": 3550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026917584909549706,
+      "loss": 0.2558,
+      "step": 3551
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026915829086983893,
+      "loss": 0.0222,
+      "step": 3552
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026914072821779977,
+      "loss": 0.249,
+      "step": 3553
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002691231611400319,
+      "loss": 0.248,
+      "step": 3554
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026910558963718793,
+      "loss": 0.2919,
+      "step": 3555
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026908801370992063,
+      "loss": 0.1375,
+      "step": 3556
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002690704333588828,
+      "loss": 0.1546,
+      "step": 3557
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026905284858472754,
+      "loss": 0.2869,
+      "step": 3558
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002690352593881081,
+      "loss": 0.2706,
+      "step": 3559
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002690176657696778,
+      "loss": 0.3337,
+      "step": 3560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002690000677300903,
+      "loss": 0.3202,
+      "step": 3561
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026898246526999914,
+      "loss": 0.3265,
+      "step": 3562
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026896485839005837,
+      "loss": 0.0931,
+      "step": 3563
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000268947247090922,
+      "loss": 0.263,
+      "step": 3564
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026892963137324416,
+      "loss": 0.1148,
+      "step": 3565
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026891201123767936,
+      "loss": 0.1303,
+      "step": 3566
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026889438668488197,
+      "loss": 0.1031,
+      "step": 3567
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688767577155068,
+      "loss": 0.1718,
+      "step": 3568
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00026885912433020865,
+      "loss": 0.1485,
+      "step": 3569
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688414865296426,
+      "loss": 0.1781,
+      "step": 3570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688238443144638,
+      "loss": 0.1434,
+      "step": 3571
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002688061976853277,
+      "loss": 0.1919,
+      "step": 3572
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002687885466428897,
+      "loss": 0.2997,
+      "step": 3573
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026877089118780556,
+      "loss": 0.1916,
+      "step": 3574
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026875323132073106,
+      "loss": 0.1202,
+      "step": 3575
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002687355670423223,
+      "loss": 0.1297,
+      "step": 3576
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002687178983532354,
+      "loss": 0.1427,
+      "step": 3577
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026870022525412674,
+      "loss": 0.262,
+      "step": 3578
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002686825477456527,
+      "loss": 0.2571,
+      "step": 3579
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002686648658284702,
+      "loss": 0.2699,
+      "step": 3580
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026864717950323585,
+      "loss": 0.0886,
+      "step": 3581
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026862948877060666,
+      "loss": 0.1484,
+      "step": 3582
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026861179363123986,
+      "loss": 0.2065,
+      "step": 3583
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002685940940857928,
+      "loss": 0.1554,
+      "step": 3584
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002685763901349229,
+      "loss": 0.1517,
+      "step": 3585
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026855868177928777,
+      "loss": 0.2371,
+      "step": 3586
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002685409690195453,
+      "loss": 0.1919,
+      "step": 3587
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026852325185635354,
+      "loss": 0.1272,
+      "step": 3588
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026850553029037047,
+      "loss": 0.0997,
+      "step": 3589
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002684878043222545,
+      "loss": 0.136,
+      "step": 3590
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000268470073952664,
+      "loss": 0.0671,
+      "step": 3591
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002684523391822577,
+      "loss": 0.1119,
+      "step": 3592
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026843460001169437,
+      "loss": 0.1314,
+      "step": 3593
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026841685644163294,
+      "loss": 0.2294,
+      "step": 3594
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002683991084727326,
+      "loss": 0.1502,
+      "step": 3595
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002683813561056525,
+      "loss": 0.1902,
+      "step": 3596
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002683635993410522,
+      "loss": 0.1619,
+      "step": 3597
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002683458381795913,
+      "loss": 0.2618,
+      "step": 3598
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026832807262192954,
+      "loss": 0.1426,
+      "step": 3599
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002683103026687269,
+      "loss": 0.1227,
+      "step": 3600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002682925283206434,
+      "loss": 0.3051,
+      "step": 3601
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026827474957833943,
+      "loss": 0.1897,
+      "step": 3602
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026825696644247534,
+      "loss": 0.2249,
+      "step": 3603
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026823917891371173,
+      "loss": 0.0778,
+      "step": 3604
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002682213869927093,
+      "loss": 0.151,
+      "step": 3605
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00026820359068012906,
+      "loss": 0.173,
+      "step": 3606
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002681857899766321,
+      "loss": 0.1524,
+      "step": 3607
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002681679848828796,
+      "loss": 0.2961,
+      "step": 3608
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026815017539953294,
+      "loss": 0.24,
+      "step": 3609
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026813236152725375,
+      "loss": 0.2898,
+      "step": 3610
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002681145432667038,
+      "loss": 0.1952,
+      "step": 3611
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026809672061854493,
+      "loss": 0.1728,
+      "step": 3612
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680788935834391,
+      "loss": 0.1653,
+      "step": 3613
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680610621620487,
+      "loss": 0.2511,
+      "step": 3614
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680432263550361,
+      "loss": 0.2217,
+      "step": 3615
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680253861630638,
+      "loss": 0.3184,
+      "step": 3616
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002680075415867944,
+      "loss": 0.3028,
+      "step": 3617
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000267989692626891,
+      "loss": 0.2388,
+      "step": 3618
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026797183928401646,
+      "loss": 0.0988,
+      "step": 3619
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002679539815588341,
+      "loss": 0.2091,
+      "step": 3620
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002679361194520072,
+      "loss": 0.0948,
+      "step": 3621
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026791825296419925,
+      "loss": 0.1498,
+      "step": 3622
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000267900382096074,
+      "loss": 0.408,
+      "step": 3623
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026788250684829537,
+      "loss": 0.1844,
+      "step": 3624
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026786462722152725,
+      "loss": 0.2828,
+      "step": 3625
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002678467432164339,
+      "loss": 0.2168,
+      "step": 3626
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026782885483367955,
+      "loss": 0.243,
+      "step": 3627
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002678109620739288,
+      "loss": 0.3528,
+      "step": 3628
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026779306493784634,
+      "loss": 0.2141,
+      "step": 3629
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002677751634260969,
+      "loss": 0.1311,
+      "step": 3630
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002677572575393455,
+      "loss": 0.1363,
+      "step": 3631
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026773934727825733,
+      "loss": 0.172,
+      "step": 3632
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002677214326434977,
+      "loss": 0.0915,
+      "step": 3633
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026770351363573203,
+      "loss": 0.1443,
+      "step": 3634
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000267685590255626,
+      "loss": 0.0763,
+      "step": 3635
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002676676625038454,
+      "loss": 0.1919,
+      "step": 3636
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026764973038105616,
+      "loss": 0.1521,
+      "step": 3637
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0002676317938879245,
+      "loss": 0.2586,
+      "step": 3638
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00026761385302511657,
+      "loss": 0.0773,
+      "step": 3639
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000267595907793299,
+      "loss": 0.1248,
+      "step": 3640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026757795819313824,
+      "loss": 0.1474,
+      "step": 3641
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675600042253011,
+      "loss": 0.2726,
+      "step": 3642
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675420458904546,
+      "loss": 0.0246,
+      "step": 3643
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675240831892658,
+      "loss": 0.1317,
+      "step": 3644
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002675061161224019,
+      "loss": 0.1106,
+      "step": 3645
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026748814469053036,
+      "loss": 0.2021,
+      "step": 3646
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026747016889431884,
+      "loss": 0.1946,
+      "step": 3647
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.000267452188734435,
+      "loss": 0.0935,
+      "step": 3648
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026743420421154675,
+      "loss": 0.1449,
+      "step": 3649
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002674162153263222,
+      "loss": 0.0623,
+      "step": 3650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002673982220794295,
+      "loss": 0.0668,
+      "step": 3651
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002673802244715372,
+      "loss": 0.2031,
+      "step": 3652
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002673622225033138,
+      "loss": 0.221,
+      "step": 3653
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002673442161754279,
+      "loss": 0.1454,
+      "step": 3654
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026732620548854856,
+      "loss": 0.2261,
+      "step": 3655
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026730819044334475,
+      "loss": 0.3088,
+      "step": 3656
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026729017104048565,
+      "loss": 0.0615,
+      "step": 3657
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672721472806406,
+      "loss": 0.1524,
+      "step": 3658
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672541191644792,
+      "loss": 0.1791,
+      "step": 3659
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672360866926711,
+      "loss": 0.1723,
+      "step": 3660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002672180498658862,
+      "loss": 0.2944,
+      "step": 3661
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026720000868479446,
+      "loss": 0.1147,
+      "step": 3662
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002671819631500661,
+      "loss": 0.0647,
+      "step": 3663
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002671639132623715,
+      "loss": 0.2812,
+      "step": 3664
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026714585902238105,
+      "loss": 0.2413,
+      "step": 3665
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026712780043076543,
+      "loss": 0.0039,
+      "step": 3666
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002671097374881956,
+      "loss": 0.1738,
+      "step": 3667
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002670916701953423,
+      "loss": 0.2237,
+      "step": 3668
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002670735985528769,
+      "loss": 0.2299,
+      "step": 3669
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002670555225614707,
+      "loss": 0.1246,
+      "step": 3670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026703744222179496,
+      "loss": 0.1813,
+      "step": 3671
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00026701935753452153,
+      "loss": 0.1145,
+      "step": 3672
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026700126850032213,
+      "loss": 0.138,
+      "step": 3673
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002669831751198687,
+      "loss": 0.1579,
+      "step": 3674
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026696507739383334,
+      "loss": 0.1209,
+      "step": 3675
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026694697532288837,
+      "loss": 0.1882,
+      "step": 3676
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002669288689077062,
+      "loss": 0.2058,
+      "step": 3677
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026691075814895944,
+      "loss": 0.1966,
+      "step": 3678
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026689264304732087,
+      "loss": 0.1074,
+      "step": 3679
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002668745236034634,
+      "loss": 0.1638,
+      "step": 3680
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002668563998180601,
+      "loss": 0.2322,
+      "step": 3681
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026683827169178423,
+      "loss": 0.1967,
+      "step": 3682
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002668201392253092,
+      "loss": 0.2587,
+      "step": 3683
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002668020024193085,
+      "loss": 0.2075,
+      "step": 3684
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026678386127445603,
+      "loss": 0.0564,
+      "step": 3685
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002667657157914255,
+      "loss": 0.1575,
+      "step": 3686
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002667475659708911,
+      "loss": 0.1333,
+      "step": 3687
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002667294118135269,
+      "loss": 0.2431,
+      "step": 3688
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002667112533200075,
+      "loss": 0.166,
+      "step": 3689
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026669309049100713,
+      "loss": 0.0729,
+      "step": 3690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002666749233272008,
+      "loss": 0.2378,
+      "step": 3691
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002666567518292631,
+      "loss": 0.1181,
+      "step": 3692
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026663857599786917,
+      "loss": 0.1184,
+      "step": 3693
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026662039583369424,
+      "loss": 0.103,
+      "step": 3694
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026660221133741356,
+      "loss": 0.126,
+      "step": 3695
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026658402250970265,
+      "loss": 0.0791,
+      "step": 3696
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026656582935123723,
+      "loss": 0.2097,
+      "step": 3697
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026654763186269304,
+      "loss": 0.0891,
+      "step": 3698
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002665294300447461,
+      "loss": 0.1146,
+      "step": 3699
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026651122389807254,
+      "loss": 0.2203,
+      "step": 3700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002664930134233487,
+      "loss": 0.2225,
+      "step": 3701
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026647479862125105,
+      "loss": 0.2219,
+      "step": 3702
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002664565794924562,
+      "loss": 0.1283,
+      "step": 3703
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026643835603764086,
+      "loss": 0.2195,
+      "step": 3704
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00026642012825748207,
+      "loss": 0.0967,
+      "step": 3705
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026640189615265695,
+      "loss": 0.1436,
+      "step": 3706
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026638365972384264,
+      "loss": 0.0532,
+      "step": 3707
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026636541897171676,
+      "loss": 0.0676,
+      "step": 3708
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002663471738969568,
+      "loss": 0.1536,
+      "step": 3709
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002663289245002405,
+      "loss": 0.1136,
+      "step": 3710
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026631067078224574,
+      "loss": 0.2113,
+      "step": 3711
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002662924127436507,
+      "loss": 0.3893,
+      "step": 3712
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002662741503851335,
+      "loss": 0.1268,
+      "step": 3713
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002662558837073726,
+      "loss": 0.2048,
+      "step": 3714
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026623761271104656,
+      "loss": 0.1537,
+      "step": 3715
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026621933739683404,
+      "loss": 0.2116,
+      "step": 3716
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026620105776541396,
+      "loss": 0.1494,
+      "step": 3717
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026618277381746535,
+      "loss": 0.1827,
+      "step": 3718
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002661644855536674,
+      "loss": 0.1376,
+      "step": 3719
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002661461929746994,
+      "loss": 0.2105,
+      "step": 3720
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000266127896081241,
+      "loss": 0.1402,
+      "step": 3721
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002661095948739718,
+      "loss": 0.2192,
+      "step": 3722
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026609128935357163,
+      "loss": 0.2155,
+      "step": 3723
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026607297952072046,
+      "loss": 0.1901,
+      "step": 3724
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002660546653760985,
+      "loss": 0.1281,
+      "step": 3725
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026603634692038607,
+      "loss": 0.2691,
+      "step": 3726
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026601802415426363,
+      "loss": 0.159,
+      "step": 3727
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026599969707841176,
+      "loss": 0.0159,
+      "step": 3728
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026598136569351143,
+      "loss": 0.1433,
+      "step": 3729
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026596303000024337,
+      "loss": 0.2711,
+      "step": 3730
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026594468999928877,
+      "loss": 0.1032,
+      "step": 3731
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.000265926345691329,
+      "loss": 0.1518,
+      "step": 3732
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00026590799707704545,
+      "loss": 0.1114,
+      "step": 3733
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002658896441571197,
+      "loss": 0.1719,
+      "step": 3734
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002658712869322335,
+      "loss": 0.2186,
+      "step": 3735
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002658529254030688,
+      "loss": 0.178,
+      "step": 3736
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002658345595703077,
+      "loss": 0.21,
+      "step": 3737
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0002658161894346323,
+      "loss": 0.1566,
+      "step": 3738
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002657978149967251,
+      "loss": 0.1673,
+      "step": 3739
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002657794362572687,
+      "loss": 0.2138,
+      "step": 3740
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002657610532169457,
+      "loss": 0.0823,
+      "step": 3741
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002657426658764391,
+      "loss": 0.1131,
+      "step": 3742
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026572427423643185,
+      "loss": 0.2023,
+      "step": 3743
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026570587829760715,
+      "loss": 0.1865,
+      "step": 3744
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026568747806064843,
+      "loss": 0.1588,
+      "step": 3745
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026566907352623914,
+      "loss": 0.1416,
+      "step": 3746
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002656506646950629,
+      "loss": 0.0707,
+      "step": 3747
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026563225156780366,
+      "loss": 0.1928,
+      "step": 3748
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026561383414514536,
+      "loss": 0.3247,
+      "step": 3749
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026559541242777214,
+      "loss": 0.1617,
+      "step": 3750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026557698641636835,
+      "loss": 0.2316,
+      "step": 3751
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026555855611161845,
+      "loss": 0.1155,
+      "step": 3752
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002655401215142071,
+      "loss": 0.1152,
+      "step": 3753
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026552168262481893,
+      "loss": 0.1809,
+      "step": 3754
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002655032394441391,
+      "loss": 0.1559,
+      "step": 3755
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026548479197285266,
+      "loss": 0.1666,
+      "step": 3756
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002654663402116448,
+      "loss": 0.1511,
+      "step": 3757
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000265447884161201,
+      "loss": 0.1595,
+      "step": 3758
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002654294238222069,
+      "loss": 0.2037,
+      "step": 3759
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026541095919534814,
+      "loss": 0.2073,
+      "step": 3760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002653924902813107,
+      "loss": 0.0684,
+      "step": 3761
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026537401708078066,
+      "loss": 0.1792,
+      "step": 3762
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002653555395944442,
+      "loss": 0.1384,
+      "step": 3763
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026533705782298776,
+      "loss": 0.1668,
+      "step": 3764
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026531857176709776,
+      "loss": 0.1861,
+      "step": 3765
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026530008142746107,
+      "loss": 0.2376,
+      "step": 3766
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00026528158680476447,
+      "loss": 0.2099,
+      "step": 3767
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652630878996949,
+      "loss": 0.2086,
+      "step": 3768
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652445847129397,
+      "loss": 0.2454,
+      "step": 3769
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652260772451861,
+      "loss": 0.1789,
+      "step": 3770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0002652075654971216,
+      "loss": 0.2834,
+      "step": 3771
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026518904946943396,
+      "loss": 0.1846,
+      "step": 3772
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026517052916281086,
+      "loss": 0.1971,
+      "step": 3773
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026515200457794035,
+      "loss": 0.2402,
+      "step": 3774
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002651334757155106,
+      "loss": 0.251,
+      "step": 3775
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026511494257620973,
+      "loss": 0.0696,
+      "step": 3776
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002650964051607264,
+      "loss": 0.1461,
+      "step": 3777
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002650778634697491,
+      "loss": 0.2822,
+      "step": 3778
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002650593175039667,
+      "loss": 0.2344,
+      "step": 3779
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000265040767264068,
+      "loss": 0.1469,
+      "step": 3780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002650222127507422,
+      "loss": 0.238,
+      "step": 3781
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026500365396467845,
+      "loss": 0.0659,
+      "step": 3782
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002649850909065662,
+      "loss": 0.1172,
+      "step": 3783
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026496652357709503,
+      "loss": 0.1031,
+      "step": 3784
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002649479519769547,
+      "loss": 0.1643,
+      "step": 3785
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000264929376106835,
+      "loss": 0.3211,
+      "step": 3786
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026491079596742596,
+      "loss": 0.2551,
+      "step": 3787
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026489221155941785,
+      "loss": 0.2387,
+      "step": 3788
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026487362288350096,
+      "loss": 0.187,
+      "step": 3789
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002648550299403659,
+      "loss": 0.2555,
+      "step": 3790
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002648364327307033,
+      "loss": 0.1631,
+      "step": 3791
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.000264817831255204,
+      "loss": 0.2536,
+      "step": 3792
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002647992255145589,
+      "loss": 0.0432,
+      "step": 3793
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002647806155094592,
+      "loss": 0.3191,
+      "step": 3794
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002647620012405963,
+      "loss": 0.2235,
+      "step": 3795
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026474338270866154,
+      "loss": 0.1158,
+      "step": 3796
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002647247599143466,
+      "loss": 0.1545,
+      "step": 3797
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002647061328583433,
+      "loss": 0.1291,
+      "step": 3798
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002646875015413435,
+      "loss": 0.2717,
+      "step": 3799
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0002646688659640393,
+      "loss": 0.2376,
+      "step": 3800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026465022612712306,
+      "loss": 0.1332,
+      "step": 3801
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026463158203128713,
+      "loss": 0.3079,
+      "step": 3802
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00026461293367722403,
+      "loss": 0.0968,
+      "step": 3803
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002645942810656266,
+      "loss": 0.195,
+      "step": 3804
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002645756241971876,
+      "loss": 0.1677,
+      "step": 3805
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002645569630726002,
+      "loss": 0.1499,
+      "step": 3806
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026453829769255754,
+      "loss": 0.0622,
+      "step": 3807
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000264519628057753,
+      "loss": 0.1424,
+      "step": 3808
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002645009541688801,
+      "loss": 0.2397,
+      "step": 3809
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002644822760266326,
+      "loss": 0.1553,
+      "step": 3810
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026446359363170417,
+      "loss": 0.1589,
+      "step": 3811
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026444490698478894,
+      "loss": 0.0727,
+      "step": 3812
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026442621608658103,
+      "loss": 0.2694,
+      "step": 3813
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002644075209377747,
+      "loss": 0.248,
+      "step": 3814
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026438882153906454,
+      "loss": 0.2028,
+      "step": 3815
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000264370117891145,
+      "loss": 0.256,
+      "step": 3816
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026435140999471105,
+      "loss": 0.17,
+      "step": 3817
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026433269785045753,
+      "loss": 0.1803,
+      "step": 3818
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026431398145907955,
+      "loss": 0.1311,
+      "step": 3819
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642952608212724,
+      "loss": 0.2244,
+      "step": 3820
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642765359377315,
+      "loss": 0.1509,
+      "step": 3821
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642578068091523,
+      "loss": 0.165,
+      "step": 3822
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002642390734362307,
+      "loss": 0.1023,
+      "step": 3823
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026422033581966257,
+      "loss": 0.1476,
+      "step": 3824
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026420159396014386,
+      "loss": 0.2295,
+      "step": 3825
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026418284785837087,
+      "loss": 0.1383,
+      "step": 3826
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002641640975150399,
+      "loss": 0.1182,
+      "step": 3827
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026414534293084743,
+      "loss": 0.1107,
+      "step": 3828
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002641265841064903,
+      "loss": 0.1112,
+      "step": 3829
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002641078210426652,
+      "loss": 0.163,
+      "step": 3830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002640890537400691,
+      "loss": 0.1887,
+      "step": 3831
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002640702821993993,
+      "loss": 0.1598,
+      "step": 3832
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000264051506421353,
+      "loss": 0.2531,
+      "step": 3833
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026403272640662765,
+      "loss": 0.2543,
+      "step": 3834
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000264013942155921,
+      "loss": 0.2386,
+      "step": 3835
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00026399515366993067,
+      "loss": 0.2864,
+      "step": 3836
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026397636094935474,
+      "loss": 0.2927,
+      "step": 3837
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002639575639948912,
+      "loss": 0.0652,
+      "step": 3838
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002639387628072383,
+      "loss": 0.2077,
+      "step": 3839
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002639199573870945,
+      "loss": 0.2993,
+      "step": 3840
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002639011477351584,
+      "loss": 0.1745,
+      "step": 3841
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026388233385212856,
+      "loss": 0.1296,
+      "step": 3842
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026386351573870405,
+      "loss": 0.2109,
+      "step": 3843
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026384469339558385,
+      "loss": 0.2261,
+      "step": 3844
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026382586682346715,
+      "loss": 0.1412,
+      "step": 3845
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002638070360230532,
+      "loss": 0.1514,
+      "step": 3846
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002637882009950417,
+      "loss": 0.1814,
+      "step": 3847
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026376936174013217,
+      "loss": 0.1747,
+      "step": 3848
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026375051825902444,
+      "loss": 0.1733,
+      "step": 3849
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026373167055241856,
+      "loss": 0.2191,
+      "step": 3850
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026371281862101466,
+      "loss": 0.2079,
+      "step": 3851
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000263693962465513,
+      "loss": 0.1111,
+      "step": 3852
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000263675102086614,
+      "loss": 0.1886,
+      "step": 3853
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026365623748501833,
+      "loss": 0.1646,
+      "step": 3854
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026363736866142674,
+      "loss": 0.1133,
+      "step": 3855
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002636184956165401,
+      "loss": 0.1718,
+      "step": 3856
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002635996183510596,
+      "loss": 0.1357,
+      "step": 3857
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002635807368656864,
+      "loss": 0.2566,
+      "step": 3858
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026356185116112183,
+      "loss": 0.2485,
+      "step": 3859
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026354296123806757,
+      "loss": 0.1726,
+      "step": 3860
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026352406709722524,
+      "loss": 0.1269,
+      "step": 3861
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002635051687392967,
+      "loss": 0.0873,
+      "step": 3862
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026348626616498405,
+      "loss": 0.1873,
+      "step": 3863
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002634673593749894,
+      "loss": 0.1285,
+      "step": 3864
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026344844837001506,
+      "loss": 0.0538,
+      "step": 3865
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00026342953315076353,
+      "loss": 0.0833,
+      "step": 3866
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002634106137179375,
+      "loss": 0.1665,
+      "step": 3867
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002633916900722397,
+      "loss": 0.1998,
+      "step": 3868
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0002633727622143732,
+      "loss": 0.1203,
+      "step": 3869
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000263353830145041,
+      "loss": 0.1223,
+      "step": 3870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026333489386494647,
+      "loss": 0.1123,
+      "step": 3871
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000263315953374793,
+      "loss": 0.215,
+      "step": 3872
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002632970086752841,
+      "loss": 0.2643,
+      "step": 3873
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002632780597671235,
+      "loss": 0.1208,
+      "step": 3874
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002632591066510153,
+      "loss": 0.1534,
+      "step": 3875
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026324014932766333,
+      "loss": 0.1991,
+      "step": 3876
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002632211877977719,
+      "loss": 0.1587,
+      "step": 3877
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002632022220620454,
+      "loss": 0.2054,
+      "step": 3878
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002631832521211883,
+      "loss": 0.1169,
+      "step": 3879
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002631642779759053,
+      "loss": 0.2059,
+      "step": 3880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026314529962690115,
+      "loss": 0.1549,
+      "step": 3881
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000263126317074881,
+      "loss": 0.1585,
+      "step": 3882
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002631073303205499,
+      "loss": 0.1481,
+      "step": 3883
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002630883393646131,
+      "loss": 0.1674,
+      "step": 3884
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002630693442077762,
+      "loss": 0.158,
+      "step": 3885
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026305034485074466,
+      "loss": 0.2185,
+      "step": 3886
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026303134129422443,
+      "loss": 0.1097,
+      "step": 3887
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002630123335389212,
+      "loss": 0.1056,
+      "step": 3888
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629933215855413,
+      "loss": 0.1212,
+      "step": 3889
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629743054347908,
+      "loss": 0.1111,
+      "step": 3890
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026295528508737617,
+      "loss": 0.0915,
+      "step": 3891
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629362605440039,
+      "loss": 0.1216,
+      "step": 3892
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002629172318053808,
+      "loss": 0.2944,
+      "step": 3893
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026289819887221364,
+      "loss": 0.2602,
+      "step": 3894
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026287916174520945,
+      "loss": 0.2067,
+      "step": 3895
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026286012042507536,
+      "loss": 0.2782,
+      "step": 3896
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002628410749125188,
+      "loss": 0.0836,
+      "step": 3897
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026282202520824723,
+      "loss": 0.1202,
+      "step": 3898
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002628029713129683,
+      "loss": 0.0783,
+      "step": 3899
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002627839132273897,
+      "loss": 0.1917,
+      "step": 3900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0002627648509522195,
+      "loss": 0.1116,
+      "step": 3901
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00026274578448816577,
+      "loss": 0.1813,
+      "step": 3902
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002627267138359368,
+      "loss": 0.1743,
+      "step": 3903
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026270763899624095,
+      "loss": 0.1523,
+      "step": 3904
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026268855996978676,
+      "loss": 0.2509,
+      "step": 3905
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002626694767572831,
+      "loss": 0.2704,
+      "step": 3906
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002626503893594388,
+      "loss": 0.1339,
+      "step": 3907
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002626312977769628,
+      "loss": 0.088,
+      "step": 3908
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026261220201056443,
+      "loss": 0.2307,
+      "step": 3909
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262593102060953,
+      "loss": 0.219,
+      "step": 3910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262573997928838,
+      "loss": 0.2331,
+      "step": 3911
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026255488961492905,
+      "loss": 0.2133,
+      "step": 3912
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262535777119936,
+      "loss": 0.3963,
+      "step": 3913
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002625166604445689,
+      "loss": 0.0361,
+      "step": 3914
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624975395895378,
+      "loss": 0.1547,
+      "step": 3915
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000262478414555553,
+      "loss": 0.0771,
+      "step": 3916
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026245928534332497,
+      "loss": 0.1299,
+      "step": 3917
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624401519535643,
+      "loss": 0.1175,
+      "step": 3918
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624210143869816,
+      "loss": 0.2305,
+      "step": 3919
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002624018726442879,
+      "loss": 0.0791,
+      "step": 3920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002623827267261943,
+      "loss": 0.118,
+      "step": 3921
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026236357663341197,
+      "loss": 0.2312,
+      "step": 3922
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026234442236665225,
+      "loss": 0.3259,
+      "step": 3923
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026232526392662664,
+      "loss": 0.1091,
+      "step": 3924
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026230610131404696,
+      "loss": 0.1878,
+      "step": 3925
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026228693452962494,
+      "loss": 0.2935,
+      "step": 3926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002622677635740725,
+      "loss": 0.2647,
+      "step": 3927
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002622485884481019,
+      "loss": 0.2307,
+      "step": 3928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002622294091524254,
+      "loss": 0.2233,
+      "step": 3929
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002622102256877555,
+      "loss": 0.0594,
+      "step": 3930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002621910380548048,
+      "loss": 0.1366,
+      "step": 3931
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026217184625428595,
+      "loss": 0.349,
+      "step": 3932
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026215265028691197,
+      "loss": 0.1407,
+      "step": 3933
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00026213345015339596,
+      "loss": 0.1181,
+      "step": 3934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0002621142458544511,
+      "loss": 0.0668,
+      "step": 3935
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002620950373907907,
+      "loss": 0.164,
+      "step": 3936
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026207582476312847,
+      "loss": 0.2102,
+      "step": 3937
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026205660797217796,
+      "loss": 0.3387,
+      "step": 3938
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002620373870186531,
+      "loss": 0.1672,
+      "step": 3939
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002620181619032678,
+      "loss": 0.2108,
+      "step": 3940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002619989326267363,
+      "loss": 0.2563,
+      "step": 3941
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002619796991897729,
+      "loss": 0.0993,
+      "step": 3942
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002619604615930921,
+      "loss": 0.2964,
+      "step": 3943
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002619412198374084,
+      "loss": 0.0285,
+      "step": 3944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026192197392343666,
+      "loss": 0.1662,
+      "step": 3945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026190272385189184,
+      "loss": 0.2739,
+      "step": 3946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026188346962348894,
+      "loss": 0.0462,
+      "step": 3947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002618642112389433,
+      "loss": 0.1726,
+      "step": 3948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002618449486989702,
+      "loss": 0.0405,
+      "step": 3949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002618256820042852,
+      "loss": 0.1674,
+      "step": 3950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002618064111556041,
+      "loss": 0.1846,
+      "step": 3951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026178713615364265,
+      "loss": 0.206,
+      "step": 3952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026176785699911695,
+      "loss": 0.1263,
+      "step": 3953
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026174857369274307,
+      "loss": 0.1085,
+      "step": 3954
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026172928623523743,
+      "loss": 0.1195,
+      "step": 3955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026170999462731643,
+      "loss": 0.2139,
+      "step": 3956
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026169069886969673,
+      "loss": 0.1215,
+      "step": 3957
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002616713989630951,
+      "loss": 0.1107,
+      "step": 3958
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026165209490822843,
+      "loss": 0.3585,
+      "step": 3959
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026163278670581386,
+      "loss": 0.1092,
+      "step": 3960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002616134743565687,
+      "loss": 0.1964,
+      "step": 3961
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0002615941578612102,
+      "loss": 0.1071,
+      "step": 3962
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000261574837220456,
+      "loss": 0.1764,
+      "step": 3963
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026155551243502373,
+      "loss": 0.1171,
+      "step": 3964
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026153618350563136,
+      "loss": 0.2148,
+      "step": 3965
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026151685043299686,
+      "loss": 0.1306,
+      "step": 3966
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026149751321783836,
+      "loss": 0.1231,
+      "step": 3967
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00026147817186087427,
+      "loss": 0.0769,
+      "step": 3968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002614588263628229,
+      "loss": 0.2408,
+      "step": 3969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026143947672440304,
+      "loss": 0.1613,
+      "step": 3970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002614201229463334,
+      "loss": 0.3067,
+      "step": 3971
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026140076502933297,
+      "loss": 0.1379,
+      "step": 3972
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002613814029741208,
+      "loss": 0.2535,
+      "step": 3973
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002613620367814161,
+      "loss": 0.2004,
+      "step": 3974
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002613426664519383,
+      "loss": 0.1115,
+      "step": 3975
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002613232919864069,
+      "loss": 0.0887,
+      "step": 3976
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026130391338554174,
+      "loss": 0.2267,
+      "step": 3977
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026128453065006255,
+      "loss": 0.0274,
+      "step": 3978
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002612651437806894,
+      "loss": 0.2667,
+      "step": 3979
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026124575277814246,
+      "loss": 0.1268,
+      "step": 3980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026122635764314197,
+      "loss": 0.0758,
+      "step": 3981
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026120695837640855,
+      "loss": 0.1685,
+      "step": 3982
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026118755497866265,
+      "loss": 0.0902,
+      "step": 3983
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026116814745062516,
+      "loss": 0.1521,
+      "step": 3984
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000261148735793017,
+      "loss": 0.1715,
+      "step": 3985
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026112932000655926,
+      "loss": 0.3303,
+      "step": 3986
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026110990009197316,
+      "loss": 0.1132,
+      "step": 3987
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002610904760499801,
+      "loss": 0.0971,
+      "step": 3988
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002610710478813016,
+      "loss": 0.1677,
+      "step": 3989
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026105161558665944,
+      "loss": 0.1902,
+      "step": 3990
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026103217916677535,
+      "loss": 0.2944,
+      "step": 3991
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026101273862237146,
+      "loss": 0.1923,
+      "step": 3992
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002609932939541698,
+      "loss": 0.0968,
+      "step": 3993
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00026097384516289286,
+      "loss": 0.0665,
+      "step": 3994
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002609543922492629,
+      "loss": 0.249,
+      "step": 3995
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002609349352140027,
+      "loss": 0.1739,
+      "step": 3996
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000260915474057835,
+      "loss": 0.1958,
+      "step": 3997
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002608960087814826,
+      "loss": 0.1598,
+      "step": 3998
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002608765393856687,
+      "loss": 0.0974,
+      "step": 3999
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002608570658711166,
+      "loss": 0.2957,
+      "step": 4000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0002608375882385495,
+      "loss": 0.0935,
+      "step": 4001
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026081810648869103,
+      "loss": 0.2236,
+      "step": 4002
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026079862062226486,
+      "loss": 0.153,
+      "step": 4003
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002607791306399949,
+      "loss": 0.2576,
+      "step": 4004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026075963654260503,
+      "loss": 0.1763,
+      "step": 4005
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002607401383308195,
+      "loss": 0.1344,
+      "step": 4006
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002607206360053626,
+      "loss": 0.2049,
+      "step": 4007
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026070112956695873,
+      "loss": 0.2285,
+      "step": 4008
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002606816190163325,
+      "loss": 0.0866,
+      "step": 4009
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026066210435420875,
+      "loss": 0.1844,
+      "step": 4010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002606425855813123,
+      "loss": 0.046,
+      "step": 4011
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002606230626983683,
+      "loss": 0.1398,
+      "step": 4012
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002606035357061018,
+      "loss": 0.2996,
+      "step": 4013
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002605840046052384,
+      "loss": 0.2093,
+      "step": 4014
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026056446939650346,
+      "loss": 0.0018,
+      "step": 4015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002605449300806228,
+      "loss": 0.1312,
+      "step": 4016
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026052538665832206,
+      "loss": 0.1281,
+      "step": 4017
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026050583913032736,
+      "loss": 0.2583,
+      "step": 4018
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026048628749736477,
+      "loss": 0.1789,
+      "step": 4019
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026046673176016065,
+      "loss": 0.191,
+      "step": 4020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002604471719194413,
+      "loss": 0.1299,
+      "step": 4021
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026042760797593346,
+      "loss": 0.1871,
+      "step": 4022
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002604080399303638,
+      "loss": 0.0448,
+      "step": 4023
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002603884677834592,
+      "loss": 0.1128,
+      "step": 4024
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026036889153594676,
+      "loss": 0.2422,
+      "step": 4025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026034931118855366,
+      "loss": 0.1204,
+      "step": 4026
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026032972674200724,
+      "loss": 0.1888,
+      "step": 4027
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026031013819703496,
+      "loss": 0.1291,
+      "step": 4028
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002602905455543646,
+      "loss": 0.1568,
+      "step": 4029
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002602709488147239,
+      "loss": 0.375,
+      "step": 4030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026025134797884076,
+      "loss": 0.1759,
+      "step": 4031
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002602317430474434,
+      "loss": 0.2703,
+      "step": 4032
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00026021213402125997,
+      "loss": 0.2689,
+      "step": 4033
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000260192520901019,
+      "loss": 0.1667,
+      "step": 4034
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026017290368744907,
+      "loss": 0.2239,
+      "step": 4035
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026015328238127873,
+      "loss": 0.278,
+      "step": 4036
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002601336569832371,
+      "loss": 0.1687,
+      "step": 4037
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026011402749405297,
+      "loss": 0.1662,
+      "step": 4038
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002600943939144557,
+      "loss": 0.2879,
+      "step": 4039
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002600747562451745,
+      "loss": 0.1574,
+      "step": 4040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002600551144869389,
+      "loss": 0.3363,
+      "step": 4041
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00026003546864047857,
+      "loss": 0.1504,
+      "step": 4042
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002600158187065232,
+      "loss": 0.185,
+      "step": 4043
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025999616468580283,
+      "loss": 0.1259,
+      "step": 4044
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025997650657904746,
+      "loss": 0.2887,
+      "step": 4045
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002599568443869874,
+      "loss": 0.1495,
+      "step": 4046
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025993717811035303,
+      "loss": 0.1491,
+      "step": 4047
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025991750774987483,
+      "loss": 0.1101,
+      "step": 4048
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002598978333062836,
+      "loss": 0.0524,
+      "step": 4049
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025987815478031016,
+      "loss": 0.3387,
+      "step": 4050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002598584721726854,
+      "loss": 0.0715,
+      "step": 4051
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002598387854841406,
+      "loss": 0.1726,
+      "step": 4052
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025981909471540707,
+      "loss": 0.2841,
+      "step": 4053
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002597993998672161,
+      "loss": 0.1178,
+      "step": 4054
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002597797009402995,
+      "loss": 0.2284,
+      "step": 4055
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002597599979353889,
+      "loss": 0.2448,
+      "step": 4056
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025974029085321624,
+      "loss": 0.2167,
+      "step": 4057
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002597205796945136,
+      "loss": 0.287,
+      "step": 4058
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002597008644600132,
+      "loss": 0.2035,
+      "step": 4059
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002596811451504474,
+      "loss": 0.2791,
+      "step": 4060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002596614217665486,
+      "loss": 0.1181,
+      "step": 4061
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002596416943090496,
+      "loss": 0.2179,
+      "step": 4062
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002596219627786832,
+      "loss": 0.1001,
+      "step": 4063
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002596022271761823,
+      "loss": 0.0858,
+      "step": 4064
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00025958248750228014,
+      "loss": 0.2815,
+      "step": 4065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002595627437577099,
+      "loss": 0.0753,
+      "step": 4066
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000259542995943205,
+      "loss": 0.1669,
+      "step": 4067
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000259523244059499,
+      "loss": 0.2817,
+      "step": 4068
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002595034881073257,
+      "loss": 0.1117,
+      "step": 4069
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025948372808741894,
+      "loss": 0.1743,
+      "step": 4070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002594639640005127,
+      "loss": 0.2263,
+      "step": 4071
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002594441958473412,
+      "loss": 0.1546,
+      "step": 4072
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025942442362863886,
+      "loss": 0.084,
+      "step": 4073
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025940464734514,
+      "loss": 0.0801,
+      "step": 4074
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002593848669975793,
+      "loss": 0.3001,
+      "step": 4075
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025936508258669156,
+      "loss": 0.2175,
+      "step": 4076
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002593452941132117,
+      "loss": 0.3239,
+      "step": 4077
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025932550157787486,
+      "loss": 0.2916,
+      "step": 4078
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002593057049814162,
+      "loss": 0.2788,
+      "step": 4079
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002592859043245711,
+      "loss": 0.2021,
+      "step": 4080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025926609960807523,
+      "loss": 0.1741,
+      "step": 4081
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025924629083266414,
+      "loss": 0.0865,
+      "step": 4082
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002592264779990737,
+      "loss": 0.1489,
+      "step": 4083
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002592066611080399,
+      "loss": 0.0856,
+      "step": 4084
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025918684016029887,
+      "loss": 0.139,
+      "step": 4085
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025916701515658696,
+      "loss": 0.1729,
+      "step": 4086
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025914718609764054,
+      "loss": 0.1573,
+      "step": 4087
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002591273529841962,
+      "loss": 0.2619,
+      "step": 4088
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025910751581699075,
+      "loss": 0.2102,
+      "step": 4089
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025908767459676107,
+      "loss": 0.1005,
+      "step": 4090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002590678293242441,
+      "loss": 0.1685,
+      "step": 4091
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025904798000017713,
+      "loss": 0.14,
+      "step": 4092
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002590281266252975,
+      "loss": 0.2498,
+      "step": 4093
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002590082692003427,
+      "loss": 0.1318,
+      "step": 4094
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025898840772605033,
+      "loss": 0.1808,
+      "step": 4095
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002589685422031583,
+      "loss": 0.0863,
+      "step": 4096
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002589486726324044,
+      "loss": 0.0017,
+      "step": 4097
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002589287990145268,
+      "loss": 0.1309,
+      "step": 4098
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0002589089213502638,
+      "loss": 0.3018,
+      "step": 4099
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00025888903964035366,
+      "loss": 0.1494,
+      "step": 4100
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025886915388553507,
+      "loss": 0.1619,
+      "step": 4101
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025884926408654665,
+      "loss": 0.1953,
+      "step": 4102
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025882937024412727,
+      "loss": 0.1316,
+      "step": 4103
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002588094723590159,
+      "loss": 0.1293,
+      "step": 4104
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002587895704319518,
+      "loss": 0.1175,
+      "step": 4105
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025876966446367413,
+      "loss": 0.1413,
+      "step": 4106
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002587497544549224,
+      "loss": 0.2212,
+      "step": 4107
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025872984040643614,
+      "loss": 0.2501,
+      "step": 4108
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002587099223189552,
+      "loss": 0.0461,
+      "step": 4109
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002586900001932194,
+      "loss": 0.2056,
+      "step": 4110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002586700740299689,
+      "loss": 0.1539,
+      "step": 4111
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002586501438299438,
+      "loss": 0.1334,
+      "step": 4112
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025863020959388444,
+      "loss": 0.2087,
+      "step": 4113
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002586102713225314,
+      "loss": 0.1274,
+      "step": 4114
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002585903290166252,
+      "loss": 0.2935,
+      "step": 4115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002585703826769067,
+      "loss": 0.0552,
+      "step": 4116
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000258550432304117,
+      "loss": 0.0206,
+      "step": 4117
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025853047789899694,
+      "loss": 0.2062,
+      "step": 4118
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025851051946228795,
+      "loss": 0.0972,
+      "step": 4119
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002584905569947313,
+      "loss": 0.2999,
+      "step": 4120
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002584705904970687,
+      "loss": 0.1394,
+      "step": 4121
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002584506199700417,
+      "loss": 0.0736,
+      "step": 4122
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025843064541439216,
+      "loss": 0.1385,
+      "step": 4123
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025841066683086216,
+      "loss": 0.1611,
+      "step": 4124
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002583906842201938,
+      "loss": 0.1,
+      "step": 4125
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025837069758312934,
+      "loss": 0.1623,
+      "step": 4126
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025835070692041133,
+      "loss": 0.4328,
+      "step": 4127
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025833071223278217,
+      "loss": 0.1126,
+      "step": 4128
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00025831071352098485,
+      "loss": 0.2002,
+      "step": 4129
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002582907107857621,
+      "loss": 0.1307,
+      "step": 4130
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000258270704027857,
+      "loss": 0.2638,
+      "step": 4131
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002582506932480127,
+      "loss": 0.2019,
+      "step": 4132
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002582306784469727,
+      "loss": 0.105,
+      "step": 4133
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002582106596254803,
+      "loss": 0.2216,
+      "step": 4134
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002581906367842792,
+      "loss": 0.1634,
+      "step": 4135
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002581706099241132,
+      "loss": 0.1515,
+      "step": 4136
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002581505790457263,
+      "loss": 0.2385,
+      "step": 4137
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025813054414986256,
+      "loss": 0.2257,
+      "step": 4138
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002581105052372662,
+      "loss": 0.2508,
+      "step": 4139
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002580904623086815,
+      "loss": 0.0496,
+      "step": 4140
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002580704153648532,
+      "loss": 0.2543,
+      "step": 4141
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002580503644065258,
+      "loss": 0.2127,
+      "step": 4142
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002580303094344443,
+      "loss": 0.1743,
+      "step": 4143
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002580102504493535,
+      "loss": 0.0996,
+      "step": 4144
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025799018745199874,
+      "loss": 0.2026,
+      "step": 4145
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025797012044312514,
+      "loss": 0.1117,
+      "step": 4146
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025795004942347823,
+      "loss": 0.2989,
+      "step": 4147
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002579299743938035,
+      "loss": 0.1531,
+      "step": 4148
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025790989535484676,
+      "loss": 0.129,
+      "step": 4149
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002578898123073538,
+      "loss": 0.1744,
+      "step": 4150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025786972525207073,
+      "loss": 0.1659,
+      "step": 4151
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002578496341897437,
+      "loss": 0.1604,
+      "step": 4152
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025782953912111895,
+      "loss": 0.2314,
+      "step": 4153
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002578094400469431,
+      "loss": 0.1702,
+      "step": 4154
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002577893369679627,
+      "loss": 0.2042,
+      "step": 4155
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002577692298849245,
+      "loss": 0.1269,
+      "step": 4156
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025774911879857546,
+      "loss": 0.2521,
+      "step": 4157
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002577290037096626,
+      "loss": 0.1738,
+      "step": 4158
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002577088846189332,
+      "loss": 0.1156,
+      "step": 4159
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002576887615271346,
+      "loss": 0.2492,
+      "step": 4160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002576686344350143,
+      "loss": 0.1232,
+      "step": 4161
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002576485033433199,
+      "loss": 0.2616,
+      "step": 4162
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025762836825279933,
+      "loss": 0.1588,
+      "step": 4163
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002576082291642005,
+      "loss": 0.1229,
+      "step": 4164
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002575880860782715,
+      "loss": 0.068,
+      "step": 4165
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00025756793899576065,
+      "loss": 0.2273,
+      "step": 4166
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025754778791741627,
+      "loss": 0.3106,
+      "step": 4167
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002575276328439869,
+      "loss": 0.3169,
+      "step": 4168
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025750747377622135,
+      "loss": 0.1428,
+      "step": 4169
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002574873107148684,
+      "loss": 0.1609,
+      "step": 4170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025746714366067704,
+      "loss": 0.3146,
+      "step": 4171
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025744697261439645,
+      "loss": 0.2274,
+      "step": 4172
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025742679757677594,
+      "loss": 0.0857,
+      "step": 4173
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002574066185485649,
+      "loss": 0.1218,
+      "step": 4174
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000257386435530513,
+      "loss": 0.2525,
+      "step": 4175
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002573662485233699,
+      "loss": 0.1785,
+      "step": 4176
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025734605752788544,
+      "loss": 0.1079,
+      "step": 4177
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002573258625448098,
+      "loss": 0.1846,
+      "step": 4178
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002573056635748931,
+      "loss": 0.0668,
+      "step": 4179
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002572854606188857,
+      "loss": 0.2226,
+      "step": 4180
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025726525367753797,
+      "loss": 0.1816,
+      "step": 4181
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002572450427516007,
+      "loss": 0.0426,
+      "step": 4182
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025722482784182446,
+      "loss": 0.1138,
+      "step": 4183
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025720460894896046,
+      "loss": 0.1624,
+      "step": 4184
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002571843860737595,
+      "loss": 0.1852,
+      "step": 4185
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025716415921697293,
+      "loss": 0.3743,
+      "step": 4186
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002571439283793521,
+      "loss": 0.2182,
+      "step": 4187
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002571236935616486,
+      "loss": 0.1438,
+      "step": 4188
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025710345476461394,
+      "loss": 0.1669,
+      "step": 4189
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025708321198900005,
+      "loss": 0.1701,
+      "step": 4190
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002570629652355588,
+      "loss": 0.1603,
+      "step": 4191
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025704271450504245,
+      "loss": 0.1338,
+      "step": 4192
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002570224597982031,
+      "loss": 0.1395,
+      "step": 4193
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002570022011157932,
+      "loss": 0.0354,
+      "step": 4194
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002569819384585653,
+      "loss": 0.1264,
+      "step": 4195
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025696167182727214,
+      "loss": 0.2002,
+      "step": 4196
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00025694140122266653,
+      "loss": 0.1809,
+      "step": 4197
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002569211266455015,
+      "loss": 0.2513,
+      "step": 4198
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0002569008480965301,
+      "loss": 0.2597,
+      "step": 4199
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568805655765057,
+      "loss": 0.2595,
+      "step": 4200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568602790861817,
+      "loss": 0.2388,
+      "step": 4201
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568399886263117,
+      "loss": 0.1399,
+      "step": 4202
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002568196941976495,
+      "loss": 0.1252,
+      "step": 4203
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002567993958009488,
+      "loss": 0.279,
+      "step": 4204
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002567790934369638,
+      "loss": 0.2732,
+      "step": 4205
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025675878710644854,
+      "loss": 0.2023,
+      "step": 4206
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002567384768101575,
+      "loss": 0.1929,
+      "step": 4207
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000256718162548845,
+      "loss": 0.2213,
+      "step": 4208
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566978443232657,
+      "loss": 0.2138,
+      "step": 4209
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025667752213417435,
+      "loss": 0.2451,
+      "step": 4210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025665719598232594,
+      "loss": 0.2367,
+      "step": 4211
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025663686586847547,
+      "loss": 0.119,
+      "step": 4212
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002566165317933781,
+      "loss": 0.2256,
+      "step": 4213
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002565961937577892,
+      "loss": 0.1422,
+      "step": 4214
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025657585176246437,
+      "loss": 0.0643,
+      "step": 4215
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025655550580815914,
+      "loss": 0.2553,
+      "step": 4216
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002565351558956293,
+      "loss": 0.1921,
+      "step": 4217
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002565148020256309,
+      "loss": 0.2913,
+      "step": 4218
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025649444419891996,
+      "loss": 0.2056,
+      "step": 4219
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025647408241625267,
+      "loss": 0.1966,
+      "step": 4220
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002564537166783855,
+      "loss": 0.2246,
+      "step": 4221
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002564333469860749,
+      "loss": 0.2523,
+      "step": 4222
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002564129733400775,
+      "loss": 0.0024,
+      "step": 4223
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002563925957411503,
+      "loss": 0.2817,
+      "step": 4224
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002563722141900501,
+      "loss": 0.1895,
+      "step": 4225
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002563518286875341,
+      "loss": 0.1255,
+      "step": 4226
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00025633143923435955,
+      "loss": 0.1655,
+      "step": 4227
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002563110458312838,
+      "loss": 0.1441,
+      "step": 4228
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002562906484790645,
+      "loss": 0.2063,
+      "step": 4229
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002562702471784593,
+      "loss": 0.2122,
+      "step": 4230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000256249841930226,
+      "loss": 0.2318,
+      "step": 4231
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002562294327351227,
+      "loss": 0.1752,
+      "step": 4232
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025620901959390747,
+      "loss": 0.3459,
+      "step": 4233
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025618860250733864,
+      "loss": 0.2903,
+      "step": 4234
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002561681814761746,
+      "loss": 0.2404,
+      "step": 4235
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025614775650117396,
+      "loss": 0.1463,
+      "step": 4236
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025612732758309545,
+      "loss": 0.2625,
+      "step": 4237
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002561068947226979,
+      "loss": 0.3019,
+      "step": 4238
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025608645792074043,
+      "loss": 0.1873,
+      "step": 4239
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025606601717798207,
+      "loss": 0.1612,
+      "step": 4240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002560455724951823,
+      "loss": 0.2073,
+      "step": 4241
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002560251238731004,
+      "loss": 0.1716,
+      "step": 4242
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002560046713124961,
+      "loss": 0.008,
+      "step": 4243
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025598421481412917,
+      "loss": 0.1354,
+      "step": 4244
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002559637543787594,
+      "loss": 0.2173,
+      "step": 4245
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025594329000714684,
+      "loss": 0.2582,
+      "step": 4246
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025592282170005176,
+      "loss": 0.1655,
+      "step": 4247
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025590234945823456,
+      "loss": 0.1738,
+      "step": 4248
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002558818732824556,
+      "loss": 0.1009,
+      "step": 4249
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025586139317347546,
+      "loss": 0.2141,
+      "step": 4250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025584090913205503,
+      "loss": 0.1251,
+      "step": 4251
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025582042115895523,
+      "loss": 0.1487,
+      "step": 4252
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002557999292549371,
+      "loss": 0.0953,
+      "step": 4253
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002557794334207619,
+      "loss": 0.2014,
+      "step": 4254
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025575893365719087,
+      "loss": 0.2347,
+      "step": 4255
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002557384299649856,
+      "loss": 0.1739,
+      "step": 4256
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025571792234490773,
+      "loss": 0.2156,
+      "step": 4257
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025569741079771907,
+      "loss": 0.095,
+      "step": 4258
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002556768953241816,
+      "loss": 0.04,
+      "step": 4259
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002556563759250573,
+      "loss": 0.3647,
+      "step": 4260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025563585260110846,
+      "loss": 0.0762,
+      "step": 4261
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002556153253530975,
+      "loss": 0.2275,
+      "step": 4262
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002555947941817869,
+      "loss": 0.2045,
+      "step": 4263
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002555742590879394,
+      "loss": 0.1859,
+      "step": 4264
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00025555372007231774,
+      "loss": 0.2858,
+      "step": 4265
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025553317713568487,
+      "loss": 0.0979,
+      "step": 4266
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025551263027880403,
+      "loss": 0.255,
+      "step": 4267
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025549207950243834,
+      "loss": 0.2457,
+      "step": 4268
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025547152480735123,
+      "loss": 0.0759,
+      "step": 4269
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002554509661943063,
+      "loss": 0.121,
+      "step": 4270
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002554304036640672,
+      "loss": 0.1347,
+      "step": 4271
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025540983721739774,
+      "loss": 0.2941,
+      "step": 4272
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000255389266855062,
+      "loss": 0.2435,
+      "step": 4273
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000255368692577824,
+      "loss": 0.1708,
+      "step": 4274
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002553481143864481,
+      "loss": 0.089,
+      "step": 4275
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025532753228169866,
+      "loss": 0.1432,
+      "step": 4276
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002553069462643403,
+      "loss": 0.1267,
+      "step": 4277
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025528635633513763,
+      "loss": 0.1915,
+      "step": 4278
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002552657624948556,
+      "loss": 0.0586,
+      "step": 4279
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025524516474425917,
+      "loss": 0.1039,
+      "step": 4280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025522456308411354,
+      "loss": 0.2094,
+      "step": 4281
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025520395751518397,
+      "loss": 0.1204,
+      "step": 4282
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025518334803823584,
+      "loss": 0.137,
+      "step": 4283
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025516273465403473,
+      "loss": 0.2003,
+      "step": 4284
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002551421173633465,
+      "loss": 0.2184,
+      "step": 4285
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025512149616693694,
+      "loss": 0.2458,
+      "step": 4286
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000255100871065572,
+      "loss": 0.2783,
+      "step": 4287
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025508024206001795,
+      "loss": 0.2701,
+      "step": 4288
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025505960915104105,
+      "loss": 0.104,
+      "step": 4289
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002550389723394078,
+      "loss": 0.2904,
+      "step": 4290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00025501833162588466,
+      "loss": 0.3025,
+      "step": 4291
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549976870112385,
+      "loss": 0.155,
+      "step": 4292
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549770384962362,
+      "loss": 0.2118,
+      "step": 4293
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549563860816447,
+      "loss": 0.0679,
+      "step": 4294
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549357297682313,
+      "loss": 0.2418,
+      "step": 4295
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002549150695567632,
+      "loss": 0.245,
+      "step": 4296
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000254894405448008,
+      "loss": 0.1558,
+      "step": 4297
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002548737374427332,
+      "loss": 0.1318,
+      "step": 4298
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002548530655417066,
+      "loss": 0.1733,
+      "step": 4299
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002548323897456961,
+      "loss": 0.3409,
+      "step": 4300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025481171005546974,
+      "loss": 0.3033,
+      "step": 4301
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025479102647179566,
+      "loss": 0.1971,
+      "step": 4302
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025477033899544226,
+      "loss": 0.1816,
+      "step": 4303
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025474964762717804,
+      "loss": 0.2536,
+      "step": 4304
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002547289523677716,
+      "loss": 0.1259,
+      "step": 4305
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002547082532179917,
+      "loss": 0.183,
+      "step": 4306
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002546875501786072,
+      "loss": 0.192,
+      "step": 4307
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002546668432503872,
+      "loss": 0.0433,
+      "step": 4308
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002546461324341009,
+      "loss": 0.2677,
+      "step": 4309
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025462541773051773,
+      "loss": 0.1748,
+      "step": 4310
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025460469914040704,
+      "loss": 0.2153,
+      "step": 4311
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025458397666453853,
+      "loss": 0.2379,
+      "step": 4312
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000254563250303682,
+      "loss": 0.2173,
+      "step": 4313
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002545425200586073,
+      "loss": 0.2508,
+      "step": 4314
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002545217859300846,
+      "loss": 0.2203,
+      "step": 4315
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025450104791888405,
+      "loss": 0.2244,
+      "step": 4316
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025448030602577597,
+      "loss": 0.0922,
+      "step": 4317
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025445956025153094,
+      "loss": 0.1845,
+      "step": 4318
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002544388105969196,
+      "loss": 0.0785,
+      "step": 4319
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002544180570627127,
+      "loss": 0.1685,
+      "step": 4320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025439729964968116,
+      "loss": 0.1556,
+      "step": 4321
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002543765383585961,
+      "loss": 0.11,
+      "step": 4322
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025435577319022875,
+      "loss": 0.2852,
+      "step": 4323
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025433500414535036,
+      "loss": 0.1023,
+      "step": 4324
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0002543142312247326,
+      "loss": 0.1445,
+      "step": 4325
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025429345442914704,
+      "loss": 0.0697,
+      "step": 4326
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025427267375936546,
+      "loss": 0.0863,
+      "step": 4327
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025425188921615984,
+      "loss": 0.2336,
+      "step": 4328
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025423110080030224,
+      "loss": 0.1306,
+      "step": 4329
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00025421030851256494,
+      "loss": 0.1255,
+      "step": 4330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025418951235372026,
+      "loss": 0.0936,
+      "step": 4331
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002541687123245407,
+      "loss": 0.2132,
+      "step": 4332
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002541479084257989,
+      "loss": 0.0508,
+      "step": 4333
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002541271006582678,
+      "loss": 0.0679,
+      "step": 4334
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025410628902272024,
+      "loss": 0.2652,
+      "step": 4335
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025408547351992935,
+      "loss": 0.2264,
+      "step": 4336
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002540646541506683,
+      "loss": 0.1389,
+      "step": 4337
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025404383091571055,
+      "loss": 0.1385,
+      "step": 4338
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002540230038158296,
+      "loss": 0.1618,
+      "step": 4339
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002540021728517991,
+      "loss": 0.3054,
+      "step": 4340
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002539813380243928,
+      "loss": 0.0295,
+      "step": 4341
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002539604993343848,
+      "loss": 0.2409,
+      "step": 4342
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025393965678254904,
+      "loss": 0.2007,
+      "step": 4343
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002539188103696599,
+      "loss": 0.0768,
+      "step": 4344
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002538979600964916,
+      "loss": 0.1539,
+      "step": 4345
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025387710596381884,
+      "loss": 0.354,
+      "step": 4346
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025385624797241615,
+      "loss": 0.1321,
+      "step": 4347
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002538353861230585,
+      "loss": 0.1979,
+      "step": 4348
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002538145204165207,
+      "loss": 0.3339,
+      "step": 4349
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025379365085357783,
+      "loss": 0.2354,
+      "step": 4350
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025377277743500535,
+      "loss": 0.1243,
+      "step": 4351
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025375190016157834,
+      "loss": 0.1598,
+      "step": 4352
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002537310190340726,
+      "loss": 0.157,
+      "step": 4353
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025371013405326364,
+      "loss": 0.1468,
+      "step": 4354
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025368924521992735,
+      "loss": 0.2653,
+      "step": 4355
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025366835253483967,
+      "loss": 0.106,
+      "step": 4356
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002536474559987767,
+      "loss": 0.3201,
+      "step": 4357
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002536265556125146,
+      "loss": 0.159,
+      "step": 4358
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002536056513768299,
+      "loss": 0.1555,
+      "step": 4359
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002535847432924991,
+      "loss": 0.2708,
+      "step": 4360
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025356383136029885,
+      "loss": 0.2348,
+      "step": 4361
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025354291558100597,
+      "loss": 0.1267,
+      "step": 4362
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00025352199595539735,
+      "loss": 0.1971,
+      "step": 4363
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025350107248425026,
+      "loss": 0.1252,
+      "step": 4364
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002534801451683417,
+      "loss": 0.1865,
+      "step": 4365
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025345921400844935,
+      "loss": 0.2762,
+      "step": 4366
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025343827900535043,
+      "loss": 0.2051,
+      "step": 4367
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002534173401598229,
+      "loss": 0.1932,
+      "step": 4368
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002533963974726444,
+      "loss": 0.1648,
+      "step": 4369
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025337545094459294,
+      "loss": 0.051,
+      "step": 4370
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002533545005764467,
+      "loss": 0.2234,
+      "step": 4371
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025333354636898376,
+      "loss": 0.1073,
+      "step": 4372
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002533125883229826,
+      "loss": 0.2127,
+      "step": 4373
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002532916264392218,
+      "loss": 0.2292,
+      "step": 4374
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002532706607184799,
+      "loss": 0.1945,
+      "step": 4375
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025324969116153587,
+      "loss": 0.1662,
+      "step": 4376
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025322871776916857,
+      "loss": 0.1819,
+      "step": 4377
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002532077405421571,
+      "loss": 0.1208,
+      "step": 4378
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025318675948128073,
+      "loss": 0.2181,
+      "step": 4379
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025316577458731884,
+      "loss": 0.1249,
+      "step": 4380
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025314478586105094,
+      "loss": 0.0371,
+      "step": 4381
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002531237933032567,
+      "loss": 0.2304,
+      "step": 4382
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025310279691471597,
+      "loss": 0.1941,
+      "step": 4383
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002530817966962088,
+      "loss": 0.1845,
+      "step": 4384
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025306079264851496,
+      "loss": 0.2078,
+      "step": 4385
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000253039784772415,
+      "loss": 0.1926,
+      "step": 4386
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025301877306868923,
+      "loss": 0.1454,
+      "step": 4387
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529977575381181,
+      "loss": 0.1107,
+      "step": 4388
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529767381814823,
+      "loss": 0.2026,
+      "step": 4389
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025295571499956267,
+      "loss": 0.2475,
+      "step": 4390
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529346879931401,
+      "loss": 0.2163,
+      "step": 4391
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002529136571629958,
+      "loss": 0.1569,
+      "step": 4392
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00025289262250991086,
+      "loss": 0.2753,
+      "step": 4393
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002528715840346668,
+      "loss": 0.2022,
+      "step": 4394
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000252850541738045,
+      "loss": 0.0909,
+      "step": 4395
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002528294956208272,
+      "loss": 0.2593,
+      "step": 4396
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002528084456837952,
+      "loss": 0.1326,
+      "step": 4397
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002527873919277309,
+      "loss": 0.1894,
+      "step": 4398
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025276633435341645,
+      "loss": 0.1443,
+      "step": 4399
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000252745272961634,
+      "loss": 0.1819,
+      "step": 4400
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000252724207753166,
+      "loss": 0.1046,
+      "step": 4401
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025270313872879485,
+      "loss": 0.1032,
+      "step": 4402
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002526820658893033,
+      "loss": 0.1733,
+      "step": 4403
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002526609892354741,
+      "loss": 0.2367,
+      "step": 4404
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002526399087680902,
+      "loss": 0.1463,
+      "step": 4405
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002526188244879347,
+      "loss": 0.175,
+      "step": 4406
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025259773639579083,
+      "loss": 0.231,
+      "step": 4407
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025257664449244185,
+      "loss": 0.2179,
+      "step": 4408
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002525555487786714,
+      "loss": 0.1904,
+      "step": 4409
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025253444925526304,
+      "loss": 0.2491,
+      "step": 4410
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025251334592300053,
+      "loss": 0.195,
+      "step": 4411
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025249223878266784,
+      "loss": 0.1501,
+      "step": 4412
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025247112783504907,
+      "loss": 0.19,
+      "step": 4413
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025245001308092836,
+      "loss": 0.161,
+      "step": 4414
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025242889452109014,
+      "loss": 0.1791,
+      "step": 4415
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002524077721563188,
+      "loss": 0.2351,
+      "step": 4416
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002523866459873991,
+      "loss": 0.1969,
+      "step": 4417
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002523655160151157,
+      "loss": 0.2822,
+      "step": 4418
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025234438224025355,
+      "loss": 0.1521,
+      "step": 4419
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025232324466359775,
+      "loss": 0.1844,
+      "step": 4420
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002523021032859335,
+      "loss": 0.1376,
+      "step": 4421
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002522809581080461,
+      "loss": 0.1478,
+      "step": 4422
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000252259809130721,
+      "loss": 0.2387,
+      "step": 4423
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002522386563547439,
+      "loss": 0.1397,
+      "step": 4424
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025221749978090053,
+      "loss": 0.0633,
+      "step": 4425
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025219633940997674,
+      "loss": 0.2911,
+      "step": 4426
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0002521751752427587,
+      "loss": 0.2087,
+      "step": 4427
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025215400728003246,
+      "loss": 0.172,
+      "step": 4428
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00025213283552258446,
+      "loss": 0.0438,
+      "step": 4429
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025211165997120113,
+      "loss": 0.2135,
+      "step": 4430
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025209048062666906,
+      "loss": 0.3009,
+      "step": 4431
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000252069297489775,
+      "loss": 0.1806,
+      "step": 4432
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002520481105613059,
+      "loss": 0.1209,
+      "step": 4433
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002520269198420487,
+      "loss": 0.2764,
+      "step": 4434
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002520057253327906,
+      "loss": 0.1753,
+      "step": 4435
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025198452703431893,
+      "loss": 0.1249,
+      "step": 4436
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025196332494742125,
+      "loss": 0.3128,
+      "step": 4437
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002519421190728849,
+      "loss": 0.1852,
+      "step": 4438
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002519209094114979,
+      "loss": 0.2082,
+      "step": 4439
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000251899695964048,
+      "loss": 0.0786,
+      "step": 4440
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002518784787313231,
+      "loss": 0.2009,
+      "step": 4441
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002518572577141115,
+      "loss": 0.1996,
+      "step": 4442
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025183603291320153,
+      "loss": 0.2742,
+      "step": 4443
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002518148043293815,
+      "loss": 0.1295,
+      "step": 4444
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002517935719634402,
+      "loss": 0.1328,
+      "step": 4445
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025177233581616604,
+      "loss": 0.0894,
+      "step": 4446
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025175109588834813,
+      "loss": 0.2079,
+      "step": 4447
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002517298521807754,
+      "loss": 0.2586,
+      "step": 4448
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002517086046942369,
+      "loss": 0.1332,
+      "step": 4449
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002516873534295221,
+      "loss": 0.0649,
+      "step": 4450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002516660983874203,
+      "loss": 0.1993,
+      "step": 4451
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000251644839568721,
+      "loss": 0.1813,
+      "step": 4452
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002516235769742141,
+      "loss": 0.0845,
+      "step": 4453
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002516023106046892,
+      "loss": 0.2498,
+      "step": 4454
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025158104046093643,
+      "loss": 0.1021,
+      "step": 4455
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002515597665437459,
+      "loss": 0.2819,
+      "step": 4456
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002515384888539079,
+      "loss": 0.2384,
+      "step": 4457
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002515172073922128,
+      "loss": 0.1955,
+      "step": 4458
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00025149592215945114,
+      "loss": 0.2392,
+      "step": 4459
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002514746331564136,
+      "loss": 0.1652,
+      "step": 4460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000251453340383891,
+      "loss": 0.2183,
+      "step": 4461
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0002514320438426743,
+      "loss": 0.1652,
+      "step": 4462
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025141074353355464,
+      "loss": 0.1744,
+      "step": 4463
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002513894394573232,
+      "loss": 0.1655,
+      "step": 4464
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025136813161477146,
+      "loss": 0.0833,
+      "step": 4465
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002513468200066909,
+      "loss": 0.1104,
+      "step": 4466
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025132550463387315,
+      "loss": 0.1719,
+      "step": 4467
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025130418549711007,
+      "loss": 0.1918,
+      "step": 4468
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002512828625971935,
+      "loss": 0.1336,
+      "step": 4469
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002512615359349156,
+      "loss": 0.0794,
+      "step": 4470
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002512402055110686,
+      "loss": 0.0685,
+      "step": 4471
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025121887132644485,
+      "loss": 0.165,
+      "step": 4472
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002511975333818369,
+      "loss": 0.2185,
+      "step": 4473
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002511761916780373,
+      "loss": 0.1996,
+      "step": 4474
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002511548462158389,
+      "loss": 0.0571,
+      "step": 4475
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025113349699603456,
+      "loss": 0.2124,
+      "step": 4476
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025111214401941744,
+      "loss": 0.1197,
+      "step": 4477
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025109078728678067,
+      "loss": 0.1872,
+      "step": 4478
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025106942679891753,
+      "loss": 0.3016,
+      "step": 4479
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002510480625566216,
+      "loss": 0.2502,
+      "step": 4480
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025102669456068653,
+      "loss": 0.1183,
+      "step": 4481
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002510053228119059,
+      "loss": 0.1569,
+      "step": 4482
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002509839473110738,
+      "loss": 0.2077,
+      "step": 4483
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025096256805898416,
+      "loss": 0.1041,
+      "step": 4484
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025094118505643125,
+      "loss": 0.2553,
+      "step": 4485
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025091979830420925,
+      "loss": 0.1364,
+      "step": 4486
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002508984078031127,
+      "loss": 0.1936,
+      "step": 4487
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025087701355393623,
+      "loss": 0.2376,
+      "step": 4488
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002508556155574745,
+      "loss": 0.1741,
+      "step": 4489
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025083421381452237,
+      "loss": 0.3077,
+      "step": 4490
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025081280832587493,
+      "loss": 0.2122,
+      "step": 4491
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025079139909232727,
+      "loss": 0.2329,
+      "step": 4492
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002507699861146747,
+      "loss": 0.0873,
+      "step": 4493
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0002507485693937127,
+      "loss": 0.2268,
+      "step": 4494
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00025072714893023675,
+      "loss": 0.3365,
+      "step": 4495
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025070572472504256,
+      "loss": 0.1961,
+      "step": 4496
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002506842967789261,
+      "loss": 0.1364,
+      "step": 4497
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025066286509268316,
+      "loss": 0.0663,
+      "step": 4498
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025064142966711007,
+      "loss": 0.1246,
+      "step": 4499
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025061999050300296,
+      "loss": 0.3263,
+      "step": 4500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002505985476011583,
+      "loss": 0.1275,
+      "step": 4501
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025057710096237256,
+      "loss": 0.2249,
+      "step": 4502
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025055565058744244,
+      "loss": 0.2208,
+      "step": 4503
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025053419647716484,
+      "loss": 0.2069,
+      "step": 4504
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002505127386323366,
+      "loss": 0.0964,
+      "step": 4505
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002504912770537549,
+      "loss": 0.1233,
+      "step": 4506
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002504698117422169,
+      "loss": 0.0977,
+      "step": 4507
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025044834269852007,
+      "loss": 0.156,
+      "step": 4508
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025042686992346187,
+      "loss": 0.1866,
+      "step": 4509
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025040539341783996,
+      "loss": 0.1176,
+      "step": 4510
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025038391318245215,
+      "loss": 0.1821,
+      "step": 4511
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025036242921809625,
+      "loss": 0.1511,
+      "step": 4512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025034094152557044,
+      "loss": 0.2218,
+      "step": 4513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025031945010567293,
+      "loss": 0.2324,
+      "step": 4514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000250297954959202,
+      "loss": 0.1676,
+      "step": 4515
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002502764560869562,
+      "loss": 0.1255,
+      "step": 4516
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002502549534897341,
+      "loss": 0.1794,
+      "step": 4517
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025023344716833447,
+      "loss": 0.2009,
+      "step": 4518
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025021193712355624,
+      "loss": 0.2092,
+      "step": 4519
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025019042335619837,
+      "loss": 0.1305,
+      "step": 4520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025016890586706007,
+      "loss": 0.2708,
+      "step": 4521
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002501473846569407,
+      "loss": 0.0325,
+      "step": 4522
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002501258597266397,
+      "loss": 0.2157,
+      "step": 4523
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002501043310769565,
+      "loss": 0.1326,
+      "step": 4524
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025008279870869103,
+      "loss": 0.2474,
+      "step": 4525
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025006126262264313,
+      "loss": 0.1245,
+      "step": 4526
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0002500397228196127,
+      "loss": 0.0861,
+      "step": 4527
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00025001817930039996,
+      "loss": 0.2007,
+      "step": 4528
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002499966320658051,
+      "loss": 0.1794,
+      "step": 4529
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002499750811166286,
+      "loss": 0.1525,
+      "step": 4530
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002499535264536711,
+      "loss": 0.1003,
+      "step": 4531
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002499319680777331,
+      "loss": 0.1705,
+      "step": 4532
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024991040598961557,
+      "loss": 0.0757,
+      "step": 4533
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002498888401901195,
+      "loss": 0.1418,
+      "step": 4534
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024986727068004584,
+      "loss": 0.231,
+      "step": 4535
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000249845697460196,
+      "loss": 0.0883,
+      "step": 4536
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002498241205313713,
+      "loss": 0.0655,
+      "step": 4537
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002498025398943732,
+      "loss": 0.2328,
+      "step": 4538
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002497809555500035,
+      "loss": 0.106,
+      "step": 4539
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024975936749906385,
+      "loss": 0.0983,
+      "step": 4540
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002497377757423562,
+      "loss": 0.1577,
+      "step": 4541
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002497161802806827,
+      "loss": 0.2543,
+      "step": 4542
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024969458111484555,
+      "loss": 0.2453,
+      "step": 4543
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000249672978245647,
+      "loss": 0.1925,
+      "step": 4544
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002496513716738896,
+      "loss": 0.0977,
+      "step": 4545
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000249629761400376,
+      "loss": 0.2412,
+      "step": 4546
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024960814742590895,
+      "loss": 0.0896,
+      "step": 4547
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002495865297512913,
+      "loss": 0.1392,
+      "step": 4548
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024956490837732613,
+      "loss": 0.1788,
+      "step": 4549
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002495432833048165,
+      "loss": 0.3039,
+      "step": 4550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024952165453456587,
+      "loss": 0.1599,
+      "step": 4551
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002495000220673776,
+      "loss": 0.0631,
+      "step": 4552
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002494783859040553,
+      "loss": 0.2225,
+      "step": 4553
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002494567460454027,
+      "loss": 0.1561,
+      "step": 4554
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024943510249222364,
+      "loss": 0.2807,
+      "step": 4555
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024941345524532205,
+      "loss": 0.0758,
+      "step": 4556
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024939180430550215,
+      "loss": 0.1559,
+      "step": 4557
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00024937014967356825,
+      "loss": 0.1936,
+      "step": 4558
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002493484913503246,
+      "loss": 0.1285,
+      "step": 4559
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002493268293365759,
+      "loss": 0.1334,
+      "step": 4560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0002493051636331267,
+      "loss": 0.3719,
+      "step": 4561
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000249283494240782,
+      "loss": 0.107,
+      "step": 4562
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002492618211603465,
+      "loss": 0.1071,
+      "step": 4563
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002492401443926255,
+      "loss": 0.1779,
+      "step": 4564
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002492184639384241,
+      "loss": 0.2345,
+      "step": 4565
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024919677979854776,
+      "loss": 0.1669,
+      "step": 4566
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024917509197380197,
+      "loss": 0.0824,
+      "step": 4567
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002491534004649923,
+      "loss": 0.1376,
+      "step": 4568
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024913170527292457,
+      "loss": 0.3177,
+      "step": 4569
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002491100063984047,
+      "loss": 0.2267,
+      "step": 4570
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002490883038422387,
+      "loss": 0.1583,
+      "step": 4571
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024906659760523287,
+      "loss": 0.1015,
+      "step": 4572
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002490448876881934,
+      "loss": 0.2095,
+      "step": 4573
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024902317409192676,
+      "loss": 0.2244,
+      "step": 4574
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024900145681723963,
+      "loss": 0.2509,
+      "step": 4575
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002489797358649386,
+      "loss": 0.1037,
+      "step": 4576
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002489580112358308,
+      "loss": 0.2451,
+      "step": 4577
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024893628293072296,
+      "loss": 0.2685,
+      "step": 4578
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024891455095042236,
+      "loss": 0.247,
+      "step": 4579
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002488928152957364,
+      "loss": 0.1864,
+      "step": 4580
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002488710759674721,
+      "loss": 0.2308,
+      "step": 4581
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024884933296643745,
+      "loss": 0.1983,
+      "step": 4582
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024882758629343987,
+      "loss": 0.1911,
+      "step": 4583
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024880583594928733,
+      "loss": 0.2323,
+      "step": 4584
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024878408193478767,
+      "loss": 0.0357,
+      "step": 4585
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002487623242507491,
+      "loss": 0.1101,
+      "step": 4586
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024874056289797975,
+      "loss": 0.1499,
+      "step": 4587
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024871879787728804,
+      "loss": 0.3232,
+      "step": 4588
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002486970291894825,
+      "loss": 0.24,
+      "step": 4589
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002486752568353718,
+      "loss": 0.2443,
+      "step": 4590
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002486534808157646,
+      "loss": 0.1499,
+      "step": 4591
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002486317011314699,
+      "loss": 0.0788,
+      "step": 4592
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002486099177832968,
+      "loss": 0.1831,
+      "step": 4593
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00024858813077205437,
+      "loss": 0.0793,
+      "step": 4594
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000248566340098552,
+      "loss": 0.1333,
+      "step": 4595
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024854454576359915,
+      "loss": 0.3416,
+      "step": 4596
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002485227477680054,
+      "loss": 0.0712,
+      "step": 4597
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024850094611258045,
+      "loss": 0.2153,
+      "step": 4598
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024847914079813424,
+      "loss": 0.1488,
+      "step": 4599
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024845733182547676,
+      "loss": 0.2401,
+      "step": 4600
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002484355191954181,
+      "loss": 0.2089,
+      "step": 4601
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002484137029087685,
+      "loss": 0.3656,
+      "step": 4602
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024839188296633854,
+      "loss": 0.1557,
+      "step": 4603
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024837005936893857,
+      "loss": 0.0458,
+      "step": 4604
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002483482321173794,
+      "loss": 0.281,
+      "step": 4605
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002483264012124718,
+      "loss": 0.1865,
+      "step": 4606
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024830456665502675,
+      "loss": 0.1382,
+      "step": 4607
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024828272844585535,
+      "loss": 0.2776,
+      "step": 4608
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024826088658576876,
+      "loss": 0.2243,
+      "step": 4609
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002482390410755784,
+      "loss": 0.1688,
+      "step": 4610
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024821719191609577,
+      "loss": 0.1907,
+      "step": 4611
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002481953391081324,
+      "loss": 0.1841,
+      "step": 4612
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002481734826525002,
+      "loss": 0.2593,
+      "step": 4613
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024815162255001095,
+      "loss": 0.2483,
+      "step": 4614
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002481297588014769,
+      "loss": 0.2913,
+      "step": 4615
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002481078914077099,
+      "loss": 0.0452,
+      "step": 4616
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024808602036952254,
+      "loss": 0.1833,
+      "step": 4617
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002480641456877271,
+      "loss": 0.242,
+      "step": 4618
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024804226736313626,
+      "loss": 0.1588,
+      "step": 4619
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024802038539656265,
+      "loss": 0.2175,
+      "step": 4620
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024799849978881925,
+      "loss": 0.2072,
+      "step": 4621
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024797661054071885,
+      "loss": 0.3147,
+      "step": 4622
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024795471765307476,
+      "loss": 0.2008,
+      "step": 4623
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002479328211267001,
+      "loss": 0.0743,
+      "step": 4624
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002479109209624084,
+      "loss": 0.1957,
+      "step": 4625
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000247889017161013,
+      "loss": 0.1676,
+      "step": 4626
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00024786710972332773,
+      "loss": 0.0548,
+      "step": 4627
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024784519865016637,
+      "loss": 0.1944,
+      "step": 4628
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024782328394234277,
+      "loss": 0.1307,
+      "step": 4629
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000247801365600671,
+      "loss": 0.2305,
+      "step": 4630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024777944362596534,
+      "loss": 0.1908,
+      "step": 4631
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024775751801904,
+      "loss": 0.2267,
+      "step": 4632
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002477355887807097,
+      "loss": 0.1953,
+      "step": 4633
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002477136559117888,
+      "loss": 0.1556,
+      "step": 4634
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002476917194130921,
+      "loss": 0.1459,
+      "step": 4635
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002476697792854345,
+      "loss": 0.211,
+      "step": 4636
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002476478355296311,
+      "loss": 0.1096,
+      "step": 4637
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024762588814649685,
+      "loss": 0.269,
+      "step": 4638
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002476039371368472,
+      "loss": 0.1464,
+      "step": 4639
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002475819825014975,
+      "loss": 0.1802,
+      "step": 4640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002475600242412633,
+      "loss": 0.1766,
+      "step": 4641
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002475380623569603,
+      "loss": 0.1388,
+      "step": 4642
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002475160968494043,
+      "loss": 0.2625,
+      "step": 4643
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002474941277194113,
+      "loss": 0.0437,
+      "step": 4644
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002474721549677973,
+      "loss": 0.1918,
+      "step": 4645
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002474501785953786,
+      "loss": 0.0012,
+      "step": 4646
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002474281986029716,
+      "loss": 0.3198,
+      "step": 4647
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002474062149913926,
+      "loss": 0.3029,
+      "step": 4648
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002473842277614584,
+      "loss": 0.3896,
+      "step": 4649
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024736223691398577,
+      "loss": 0.2223,
+      "step": 4650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002473402424497915,
+      "loss": 0.2823,
+      "step": 4651
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024731824436969273,
+      "loss": 0.165,
+      "step": 4652
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002472962426745065,
+      "loss": 0.1656,
+      "step": 4653
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002472742373650502,
+      "loss": 0.2319,
+      "step": 4654
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024725222844214123,
+      "loss": 0.1988,
+      "step": 4655
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002472302159065972,
+      "loss": 0.1534,
+      "step": 4656
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002472081997592357,
+      "loss": 0.145,
+      "step": 4657
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0002471861800008747,
+      "loss": 0.2027,
+      "step": 4658
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00024716415663233207,
+      "loss": 0.1636,
+      "step": 4659
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000247142129654426,
+      "loss": 0.2753,
+      "step": 4660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002471200990679746,
+      "loss": 0.2453,
+      "step": 4661
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002470980648737964,
+      "loss": 0.1659,
+      "step": 4662
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024707602707270984,
+      "loss": 0.0909,
+      "step": 4663
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024705398566553347,
+      "loss": 0.1746,
+      "step": 4664
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024703194065308617,
+      "loss": 0.0105,
+      "step": 4665
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024700989203618685,
+      "loss": 0.2931,
+      "step": 4666
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024698783981565446,
+      "loss": 0.1265,
+      "step": 4667
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002469657839923083,
+      "loss": 0.1212,
+      "step": 4668
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002469437245669676,
+      "loss": 0.0003,
+      "step": 4669
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002469216615404518,
+      "loss": 0.1713,
+      "step": 4670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002468995949135805,
+      "loss": 0.201,
+      "step": 4671
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024687752468717343,
+      "loss": 0.2364,
+      "step": 4672
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024685545086205037,
+      "loss": 0.3092,
+      "step": 4673
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024683337343903134,
+      "loss": 0.2201,
+      "step": 4674
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002468112924189365,
+      "loss": 0.2696,
+      "step": 4675
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000246789207802586,
+      "loss": 0.0685,
+      "step": 4676
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002467671195908002,
+      "loss": 0.2495,
+      "step": 4677
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024674502778439976,
+      "loss": 0.1893,
+      "step": 4678
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002467229323842052,
+      "loss": 0.2198,
+      "step": 4679
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024670083339103737,
+      "loss": 0.2061,
+      "step": 4680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024667873080571717,
+      "loss": 0.2331,
+      "step": 4681
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002466566246290656,
+      "loss": 0.1093,
+      "step": 4682
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024663451486190384,
+      "loss": 0.2158,
+      "step": 4683
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002466124015050533,
+      "loss": 0.0633,
+      "step": 4684
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002465902845593353,
+      "loss": 0.2031,
+      "step": 4685
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002465681640255715,
+      "loss": 0.0305,
+      "step": 4686
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002465460399045836,
+      "loss": 0.2408,
+      "step": 4687
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002465239121971934,
+      "loss": 0.2876,
+      "step": 4688
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024650178090422295,
+      "loss": 0.2108,
+      "step": 4689
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002464796460264943,
+      "loss": 0.1216,
+      "step": 4690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024645750756482974,
+      "loss": 0.2549,
+      "step": 4691
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0002464353655200516,
+      "loss": 0.2445,
+      "step": 4692
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00024641321989298247,
+      "loss": 0.1324,
+      "step": 4693
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000246391070684445,
+      "loss": 0.3039,
+      "step": 4694
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002463689178952618,
+      "loss": 0.2079,
+      "step": 4695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000246346761526256,
+      "loss": 0.0796,
+      "step": 4696
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024632460157825045,
+      "loss": 0.1637,
+      "step": 4697
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024630243805206847,
+      "loss": 0.0894,
+      "step": 4698
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024628027094853334,
+      "loss": 0.105,
+      "step": 4699
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002462581002684685,
+      "loss": 0.2078,
+      "step": 4700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024623592601269743,
+      "loss": 0.2091,
+      "step": 4701
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000246213748182044,
+      "loss": 0.1198,
+      "step": 4702
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002461915667773319,
+      "loss": 0.1615,
+      "step": 4703
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002461693817993852,
+      "loss": 0.2517,
+      "step": 4704
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000246147193249028,
+      "loss": 0.2487,
+      "step": 4705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002461250011270845,
+      "loss": 0.2083,
+      "step": 4706
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024610280543437913,
+      "loss": 0.0739,
+      "step": 4707
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002460806061717363,
+      "loss": 0.229,
+      "step": 4708
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002460584033399808,
+      "loss": 0.2723,
+      "step": 4709
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002460361969399372,
+      "loss": 0.2879,
+      "step": 4710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024601398697243055,
+      "loss": 0.1714,
+      "step": 4711
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024599177343828584,
+      "loss": 0.184,
+      "step": 4712
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002459695563383282,
+      "loss": 0.297,
+      "step": 4713
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000245947335673383,
+      "loss": 0.1436,
+      "step": 4714
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024592511144427566,
+      "loss": 0.1914,
+      "step": 4715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002459028836518317,
+      "loss": 0.1877,
+      "step": 4716
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002458806522968769,
+      "loss": 0.1877,
+      "step": 4717
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000245858417380237,
+      "loss": 0.2509,
+      "step": 4718
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024583617890273804,
+      "loss": 0.1632,
+      "step": 4719
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024581393686520596,
+      "loss": 0.1159,
+      "step": 4720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024579169126846714,
+      "loss": 0.1338,
+      "step": 4721
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024576944211334795,
+      "loss": 0.1281,
+      "step": 4722
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002457471894006748,
+      "loss": 0.1804,
+      "step": 4723
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024572493313127435,
+      "loss": 0.2675,
+      "step": 4724
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024570267330597335,
+      "loss": 0.212,
+      "step": 4725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00024568040992559864,
+      "loss": 0.1415,
+      "step": 4726
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024565814299097735,
+      "loss": 0.1386,
+      "step": 4727
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002456358725029365,
+      "loss": 0.2778,
+      "step": 4728
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002456135984623034,
+      "loss": 0.1808,
+      "step": 4729
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002455913208699056,
+      "loss": 0.0621,
+      "step": 4730
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002455690397265705,
+      "loss": 0.2178,
+      "step": 4731
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024554675503312583,
+      "loss": 0.2311,
+      "step": 4732
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002455244667903994,
+      "loss": 0.0812,
+      "step": 4733
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002455021749992192,
+      "loss": 0.1104,
+      "step": 4734
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002454798796604132,
+      "loss": 0.1513,
+      "step": 4735
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002454575807748097,
+      "loss": 0.3219,
+      "step": 4736
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002454352783432369,
+      "loss": 0.1014,
+      "step": 4737
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002454129723665235,
+      "loss": 0.218,
+      "step": 4738
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002453906628454979,
+      "loss": 0.0814,
+      "step": 4739
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024536834978098894,
+      "loss": 0.1312,
+      "step": 4740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002453460331738254,
+      "loss": 0.1668,
+      "step": 4741
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002453237130248364,
+      "loss": 0.2353,
+      "step": 4742
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000245301389334851,
+      "loss": 0.1795,
+      "step": 4743
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002452790621046984,
+      "loss": 0.1822,
+      "step": 4744
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024525673133520806,
+      "loss": 0.2078,
+      "step": 4745
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024523439702720944,
+      "loss": 0.2235,
+      "step": 4746
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002452120591815323,
+      "loss": 0.1947,
+      "step": 4747
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024518971779900636,
+      "loss": 0.0717,
+      "step": 4748
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024516737288046155,
+      "loss": 0.1344,
+      "step": 4749
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024514502442672795,
+      "loss": 0.1137,
+      "step": 4750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024512267243863566,
+      "loss": 0.3489,
+      "step": 4751
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024510031691701503,
+      "loss": 0.1912,
+      "step": 4752
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002450779578626965,
+      "loss": 0.1609,
+      "step": 4753
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002450555952765107,
+      "loss": 0.1677,
+      "step": 4754
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002450332291592882,
+      "loss": 0.1153,
+      "step": 4755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024501085951186,
+      "loss": 0.1126,
+      "step": 4756
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00024498848633505694,
+      "loss": 0.08,
+      "step": 4757
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002449661096297102,
+      "loss": 0.2,
+      "step": 4758
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0002449437293966509,
+      "loss": 0.1255,
+      "step": 4759
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002449213456367106,
+      "loss": 0.0787,
+      "step": 4760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002448989583507205,
+      "loss": 0.1561,
+      "step": 4761
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002448765675395125,
+      "loss": 0.1256,
+      "step": 4762
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024485417320391825,
+      "loss": 0.1889,
+      "step": 4763
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024483177534476955,
+      "loss": 0.1415,
+      "step": 4764
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024480937396289854,
+      "loss": 0.2955,
+      "step": 4765
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002447869690591373,
+      "loss": 0.2995,
+      "step": 4766
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002447645606343181,
+      "loss": 0.2161,
+      "step": 4767
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002447421486892734,
+      "loss": 0.1068,
+      "step": 4768
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002447197332248357,
+      "loss": 0.1811,
+      "step": 4769
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002446973142418376,
+      "loss": 0.0418,
+      "step": 4770
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024467489174111206,
+      "loss": 0.1224,
+      "step": 4771
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024465246572349193,
+      "loss": 0.2822,
+      "step": 4772
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024463003618981023,
+      "loss": 0.0779,
+      "step": 4773
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024460760314090016,
+      "loss": 0.3276,
+      "step": 4774
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024458516657759516,
+      "loss": 0.2812,
+      "step": 4775
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002445627265007285,
+      "loss": 0.1717,
+      "step": 4776
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002445402829111339,
+      "loss": 0.031,
+      "step": 4777
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000244517835809645,
+      "loss": 0.1756,
+      "step": 4778
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024449538519709564,
+      "loss": 0.2788,
+      "step": 4779
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002444729310743199,
+      "loss": 0.1941,
+      "step": 4780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024445047344215175,
+      "loss": 0.127,
+      "step": 4781
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002444280123014255,
+      "loss": 0.1348,
+      "step": 4782
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024440554765297554,
+      "loss": 0.1381,
+      "step": 4783
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024438307949763627,
+      "loss": 0.2509,
+      "step": 4784
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024436060783624244,
+      "loss": 0.1676,
+      "step": 4785
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002443381326696287,
+      "loss": 0.2935,
+      "step": 4786
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024431565399862993,
+      "loss": 0.1558,
+      "step": 4787
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024429317182408125,
+      "loss": 0.2759,
+      "step": 4788
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024427068614681774,
+      "loss": 0.1641,
+      "step": 4789
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00024424819696767467,
+      "loss": 0.2955,
+      "step": 4790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002442257042874875,
+      "loss": 0.0562,
+      "step": 4791
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002442032081070917,
+      "loss": 0.2184,
+      "step": 4792
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000244180708427323,
+      "loss": 0.1941,
+      "step": 4793
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002441582052490171,
+      "loss": 0.1586,
+      "step": 4794
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024413569857301005,
+      "loss": 0.1463,
+      "step": 4795
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024411318840013784,
+      "loss": 0.045,
+      "step": 4796
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024409067473123662,
+      "loss": 0.1648,
+      "step": 4797
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002440681575671428,
+      "loss": 0.1849,
+      "step": 4798
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024404563690869278,
+      "loss": 0.0893,
+      "step": 4799
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024402311275672313,
+      "loss": 0.188,
+      "step": 4800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024400058511207055,
+      "loss": 0.1182,
+      "step": 4801
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002439780539755719,
+      "loss": 0.1262,
+      "step": 4802
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024395551934806414,
+      "loss": 0.0496,
+      "step": 4803
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002439329812303843,
+      "loss": 0.1644,
+      "step": 4804
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024391043962336973,
+      "loss": 0.2357,
+      "step": 4805
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024388789452785766,
+      "loss": 0.2399,
+      "step": 4806
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002438653459446857,
+      "loss": 0.1819,
+      "step": 4807
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024384279387469135,
+      "loss": 0.2445,
+      "step": 4808
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002438202383187124,
+      "loss": 0.0869,
+      "step": 4809
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024379767927758673,
+      "loss": 0.2523,
+      "step": 4810
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024377511675215236,
+      "loss": 0.108,
+      "step": 4811
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024375255074324736,
+      "loss": 0.1193,
+      "step": 4812
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024372998125171005,
+      "loss": 0.0545,
+      "step": 4813
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024370740827837881,
+      "loss": 0.2476,
+      "step": 4814
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002436848318240921,
+      "loss": 0.243,
+      "step": 4815
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024366225188968864,
+      "loss": 0.1162,
+      "step": 4816
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002436396684760072,
+      "loss": 0.1327,
+      "step": 4817
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024361708158388666,
+      "loss": 0.1892,
+      "step": 4818
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024359449121416602,
+      "loss": 0.2925,
+      "step": 4819
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024357189736768457,
+      "loss": 0.0239,
+      "step": 4820
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002435493000452815,
+      "loss": 0.2373,
+      "step": 4821
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002435266992477963,
+      "loss": 0.2658,
+      "step": 4822
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002435040949760684,
+      "loss": 0.1807,
+      "step": 4823
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00024348148723093765,
+      "loss": 0.2083,
+      "step": 4824
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024345887601324374,
+      "loss": 0.2167,
+      "step": 4825
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002434362613238267,
+      "loss": 0.263,
+      "step": 4826
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024341364316352647,
+      "loss": 0.1788,
+      "step": 4827
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002433910215331834,
+      "loss": 0.28,
+      "step": 4828
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024336839643363773,
+      "loss": 0.2452,
+      "step": 4829
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002433457678657299,
+      "loss": 0.1462,
+      "step": 4830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024332313583030058,
+      "loss": 0.1457,
+      "step": 4831
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024330050032819045,
+      "loss": 0.2081,
+      "step": 4832
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002432778613602403,
+      "loss": 0.1631,
+      "step": 4833
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002432552189272911,
+      "loss": 0.219,
+      "step": 4834
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024323257303018403,
+      "loss": 0.1134,
+      "step": 4835
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024320992366976023,
+      "loss": 0.1245,
+      "step": 4836
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024318727084686115,
+      "loss": 0.0891,
+      "step": 4837
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024316461456232817,
+      "loss": 0.1657,
+      "step": 4838
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024314195481700302,
+      "loss": 0.2089,
+      "step": 4839
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024311929161172733,
+      "loss": 0.2128,
+      "step": 4840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002430966249473431,
+      "loss": 0.2262,
+      "step": 4841
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024307395482469222,
+      "loss": 0.3254,
+      "step": 4842
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024305128124461683,
+      "loss": 0.1541,
+      "step": 4843
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024302860420795925,
+      "loss": 0.213,
+      "step": 4844
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002430059237155618,
+      "loss": 0.2919,
+      "step": 4845
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024298323976826706,
+      "loss": 0.1862,
+      "step": 4846
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002429605523669176,
+      "loss": 0.1482,
+      "step": 4847
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024293786151235623,
+      "loss": 0.2905,
+      "step": 4848
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024291516720542587,
+      "loss": 0.3043,
+      "step": 4849
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024289246944696945,
+      "loss": 0.1277,
+      "step": 4850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024286976823783028,
+      "loss": 0.1878,
+      "step": 4851
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024284706357885152,
+      "loss": 0.1704,
+      "step": 4852
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024282435547087658,
+      "loss": 0.1191,
+      "step": 4853
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024280164391474909,
+      "loss": 0.0757,
+      "step": 4854
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024277892891131264,
+      "loss": 0.2662,
+      "step": 4855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024275621046141108,
+      "loss": 0.1093,
+      "step": 4856
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00024273348856588832,
+      "loss": 0.1464,
+      "step": 4857
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024271076322558835,
+      "loss": 0.1478,
+      "step": 4858
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002426880344413554,
+      "loss": 0.2105,
+      "step": 4859
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002426653022140338,
+      "loss": 0.0678,
+      "step": 4860
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000242642566544468,
+      "loss": 0.1581,
+      "step": 4861
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024261982743350248,
+      "loss": 0.0523,
+      "step": 4862
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024259708488198197,
+      "loss": 0.27,
+      "step": 4863
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002425743388907513,
+      "loss": 0.2369,
+      "step": 4864
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002425515894606554,
+      "loss": 0.1578,
+      "step": 4865
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024252883659253938,
+      "loss": 0.1166,
+      "step": 4866
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002425060802872484,
+      "loss": 0.2162,
+      "step": 4867
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024248332054562784,
+      "loss": 0.1946,
+      "step": 4868
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002424605573685231,
+      "loss": 0.0479,
+      "step": 4869
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024243779075677982,
+      "loss": 0.1106,
+      "step": 4870
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024241502071124365,
+      "loss": 0.1466,
+      "step": 4871
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024239224723276046,
+      "loss": 0.0656,
+      "step": 4872
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024236947032217626,
+      "loss": 0.1563,
+      "step": 4873
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002423466899803371,
+      "loss": 0.1884,
+      "step": 4874
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002423239062080892,
+      "loss": 0.0719,
+      "step": 4875
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024230111900627892,
+      "loss": 0.1252,
+      "step": 4876
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002422783283757528,
+      "loss": 0.2118,
+      "step": 4877
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002422555343173573,
+      "loss": 0.1459,
+      "step": 4878
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002422327368319393,
+      "loss": 0.1147,
+      "step": 4879
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024220993592034552,
+      "loss": 0.2229,
+      "step": 4880
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024218713158342308,
+      "loss": 0.2909,
+      "step": 4881
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024216432382201904,
+      "loss": 0.2582,
+      "step": 4882
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024214151263698063,
+      "loss": 0.1231,
+      "step": 4883
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002421186980291552,
+      "loss": 0.1089,
+      "step": 4884
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002420958799993903,
+      "loss": 0.2527,
+      "step": 4885
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024207305854853355,
+      "loss": 0.1949,
+      "step": 4886
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024205023367743263,
+      "loss": 0.1793,
+      "step": 4887
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002420274053869355,
+      "loss": 0.2496,
+      "step": 4888
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00024200457367789011,
+      "loss": 0.1955,
+      "step": 4889
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0002419817385511446,
+      "loss": 0.1759,
+      "step": 4890
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024195890000754724,
+      "loss": 0.2314,
+      "step": 4891
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024193605804794646,
+      "loss": 0.1863,
+      "step": 4892
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024191321267319065,
+      "loss": 0.0854,
+      "step": 4893
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002418903638841286,
+      "loss": 0.1258,
+      "step": 4894
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000241867511681609,
+      "loss": 0.0963,
+      "step": 4895
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002418446560664806,
+      "loss": 0.1528,
+      "step": 4896
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024182179703959272,
+      "loss": 0.2272,
+      "step": 4897
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024179893460179428,
+      "loss": 0.1176,
+      "step": 4898
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002417760687539346,
+      "loss": 0.0736,
+      "step": 4899
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002417531994968631,
+      "loss": 0.1079,
+      "step": 4900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024173032683142933,
+      "loss": 0.202,
+      "step": 4901
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024170745075848292,
+      "loss": 0.2651,
+      "step": 4902
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024168457127887364,
+      "loss": 0.1272,
+      "step": 4903
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002416616883934514,
+      "loss": 0.2007,
+      "step": 4904
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024163880210306622,
+      "loss": 0.1778,
+      "step": 4905
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024161591240856825,
+      "loss": 0.3051,
+      "step": 4906
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024159301931080782,
+      "loss": 0.0878,
+      "step": 4907
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002415701228106353,
+      "loss": 0.1748,
+      "step": 4908
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024154722290890127,
+      "loss": 0.0855,
+      "step": 4909
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024152431960645636,
+      "loss": 0.1633,
+      "step": 4910
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024150141290415136,
+      "loss": 0.0559,
+      "step": 4911
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002414785028028372,
+      "loss": 0.1372,
+      "step": 4912
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024145558930336493,
+      "loss": 0.1148,
+      "step": 4913
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024143267240658568,
+      "loss": 0.1413,
+      "step": 4914
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024140975211335084,
+      "loss": 0.1947,
+      "step": 4915
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002413868284245117,
+      "loss": 0.0605,
+      "step": 4916
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002413639013409199,
+      "loss": 0.2568,
+      "step": 4917
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002413409708634271,
+      "loss": 0.161,
+      "step": 4918
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024131803699288504,
+      "loss": 0.1572,
+      "step": 4919
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024129509973014576,
+      "loss": 0.2067,
+      "step": 4920
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002412721590760612,
+      "loss": 0.172,
+      "step": 4921
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024124921503148358,
+      "loss": 0.0394,
+      "step": 4922
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00024122626759726524,
+      "loss": 0.2853,
+      "step": 4923
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024120331677425855,
+      "loss": 0.2808,
+      "step": 4924
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024118036256331609,
+      "loss": 0.1654,
+      "step": 4925
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024115740496529056,
+      "loss": 0.2078,
+      "step": 4926
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024113444398103474,
+      "loss": 0.1198,
+      "step": 4927
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002411114796114016,
+      "loss": 0.1519,
+      "step": 4928
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024108851185724415,
+      "loss": 0.1121,
+      "step": 4929
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002410655407194156,
+      "loss": 0.1119,
+      "step": 4930
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024104256619876928,
+      "loss": 0.1005,
+      "step": 4931
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002410195882961586,
+      "loss": 0.2238,
+      "step": 4932
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002409966070124371,
+      "loss": 0.2841,
+      "step": 4933
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024097362234845853,
+      "loss": 0.0564,
+      "step": 4934
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024095063430507667,
+      "loss": 0.1973,
+      "step": 4935
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024092764288314545,
+      "loss": 0.2603,
+      "step": 4936
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002409046480835189,
+      "loss": 0.2984,
+      "step": 4937
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024088164990705127,
+      "loss": 0.2175,
+      "step": 4938
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002408586483545969,
+      "loss": 0.2486,
+      "step": 4939
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024083564342701014,
+      "loss": 0.1566,
+      "step": 4940
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002408126351251456,
+      "loss": 0.2282,
+      "step": 4941
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024078962344985797,
+      "loss": 0.2482,
+      "step": 4942
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002407666084020021,
+      "loss": 0.2571,
+      "step": 4943
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024074358998243283,
+      "loss": 0.2135,
+      "step": 4944
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024072056819200537,
+      "loss": 0.1883,
+      "step": 4945
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024069754303157481,
+      "loss": 0.2808,
+      "step": 4946
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024067451450199647,
+      "loss": 0.2682,
+      "step": 4947
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024065148260412586,
+      "loss": 0.1381,
+      "step": 4948
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002406284473388185,
+      "loss": 0.1786,
+      "step": 4949
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024060540870693005,
+      "loss": 0.1161,
+      "step": 4950
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024058236670931638,
+      "loss": 0.242,
+      "step": 4951
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024055932134683345,
+      "loss": 0.166,
+      "step": 4952
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002405362726203373,
+      "loss": 0.1391,
+      "step": 4953
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024051322053068403,
+      "loss": 0.1153,
+      "step": 4954
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002404901650787301,
+      "loss": 0.2139,
+      "step": 4955
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00024046710626533192,
+      "loss": 0.1491,
+      "step": 4956
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000240444044091346,
+      "loss": 0.2523,
+      "step": 4957
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024042097855762908,
+      "loss": 0.0984,
+      "step": 4958
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024039790966503795,
+      "loss": 0.1191,
+      "step": 4959
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002403748374144296,
+      "loss": 0.2819,
+      "step": 4960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024035176180666105,
+      "loss": 0.1829,
+      "step": 4961
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024032868284258949,
+      "loss": 0.2079,
+      "step": 4962
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024030560052307225,
+      "loss": 0.1978,
+      "step": 4963
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024028251484896676,
+      "loss": 0.214,
+      "step": 4964
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024025942582113065,
+      "loss": 0.149,
+      "step": 4965
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024023633344042147,
+      "loss": 0.1049,
+      "step": 4966
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002402132377076972,
+      "loss": 0.1477,
+      "step": 4967
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024019013862381568,
+      "loss": 0.2492,
+      "step": 4968
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002401670361896349,
+      "loss": 0.2476,
+      "step": 4969
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024014393040601323,
+      "loss": 0.1432,
+      "step": 4970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024012082127380886,
+      "loss": 0.2337,
+      "step": 4971
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002400977087938803,
+      "loss": 0.1171,
+      "step": 4972
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024007459296708606,
+      "loss": 0.1095,
+      "step": 4973
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024005147379428475,
+      "loss": 0.1287,
+      "step": 4974
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024002835127633532,
+      "loss": 0.1915,
+      "step": 4975
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00024000522541409668,
+      "loss": 0.0828,
+      "step": 4976
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002399820962084278,
+      "loss": 0.1927,
+      "step": 4977
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002399589636601879,
+      "loss": 0.0768,
+      "step": 4978
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002399358277702364,
+      "loss": 0.172,
+      "step": 4979
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023991268853943256,
+      "loss": 0.2276,
+      "step": 4980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023988954596863596,
+      "loss": 0.2071,
+      "step": 4981
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023986640005870641,
+      "loss": 0.2261,
+      "step": 4982
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002398432508105036,
+      "loss": 0.2018,
+      "step": 4983
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002398200982248875,
+      "loss": 0.2071,
+      "step": 4984
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0002397969423027181,
+      "loss": 0.304,
+      "step": 4985
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023977378304485566,
+      "loss": 0.1309,
+      "step": 4986
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023975062045216046,
+      "loss": 0.2296,
+      "step": 4987
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023972745452549284,
+      "loss": 0.1748,
+      "step": 4988
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00023970428526571346,
+      "loss": 0.2961,
+      "step": 4989
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002396811126736829,
+      "loss": 0.2186,
+      "step": 4990
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023965793675026205,
+      "loss": 0.1569,
+      "step": 4991
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023963475749631173,
+      "loss": 0.0596,
+      "step": 4992
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023961157491269303,
+      "loss": 0.1762,
+      "step": 4993
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023958838900026708,
+      "loss": 0.0936,
+      "step": 4994
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023956519975989524,
+      "loss": 0.1699,
+      "step": 4995
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023954200719243884,
+      "loss": 0.2865,
+      "step": 4996
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023951881129875946,
+      "loss": 0.1818,
+      "step": 4997
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023949561207971875,
+      "loss": 0.2496,
+      "step": 4998
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002394724095361785,
+      "loss": 0.2167,
+      "step": 4999
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023944920366900061,
+      "loss": 0.0765,
+      "step": 5000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023942599447904713,
+      "loss": 0.245,
+      "step": 5001
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023940278196718016,
+      "loss": 0.1438,
+      "step": 5002
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023937956613426201,
+      "loss": 0.2264,
+      "step": 5003
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002393563469811551,
+      "loss": 0.2475,
+      "step": 5004
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023933312450872196,
+      "loss": 0.1016,
+      "step": 5005
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002393098987178252,
+      "loss": 0.2269,
+      "step": 5006
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023928666960932755,
+      "loss": 0.153,
+      "step": 5007
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023926343718409202,
+      "loss": 0.0776,
+      "step": 5008
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023924020144298154,
+      "loss": 0.1575,
+      "step": 5009
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023921696238685925,
+      "loss": 0.2811,
+      "step": 5010
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023919372001658846,
+      "loss": 0.2035,
+      "step": 5011
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023917047433303254,
+      "loss": 0.3496,
+      "step": 5012
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023914722533705494,
+      "loss": 0.2451,
+      "step": 5013
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002391239730295194,
+      "loss": 0.2767,
+      "step": 5014
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023910071741128954,
+      "loss": 0.1617,
+      "step": 5015
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023907745848322938,
+      "loss": 0.1488,
+      "step": 5016
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023905419624620285,
+      "loss": 0.2379,
+      "step": 5017
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023903093070107404,
+      "loss": 0.2984,
+      "step": 5018
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023900766184870725,
+      "loss": 0.1839,
+      "step": 5019
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023898438968996685,
+      "loss": 0.3119,
+      "step": 5020
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00023896111422571734,
+      "loss": 0.2784,
+      "step": 5021
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002389378354568233,
+      "loss": 0.1766,
+      "step": 5022
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023891455338414941,
+      "loss": 0.238,
+      "step": 5023
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023889126800856066,
+      "loss": 0.2001,
+      "step": 5024
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000238867979330922,
+      "loss": 0.1802,
+      "step": 5025
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023884468735209846,
+      "loss": 0.2057,
+      "step": 5026
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023882139207295536,
+      "loss": 0.136,
+      "step": 5027
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023879809349435798,
+      "loss": 0.2949,
+      "step": 5028
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023877479161717183,
+      "loss": 0.1228,
+      "step": 5029
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023875148644226255,
+      "loss": 0.1453,
+      "step": 5030
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023872817797049572,
+      "loss": 0.1593,
+      "step": 5031
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023870486620273737,
+      "loss": 0.2175,
+      "step": 5032
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002386815511398533,
+      "loss": 0.0797,
+      "step": 5033
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023865823278270966,
+      "loss": 0.1547,
+      "step": 5034
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023863491113217269,
+      "loss": 0.2597,
+      "step": 5035
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023861158618910868,
+      "loss": 0.1262,
+      "step": 5036
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002385882579543841,
+      "loss": 0.1672,
+      "step": 5037
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002385649264288655,
+      "loss": 0.1981,
+      "step": 5038
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023854159161341962,
+      "loss": 0.1412,
+      "step": 5039
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023851825350891323,
+      "loss": 0.185,
+      "step": 5040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002384949121162133,
+      "loss": 0.2069,
+      "step": 5041
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023847156743618693,
+      "loss": 0.1512,
+      "step": 5042
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023844821946970122,
+      "loss": 0.2143,
+      "step": 5043
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023842486821762357,
+      "loss": 0.1733,
+      "step": 5044
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023840151368082139,
+      "loss": 0.1687,
+      "step": 5045
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002383781558601622,
+      "loss": 0.1792,
+      "step": 5046
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023835479475651367,
+      "loss": 0.1464,
+      "step": 5047
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002383314303707436,
+      "loss": 0.2294,
+      "step": 5048
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002383080627037199,
+      "loss": 0.0703,
+      "step": 5049
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023828469175631072,
+      "loss": 0.2708,
+      "step": 5050
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023826131752938405,
+      "loss": 0.1855,
+      "step": 5051
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023823794002380828,
+      "loss": 0.2047,
+      "step": 5052
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023821455924045182,
+      "loss": 0.1857,
+      "step": 5053
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023819117518018317,
+      "loss": 0.2015,
+      "step": 5054
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00023816778784387094,
+      "loss": 0.1357,
+      "step": 5055
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023814439723238398,
+      "loss": 0.1997,
+      "step": 5056
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023812100334659114,
+      "loss": 0.1263,
+      "step": 5057
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023809760618736142,
+      "loss": 0.0634,
+      "step": 5058
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000238074205755564,
+      "loss": 0.0923,
+      "step": 5059
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002380508020520681,
+      "loss": 0.289,
+      "step": 5060
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023802739507774307,
+      "loss": 0.0592,
+      "step": 5061
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002380039848334585,
+      "loss": 0.2292,
+      "step": 5062
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023798057132008393,
+      "loss": 0.1386,
+      "step": 5063
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023795715453848914,
+      "loss": 0.0387,
+      "step": 5064
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023793373448954401,
+      "loss": 0.1399,
+      "step": 5065
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023791031117411854,
+      "loss": 0.1914,
+      "step": 5066
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002378868845930828,
+      "loss": 0.1998,
+      "step": 5067
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023786345474730692,
+      "loss": 0.1886,
+      "step": 5068
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023784002163766143,
+      "loss": 0.2196,
+      "step": 5069
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023781658526501675,
+      "loss": 0.2483,
+      "step": 5070
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002377931456302434,
+      "loss": 0.1262,
+      "step": 5071
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023776970273421216,
+      "loss": 0.1282,
+      "step": 5072
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023774625657779387,
+      "loss": 0.2021,
+      "step": 5073
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002377228071618594,
+      "loss": 0.1006,
+      "step": 5074
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023769935448727996,
+      "loss": 0.0727,
+      "step": 5075
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002376758985549266,
+      "loss": 0.2057,
+      "step": 5076
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023765243936567078,
+      "loss": 0.0388,
+      "step": 5077
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023762897692038387,
+      "loss": 0.1072,
+      "step": 5078
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002376055112199374,
+      "loss": 0.1219,
+      "step": 5079
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023758204226520315,
+      "loss": 0.065,
+      "step": 5080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023755857005705282,
+      "loss": 0.0591,
+      "step": 5081
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023753509459635834,
+      "loss": 0.2352,
+      "step": 5082
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023751161588399185,
+      "loss": 0.1683,
+      "step": 5083
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002374881339208254,
+      "loss": 0.1731,
+      "step": 5084
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023746464870773144,
+      "loss": 0.2199,
+      "step": 5085
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023744116024558215,
+      "loss": 0.2127,
+      "step": 5086
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0002374176685352502,
+      "loss": 0.3046,
+      "step": 5087
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00023739417357760827,
+      "loss": 0.1148,
+      "step": 5088
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000237370675373529,
+      "loss": 0.0875,
+      "step": 5089
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023734717392388541,
+      "loss": 0.2635,
+      "step": 5090
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023732366922955044,
+      "loss": 0.0394,
+      "step": 5091
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002373001612913972,
+      "loss": 0.1275,
+      "step": 5092
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023727665011029903,
+      "loss": 0.1585,
+      "step": 5093
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002372531356871292,
+      "loss": 0.1909,
+      "step": 5094
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023722961802276126,
+      "loss": 0.1861,
+      "step": 5095
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002372060971180688,
+      "loss": 0.0451,
+      "step": 5096
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023718257297392556,
+      "loss": 0.1527,
+      "step": 5097
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023715904559120545,
+      "loss": 0.1409,
+      "step": 5098
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023713551497078232,
+      "loss": 0.0657,
+      "step": 5099
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023711198111353037,
+      "loss": 0.2353,
+      "step": 5100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023708844402032375,
+      "loss": 0.1664,
+      "step": 5101
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002370649036920368,
+      "loss": 0.3046,
+      "step": 5102
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002370413601295441,
+      "loss": 0.1967,
+      "step": 5103
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023701781333372002,
+      "loss": 0.2013,
+      "step": 5104
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023699426330543933,
+      "loss": 0.0777,
+      "step": 5105
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023697071004557692,
+      "loss": 0.25,
+      "step": 5106
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023694715355500768,
+      "loss": 0.1802,
+      "step": 5107
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023692359383460665,
+      "loss": 0.1548,
+      "step": 5108
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000236900030885249,
+      "loss": 0.3857,
+      "step": 5109
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023687646470781,
+      "loss": 0.2961,
+      "step": 5110
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023685289530316517,
+      "loss": 0.2265,
+      "step": 5111
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002368293226721899,
+      "loss": 0.1199,
+      "step": 5112
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023680574681575998,
+      "loss": 0.3478,
+      "step": 5113
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002367821677347511,
+      "loss": 0.1874,
+      "step": 5114
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023675858543003916,
+      "loss": 0.1617,
+      "step": 5115
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002367349999025002,
+      "loss": 0.2252,
+      "step": 5116
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023671141115301034,
+      "loss": 0.2811,
+      "step": 5117
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023668781918244585,
+      "loss": 0.0342,
+      "step": 5118
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00023666422399168307,
+      "loss": 0.1806,
+      "step": 5119
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002366406255815985,
+      "loss": 0.2179,
+      "step": 5120
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0002366170239530688,
+      "loss": 0.2683,
+      "step": 5121
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023659341910697062,
+      "loss": 0.1873,
+      "step": 5122
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023656981104418088,
+      "loss": 0.2122,
+      "step": 5123
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002365461997655765,
+      "loss": 0.2207,
+      "step": 5124
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023652258527203458,
+      "loss": 0.2362,
+      "step": 5125
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002364989675644324,
+      "loss": 0.1312,
+      "step": 5126
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002364753466436472,
+      "loss": 0.1584,
+      "step": 5127
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023645172251055645,
+      "loss": 0.17,
+      "step": 5128
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023642809516603776,
+      "loss": 0.1733,
+      "step": 5129
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002364044646109687,
+      "loss": 0.2235,
+      "step": 5130
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023638083084622728,
+      "loss": 0.2873,
+      "step": 5131
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002363571938726912,
+      "loss": 0.1828,
+      "step": 5132
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023633355369123867,
+      "loss": 0.1316,
+      "step": 5133
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023630991030274776,
+      "loss": 0.1998,
+      "step": 5134
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002362862637080968,
+      "loss": 0.0698,
+      "step": 5135
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023626261390816418,
+      "loss": 0.1407,
+      "step": 5136
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002362389609038284,
+      "loss": 0.1295,
+      "step": 5137
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002362153046959681,
+      "loss": 0.2411,
+      "step": 5138
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023619164528546207,
+      "loss": 0.1412,
+      "step": 5139
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002361679826731892,
+      "loss": 0.1354,
+      "step": 5140
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023614431686002846,
+      "loss": 0.2557,
+      "step": 5141
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002361206478468589,
+      "loss": 0.1248,
+      "step": 5142
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023609697563455986,
+      "loss": 0.2325,
+      "step": 5143
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002360733002240107,
+      "loss": 0.1865,
+      "step": 5144
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002360496216160908,
+      "loss": 0.1871,
+      "step": 5145
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023602593981167982,
+      "loss": 0.1527,
+      "step": 5146
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002360022548116574,
+      "loss": 0.2027,
+      "step": 5147
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023597856661690343,
+      "loss": 0.2909,
+      "step": 5148
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023595487522829785,
+      "loss": 0.0884,
+      "step": 5149
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002359311806467207,
+      "loss": 0.1674,
+      "step": 5150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002359074828730522,
+      "loss": 0.2346,
+      "step": 5151
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002358837819081726,
+      "loss": 0.1567,
+      "step": 5152
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002358600777529624,
+      "loss": 0.2658,
+      "step": 5153
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00023583637040830205,
+      "loss": 0.1875,
+      "step": 5154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023581265987507224,
+      "loss": 0.2446,
+      "step": 5155
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023578894615415384,
+      "loss": 0.2497,
+      "step": 5156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002357652292464276,
+      "loss": 0.0495,
+      "step": 5157
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002357415091527746,
+      "loss": 0.2376,
+      "step": 5158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000235717785874076,
+      "loss": 0.1314,
+      "step": 5159
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000235694059411213,
+      "loss": 0.1556,
+      "step": 5160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023567032976506698,
+      "loss": 0.3135,
+      "step": 5161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023564659693651945,
+      "loss": 0.2418,
+      "step": 5162
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000235622860926452,
+      "loss": 0.1955,
+      "step": 5163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002355991217357464,
+      "loss": 0.1771,
+      "step": 5164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023557537936528438,
+      "loss": 0.1426,
+      "step": 5165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023555163381594794,
+      "loss": 0.2492,
+      "step": 5166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023552788508861925,
+      "loss": 0.2686,
+      "step": 5167
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002355041331841804,
+      "loss": 0.244,
+      "step": 5168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002354803781035138,
+      "loss": 0.2981,
+      "step": 5169
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023545661984750174,
+      "loss": 0.1789,
+      "step": 5170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002354328584170269,
+      "loss": 0.2373,
+      "step": 5171
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023540909381297189,
+      "loss": 0.2449,
+      "step": 5172
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002353853260362195,
+      "loss": 0.1628,
+      "step": 5173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002353615550876526,
+      "loss": 0.1879,
+      "step": 5174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002353377809681543,
+      "loss": 0.0188,
+      "step": 5175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002353140036786077,
+      "loss": 0.0523,
+      "step": 5176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023529022321989602,
+      "loss": 0.0973,
+      "step": 5177
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023526643959290264,
+      "loss": 0.2983,
+      "step": 5178
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002352426527985111,
+      "loss": 0.2471,
+      "step": 5179
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000235218862837605,
+      "loss": 0.1444,
+      "step": 5180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000235195069711068,
+      "loss": 0.0865,
+      "step": 5181
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023517127341978402,
+      "loss": 0.1156,
+      "step": 5182
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000235147473964637,
+      "loss": 0.2391,
+      "step": 5183
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000235123671346511,
+      "loss": 0.1601,
+      "step": 5184
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00023509986556629023,
+      "loss": 0.1615,
+      "step": 5185
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000235076056624859,
+      "loss": 0.1859,
+      "step": 5186
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002350522445231018,
+      "loss": 0.0989,
+      "step": 5187
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023502842926190304,
+      "loss": 0.1609,
+      "step": 5188
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023500461084214756,
+      "loss": 0.1585,
+      "step": 5189
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023498078926472,
+      "loss": 0.1263,
+      "step": 5190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023495696453050536,
+      "loss": 0.1606,
+      "step": 5191
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023493313664038864,
+      "loss": 0.1483,
+      "step": 5192
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002349093055952549,
+      "loss": 0.3372,
+      "step": 5193
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002348854713959895,
+      "loss": 0.1067,
+      "step": 5194
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023486163404347774,
+      "loss": 0.1232,
+      "step": 5195
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023483779353860513,
+      "loss": 0.2353,
+      "step": 5196
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002348139498822573,
+      "loss": 0.2022,
+      "step": 5197
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023479010307531992,
+      "loss": 0.3067,
+      "step": 5198
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023476625311867882,
+      "loss": 0.1425,
+      "step": 5199
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023474240001322005,
+      "loss": 0.1639,
+      "step": 5200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023471854375982967,
+      "loss": 0.0642,
+      "step": 5201
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023469468435939375,
+      "loss": 0.1281,
+      "step": 5202
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023467082181279867,
+      "loss": 0.2481,
+      "step": 5203
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023464695612093086,
+      "loss": 0.2231,
+      "step": 5204
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002346230872846769,
+      "loss": 0.1351,
+      "step": 5205
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023459921530492333,
+      "loss": 0.1821,
+      "step": 5206
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023457534018255706,
+      "loss": 0.2734,
+      "step": 5207
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023455146191846487,
+      "loss": 0.1711,
+      "step": 5208
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002345275805135338,
+      "loss": 0.1194,
+      "step": 5209
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023450369596865106,
+      "loss": 0.1934,
+      "step": 5210
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023447980828470376,
+      "loss": 0.0135,
+      "step": 5211
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002344559174625794,
+      "loss": 0.2411,
+      "step": 5212
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002344320235031653,
+      "loss": 0.1751,
+      "step": 5213
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023440812640734913,
+      "loss": 0.2132,
+      "step": 5214
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002343842261760186,
+      "loss": 0.2038,
+      "step": 5215
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0002343603228100615,
+      "loss": 0.1714,
+      "step": 5216
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023433641631036585,
+      "loss": 0.1407,
+      "step": 5217
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023431250667781958,
+      "loss": 0.1532,
+      "step": 5218
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023428859391331095,
+      "loss": 0.2543,
+      "step": 5219
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00023426467801772826,
+      "loss": 0.1301,
+      "step": 5220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023424075899195984,
+      "loss": 0.2581,
+      "step": 5221
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002342168368368943,
+      "loss": 0.1401,
+      "step": 5222
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023419291155342019,
+      "loss": 0.1674,
+      "step": 5223
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023416898314242637,
+      "loss": 0.1811,
+      "step": 5224
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002341450516048016,
+      "loss": 0.1314,
+      "step": 5225
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023412111694143492,
+      "loss": 0.2705,
+      "step": 5226
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002340971791532155,
+      "loss": 0.1528,
+      "step": 5227
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023407323824103238,
+      "loss": 0.2879,
+      "step": 5228
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023404929420577504,
+      "loss": 0.2027,
+      "step": 5229
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023402534704833293,
+      "loss": 0.178,
+      "step": 5230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023400139676959554,
+      "loss": 0.1783,
+      "step": 5231
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002339774433704526,
+      "loss": 0.1422,
+      "step": 5232
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023395348685179392,
+      "loss": 0.2072,
+      "step": 5233
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023392952721450937,
+      "loss": 0.1984,
+      "step": 5234
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023390556445948904,
+      "loss": 0.2829,
+      "step": 5235
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000233881598587623,
+      "loss": 0.213,
+      "step": 5236
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023385762959980154,
+      "loss": 0.2701,
+      "step": 5237
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023383365749691514,
+      "loss": 0.2015,
+      "step": 5238
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023380968227985408,
+      "loss": 0.1391,
+      "step": 5239
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002337857039495092,
+      "loss": 0.1557,
+      "step": 5240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002337617225067711,
+      "loss": 0.2871,
+      "step": 5241
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023373773795253057,
+      "loss": 0.319,
+      "step": 5242
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002337137502876787,
+      "loss": 0.1167,
+      "step": 5243
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002336897595131065,
+      "loss": 0.1444,
+      "step": 5244
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002336657656297051,
+      "loss": 0.1879,
+      "step": 5245
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023364176863836588,
+      "loss": 0.2903,
+      "step": 5246
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023361776853998023,
+      "loss": 0.1307,
+      "step": 5247
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023359376533543967,
+      "loss": 0.1924,
+      "step": 5248
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002335697590256359,
+      "loss": 0.0748,
+      "step": 5249
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023354574961146065,
+      "loss": 0.0964,
+      "step": 5250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023352173709380577,
+      "loss": 0.1264,
+      "step": 5251
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023349772147356328,
+      "loss": 0.1918,
+      "step": 5252
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00023347370275162533,
+      "loss": 0.1046,
+      "step": 5253
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023344968092888405,
+      "loss": 0.2906,
+      "step": 5254
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002334256560062319,
+      "loss": 0.1945,
+      "step": 5255
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002334016279845612,
+      "loss": 0.1806,
+      "step": 5256
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023337759686476464,
+      "loss": 0.214,
+      "step": 5257
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023335356264773488,
+      "loss": 0.1464,
+      "step": 5258
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023332952533436466,
+      "loss": 0.0836,
+      "step": 5259
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023330548492554696,
+      "loss": 0.2864,
+      "step": 5260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023328144142217483,
+      "loss": 0.1875,
+      "step": 5261
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002332573948251413,
+      "loss": 0.3533,
+      "step": 5262
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002332333451353397,
+      "loss": 0.1492,
+      "step": 5263
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023320929235366342,
+      "loss": 0.0532,
+      "step": 5264
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023318523648100596,
+      "loss": 0.1438,
+      "step": 5265
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002331611775182609,
+      "loss": 0.2041,
+      "step": 5266
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023313711546632196,
+      "loss": 0.1008,
+      "step": 5267
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023311305032608297,
+      "loss": 0.2681,
+      "step": 5268
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002330889820984379,
+      "loss": 0.1793,
+      "step": 5269
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002330649107842808,
+      "loss": 0.2443,
+      "step": 5270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023304083638450582,
+      "loss": 0.0368,
+      "step": 5271
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002330167589000073,
+      "loss": 0.0445,
+      "step": 5272
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023299267833167968,
+      "loss": 0.214,
+      "step": 5273
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023296859468041736,
+      "loss": 0.2037,
+      "step": 5274
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023294450794711507,
+      "loss": 0.1704,
+      "step": 5275
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023292041813266755,
+      "loss": 0.1731,
+      "step": 5276
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023289632523796968,
+      "loss": 0.2301,
+      "step": 5277
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002328722292639164,
+      "loss": 0.2958,
+      "step": 5278
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023284813021140282,
+      "loss": 0.1397,
+      "step": 5279
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023282402808132415,
+      "loss": 0.1707,
+      "step": 5280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023279992287457573,
+      "loss": 0.1606,
+      "step": 5281
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023277581459205297,
+      "loss": 0.0472,
+      "step": 5282
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023275170323465145,
+      "loss": 0.161,
+      "step": 5283
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023272758880326681,
+      "loss": 0.2294,
+      "step": 5284
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023270347129879487,
+      "loss": 0.1491,
+      "step": 5285
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00023267935072213147,
+      "loss": 0.2489,
+      "step": 5286
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023265522707417266,
+      "loss": 0.3624,
+      "step": 5287
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023263110035581455,
+      "loss": 0.1972,
+      "step": 5288
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023260697056795335,
+      "loss": 0.1642,
+      "step": 5289
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023258283771148547,
+      "loss": 0.1297,
+      "step": 5290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023255870178730734,
+      "loss": 0.1705,
+      "step": 5291
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002325345627963155,
+      "loss": 0.065,
+      "step": 5292
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023251042073940667,
+      "loss": 0.1461,
+      "step": 5293
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023248627561747777,
+      "loss": 0.0694,
+      "step": 5294
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002324621274314255,
+      "loss": 0.0739,
+      "step": 5295
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023243797618214713,
+      "loss": 0.1937,
+      "step": 5296
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023241382187053961,
+      "loss": 0.1993,
+      "step": 5297
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002323896644975003,
+      "loss": 0.0703,
+      "step": 5298
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023236550406392656,
+      "loss": 0.0961,
+      "step": 5299
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023234134057071587,
+      "loss": 0.1755,
+      "step": 5300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023231717401876587,
+      "loss": 0.2523,
+      "step": 5301
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023229300440897425,
+      "loss": 0.2397,
+      "step": 5302
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023226883174223882,
+      "loss": 0.1704,
+      "step": 5303
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023224465601945753,
+      "loss": 0.1975,
+      "step": 5304
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002322204772415285,
+      "loss": 0.2197,
+      "step": 5305
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002321962954093498,
+      "loss": 0.196,
+      "step": 5306
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023217211052381978,
+      "loss": 0.1307,
+      "step": 5307
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023214792258583679,
+      "loss": 0.197,
+      "step": 5308
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023212373159629942,
+      "loss": 0.3414,
+      "step": 5309
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023209953755610622,
+      "loss": 0.221,
+      "step": 5310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000232075340466156,
+      "loss": 0.2398,
+      "step": 5311
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002320511403273475,
+      "loss": 0.1026,
+      "step": 5312
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002320269371405798,
+      "loss": 0.2256,
+      "step": 5313
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002320027309067519,
+      "loss": 0.0585,
+      "step": 5314
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023197852162676306,
+      "loss": 0.1671,
+      "step": 5315
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023195430930151247,
+      "loss": 0.1523,
+      "step": 5316
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00023193009393189968,
+      "loss": 0.1844,
+      "step": 5317
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002319058755188241,
+      "loss": 0.237,
+      "step": 5318
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002318816540631855,
+      "loss": 0.2037,
+      "step": 5319
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023185742956588354,
+      "loss": 0.288,
+      "step": 5320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002318332020278181,
+      "loss": 0.1655,
+      "step": 5321
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002318089714498892,
+      "loss": 0.2336,
+      "step": 5322
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002317847378329969,
+      "loss": 0.0607,
+      "step": 5323
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002317605011780414,
+      "loss": 0.1321,
+      "step": 5324
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023173626148592305,
+      "loss": 0.1516,
+      "step": 5325
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002317120187575423,
+      "loss": 0.1435,
+      "step": 5326
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023168777299379967,
+      "loss": 0.2422,
+      "step": 5327
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002316635241955958,
+      "loss": 0.1954,
+      "step": 5328
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023163927236383153,
+      "loss": 0.1268,
+      "step": 5329
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023161501749940764,
+      "loss": 0.1842,
+      "step": 5330
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023159075960322518,
+      "loss": 0.1777,
+      "step": 5331
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002315664986761853,
+      "loss": 0.1329,
+      "step": 5332
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023154223471918915,
+      "loss": 0.271,
+      "step": 5333
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023151796773313816,
+      "loss": 0.1516,
+      "step": 5334
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023149369771893372,
+      "loss": 0.1383,
+      "step": 5335
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002314694246774773,
+      "loss": 0.0993,
+      "step": 5336
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023144514860967073,
+      "loss": 0.2177,
+      "step": 5337
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023142086951641571,
+      "loss": 0.2265,
+      "step": 5338
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023139658739861416,
+      "loss": 0.2654,
+      "step": 5339
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023137230225716806,
+      "loss": 0.2804,
+      "step": 5340
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023134801409297959,
+      "loss": 0.1506,
+      "step": 5341
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002313237229069509,
+      "loss": 0.1795,
+      "step": 5342
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002312994286999844,
+      "loss": 0.2252,
+      "step": 5343
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023127513147298256,
+      "loss": 0.1698,
+      "step": 5344
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023125083122684792,
+      "loss": 0.1044,
+      "step": 5345
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002312265279624831,
+      "loss": 0.1911,
+      "step": 5346
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023120222168079103,
+      "loss": 0.3269,
+      "step": 5347
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002311779123826745,
+      "loss": 0.2422,
+      "step": 5348
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023115360006903664,
+      "loss": 0.2011,
+      "step": 5349
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00023112928474078046,
+      "loss": 0.1085,
+      "step": 5350
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0002311049663988093,
+      "loss": 0.2305,
+      "step": 5351
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023108064504402646,
+      "loss": 0.1898,
+      "step": 5352
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023105632067733543,
+      "loss": 0.216,
+      "step": 5353
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023103199329963976,
+      "loss": 0.1489,
+      "step": 5354
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023100766291184318,
+      "loss": 0.1888,
+      "step": 5355
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023098332951484947,
+      "loss": 0.3072,
+      "step": 5356
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023095899310956257,
+      "loss": 0.1496,
+      "step": 5357
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023093465369688647,
+      "loss": 0.2232,
+      "step": 5358
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023091031127772533,
+      "loss": 0.1596,
+      "step": 5359
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023088596585298343,
+      "loss": 0.155,
+      "step": 5360
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000230861617423565,
+      "loss": 0.0705,
+      "step": 5361
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002308372659903747,
+      "loss": 0.2638,
+      "step": 5362
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023081291155431695,
+      "loss": 0.121,
+      "step": 5363
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023078855411629655,
+      "loss": 0.253,
+      "step": 5364
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002307641936772183,
+      "loss": 0.1411,
+      "step": 5365
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023073983023798704,
+      "loss": 0.2953,
+      "step": 5366
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002307154637995079,
+      "loss": 0.1674,
+      "step": 5367
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023069109436268595,
+      "loss": 0.2748,
+      "step": 5368
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002306667219284265,
+      "loss": 0.213,
+      "step": 5369
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002306423464976349,
+      "loss": 0.2681,
+      "step": 5370
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023061796807121652,
+      "loss": 0.1329,
+      "step": 5371
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002305935866500771,
+      "loss": 0.1266,
+      "step": 5372
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023056920223512228,
+      "loss": 0.1553,
+      "step": 5373
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023054481482725784,
+      "loss": 0.2031,
+      "step": 5374
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023052042442738973,
+      "loss": 0.1137,
+      "step": 5375
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023049603103642398,
+      "loss": 0.1303,
+      "step": 5376
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023047163465526676,
+      "loss": 0.2593,
+      "step": 5377
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023044723528482425,
+      "loss": 0.1185,
+      "step": 5378
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023042283292600287,
+      "loss": 0.1413,
+      "step": 5379
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023039842757970906,
+      "loss": 0.0738,
+      "step": 5380
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023037401924684946,
+      "loss": 0.1574,
+      "step": 5381
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023034960792833075,
+      "loss": 0.1416,
+      "step": 5382
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00023032519362505975,
+      "loss": 0.1774,
+      "step": 5383
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002303007763379433,
+      "loss": 0.1759,
+      "step": 5384
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023027635606788854,
+      "loss": 0.2708,
+      "step": 5385
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023025193281580252,
+      "loss": 0.2315,
+      "step": 5386
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023022750658259253,
+      "loss": 0.1745,
+      "step": 5387
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023020307736916596,
+      "loss": 0.227,
+      "step": 5388
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023017864517643026,
+      "loss": 0.2067,
+      "step": 5389
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000230154210005293,
+      "loss": 0.1899,
+      "step": 5390
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002301297718566619,
+      "loss": 0.1882,
+      "step": 5391
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023010533073144476,
+      "loss": 0.215,
+      "step": 5392
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023008088663054946,
+      "loss": 0.0865,
+      "step": 5393
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023005643955488408,
+      "loss": 0.1702,
+      "step": 5394
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023003198950535676,
+      "loss": 0.1815,
+      "step": 5395
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00023000753648287566,
+      "loss": 0.2258,
+      "step": 5396
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022998308048834923,
+      "loss": 0.1958,
+      "step": 5397
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022995862152268593,
+      "loss": 0.1585,
+      "step": 5398
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022993415958679427,
+      "loss": 0.1778,
+      "step": 5399
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022990969468158302,
+      "loss": 0.1255,
+      "step": 5400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022988522680796096,
+      "loss": 0.184,
+      "step": 5401
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022986075596683698,
+      "loss": 0.0975,
+      "step": 5402
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002298362821591201,
+      "loss": 0.171,
+      "step": 5403
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022981180538571945,
+      "loss": 0.2191,
+      "step": 5404
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022978732564754424,
+      "loss": 0.1046,
+      "step": 5405
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022976284294550387,
+      "loss": 0.1917,
+      "step": 5406
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022973835728050783,
+      "loss": 0.2924,
+      "step": 5407
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022971386865346563,
+      "loss": 0.1816,
+      "step": 5408
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022968937706528694,
+      "loss": 0.2419,
+      "step": 5409
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022966488251688162,
+      "loss": 0.1828,
+      "step": 5410
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002296403850091595,
+      "loss": 0.0825,
+      "step": 5411
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022961588454303066,
+      "loss": 0.1165,
+      "step": 5412
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022959138111940512,
+      "loss": 0.0548,
+      "step": 5413
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002295668747391932,
+      "loss": 0.1963,
+      "step": 5414
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022954236540330525,
+      "loss": 0.1051,
+      "step": 5415
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022951785311265164,
+      "loss": 0.1832,
+      "step": 5416
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00022949333786814296,
+      "loss": 0.1912,
+      "step": 5417
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022946881967068992,
+      "loss": 0.2017,
+      "step": 5418
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022944429852120326,
+      "loss": 0.1537,
+      "step": 5419
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022941977442059386,
+      "loss": 0.1892,
+      "step": 5420
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022939524736977277,
+      "loss": 0.1447,
+      "step": 5421
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022937071736965105,
+      "loss": 0.2137,
+      "step": 5422
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022934618442113994,
+      "loss": 0.1935,
+      "step": 5423
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022932164852515076,
+      "loss": 0.1043,
+      "step": 5424
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002292971096825949,
+      "loss": 0.1822,
+      "step": 5425
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000229272567894384,
+      "loss": 0.3027,
+      "step": 5426
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022924802316142968,
+      "loss": 0.1693,
+      "step": 5427
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022922347548464365,
+      "loss": 0.1392,
+      "step": 5428
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022919892486493782,
+      "loss": 0.1333,
+      "step": 5429
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002291743713032242,
+      "loss": 0.1488,
+      "step": 5430
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022914981480041488,
+      "loss": 0.1532,
+      "step": 5431
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022912525535742203,
+      "loss": 0.1652,
+      "step": 5432
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000229100692975158,
+      "loss": 0.173,
+      "step": 5433
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022907612765453518,
+      "loss": 0.0571,
+      "step": 5434
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022905155939646608,
+      "loss": 0.1208,
+      "step": 5435
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022902698820186337,
+      "loss": 0.039,
+      "step": 5436
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022900241407163978,
+      "loss": 0.1962,
+      "step": 5437
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022897783700670822,
+      "loss": 0.1832,
+      "step": 5438
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002289532570079816,
+      "loss": 0.1943,
+      "step": 5439
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000228928674076373,
+      "loss": 0.1142,
+      "step": 5440
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022890408821279565,
+      "loss": 0.2505,
+      "step": 5441
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022887949941816278,
+      "loss": 0.1559,
+      "step": 5442
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022885490769338783,
+      "loss": 0.1074,
+      "step": 5443
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022883031303938428,
+      "loss": 0.0705,
+      "step": 5444
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022880571545706577,
+      "loss": 0.1852,
+      "step": 5445
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022878111494734602,
+      "loss": 0.22,
+      "step": 5446
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022875651151113888,
+      "loss": 0.2817,
+      "step": 5447
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0002287319051493583,
+      "loss": 0.1371,
+      "step": 5448
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022870729586291832,
+      "loss": 0.1621,
+      "step": 5449
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00022868268365273307,
+      "loss": 0.17,
+      "step": 5450
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022865806851971694,
+      "loss": 0.1525,
+      "step": 5451
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022863345046478419,
+      "loss": 0.2471,
+      "step": 5452
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022860882948884932,
+      "loss": 0.0706,
+      "step": 5453
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022858420559282696,
+      "loss": 0.1443,
+      "step": 5454
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002285595787776318,
+      "loss": 0.19,
+      "step": 5455
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022853494904417868,
+      "loss": 0.1061,
+      "step": 5456
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022851031639338251,
+      "loss": 0.0992,
+      "step": 5457
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022848568082615831,
+      "loss": 0.1572,
+      "step": 5458
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022846104234342123,
+      "loss": 0.2101,
+      "step": 5459
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022843640094608652,
+      "loss": 0.2091,
+      "step": 5460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022841175663506946,
+      "loss": 0.2354,
+      "step": 5461
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022838710941128568,
+      "loss": 0.3019,
+      "step": 5462
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002283624592756506,
+      "loss": 0.2564,
+      "step": 5463
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022833780622907995,
+      "loss": 0.3279,
+      "step": 5464
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022831315027248956,
+      "loss": 0.1277,
+      "step": 5465
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022828849140679527,
+      "loss": 0.1443,
+      "step": 5466
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022826382963291307,
+      "loss": 0.317,
+      "step": 5467
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022823916495175913,
+      "loss": 0.193,
+      "step": 5468
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022821449736424963,
+      "loss": 0.1554,
+      "step": 5469
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022818982687130096,
+      "loss": 0.1505,
+      "step": 5470
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022816515347382952,
+      "loss": 0.179,
+      "step": 5471
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022814047717275177,
+      "loss": 0.1434,
+      "step": 5472
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002281157979689845,
+      "loss": 0.1027,
+      "step": 5473
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002280911158634444,
+      "loss": 0.0922,
+      "step": 5474
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022806643085704833,
+      "loss": 0.2323,
+      "step": 5475
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002280417429507133,
+      "loss": 0.2096,
+      "step": 5476
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022801705214535632,
+      "loss": 0.1692,
+      "step": 5477
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022799235844189465,
+      "loss": 0.1307,
+      "step": 5478
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022796766184124563,
+      "loss": 0.1327,
+      "step": 5479
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022794296234432656,
+      "loss": 0.1446,
+      "step": 5480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022791825995205505,
+      "loss": 0.2114,
+      "step": 5481
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00022789355466534865,
+      "loss": 0.2348,
+      "step": 5482
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002278688464851251,
+      "loss": 0.1768,
+      "step": 5483
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022784413541230233,
+      "loss": 0.1451,
+      "step": 5484
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002278194214477981,
+      "loss": 0.2417,
+      "step": 5485
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002277947045925306,
+      "loss": 0.3072,
+      "step": 5486
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022776998484741798,
+      "loss": 0.2184,
+      "step": 5487
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022774526221337846,
+      "loss": 0.1253,
+      "step": 5488
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022772053669133045,
+      "loss": 0.1839,
+      "step": 5489
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022769580828219243,
+      "loss": 0.2429,
+      "step": 5490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022767107698688297,
+      "loss": 0.1177,
+      "step": 5491
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022764634280632073,
+      "loss": 0.188,
+      "step": 5492
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022762160574142458,
+      "loss": 0.2874,
+      "step": 5493
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002275968657931134,
+      "loss": 0.2323,
+      "step": 5494
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022757212296230619,
+      "loss": 0.2153,
+      "step": 5495
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022754737724992204,
+      "loss": 0.3515,
+      "step": 5496
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022752262865688034,
+      "loss": 0.1379,
+      "step": 5497
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022749787718410024,
+      "loss": 0.2434,
+      "step": 5498
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022747312283250134,
+      "loss": 0.2472,
+      "step": 5499
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022744836560300303,
+      "loss": 0.1718,
+      "step": 5500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002274236054965251,
+      "loss": 0.2945,
+      "step": 5501
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022739884251398728,
+      "loss": 0.1162,
+      "step": 5502
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002273740766563094,
+      "loss": 0.1724,
+      "step": 5503
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022734930792441148,
+      "loss": 0.2083,
+      "step": 5504
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002273245363192136,
+      "loss": 0.1233,
+      "step": 5505
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002272997618416359,
+      "loss": 0.0848,
+      "step": 5506
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002272749844925988,
+      "loss": 0.1178,
+      "step": 5507
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022725020427302262,
+      "loss": 0.2878,
+      "step": 5508
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022722542118382784,
+      "loss": 0.2094,
+      "step": 5509
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022720063522593516,
+      "loss": 0.1794,
+      "step": 5510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022717584640026526,
+      "loss": 0.1792,
+      "step": 5511
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.000227151054707739,
+      "loss": 0.1802,
+      "step": 5512
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022712626014927728,
+      "loss": 0.1388,
+      "step": 5513
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022710146272580118,
+      "loss": 0.0617,
+      "step": 5514
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022707666243823184,
+      "loss": 0.2702,
+      "step": 5515
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00022705185928749048,
+      "loss": 0.1959,
+      "step": 5516
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022702705327449856,
+      "loss": 0.152,
+      "step": 5517
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022700224440017747,
+      "loss": 0.252,
+      "step": 5518
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022697743266544875,
+      "loss": 0.1385,
+      "step": 5519
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002269526180712342,
+      "loss": 0.1902,
+      "step": 5520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022692780061845554,
+      "loss": 0.2582,
+      "step": 5521
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022690298030803466,
+      "loss": 0.2341,
+      "step": 5522
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022687815714089358,
+      "loss": 0.2845,
+      "step": 5523
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022685333111795446,
+      "loss": 0.162,
+      "step": 5524
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022682850224013941,
+      "loss": 0.3112,
+      "step": 5525
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022680367050837083,
+      "loss": 0.23,
+      "step": 5526
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022677883592357108,
+      "loss": 0.2386,
+      "step": 5527
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022675399848666268,
+      "loss": 0.2431,
+      "step": 5528
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002267291581985684,
+      "loss": 0.1023,
+      "step": 5529
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022670431506021087,
+      "loss": 0.2842,
+      "step": 5530
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022667946907251299,
+      "loss": 0.0973,
+      "step": 5531
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002266546202363977,
+      "loss": 0.3037,
+      "step": 5532
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000226629768552788,
+      "loss": 0.098,
+      "step": 5533
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022660491402260715,
+      "loss": 0.1653,
+      "step": 5534
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022658005664677838,
+      "loss": 0.2562,
+      "step": 5535
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022655519642622506,
+      "loss": 0.0628,
+      "step": 5536
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022653033336187073,
+      "loss": 0.2001,
+      "step": 5537
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002265054674546389,
+      "loss": 0.3274,
+      "step": 5538
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022648059870545334,
+      "loss": 0.147,
+      "step": 5539
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002264557271152378,
+      "loss": 0.2276,
+      "step": 5540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002264308526849162,
+      "loss": 0.2087,
+      "step": 5541
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022640597541541253,
+      "loss": 0.2986,
+      "step": 5542
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022638109530765093,
+      "loss": 0.1003,
+      "step": 5543
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022635621236255567,
+      "loss": 0.1924,
+      "step": 5544
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022633132658105105,
+      "loss": 0.1579,
+      "step": 5545
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022630643796406143,
+      "loss": 0.1522,
+      "step": 5546
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022628154651251143,
+      "loss": 0.1121,
+      "step": 5547
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00022625665222732565,
+      "loss": 0.1988,
+      "step": 5548
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002262317551094289,
+      "loss": 0.1535,
+      "step": 5549
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022620685515974598,
+      "loss": 0.2663,
+      "step": 5550
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022618195237920184,
+      "loss": 0.1757,
+      "step": 5551
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022615704676872157,
+      "loss": 0.1129,
+      "step": 5552
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022613213832923038,
+      "loss": 0.357,
+      "step": 5553
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022610722706165344,
+      "loss": 0.2189,
+      "step": 5554
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022608231296691628,
+      "loss": 0.0342,
+      "step": 5555
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022605739604594427,
+      "loss": 0.1809,
+      "step": 5556
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022603247629966302,
+      "loss": 0.224,
+      "step": 5557
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002260075537289983,
+      "loss": 0.0821,
+      "step": 5558
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022598262833487575,
+      "loss": 0.1556,
+      "step": 5559
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022595770011822145,
+      "loss": 0.1681,
+      "step": 5560
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022593276907996126,
+      "loss": 0.2349,
+      "step": 5561
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022590783522102143,
+      "loss": 0.229,
+      "step": 5562
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022588289854232814,
+      "loss": 0.0549,
+      "step": 5563
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022585795904480764,
+      "loss": 0.2592,
+      "step": 5564
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022583301672938645,
+      "loss": 0.1757,
+      "step": 5565
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0002258080715969911,
+      "loss": 0.1048,
+      "step": 5566
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022578312364854812,
+      "loss": 0.224,
+      "step": 5567
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022575817288498435,
+      "loss": 0.2521,
+      "step": 5568
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022573321930722656,
+      "loss": 0.0911,
+      "step": 5569
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022570826291620186,
+      "loss": 0.1162,
+      "step": 5570
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022568330371283717,
+      "loss": 0.1765,
+      "step": 5571
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022565834169805969,
+      "loss": 0.1026,
+      "step": 5572
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022563337687279666,
+      "loss": 0.2524,
+      "step": 5573
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022560840923797548,
+      "loss": 0.1537,
+      "step": 5574
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022558343879452365,
+      "loss": 0.1657,
+      "step": 5575
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022555846554336866,
+      "loss": 0.1063,
+      "step": 5576
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022553348948543825,
+      "loss": 0.2179,
+      "step": 5577
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022550851062166027,
+      "loss": 0.1639,
+      "step": 5578
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022548352895296252,
+      "loss": 0.1409,
+      "step": 5579
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022545854448027298,
+      "loss": 0.1295,
+      "step": 5580
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022543355720451986,
+      "loss": 0.1275,
+      "step": 5581
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00022540856712663124,
+      "loss": 0.1626,
+      "step": 5582
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022538357424753552,
+      "loss": 0.2818,
+      "step": 5583
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022535857856816104,
+      "loss": 0.1829,
+      "step": 5584
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022533358008943637,
+      "loss": 0.2435,
+      "step": 5585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022530857881229016,
+      "loss": 0.2539,
+      "step": 5586
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022528357473765103,
+      "loss": 0.2051,
+      "step": 5587
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002252585678664479,
+      "loss": 0.1013,
+      "step": 5588
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022523355819960964,
+      "loss": 0.1725,
+      "step": 5589
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022520854573806535,
+      "loss": 0.0164,
+      "step": 5590
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002251835304827441,
+      "loss": 0.1434,
+      "step": 5591
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022515851243457515,
+      "loss": 0.1873,
+      "step": 5592
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002251334915944879,
+      "loss": 0.1677,
+      "step": 5593
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022510846796341177,
+      "loss": 0.1181,
+      "step": 5594
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022508344154227627,
+      "loss": 0.2315,
+      "step": 5595
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022505841233201116,
+      "loss": 0.2881,
+      "step": 5596
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022503338033354604,
+      "loss": 0.2503,
+      "step": 5597
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000225008345547811,
+      "loss": 0.2352,
+      "step": 5598
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022498330797573579,
+      "loss": 0.104,
+      "step": 5599
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002249582676182506,
+      "loss": 0.194,
+      "step": 5600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022493322447628556,
+      "loss": 0.2312,
+      "step": 5601
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022490817855077102,
+      "loss": 0.0721,
+      "step": 5602
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002248831298426373,
+      "loss": 0.3291,
+      "step": 5603
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022485807835281489,
+      "loss": 0.1244,
+      "step": 5604
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022483302408223436,
+      "loss": 0.1643,
+      "step": 5605
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022480796703182646,
+      "loss": 0.1954,
+      "step": 5606
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022478290720252196,
+      "loss": 0.1557,
+      "step": 5607
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022475784459525172,
+      "loss": 0.144,
+      "step": 5608
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022473277921094683,
+      "loss": 0.2848,
+      "step": 5609
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022470771105053832,
+      "loss": 0.1626,
+      "step": 5610
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022468264011495743,
+      "loss": 0.1027,
+      "step": 5611
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022465756640513545,
+      "loss": 0.1864,
+      "step": 5612
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002246324899220038,
+      "loss": 0.1997,
+      "step": 5613
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022460741066649407,
+      "loss": 0.1676,
+      "step": 5614
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00022458232863953776,
+      "loss": 0.2615,
+      "step": 5615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022455724384206664,
+      "loss": 0.0773,
+      "step": 5616
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022453215627501257,
+      "loss": 0.1755,
+      "step": 5617
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022450706593930746,
+      "loss": 0.1659,
+      "step": 5618
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022448197283588336,
+      "loss": 0.1982,
+      "step": 5619
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022445687696567236,
+      "loss": 0.1512,
+      "step": 5620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022443177832960677,
+      "loss": 0.12,
+      "step": 5621
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022440667692861884,
+      "loss": 0.1743,
+      "step": 5622
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022438157276364107,
+      "loss": 0.1985,
+      "step": 5623
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022435646583560597,
+      "loss": 0.2644,
+      "step": 5624
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022433135614544623,
+      "loss": 0.1592,
+      "step": 5625
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002243062436940946,
+      "loss": 0.1527,
+      "step": 5626
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002242811284824839,
+      "loss": 0.1334,
+      "step": 5627
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002242560105115471,
+      "loss": 0.1007,
+      "step": 5628
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002242308897822173,
+      "loss": 0.0726,
+      "step": 5629
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002242057662954276,
+      "loss": 0.1442,
+      "step": 5630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002241806400521113,
+      "loss": 0.1229,
+      "step": 5631
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022415551105320174,
+      "loss": 0.0973,
+      "step": 5632
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022413037929963238,
+      "loss": 0.1055,
+      "step": 5633
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022410524479233684,
+      "loss": 0.1201,
+      "step": 5634
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002240801075322488,
+      "loss": 0.1595,
+      "step": 5635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022405496752030192,
+      "loss": 0.0657,
+      "step": 5636
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022402982475743023,
+      "loss": 0.1303,
+      "step": 5637
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022400467924456763,
+      "loss": 0.1389,
+      "step": 5638
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022397953098264825,
+      "loss": 0.3142,
+      "step": 5639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022395437997260617,
+      "loss": 0.2238,
+      "step": 5640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022392922621537577,
+      "loss": 0.1279,
+      "step": 5641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002239040697118914,
+      "loss": 0.2859,
+      "step": 5642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022387891046308755,
+      "loss": 0.1215,
+      "step": 5643
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022385374846989887,
+      "loss": 0.107,
+      "step": 5644
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022382858373325997,
+      "loss": 0.2416,
+      "step": 5645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022380341625410569,
+      "loss": 0.1045,
+      "step": 5646
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00022377824603337095,
+      "loss": 0.0686,
+      "step": 5647
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002237530730719907,
+      "loss": 0.334,
+      "step": 5648
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022372789737090006,
+      "loss": 0.1753,
+      "step": 5649
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022370271893103425,
+      "loss": 0.1634,
+      "step": 5650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022367753775332853,
+      "loss": 0.1206,
+      "step": 5651
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002236523538387184,
+      "loss": 0.1903,
+      "step": 5652
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002236271671881393,
+      "loss": 0.0468,
+      "step": 5653
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022360197780252683,
+      "loss": 0.238,
+      "step": 5654
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022357678568281667,
+      "loss": 0.1713,
+      "step": 5655
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022355159082994476,
+      "loss": 0.1206,
+      "step": 5656
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002235263932448469,
+      "loss": 0.1094,
+      "step": 5657
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022350119292845912,
+      "loss": 0.0366,
+      "step": 5658
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022347598988171763,
+      "loss": 0.137,
+      "step": 5659
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022345078410555857,
+      "loss": 0.1322,
+      "step": 5660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022342557560091824,
+      "loss": 0.2271,
+      "step": 5661
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002234003643687331,
+      "loss": 0.2333,
+      "step": 5662
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002233751504099397,
+      "loss": 0.0265,
+      "step": 5663
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022334993372547462,
+      "loss": 0.0879,
+      "step": 5664
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022332471431627457,
+      "loss": 0.1389,
+      "step": 5665
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022329949218327638,
+      "loss": 0.1582,
+      "step": 5666
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002232742673274171,
+      "loss": 0.2132,
+      "step": 5667
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022324903974963361,
+      "loss": 0.1325,
+      "step": 5668
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022322380945086307,
+      "loss": 0.1228,
+      "step": 5669
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0002231985764320428,
+      "loss": 0.0741,
+      "step": 5670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022317334069411004,
+      "loss": 0.1901,
+      "step": 5671
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022314810223800224,
+      "loss": 0.2569,
+      "step": 5672
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022312286106465696,
+      "loss": 0.2421,
+      "step": 5673
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022309761717501176,
+      "loss": 0.2047,
+      "step": 5674
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022307237057000453,
+      "loss": 0.1843,
+      "step": 5675
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022304712125057296,
+      "loss": 0.1487,
+      "step": 5676
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022302186921765508,
+      "loss": 0.2109,
+      "step": 5677
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022299661447218892,
+      "loss": 0.1302,
+      "step": 5678
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022297135701511252,
+      "loss": 0.1671,
+      "step": 5679
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022294609684736424,
+      "loss": 0.123,
+      "step": 5680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00022292083396988237,
+      "loss": 0.0747,
+      "step": 5681
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022289556838360532,
+      "loss": 0.2025,
+      "step": 5682
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022287030008947168,
+      "loss": 0.1124,
+      "step": 5683
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002228450290884201,
+      "loss": 0.2547,
+      "step": 5684
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022281975538138932,
+      "loss": 0.1366,
+      "step": 5685
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022279447896931815,
+      "loss": 0.2608,
+      "step": 5686
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022276919985314558,
+      "loss": 0.1872,
+      "step": 5687
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022274391803381058,
+      "loss": 0.2987,
+      "step": 5688
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022271863351225237,
+      "loss": 0.2127,
+      "step": 5689
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022269334628941017,
+      "loss": 0.2432,
+      "step": 5690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022266805636622332,
+      "loss": 0.0165,
+      "step": 5691
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022264276374363133,
+      "loss": 0.2366,
+      "step": 5692
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002226174684225736,
+      "loss": 0.1724,
+      "step": 5693
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002225921704039899,
+      "loss": 0.1177,
+      "step": 5694
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022256686968881998,
+      "loss": 0.2357,
+      "step": 5695
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002225415662780036,
+      "loss": 0.0829,
+      "step": 5696
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022251626017248078,
+      "loss": 0.132,
+      "step": 5697
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022249095137319156,
+      "loss": 0.2055,
+      "step": 5698
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022246563988107606,
+      "loss": 0.1504,
+      "step": 5699
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022244032569707458,
+      "loss": 0.1899,
+      "step": 5700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022241500882212739,
+      "loss": 0.1835,
+      "step": 5701
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022238968925717496,
+      "loss": 0.1529,
+      "step": 5702
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022236436700315789,
+      "loss": 0.0917,
+      "step": 5703
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002223390420610168,
+      "loss": 0.1244,
+      "step": 5704
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022231371443169243,
+      "loss": 0.0672,
+      "step": 5705
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022228838411612558,
+      "loss": 0.1729,
+      "step": 5706
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022226305111525726,
+      "loss": 0.3468,
+      "step": 5707
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022223771543002853,
+      "loss": 0.1895,
+      "step": 5708
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002222123770613805,
+      "loss": 0.1342,
+      "step": 5709
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022218703601025442,
+      "loss": 0.1795,
+      "step": 5710
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002221616922775916,
+      "loss": 0.2486,
+      "step": 5711
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022213634586433357,
+      "loss": 0.1358,
+      "step": 5712
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00022211099677142183,
+      "loss": 0.1062,
+      "step": 5713
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000222085644999798,
+      "loss": 0.1281,
+      "step": 5714
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022206029055040388,
+      "loss": 0.1494,
+      "step": 5715
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022203493342418126,
+      "loss": 0.1772,
+      "step": 5716
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022200957362207208,
+      "loss": 0.0848,
+      "step": 5717
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022198421114501846,
+      "loss": 0.1503,
+      "step": 5718
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022195884599396246,
+      "loss": 0.1618,
+      "step": 5719
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022193347816984633,
+      "loss": 0.161,
+      "step": 5720
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022190810767361248,
+      "loss": 0.1217,
+      "step": 5721
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022188273450620324,
+      "loss": 0.0522,
+      "step": 5722
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022185735866856121,
+      "loss": 0.0955,
+      "step": 5723
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022183198016162902,
+      "loss": 0.1529,
+      "step": 5724
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002218065989863495,
+      "loss": 0.084,
+      "step": 5725
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022178121514366533,
+      "loss": 0.2024,
+      "step": 5726
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002217558286345195,
+      "loss": 0.2164,
+      "step": 5727
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022173043945985512,
+      "loss": 0.2277,
+      "step": 5728
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002217050476206152,
+      "loss": 0.2605,
+      "step": 5729
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022167965311774304,
+      "loss": 0.0805,
+      "step": 5730
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000221654255952182,
+      "loss": 0.2349,
+      "step": 5731
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022162885612487542,
+      "loss": 0.1288,
+      "step": 5732
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022160345363676697,
+      "loss": 0.1827,
+      "step": 5733
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022157804848880013,
+      "loss": 0.1814,
+      "step": 5734
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002215526406819187,
+      "loss": 0.2622,
+      "step": 5735
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022152723021706653,
+      "loss": 0.1571,
+      "step": 5736
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022150181709518744,
+      "loss": 0.089,
+      "step": 5737
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022147640131722556,
+      "loss": 0.1592,
+      "step": 5738
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000221450982884125,
+      "loss": 0.3107,
+      "step": 5739
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002214255617968299,
+      "loss": 0.1476,
+      "step": 5740
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022140013805628468,
+      "loss": 0.1726,
+      "step": 5741
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022137471166343368,
+      "loss": 0.1851,
+      "step": 5742
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0002213492826192215,
+      "loss": 0.0926,
+      "step": 5743
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022132385092459266,
+      "loss": 0.1282,
+      "step": 5744
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022129841658049187,
+      "loss": 0.1026,
+      "step": 5745
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022127297958786406,
+      "loss": 0.1477,
+      "step": 5746
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00022124753994765402,
+      "loss": 0.1815,
+      "step": 5747
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022122209766080683,
+      "loss": 0.1502,
+      "step": 5748
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022119665272826754,
+      "loss": 0.1554,
+      "step": 5749
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022117120515098142,
+      "loss": 0.1553,
+      "step": 5750
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002211457549298937,
+      "loss": 0.237,
+      "step": 5751
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022112030206594983,
+      "loss": 0.1517,
+      "step": 5752
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022109484656009532,
+      "loss": 0.1898,
+      "step": 5753
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022106938841327573,
+      "loss": 0.1777,
+      "step": 5754
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022104392762643677,
+      "loss": 0.2366,
+      "step": 5755
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002210184642005242,
+      "loss": 0.1038,
+      "step": 5756
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022099299813648396,
+      "loss": 0.1902,
+      "step": 5757
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022096752943526207,
+      "loss": 0.1369,
+      "step": 5758
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022094205809780452,
+      "loss": 0.2965,
+      "step": 5759
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022091658412505755,
+      "loss": 0.1144,
+      "step": 5760
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002208911075179675,
+      "loss": 0.1517,
+      "step": 5761
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022086562827748065,
+      "loss": 0.1231,
+      "step": 5762
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002208401464045435,
+      "loss": 0.1907,
+      "step": 5763
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022081466190010266,
+      "loss": 0.1601,
+      "step": 5764
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002207891747651048,
+      "loss": 0.218,
+      "step": 5765
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002207636850004967,
+      "loss": 0.1275,
+      "step": 5766
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022073819260722517,
+      "loss": 0.2764,
+      "step": 5767
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022071269758623725,
+      "loss": 0.0712,
+      "step": 5768
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022068719993848,
+      "loss": 0.0713,
+      "step": 5769
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022066169966490054,
+      "loss": 0.2222,
+      "step": 5770
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022063619676644615,
+      "loss": 0.0889,
+      "step": 5771
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022061069124406418,
+      "loss": 0.194,
+      "step": 5772
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002205851830987021,
+      "loss": 0.0966,
+      "step": 5773
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002205596723313075,
+      "loss": 0.2432,
+      "step": 5774
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022053415894282793,
+      "loss": 0.0449,
+      "step": 5775
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022050864293421122,
+      "loss": 0.1714,
+      "step": 5776
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0002204831243064052,
+      "loss": 0.2422,
+      "step": 5777
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022045760306035778,
+      "loss": 0.1671,
+      "step": 5778
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000220432079197017,
+      "loss": 0.2976,
+      "step": 5779
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00022040655271733105,
+      "loss": 0.1397,
+      "step": 5780
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022038102362224816,
+      "loss": 0.1271,
+      "step": 5781
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022035549191271657,
+      "loss": 0.206,
+      "step": 5782
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022032995758968483,
+      "loss": 0.1155,
+      "step": 5783
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022030442065410142,
+      "loss": 0.1356,
+      "step": 5784
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022027888110691488,
+      "loss": 0.2097,
+      "step": 5785
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022025333894907406,
+      "loss": 0.1765,
+      "step": 5786
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022022779418152767,
+      "loss": 0.1989,
+      "step": 5787
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022020224680522466,
+      "loss": 0.3576,
+      "step": 5788
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022017669682111406,
+      "loss": 0.1131,
+      "step": 5789
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022015114423014494,
+      "loss": 0.213,
+      "step": 5790
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002201255890332666,
+      "loss": 0.1494,
+      "step": 5791
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022010003123142823,
+      "loss": 0.1606,
+      "step": 5792
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022007447082557927,
+      "loss": 0.1402,
+      "step": 5793
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022004890781666923,
+      "loss": 0.2842,
+      "step": 5794
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00022002334220564763,
+      "loss": 0.2035,
+      "step": 5795
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021999777399346422,
+      "loss": 0.3049,
+      "step": 5796
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021997220318106879,
+      "loss": 0.2839,
+      "step": 5797
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002199466297694112,
+      "loss": 0.2056,
+      "step": 5798
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021992105375944144,
+      "loss": 0.2256,
+      "step": 5799
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021989547515210956,
+      "loss": 0.2402,
+      "step": 5800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021986989394836575,
+      "loss": 0.3575,
+      "step": 5801
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002198443101491603,
+      "loss": 0.0803,
+      "step": 5802
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002198187237554435,
+      "loss": 0.1903,
+      "step": 5803
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021979313476816587,
+      "loss": 0.0983,
+      "step": 5804
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021976754318827797,
+      "loss": 0.134,
+      "step": 5805
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002197419490167304,
+      "loss": 0.216,
+      "step": 5806
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021971635225447397,
+      "loss": 0.1114,
+      "step": 5807
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021969075290245949,
+      "loss": 0.1945,
+      "step": 5808
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0002196651509616379,
+      "loss": 0.162,
+      "step": 5809
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021963954643296026,
+      "loss": 0.2444,
+      "step": 5810
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021961393931737764,
+      "loss": 0.2279,
+      "step": 5811
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021958832961584132,
+      "loss": 0.1555,
+      "step": 5812
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00021956271732930264,
+      "loss": 0.1934,
+      "step": 5813
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021953710245871302,
+      "loss": 0.2279,
+      "step": 5814
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021951148500502394,
+      "loss": 0.2681,
+      "step": 5815
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021948586496918702,
+      "loss": 0.2274,
+      "step": 5816
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021946024235215404,
+      "loss": 0.0598,
+      "step": 5817
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002194346171548767,
+      "loss": 0.0697,
+      "step": 5818
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021940898937830697,
+      "loss": 0.1889,
+      "step": 5819
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002193833590233968,
+      "loss": 0.2501,
+      "step": 5820
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002193577260910983,
+      "loss": 0.3253,
+      "step": 5821
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002193320905823637,
+      "loss": 0.2238,
+      "step": 5822
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002193064524981452,
+      "loss": 0.2848,
+      "step": 5823
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021928081183939524,
+      "loss": 0.1997,
+      "step": 5824
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021925516860706637,
+      "loss": 0.0751,
+      "step": 5825
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021922952280211097,
+      "loss": 0.2819,
+      "step": 5826
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002192038744254819,
+      "loss": 0.1599,
+      "step": 5827
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021917822347813177,
+      "loss": 0.0993,
+      "step": 5828
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002191525699610135,
+      "loss": 0.125,
+      "step": 5829
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002191269138750801,
+      "loss": 0.1192,
+      "step": 5830
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021910125522128448,
+      "loss": 0.1709,
+      "step": 5831
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021907559400057992,
+      "loss": 0.2382,
+      "step": 5832
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002190499302139196,
+      "loss": 0.2527,
+      "step": 5833
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021902426386225684,
+      "loss": 0.1751,
+      "step": 5834
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021899859494654513,
+      "loss": 0.1952,
+      "step": 5835
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021897292346773794,
+      "loss": 0.1866,
+      "step": 5836
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021894724942678886,
+      "loss": 0.245,
+      "step": 5837
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002189215728246517,
+      "loss": 0.172,
+      "step": 5838
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002188958936622802,
+      "loss": 0.1802,
+      "step": 5839
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002188702119406283,
+      "loss": 0.1246,
+      "step": 5840
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021884452766064995,
+      "loss": 0.048,
+      "step": 5841
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021881884082329932,
+      "loss": 0.0424,
+      "step": 5842
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021879315142953057,
+      "loss": 0.0916,
+      "step": 5843
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021876745948029797,
+      "loss": 0.2559,
+      "step": 5844
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00021874176497655587,
+      "loss": 0.1281,
+      "step": 5845
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002187160679192588,
+      "loss": 0.095,
+      "step": 5846
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021869036830936134,
+      "loss": 0.1816,
+      "step": 5847
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021866466614781815,
+      "loss": 0.2039,
+      "step": 5848
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021863896143558393,
+      "loss": 0.1833,
+      "step": 5849
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002186132541736136,
+      "loss": 0.0906,
+      "step": 5850
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021858754436286206,
+      "loss": 0.1499,
+      "step": 5851
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002185618320042844,
+      "loss": 0.0942,
+      "step": 5852
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002185361170988357,
+      "loss": 0.0678,
+      "step": 5853
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021851039964747125,
+      "loss": 0.2739,
+      "step": 5854
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002184846796511464,
+      "loss": 0.1983,
+      "step": 5855
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021845895711081645,
+      "loss": 0.2356,
+      "step": 5856
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021843323202743706,
+      "loss": 0.1764,
+      "step": 5857
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021840750440196374,
+      "loss": 0.2491,
+      "step": 5858
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021838177423535228,
+      "loss": 0.1735,
+      "step": 5859
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002183560415285584,
+      "loss": 0.2061,
+      "step": 5860
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021833030628253805,
+      "loss": 0.2307,
+      "step": 5861
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021830456849824715,
+      "loss": 0.1803,
+      "step": 5862
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002182788281766419,
+      "loss": 0.265,
+      "step": 5863
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021825308531867838,
+      "loss": 0.2297,
+      "step": 5864
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002182273399253129,
+      "loss": 0.0805,
+      "step": 5865
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021820159199750178,
+      "loss": 0.1197,
+      "step": 5866
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021817584153620157,
+      "loss": 0.2298,
+      "step": 5867
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021815008854236877,
+      "loss": 0.1966,
+      "step": 5868
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021812433301695997,
+      "loss": 0.2455,
+      "step": 5869
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021809857496093199,
+      "loss": 0.0707,
+      "step": 5870
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021807281437524168,
+      "loss": 0.1909,
+      "step": 5871
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002180470512608459,
+      "loss": 0.1268,
+      "step": 5872
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021802128561870175,
+      "loss": 0.1202,
+      "step": 5873
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002179955174497663,
+      "loss": 0.1318,
+      "step": 5874
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002179697467549968,
+      "loss": 0.172,
+      "step": 5875
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021794397353535044,
+      "loss": 0.1574,
+      "step": 5876
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021791819779178476,
+      "loss": 0.2678,
+      "step": 5877
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00021789241952525722,
+      "loss": 0.2206,
+      "step": 5878
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021786663873672534,
+      "loss": 0.1529,
+      "step": 5879
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021784085542714688,
+      "loss": 0.1129,
+      "step": 5880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021781506959747958,
+      "loss": 0.1651,
+      "step": 5881
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021778928124868132,
+      "loss": 0.2344,
+      "step": 5882
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021776349038171003,
+      "loss": 0.2285,
+      "step": 5883
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021773769699752382,
+      "loss": 0.2856,
+      "step": 5884
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021771190109708079,
+      "loss": 0.0709,
+      "step": 5885
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021768610268133916,
+      "loss": 0.3945,
+      "step": 5886
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021766030175125734,
+      "loss": 0.2398,
+      "step": 5887
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002176344983077937,
+      "loss": 0.2659,
+      "step": 5888
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002176086923519068,
+      "loss": 0.1727,
+      "step": 5889
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021758288388455525,
+      "loss": 0.2139,
+      "step": 5890
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021755707290669777,
+      "loss": 0.1572,
+      "step": 5891
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002175312594192931,
+      "loss": 0.0946,
+      "step": 5892
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021750544342330022,
+      "loss": 0.3522,
+      "step": 5893
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000217479624919678,
+      "loss": 0.1314,
+      "step": 5894
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021745380390938561,
+      "loss": 0.1762,
+      "step": 5895
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021742798039338227,
+      "loss": 0.0695,
+      "step": 5896
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021740215437262715,
+      "loss": 0.23,
+      "step": 5897
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002173763258480797,
+      "loss": 0.1213,
+      "step": 5898
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021735049482069928,
+      "loss": 0.2477,
+      "step": 5899
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021732466129144545,
+      "loss": 0.1821,
+      "step": 5900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021729882526127793,
+      "loss": 0.2033,
+      "step": 5901
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002172729867311564,
+      "loss": 0.149,
+      "step": 5902
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021724714570204069,
+      "loss": 0.1258,
+      "step": 5903
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021722130217489069,
+      "loss": 0.1416,
+      "step": 5904
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021719545615066644,
+      "loss": 0.1533,
+      "step": 5905
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002171696076303281,
+      "loss": 0.203,
+      "step": 5906
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021714375661483572,
+      "loss": 0.1647,
+      "step": 5907
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021711790310514976,
+      "loss": 0.1808,
+      "step": 5908
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00021709204710223048,
+      "loss": 0.0574,
+      "step": 5909
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002170661886070384,
+      "loss": 0.2238,
+      "step": 5910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0002170403276205341,
+      "loss": 0.0929,
+      "step": 5911
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021701446414367818,
+      "loss": 0.1834,
+      "step": 5912
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021698859817743151,
+      "loss": 0.1721,
+      "step": 5913
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002169627297227548,
+      "loss": 0.2569,
+      "step": 5914
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021693685878060906,
+      "loss": 0.1728,
+      "step": 5915
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002169109853519553,
+      "loss": 0.1273,
+      "step": 5916
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021688510943775467,
+      "loss": 0.1543,
+      "step": 5917
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021685923103896836,
+      "loss": 0.1435,
+      "step": 5918
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021683335015655766,
+      "loss": 0.2009,
+      "step": 5919
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000216807466791484,
+      "loss": 0.2158,
+      "step": 5920
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021678158094470885,
+      "loss": 0.1978,
+      "step": 5921
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021675569261719383,
+      "loss": 0.1632,
+      "step": 5922
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021672980180990057,
+      "loss": 0.2269,
+      "step": 5923
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021670390852379087,
+      "loss": 0.2781,
+      "step": 5924
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021667801275982655,
+      "loss": 0.1299,
+      "step": 5925
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021665211451896962,
+      "loss": 0.0931,
+      "step": 5926
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021662621380218208,
+      "loss": 0.1162,
+      "step": 5927
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021660031061042606,
+      "loss": 0.0706,
+      "step": 5928
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002165744049446638,
+      "loss": 0.3002,
+      "step": 5929
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021654849680585766,
+      "loss": 0.1337,
+      "step": 5930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021652258619497003,
+      "loss": 0.2706,
+      "step": 5931
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021649667311296334,
+      "loss": 0.1576,
+      "step": 5932
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021647075756080027,
+      "loss": 0.1372,
+      "step": 5933
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021644483953944352,
+      "loss": 0.1421,
+      "step": 5934
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021641891904985574,
+      "loss": 0.2958,
+      "step": 5935
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021639299609299995,
+      "loss": 0.2198,
+      "step": 5936
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021636707066983904,
+      "loss": 0.236,
+      "step": 5937
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021634114278133607,
+      "loss": 0.1391,
+      "step": 5938
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021631521242845419,
+      "loss": 0.0204,
+      "step": 5939
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021628927961215663,
+      "loss": 0.1734,
+      "step": 5940
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021626334433340674,
+      "loss": 0.0816,
+      "step": 5941
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021623740659316792,
+      "loss": 0.0423,
+      "step": 5942
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021621146639240368,
+      "loss": 0.1337,
+      "step": 5943
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00021618552373207762,
+      "loss": 0.1474,
+      "step": 5944
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021615957861315343,
+      "loss": 0.033,
+      "step": 5945
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021613363103659497,
+      "loss": 0.0924,
+      "step": 5946
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021610768100336597,
+      "loss": 0.2956,
+      "step": 5947
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021608172851443052,
+      "loss": 0.1518,
+      "step": 5948
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021605577357075267,
+      "loss": 0.1525,
+      "step": 5949
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021602981617329652,
+      "loss": 0.276,
+      "step": 5950
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021600385632302628,
+      "loss": 0.155,
+      "step": 5951
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021597789402090636,
+      "loss": 0.1636,
+      "step": 5952
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021595192926790116,
+      "loss": 0.1164,
+      "step": 5953
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002159259620649752,
+      "loss": 0.1425,
+      "step": 5954
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021589999241309308,
+      "loss": 0.1369,
+      "step": 5955
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021587402031321946,
+      "loss": 0.1217,
+      "step": 5956
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021584804576631919,
+      "loss": 0.1003,
+      "step": 5957
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002158220687733571,
+      "loss": 0.2664,
+      "step": 5958
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002157960893352982,
+      "loss": 0.0205,
+      "step": 5959
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021577010745310748,
+      "loss": 0.3149,
+      "step": 5960
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021574412312775016,
+      "loss": 0.2539,
+      "step": 5961
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002157181363601915,
+      "loss": 0.1594,
+      "step": 5962
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021569214715139675,
+      "loss": 0.1941,
+      "step": 5963
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002156661555023314,
+      "loss": 0.2426,
+      "step": 5964
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002156401614139609,
+      "loss": 0.1461,
+      "step": 5965
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002156141648872509,
+      "loss": 0.109,
+      "step": 5966
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002155881659231671,
+      "loss": 0.1087,
+      "step": 5967
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021556216452267525,
+      "loss": 0.1651,
+      "step": 5968
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021553616068674132,
+      "loss": 0.238,
+      "step": 5969
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021551015441633112,
+      "loss": 0.1456,
+      "step": 5970
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002154841457124108,
+      "loss": 0.0838,
+      "step": 5971
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021545813457594655,
+      "loss": 0.2479,
+      "step": 5972
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002154321210079045,
+      "loss": 0.1923,
+      "step": 5973
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002154061050092511,
+      "loss": 0.1761,
+      "step": 5974
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021538008658095265,
+      "loss": 0.2098,
+      "step": 5975
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021535406572397567,
+      "loss": 0.1744,
+      "step": 5976
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00021532804243928685,
+      "loss": 0.1363,
+      "step": 5977
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021530201672785283,
+      "loss": 0.2171,
+      "step": 5978
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021527598859064038,
+      "loss": 0.1733,
+      "step": 5979
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021524995802861636,
+      "loss": 0.2474,
+      "step": 5980
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021522392504274776,
+      "loss": 0.1907,
+      "step": 5981
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021519788963400162,
+      "loss": 0.2411,
+      "step": 5982
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021517185180334504,
+      "loss": 0.107,
+      "step": 5983
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002151458115517453,
+      "loss": 0.0581,
+      "step": 5984
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021511976888016965,
+      "loss": 0.1249,
+      "step": 5985
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021509372378958563,
+      "loss": 0.3107,
+      "step": 5986
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021506767628096061,
+      "loss": 0.1429,
+      "step": 5987
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021504162635526224,
+      "loss": 0.1627,
+      "step": 5988
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021501557401345817,
+      "loss": 0.2253,
+      "step": 5989
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002149895192565162,
+      "loss": 0.1255,
+      "step": 5990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021496346208540418,
+      "loss": 0.2594,
+      "step": 5991
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021493740250109,
+      "loss": 0.1432,
+      "step": 5992
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021491134050454177,
+      "loss": 0.1458,
+      "step": 5993
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021488527609672757,
+      "loss": 0.2767,
+      "step": 5994
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002148592092786157,
+      "loss": 0.1924,
+      "step": 5995
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021483314005117435,
+      "loss": 0.2129,
+      "step": 5996
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021480706841537202,
+      "loss": 0.1935,
+      "step": 5997
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002147809943721771,
+      "loss": 0.1043,
+      "step": 5998
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002147549179225582,
+      "loss": 0.1529,
+      "step": 5999
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000214728839067484,
+      "loss": 0.0977,
+      "step": 6000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021470275780792323,
+      "loss": 0.1541,
+      "step": 6001
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021467667414484478,
+      "loss": 0.2207,
+      "step": 6002
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021465058807921755,
+      "loss": 0.0448,
+      "step": 6003
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021462449961201052,
+      "loss": 0.2815,
+      "step": 6004
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021459840874419288,
+      "loss": 0.0961,
+      "step": 6005
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021457231547673377,
+      "loss": 0.2052,
+      "step": 6006
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021454621981060253,
+      "loss": 0.06,
+      "step": 6007
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021452012174676843,
+      "loss": 0.0502,
+      "step": 6008
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021449402128620104,
+      "loss": 0.1881,
+      "step": 6009
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00021446791842986995,
+      "loss": 0.1512,
+      "step": 6010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021444181317874467,
+      "loss": 0.2765,
+      "step": 6011
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021441570553379501,
+      "loss": 0.147,
+      "step": 6012
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021438959549599084,
+      "loss": 0.2724,
+      "step": 6013
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021436348306630196,
+      "loss": 0.1305,
+      "step": 6014
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021433736824569847,
+      "loss": 0.2101,
+      "step": 6015
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002143112510351504,
+      "loss": 0.2675,
+      "step": 6016
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002142851314356279,
+      "loss": 0.1694,
+      "step": 6017
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021425900944810133,
+      "loss": 0.19,
+      "step": 6018
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021423288507354098,
+      "loss": 0.2058,
+      "step": 6019
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021420675831291733,
+      "loss": 0.0807,
+      "step": 6020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002141806291672009,
+      "loss": 0.1672,
+      "step": 6021
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002141544976373623,
+      "loss": 0.0903,
+      "step": 6022
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021412836372437224,
+      "loss": 0.0357,
+      "step": 6023
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002141022274292015,
+      "loss": 0.2242,
+      "step": 6024
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000214076088752821,
+      "loss": 0.1707,
+      "step": 6025
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002140499476962017,
+      "loss": 0.1977,
+      "step": 6026
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021402380426031468,
+      "loss": 0.1014,
+      "step": 6027
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002139976584461311,
+      "loss": 0.1686,
+      "step": 6028
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021397151025462217,
+      "loss": 0.1221,
+      "step": 6029
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021394535968675922,
+      "loss": 0.2597,
+      "step": 6030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021391920674351366,
+      "loss": 0.1946,
+      "step": 6031
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000213893051425857,
+      "loss": 0.2997,
+      "step": 6032
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021386689373476087,
+      "loss": 0.2794,
+      "step": 6033
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021384073367119693,
+      "loss": 0.3307,
+      "step": 6034
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021381457123613697,
+      "loss": 0.1535,
+      "step": 6035
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021378840643055278,
+      "loss": 0.2044,
+      "step": 6036
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021376223925541633,
+      "loss": 0.1662,
+      "step": 6037
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021373606971169975,
+      "loss": 0.1494,
+      "step": 6038
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021370989780037503,
+      "loss": 0.1344,
+      "step": 6039
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021368372352241442,
+      "loss": 0.1454,
+      "step": 6040
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021365754687879026,
+      "loss": 0.1026,
+      "step": 6041
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0002136313678704749,
+      "loss": 0.1371,
+      "step": 6042
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00021360518649844083,
+      "loss": 0.1164,
+      "step": 6043
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002135790027636606,
+      "loss": 0.1423,
+      "step": 6044
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021355281666710687,
+      "loss": 0.0586,
+      "step": 6045
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021352662820975238,
+      "loss": 0.1342,
+      "step": 6046
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021350043739256988,
+      "loss": 0.2121,
+      "step": 6047
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002134742442165324,
+      "loss": 0.1582,
+      "step": 6048
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021344804868261285,
+      "loss": 0.2792,
+      "step": 6049
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021342185079178434,
+      "loss": 0.2567,
+      "step": 6050
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002133956505450201,
+      "loss": 0.1367,
+      "step": 6051
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021336944794329333,
+      "loss": 0.2173,
+      "step": 6052
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021334324298757744,
+      "loss": 0.1423,
+      "step": 6053
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002133170356788458,
+      "loss": 0.0589,
+      "step": 6054
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002132908260180719,
+      "loss": 0.2106,
+      "step": 6055
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002132646140062295,
+      "loss": 0.1608,
+      "step": 6056
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021323839964429217,
+      "loss": 0.1214,
+      "step": 6057
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021321218293323372,
+      "loss": 0.0529,
+      "step": 6058
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021318596387402805,
+      "loss": 0.3355,
+      "step": 6059
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002131597424676491,
+      "loss": 0.1565,
+      "step": 6060
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000213133518715071,
+      "loss": 0.2941,
+      "step": 6061
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021310729261726774,
+      "loss": 0.1134,
+      "step": 6062
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002130810641752137,
+      "loss": 0.1839,
+      "step": 6063
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021305483338988307,
+      "loss": 0.1002,
+      "step": 6064
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021302860026225026,
+      "loss": 0.197,
+      "step": 6065
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002130023647932898,
+      "loss": 0.2482,
+      "step": 6066
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002129761269839762,
+      "loss": 0.1369,
+      "step": 6067
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021294988683528426,
+      "loss": 0.163,
+      "step": 6068
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021292364434818853,
+      "loss": 0.1853,
+      "step": 6069
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000212897399523664,
+      "loss": 0.0886,
+      "step": 6070
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021287115236268546,
+      "loss": 0.1528,
+      "step": 6071
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.000212844902866228,
+      "loss": 0.2274,
+      "step": 6072
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021281865103526672,
+      "loss": 0.1732,
+      "step": 6073
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021279239687077673,
+      "loss": 0.2736,
+      "step": 6074
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00021276614037373334,
+      "loss": 0.2093,
+      "step": 6075
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002127398815451119,
+      "loss": 0.1365,
+      "step": 6076
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021271362038588785,
+      "loss": 0.2324,
+      "step": 6077
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002126873568970367,
+      "loss": 0.2495,
+      "step": 6078
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002126610910795341,
+      "loss": 0.2085,
+      "step": 6079
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002126348229343557,
+      "loss": 0.2818,
+      "step": 6080
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021260855246247728,
+      "loss": 0.1716,
+      "step": 6081
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021258227966487476,
+      "loss": 0.1638,
+      "step": 6082
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021255600454252402,
+      "loss": 0.1626,
+      "step": 6083
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021252972709640125,
+      "loss": 0.1961,
+      "step": 6084
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021250344732748244,
+      "loss": 0.2072,
+      "step": 6085
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021247716523674384,
+      "loss": 0.103,
+      "step": 6086
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021245088082516176,
+      "loss": 0.1798,
+      "step": 6087
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021242459409371263,
+      "loss": 0.0549,
+      "step": 6088
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021239830504337284,
+      "loss": 0.0798,
+      "step": 6089
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000212372013675119,
+      "loss": 0.2627,
+      "step": 6090
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002123457199899278,
+      "loss": 0.2326,
+      "step": 6091
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002123194239887759,
+      "loss": 0.1292,
+      "step": 6092
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021229312567264014,
+      "loss": 0.199,
+      "step": 6093
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021226682504249747,
+      "loss": 0.1373,
+      "step": 6094
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021224052209932483,
+      "loss": 0.3611,
+      "step": 6095
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021221421684409933,
+      "loss": 0.2756,
+      "step": 6096
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021218790927779807,
+      "loss": 0.1997,
+      "step": 6097
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021216159940139838,
+      "loss": 0.2479,
+      "step": 6098
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002121352872158775,
+      "loss": 0.0597,
+      "step": 6099
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021210897272221297,
+      "loss": 0.1437,
+      "step": 6100
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021208265592138225,
+      "loss": 0.1264,
+      "step": 6101
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021205633681436288,
+      "loss": 0.2515,
+      "step": 6102
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021203001540213257,
+      "loss": 0.1938,
+      "step": 6103
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021200369168566913,
+      "loss": 0.1109,
+      "step": 6104
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021197736566595034,
+      "loss": 0.2323,
+      "step": 6105
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0002119510373439542,
+      "loss": 0.3681,
+      "step": 6106
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021192470672065864,
+      "loss": 0.1943,
+      "step": 6107
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021189837379704182,
+      "loss": 0.1584,
+      "step": 6108
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00021187203857408196,
+      "loss": 0.1612,
+      "step": 6109
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002118457010527573,
+      "loss": 0.1071,
+      "step": 6110
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002118193612340462,
+      "loss": 0.213,
+      "step": 6111
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021179301911892714,
+      "loss": 0.2013,
+      "step": 6112
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021176667470837858,
+      "loss": 0.2826,
+      "step": 6113
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002117403280033792,
+      "loss": 0.0983,
+      "step": 6114
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021171397900490767,
+      "loss": 0.021,
+      "step": 6115
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021168762771394282,
+      "loss": 0.1838,
+      "step": 6116
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021166127413146352,
+      "loss": 0.1424,
+      "step": 6117
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002116349182584487,
+      "loss": 0.2242,
+      "step": 6118
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021160856009587738,
+      "loss": 0.1474,
+      "step": 6119
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021158219964472874,
+      "loss": 0.1875,
+      "step": 6120
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021155583690598195,
+      "loss": 0.1616,
+      "step": 6121
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002115294718806163,
+      "loss": 0.2491,
+      "step": 6122
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021150310456961125,
+      "loss": 0.137,
+      "step": 6123
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021147673497394625,
+      "loss": 0.1071,
+      "step": 6124
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002114503630946008,
+      "loss": 0.2151,
+      "step": 6125
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021142398893255455,
+      "loss": 0.1334,
+      "step": 6126
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021139761248878722,
+      "loss": 0.2087,
+      "step": 6127
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021137123376427865,
+      "loss": 0.101,
+      "step": 6128
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021134485276000877,
+      "loss": 0.1364,
+      "step": 6129
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021131846947695742,
+      "loss": 0.0325,
+      "step": 6130
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021129208391610477,
+      "loss": 0.1458,
+      "step": 6131
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021126569607843093,
+      "loss": 0.1545,
+      "step": 6132
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021123930596491616,
+      "loss": 0.157,
+      "step": 6133
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021121291357654074,
+      "loss": 0.2621,
+      "step": 6134
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021118651891428508,
+      "loss": 0.13,
+      "step": 6135
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002111601219791297,
+      "loss": 0.2526,
+      "step": 6136
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021113372277205514,
+      "loss": 0.1242,
+      "step": 6137
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021110732129404203,
+      "loss": 0.149,
+      "step": 6138
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021108091754607115,
+      "loss": 0.2867,
+      "step": 6139
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021105451152912327,
+      "loss": 0.2436,
+      "step": 6140
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0002110281032441793,
+      "loss": 0.1719,
+      "step": 6141
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00021100169269222034,
+      "loss": 0.1219,
+      "step": 6142
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021097527987422736,
+      "loss": 0.0953,
+      "step": 6143
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002109488647911815,
+      "loss": 0.239,
+      "step": 6144
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002109224474440641,
+      "loss": 0.1955,
+      "step": 6145
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002108960278338564,
+      "loss": 0.182,
+      "step": 6146
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021086960596153985,
+      "loss": 0.3709,
+      "step": 6147
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002108431818280959,
+      "loss": 0.2108,
+      "step": 6148
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002108167554345062,
+      "loss": 0.175,
+      "step": 6149
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002107903267817524,
+      "loss": 0.1693,
+      "step": 6150
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002107638958708162,
+      "loss": 0.0912,
+      "step": 6151
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021073746270267946,
+      "loss": 0.1952,
+      "step": 6152
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021071102727832409,
+      "loss": 0.3249,
+      "step": 6153
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021068458959873208,
+      "loss": 0.0757,
+      "step": 6154
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021065814966488553,
+      "loss": 0.1489,
+      "step": 6155
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002106317074777666,
+      "loss": 0.3395,
+      "step": 6156
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021060526303835755,
+      "loss": 0.0592,
+      "step": 6157
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021057881634764072,
+      "loss": 0.2206,
+      "step": 6158
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021055236740659847,
+      "loss": 0.1762,
+      "step": 6159
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021052591621621337,
+      "loss": 0.1358,
+      "step": 6160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002104994627774679,
+      "loss": 0.2956,
+      "step": 6161
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002104730070913449,
+      "loss": 0.2088,
+      "step": 6162
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021044654915882698,
+      "loss": 0.0708,
+      "step": 6163
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021042008898089699,
+      "loss": 0.1702,
+      "step": 6164
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021039362655853793,
+      "loss": 0.1866,
+      "step": 6165
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021036716189273269,
+      "loss": 0.1375,
+      "step": 6166
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021034069498446445,
+      "loss": 0.1591,
+      "step": 6167
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021031422583471632,
+      "loss": 0.2729,
+      "step": 6168
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021028775444447158,
+      "loss": 0.1461,
+      "step": 6169
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021026128081471355,
+      "loss": 0.0597,
+      "step": 6170
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021023480494642566,
+      "loss": 0.2576,
+      "step": 6171
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021020832684059138,
+      "loss": 0.0825,
+      "step": 6172
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021018184649819433,
+      "loss": 0.1487,
+      "step": 6173
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021015536392021813,
+      "loss": 0.0573,
+      "step": 6174
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00021012887910764664,
+      "loss": 0.1608,
+      "step": 6175
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002101023920614636,
+      "loss": 0.2233,
+      "step": 6176
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00021007590278265288,
+      "loss": 0.1721,
+      "step": 6177
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002100494112721986,
+      "loss": 0.2253,
+      "step": 6178
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00021002291753108478,
+      "loss": 0.2315,
+      "step": 6179
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020999642156029557,
+      "loss": 0.2823,
+      "step": 6180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020996992336081525,
+      "loss": 0.1931,
+      "step": 6181
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020994342293362812,
+      "loss": 0.0637,
+      "step": 6182
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020991692027971866,
+      "loss": 0.2596,
+      "step": 6183
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020989041540007126,
+      "loss": 0.1239,
+      "step": 6184
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002098639082956706,
+      "loss": 0.1099,
+      "step": 6185
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020983739896750128,
+      "loss": 0.1458,
+      "step": 6186
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020981088741654804,
+      "loss": 0.1151,
+      "step": 6187
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020978437364379572,
+      "loss": 0.2123,
+      "step": 6188
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020975785765022927,
+      "loss": 0.2184,
+      "step": 6189
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020973133943683365,
+      "loss": 0.2489,
+      "step": 6190
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020970481900459392,
+      "loss": 0.1779,
+      "step": 6191
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002096782963544952,
+      "loss": 0.1573,
+      "step": 6192
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020965177148752283,
+      "loss": 0.0953,
+      "step": 6193
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020962524440466208,
+      "loss": 0.2041,
+      "step": 6194
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020959871510689836,
+      "loss": 0.0964,
+      "step": 6195
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020957218359521706,
+      "loss": 0.1023,
+      "step": 6196
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002095456498706039,
+      "loss": 0.1659,
+      "step": 6197
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002095191139340445,
+      "loss": 0.1663,
+      "step": 6198
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020949257578652448,
+      "loss": 0.204,
+      "step": 6199
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020946603542902975,
+      "loss": 0.1106,
+      "step": 6200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002094394928625462,
+      "loss": 0.1515,
+      "step": 6201
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020941294808805977,
+      "loss": 0.1559,
+      "step": 6202
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020938640110655662,
+      "loss": 0.135,
+      "step": 6203
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020935985191902277,
+      "loss": 0.1637,
+      "step": 6204
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020933330052644444,
+      "loss": 0.1538,
+      "step": 6205
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020930674692980805,
+      "loss": 0.2515,
+      "step": 6206
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0002092801911300999,
+      "loss": 0.1521,
+      "step": 6207
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00020925363312830652,
+      "loss": 0.2207,
+      "step": 6208
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002092270729254144,
+      "loss": 0.1982,
+      "step": 6209
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002092005105224102,
+      "loss": 0.1345,
+      "step": 6210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002091739459202807,
+      "loss": 0.2749,
+      "step": 6211
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020914737912001256,
+      "loss": 0.2979,
+      "step": 6212
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020912081012259274,
+      "loss": 0.1726,
+      "step": 6213
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002090942389290082,
+      "loss": 0.0907,
+      "step": 6214
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000209067665540246,
+      "loss": 0.2007,
+      "step": 6215
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020904108995729325,
+      "loss": 0.326,
+      "step": 6216
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020901451218113714,
+      "loss": 0.0474,
+      "step": 6217
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020898793221276493,
+      "loss": 0.1851,
+      "step": 6218
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020896135005316405,
+      "loss": 0.0827,
+      "step": 6219
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002089347657033219,
+      "loss": 0.1809,
+      "step": 6220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020890817916422604,
+      "loss": 0.263,
+      "step": 6221
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020888159043686404,
+      "loss": 0.2434,
+      "step": 6222
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020885499952222363,
+      "loss": 0.1808,
+      "step": 6223
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020882840642129263,
+      "loss": 0.2758,
+      "step": 6224
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002088018111350588,
+      "loss": 0.1901,
+      "step": 6225
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020877521366451015,
+      "loss": 0.265,
+      "step": 6226
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020874861401063463,
+      "loss": 0.1941,
+      "step": 6227
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002087220121744204,
+      "loss": 0.2296,
+      "step": 6228
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002086954081568556,
+      "loss": 0.0618,
+      "step": 6229
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020866880195892853,
+      "loss": 0.1813,
+      "step": 6230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020864219358162752,
+      "loss": 0.1206,
+      "step": 6231
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020861558302594097,
+      "loss": 0.1457,
+      "step": 6232
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002085889702928574,
+      "loss": 0.3555,
+      "step": 6233
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020856235538336542,
+      "loss": 0.164,
+      "step": 6234
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020853573829845363,
+      "loss": 0.2991,
+      "step": 6235
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020850911903911085,
+      "loss": 0.2398,
+      "step": 6236
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020848249760632581,
+      "loss": 0.1186,
+      "step": 6237
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020845587400108754,
+      "loss": 0.1289,
+      "step": 6238
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020842924822438497,
+      "loss": 0.1985,
+      "step": 6239
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020840262027720718,
+      "loss": 0.1298,
+      "step": 6240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00020837599016054333,
+      "loss": 0.2804,
+      "step": 6241
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020834935787538257,
+      "loss": 0.0358,
+      "step": 6242
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002083227234227143,
+      "loss": 0.0555,
+      "step": 6243
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002082960868035279,
+      "loss": 0.2138,
+      "step": 6244
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002082694480188128,
+      "loss": 0.0289,
+      "step": 6245
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002082428070695586,
+      "loss": 0.1741,
+      "step": 6246
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002082161639567549,
+      "loss": 0.1409,
+      "step": 6247
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002081895186813914,
+      "loss": 0.1909,
+      "step": 6248
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.000208162871244458,
+      "loss": 0.1521,
+      "step": 6249
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002081362216469444,
+      "loss": 0.3363,
+      "step": 6250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020810956988984074,
+      "loss": 0.1819,
+      "step": 6251
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020808291597413693,
+      "loss": 0.143,
+      "step": 6252
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002080562599008231,
+      "loss": 0.2494,
+      "step": 6253
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020802960167088944,
+      "loss": 0.1312,
+      "step": 6254
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002080029412853263,
+      "loss": 0.2637,
+      "step": 6255
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020797627874512395,
+      "loss": 0.2539,
+      "step": 6256
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020794961405127288,
+      "loss": 0.1397,
+      "step": 6257
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020792294720476358,
+      "loss": 0.2001,
+      "step": 6258
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020789627820658665,
+      "loss": 0.1652,
+      "step": 6259
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020786960705773273,
+      "loss": 0.14,
+      "step": 6260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020784293375919265,
+      "loss": 0.0995,
+      "step": 6261
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020781625831195718,
+      "loss": 0.1578,
+      "step": 6262
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020778958071701725,
+      "loss": 0.0873,
+      "step": 6263
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002077629009753639,
+      "loss": 0.2898,
+      "step": 6264
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020773621908798815,
+      "loss": 0.0643,
+      "step": 6265
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020770953505588115,
+      "loss": 0.1797,
+      "step": 6266
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002076828488800342,
+      "loss": 0.2182,
+      "step": 6267
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020765616056143854,
+      "loss": 0.0965,
+      "step": 6268
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020762947010108558,
+      "loss": 0.0791,
+      "step": 6269
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0002076027774999668,
+      "loss": 0.2375,
+      "step": 6270
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020757608275907377,
+      "loss": 0.1108,
+      "step": 6271
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020754938587939814,
+      "loss": 0.2706,
+      "step": 6272
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020752268686193153,
+      "loss": 0.3907,
+      "step": 6273
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00020749598570766583,
+      "loss": 0.1026,
+      "step": 6274
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002074692824175928,
+      "loss": 0.2336,
+      "step": 6275
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020744257699270455,
+      "loss": 0.2044,
+      "step": 6276
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020741586943399295,
+      "loss": 0.208,
+      "step": 6277
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002073891597424502,
+      "loss": 0.1903,
+      "step": 6278
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020736244791906845,
+      "loss": 0.1278,
+      "step": 6279
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020733573396483992,
+      "loss": 0.1806,
+      "step": 6280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020730901788075707,
+      "loss": 0.1493,
+      "step": 6281
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020728229966781228,
+      "loss": 0.0639,
+      "step": 6282
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000207255579326998,
+      "loss": 0.2199,
+      "step": 6283
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020722885685930687,
+      "loss": 0.1038,
+      "step": 6284
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002072021322657315,
+      "loss": 0.1604,
+      "step": 6285
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002071754055472647,
+      "loss": 0.0421,
+      "step": 6286
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002071486767048992,
+      "loss": 0.346,
+      "step": 6287
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020712194573962794,
+      "loss": 0.0872,
+      "step": 6288
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000207095212652444,
+      "loss": 0.1005,
+      "step": 6289
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020706847744434025,
+      "loss": 0.1069,
+      "step": 6290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002070417401163099,
+      "loss": 0.3206,
+      "step": 6291
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020701500066934623,
+      "loss": 0.1434,
+      "step": 6292
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020698825910444248,
+      "loss": 0.2063,
+      "step": 6293
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020696151542259203,
+      "loss": 0.1381,
+      "step": 6294
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020693476962478828,
+      "loss": 0.1178,
+      "step": 6295
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020690802171202482,
+      "loss": 0.1075,
+      "step": 6296
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020688127168529526,
+      "loss": 0.0607,
+      "step": 6297
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020685451954559323,
+      "loss": 0.1756,
+      "step": 6298
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020682776529391247,
+      "loss": 0.1721,
+      "step": 6299
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020680100893124696,
+      "loss": 0.2373,
+      "step": 6300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0002067742504585905,
+      "loss": 0.1677,
+      "step": 6301
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020674748987693713,
+      "loss": 0.146,
+      "step": 6302
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020672072718728093,
+      "loss": 0.1463,
+      "step": 6303
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020669396239061602,
+      "loss": 0.1592,
+      "step": 6304
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020666719548793672,
+      "loss": 0.1352,
+      "step": 6305
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020664042648023723,
+      "loss": 0.1908,
+      "step": 6306
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00020661365536851203,
+      "loss": 0.1919,
+      "step": 6307
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020658688215375554,
+      "loss": 0.3064,
+      "step": 6308
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002065601068369623,
+      "loss": 0.1196,
+      "step": 6309
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020653332941912694,
+      "loss": 0.194,
+      "step": 6310
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020650654990124416,
+      "loss": 0.1785,
+      "step": 6311
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002064797682843088,
+      "loss": 0.1109,
+      "step": 6312
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020645298456931568,
+      "loss": 0.1699,
+      "step": 6313
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002064261987572597,
+      "loss": 0.1199,
+      "step": 6314
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020639941084913595,
+      "loss": 0.1841,
+      "step": 6315
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020637262084593943,
+      "loss": 0.2663,
+      "step": 6316
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002063458287486654,
+      "loss": 0.1202,
+      "step": 6317
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020631903455830902,
+      "loss": 0.1357,
+      "step": 6318
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020629223827586563,
+      "loss": 0.1878,
+      "step": 6319
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020626543990233075,
+      "loss": 0.0973,
+      "step": 6320
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020623863943869972,
+      "loss": 0.2797,
+      "step": 6321
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002062118368859682,
+      "loss": 0.1802,
+      "step": 6322
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020618503224513173,
+      "loss": 0.2515,
+      "step": 6323
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002061582255171861,
+      "loss": 0.2363,
+      "step": 6324
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020613141670312708,
+      "loss": 0.0688,
+      "step": 6325
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002061046058039505,
+      "loss": 0.1924,
+      "step": 6326
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020607779282065238,
+      "loss": 0.2286,
+      "step": 6327
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020605097775422865,
+      "loss": 0.1611,
+      "step": 6328
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002060241606056755,
+      "loss": 0.2212,
+      "step": 6329
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020599734137598905,
+      "loss": 0.0971,
+      "step": 6330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020597052006616562,
+      "loss": 0.1745,
+      "step": 6331
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020594369667720146,
+      "loss": 0.1728,
+      "step": 6332
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020591687121009308,
+      "loss": 0.0989,
+      "step": 6333
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020589004366583687,
+      "loss": 0.1762,
+      "step": 6334
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00020586321404542942,
+      "loss": 0.2358,
+      "step": 6335
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002058363823498674,
+      "loss": 0.1469,
+      "step": 6336
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002058095485801475,
+      "loss": 0.1146,
+      "step": 6337
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002057827127372666,
+      "loss": 0.1343,
+      "step": 6338
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002057558748222214,
+      "loss": 0.1324,
+      "step": 6339
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020572903483600903,
+      "loss": 0.1139,
+      "step": 6340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020570219277962643,
+      "loss": 0.0714,
+      "step": 6341
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020567534865407065,
+      "loss": 0.159,
+      "step": 6342
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000205648502460339,
+      "loss": 0.2187,
+      "step": 6343
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002056216541994286,
+      "loss": 0.1399,
+      "step": 6344
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020559480387233697,
+      "loss": 0.1115,
+      "step": 6345
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020556795148006133,
+      "loss": 0.1228,
+      "step": 6346
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020554109702359925,
+      "loss": 0.119,
+      "step": 6347
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002055142405039483,
+      "loss": 0.1526,
+      "step": 6348
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020548738192210613,
+      "loss": 0.1323,
+      "step": 6349
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020546052127907045,
+      "loss": 0.2053,
+      "step": 6350
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020543365857583895,
+      "loss": 0.164,
+      "step": 6351
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002054067938134096,
+      "loss": 0.1912,
+      "step": 6352
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020537992699278044,
+      "loss": 0.2285,
+      "step": 6353
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020535305811494932,
+      "loss": 0.1595,
+      "step": 6354
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020532618718091446,
+      "loss": 0.094,
+      "step": 6355
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020529931419167397,
+      "loss": 0.073,
+      "step": 6356
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020527243914822608,
+      "loss": 0.11,
+      "step": 6357
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020524556205156922,
+      "loss": 0.0791,
+      "step": 6358
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002052186829027017,
+      "loss": 0.2017,
+      "step": 6359
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000205191801702622,
+      "loss": 0.1569,
+      "step": 6360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002051649184523288,
+      "loss": 0.0704,
+      "step": 6361
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020513803315282058,
+      "loss": 0.2288,
+      "step": 6362
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020511114580509617,
+      "loss": 0.2907,
+      "step": 6363
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020508425641015428,
+      "loss": 0.0987,
+      "step": 6364
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002050573649689938,
+      "loss": 0.1314,
+      "step": 6365
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020503047148261367,
+      "loss": 0.2056,
+      "step": 6366
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020500357595201288,
+      "loss": 0.1796,
+      "step": 6367
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0002049766783781906,
+      "loss": 0.2586,
+      "step": 6368
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020494977876214587,
+      "loss": 0.158,
+      "step": 6369
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000204922877104878,
+      "loss": 0.1615,
+      "step": 6370
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020489597340738635,
+      "loss": 0.142,
+      "step": 6371
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00020486906767067025,
+      "loss": 0.2214,
+      "step": 6372
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002048421598957292,
+      "loss": 0.1867,
+      "step": 6373
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002048152500835627,
+      "loss": 0.1655,
+      "step": 6374
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002047883382351704,
+      "loss": 0.2072,
+      "step": 6375
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020476142435155205,
+      "loss": 0.1428,
+      "step": 6376
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002047345084337073,
+      "loss": 0.0608,
+      "step": 6377
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020470759048263617,
+      "loss": 0.134,
+      "step": 6378
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002046806704993384,
+      "loss": 0.175,
+      "step": 6379
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002046537484848141,
+      "loss": 0.2568,
+      "step": 6380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020462682444006333,
+      "loss": 0.2578,
+      "step": 6381
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002045998983660862,
+      "loss": 0.256,
+      "step": 6382
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020457297026388294,
+      "loss": 0.0479,
+      "step": 6383
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002045460401344539,
+      "loss": 0.1714,
+      "step": 6384
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002045191079787994,
+      "loss": 0.1311,
+      "step": 6385
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020449217379791995,
+      "loss": 0.0959,
+      "step": 6386
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000204465237592816,
+      "loss": 0.1153,
+      "step": 6387
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002044382993644882,
+      "loss": 0.1159,
+      "step": 6388
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020441135911393725,
+      "loss": 0.1247,
+      "step": 6389
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020438441684216387,
+      "loss": 0.1143,
+      "step": 6390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020435747255016887,
+      "loss": 0.3133,
+      "step": 6391
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020433052623895312,
+      "loss": 0.172,
+      "step": 6392
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020430357790951766,
+      "loss": 0.1445,
+      "step": 6393
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020427662756286358,
+      "loss": 0.1918,
+      "step": 6394
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002042496751999919,
+      "loss": 0.1523,
+      "step": 6395
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020422272082190393,
+      "loss": 0.1374,
+      "step": 6396
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020419576442960082,
+      "loss": 0.1737,
+      "step": 6397
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000204168806024084,
+      "loss": 0.2275,
+      "step": 6398
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020414184560635495,
+      "loss": 0.1795,
+      "step": 6399
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020411488317741505,
+      "loss": 0.3322,
+      "step": 6400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020408791873826592,
+      "loss": 0.0546,
+      "step": 6401
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020406095228990925,
+      "loss": 0.1813,
+      "step": 6402
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020403398383334672,
+      "loss": 0.1609,
+      "step": 6403
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00020400701336958016,
+      "loss": 0.294,
+      "step": 6404
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0002039800408996114,
+      "loss": 0.1624,
+      "step": 6405
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020395306642444246,
+      "loss": 0.1171,
+      "step": 6406
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020392608994507526,
+      "loss": 0.1927,
+      "step": 6407
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000203899111462512,
+      "loss": 0.1164,
+      "step": 6408
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002038721309777548,
+      "loss": 0.0979,
+      "step": 6409
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020384514849180587,
+      "loss": 0.1698,
+      "step": 6410
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020381816400566764,
+      "loss": 0.1349,
+      "step": 6411
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002037911775203424,
+      "loss": 0.0747,
+      "step": 6412
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020376418903683268,
+      "loss": 0.2688,
+      "step": 6413
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020373719855614096,
+      "loss": 0.1941,
+      "step": 6414
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020371020607926995,
+      "loss": 0.1179,
+      "step": 6415
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020368321160722222,
+      "loss": 0.1314,
+      "step": 6416
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020365621514100067,
+      "loss": 0.1471,
+      "step": 6417
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020362921668160802,
+      "loss": 0.0767,
+      "step": 6418
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002036022162300473,
+      "loss": 0.1604,
+      "step": 6419
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020357521378732134,
+      "loss": 0.1403,
+      "step": 6420
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020354820935443335,
+      "loss": 0.0549,
+      "step": 6421
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020352120293238646,
+      "loss": 0.1553,
+      "step": 6422
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020349419452218373,
+      "loss": 0.1597,
+      "step": 6423
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020346718412482862,
+      "loss": 0.137,
+      "step": 6424
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020344017174132439,
+      "loss": 0.1949,
+      "step": 6425
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020341315737267446,
+      "loss": 0.105,
+      "step": 6426
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020338614101988244,
+      "loss": 0.1853,
+      "step": 6427
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020335912268395175,
+      "loss": 0.3262,
+      "step": 6428
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020333210236588622,
+      "loss": 0.139,
+      "step": 6429
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020330508006668942,
+      "loss": 0.153,
+      "step": 6430
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002032780557873652,
+      "loss": 0.0962,
+      "step": 6431
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002032510295289175,
+      "loss": 0.2802,
+      "step": 6432
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020322400129235013,
+      "loss": 0.2325,
+      "step": 6433
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020319697107866724,
+      "loss": 0.1981,
+      "step": 6434
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020316993888887285,
+      "loss": 0.0954,
+      "step": 6435
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020314290472397118,
+      "loss": 0.206,
+      "step": 6436
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020311586858496645,
+      "loss": 0.1859,
+      "step": 6437
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00020308883047286293,
+      "loss": 0.0999,
+      "step": 6438
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020306179038866507,
+      "loss": 0.1451,
+      "step": 6439
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020303474833337727,
+      "loss": 0.2803,
+      "step": 6440
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002030077043080041,
+      "loss": 0.1166,
+      "step": 6441
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020298065831355014,
+      "loss": 0.2038,
+      "step": 6442
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002029536103510201,
+      "loss": 0.0767,
+      "step": 6443
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020292656042141877,
+      "loss": 0.2054,
+      "step": 6444
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020289950852575084,
+      "loss": 0.1778,
+      "step": 6445
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020287245466502136,
+      "loss": 0.1615,
+      "step": 6446
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020284539884023523,
+      "loss": 0.2153,
+      "step": 6447
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020281834105239746,
+      "loss": 0.1042,
+      "step": 6448
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020279128130251326,
+      "loss": 0.0544,
+      "step": 6449
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020276421959158775,
+      "loss": 0.212,
+      "step": 6450
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002027371559206262,
+      "loss": 0.1955,
+      "step": 6451
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020271009029063398,
+      "loss": 0.1399,
+      "step": 6452
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020268302270261645,
+      "loss": 0.1421,
+      "step": 6453
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002026559531575791,
+      "loss": 0.214,
+      "step": 6454
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020262888165652754,
+      "loss": 0.2137,
+      "step": 6455
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002026018082004674,
+      "loss": 0.1465,
+      "step": 6456
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020257473279040425,
+      "loss": 0.2251,
+      "step": 6457
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020254765542734396,
+      "loss": 0.2908,
+      "step": 6458
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002025205761122924,
+      "loss": 0.1883,
+      "step": 6459
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020249349484625548,
+      "loss": 0.1581,
+      "step": 6460
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020246641163023914,
+      "loss": 0.1487,
+      "step": 6461
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020243932646524945,
+      "loss": 0.1789,
+      "step": 6462
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020241223935229258,
+      "loss": 0.1321,
+      "step": 6463
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020238515029237468,
+      "loss": 0.1683,
+      "step": 6464
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020235805928650212,
+      "loss": 0.2893,
+      "step": 6465
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020233096633568112,
+      "loss": 0.2507,
+      "step": 6466
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020230387144091826,
+      "loss": 0.2429,
+      "step": 6467
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002022767746032199,
+      "loss": 0.1395,
+      "step": 6468
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020224967582359267,
+      "loss": 0.1323,
+      "step": 6469
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020222257510304323,
+      "loss": 0.1403,
+      "step": 6470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00020219547244257826,
+      "loss": 0.2268,
+      "step": 6471
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002021683678432046,
+      "loss": 0.1705,
+      "step": 6472
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020214126130592894,
+      "loss": 0.2158,
+      "step": 6473
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020211415283175838,
+      "loss": 0.1808,
+      "step": 6474
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020208704242169988,
+      "loss": 0.2172,
+      "step": 6475
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002020599300767605,
+      "loss": 0.2252,
+      "step": 6476
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002020328157979474,
+      "loss": 0.2293,
+      "step": 6477
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020200569958626774,
+      "loss": 0.1629,
+      "step": 6478
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002019785814427288,
+      "loss": 0.0803,
+      "step": 6479
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020195146136833808,
+      "loss": 0.1036,
+      "step": 6480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020192433936410287,
+      "loss": 0.0399,
+      "step": 6481
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020189721543103072,
+      "loss": 0.0458,
+      "step": 6482
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020187008957012922,
+      "loss": 0.2962,
+      "step": 6483
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020184296178240598,
+      "loss": 0.14,
+      "step": 6484
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020181583206886877,
+      "loss": 0.2634,
+      "step": 6485
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020178870043052532,
+      "loss": 0.1354,
+      "step": 6486
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020176156686838351,
+      "loss": 0.258,
+      "step": 6487
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020173443138345133,
+      "loss": 0.0825,
+      "step": 6488
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002017072939767367,
+      "loss": 0.1078,
+      "step": 6489
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002016801546492477,
+      "loss": 0.0981,
+      "step": 6490
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020165301340199253,
+      "loss": 0.0807,
+      "step": 6491
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020162587023597937,
+      "loss": 0.0905,
+      "step": 6492
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020159872515221656,
+      "loss": 0.2002,
+      "step": 6493
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002015715781517124,
+      "loss": 0.2222,
+      "step": 6494
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020154442923547535,
+      "loss": 0.2245,
+      "step": 6495
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020151727840451392,
+      "loss": 0.1424,
+      "step": 6496
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020149012565983665,
+      "loss": 0.106,
+      "step": 6497
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020146297100245218,
+      "loss": 0.185,
+      "step": 6498
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002014358144333693,
+      "loss": 0.0394,
+      "step": 6499
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020140865595359673,
+      "loss": 0.2461,
+      "step": 6500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0002013814955641433,
+      "loss": 0.1448,
+      "step": 6501
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020135433326601805,
+      "loss": 0.1059,
+      "step": 6502
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020132716906022993,
+      "loss": 0.104,
+      "step": 6503
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00020130000294778794,
+      "loss": 0.2948,
+      "step": 6504
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002012728349297013,
+      "loss": 0.1422,
+      "step": 6505
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002012456650069792,
+      "loss": 0.1699,
+      "step": 6506
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020121849318063085,
+      "loss": 0.2046,
+      "step": 6507
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002011913194516658,
+      "loss": 0.2251,
+      "step": 6508
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020116414382109326,
+      "loss": 0.2199,
+      "step": 6509
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020113696628992287,
+      "loss": 0.1794,
+      "step": 6510
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002011097868591641,
+      "loss": 0.1328,
+      "step": 6511
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020108260552982664,
+      "loss": 0.1743,
+      "step": 6512
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020105542230292023,
+      "loss": 0.1892,
+      "step": 6513
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020102823717945453,
+      "loss": 0.2045,
+      "step": 6514
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020100105016043946,
+      "loss": 0.0826,
+      "step": 6515
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020097386124688497,
+      "loss": 0.2505,
+      "step": 6516
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020094667043980098,
+      "loss": 0.1627,
+      "step": 6517
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020091947774019764,
+      "loss": 0.392,
+      "step": 6518
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020089228314908498,
+      "loss": 0.1916,
+      "step": 6519
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020086508666747324,
+      "loss": 0.19,
+      "step": 6520
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020083788829637273,
+      "loss": 0.1548,
+      "step": 6521
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020081068803679371,
+      "loss": 0.2684,
+      "step": 6522
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020078348588974666,
+      "loss": 0.1503,
+      "step": 6523
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020075628185624197,
+      "loss": 0.117,
+      "step": 6524
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020072907593729028,
+      "loss": 0.2031,
+      "step": 6525
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020070186813390224,
+      "loss": 0.1588,
+      "step": 6526
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002006746584470884,
+      "loss": 0.2152,
+      "step": 6527
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020064744687785967,
+      "loss": 0.1852,
+      "step": 6528
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020062023342722675,
+      "loss": 0.1448,
+      "step": 6529
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020059301809620064,
+      "loss": 0.1366,
+      "step": 6530
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020056580088579222,
+      "loss": 0.1287,
+      "step": 6531
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020053858179701264,
+      "loss": 0.2308,
+      "step": 6532
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020051136083087294,
+      "loss": 0.1163,
+      "step": 6533
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002004841379883843,
+      "loss": 0.1288,
+      "step": 6534
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00020045691327055793,
+      "loss": 0.1183,
+      "step": 6535
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002004296866784053,
+      "loss": 0.2121,
+      "step": 6536
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002004024582129376,
+      "loss": 0.0691,
+      "step": 6537
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020037522787516644,
+      "loss": 0.2221,
+      "step": 6538
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020034799566610322,
+      "loss": 0.1816,
+      "step": 6539
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020032076158675962,
+      "loss": 0.2292,
+      "step": 6540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020029352563814736,
+      "loss": 0.1559,
+      "step": 6541
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020026628782127807,
+      "loss": 0.1661,
+      "step": 6542
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020023904813716359,
+      "loss": 0.1828,
+      "step": 6543
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020021180658681578,
+      "loss": 0.2098,
+      "step": 6544
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020018456317124662,
+      "loss": 0.1811,
+      "step": 6545
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020015731789146813,
+      "loss": 0.2079,
+      "step": 6546
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020013007074849232,
+      "loss": 0.2222,
+      "step": 6547
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020010282174333136,
+      "loss": 0.0585,
+      "step": 6548
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020007557087699753,
+      "loss": 0.0898,
+      "step": 6549
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002000483181505031,
+      "loss": 0.1046,
+      "step": 6550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00020002106356486045,
+      "loss": 0.1029,
+      "step": 6551
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019999380712108192,
+      "loss": 0.2901,
+      "step": 6552
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019996654882018007,
+      "loss": 0.144,
+      "step": 6553
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019993928866316748,
+      "loss": 0.206,
+      "step": 6554
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019991202665105672,
+      "loss": 0.0751,
+      "step": 6555
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019988476278486058,
+      "loss": 0.0762,
+      "step": 6556
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019985749706559177,
+      "loss": 0.1997,
+      "step": 6557
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019983022949426313,
+      "loss": 0.2234,
+      "step": 6558
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019980296007188764,
+      "loss": 0.0588,
+      "step": 6559
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001997756887994782,
+      "loss": 0.1718,
+      "step": 6560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019974841567804786,
+      "loss": 0.0743,
+      "step": 6561
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001997211407086098,
+      "loss": 0.1517,
+      "step": 6562
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019969386389217714,
+      "loss": 0.1824,
+      "step": 6563
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001996665852297632,
+      "loss": 0.0874,
+      "step": 6564
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019963930472238124,
+      "loss": 0.1585,
+      "step": 6565
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019961202237104469,
+      "loss": 0.2098,
+      "step": 6566
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000199584738176767,
+      "loss": 0.1497,
+      "step": 6567
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019955745214056162,
+      "loss": 0.0977,
+      "step": 6568
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001995301642634423,
+      "loss": 0.2779,
+      "step": 6569
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019950287454642261,
+      "loss": 0.1236,
+      "step": 6570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019947558299051628,
+      "loss": 0.2935,
+      "step": 6571
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001994482895967371,
+      "loss": 0.1249,
+      "step": 6572
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019942099436609902,
+      "loss": 0.2706,
+      "step": 6573
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019939369729961594,
+      "loss": 0.1159,
+      "step": 6574
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001993663983983018,
+      "loss": 0.16,
+      "step": 6575
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019933909766317075,
+      "loss": 0.215,
+      "step": 6576
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001993117950952369,
+      "loss": 0.1649,
+      "step": 6577
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019928449069551445,
+      "loss": 0.0951,
+      "step": 6578
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019925718446501777,
+      "loss": 0.1392,
+      "step": 6579
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019922987640476106,
+      "loss": 0.1139,
+      "step": 6580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001992025665157588,
+      "loss": 0.1607,
+      "step": 6581
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001991752547990255,
+      "loss": 0.21,
+      "step": 6582
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001991479412555757,
+      "loss": 0.192,
+      "step": 6583
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019912062588642402,
+      "loss": 0.0858,
+      "step": 6584
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019909330869258512,
+      "loss": 0.2061,
+      "step": 6585
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019906598967507378,
+      "loss": 0.163,
+      "step": 6586
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001990386688349048,
+      "loss": 0.1742,
+      "step": 6587
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019901134617309304,
+      "loss": 0.152,
+      "step": 6588
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001989840216906535,
+      "loss": 0.099,
+      "step": 6589
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019895669538860122,
+      "loss": 0.1044,
+      "step": 6590
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019892936726795127,
+      "loss": 0.081,
+      "step": 6591
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019890203732971887,
+      "loss": 0.155,
+      "step": 6592
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019887470557491908,
+      "loss": 0.1077,
+      "step": 6593
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019884737200456738,
+      "loss": 0.1093,
+      "step": 6594
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019882003661967902,
+      "loss": 0.0487,
+      "step": 6595
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019879269942126945,
+      "loss": 0.1655,
+      "step": 6596
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001987653604103542,
+      "loss": 0.0898,
+      "step": 6597
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019873801958794878,
+      "loss": 0.1563,
+      "step": 6598
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019871067695506896,
+      "loss": 0.0838,
+      "step": 6599
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019868333251273022,
+      "loss": 0.0995,
+      "step": 6600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019865598626194847,
+      "loss": 0.0735,
+      "step": 6601
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019862863820373954,
+      "loss": 0.0616,
+      "step": 6602
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00019860128833911928,
+      "loss": 0.1407,
+      "step": 6603
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001985739366691037,
+      "loss": 0.1303,
+      "step": 6604
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001985465831947088,
+      "loss": 0.0746,
+      "step": 6605
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019851922791695067,
+      "loss": 0.0898,
+      "step": 6606
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019849187083684555,
+      "loss": 0.0989,
+      "step": 6607
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001984645119554096,
+      "loss": 0.0604,
+      "step": 6608
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019843715127365916,
+      "loss": 0.0652,
+      "step": 6609
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001984097887926106,
+      "loss": 0.107,
+      "step": 6610
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019838242451328036,
+      "loss": 0.1094,
+      "step": 6611
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001983550584366849,
+      "loss": 0.0517,
+      "step": 6612
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019832769056384082,
+      "loss": 0.1035,
+      "step": 6613
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019830032089576476,
+      "loss": 0.0379,
+      "step": 6614
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019827294943347345,
+      "loss": 0.1229,
+      "step": 6615
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019824557617798358,
+      "loss": 0.1159,
+      "step": 6616
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001982182011303121,
+      "loss": 0.0655,
+      "step": 6617
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019819082429147579,
+      "loss": 0.0521,
+      "step": 6618
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019816344566249167,
+      "loss": 0.1184,
+      "step": 6619
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001981360652443768,
+      "loss": 0.0613,
+      "step": 6620
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019810868303814824,
+      "loss": 0.129,
+      "step": 6621
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019808129904482324,
+      "loss": 0.0893,
+      "step": 6622
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019805391326541892,
+      "loss": 0.088,
+      "step": 6623
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019802652570095268,
+      "loss": 0.0892,
+      "step": 6624
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019799913635244182,
+      "loss": 0.0649,
+      "step": 6625
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019797174522090383,
+      "loss": 0.0533,
+      "step": 6626
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019794435230735617,
+      "loss": 0.064,
+      "step": 6627
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019791695761281638,
+      "loss": 0.058,
+      "step": 6628
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001978895611383022,
+      "loss": 0.1334,
+      "step": 6629
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019786216288483118,
+      "loss": 0.1659,
+      "step": 6630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001978347628534212,
+      "loss": 0.055,
+      "step": 6631
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019780736104509008,
+      "loss": 0.1252,
+      "step": 6632
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019777995746085564,
+      "loss": 0.076,
+      "step": 6633
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019775255210173592,
+      "loss": 0.1098,
+      "step": 6634
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019772514496874897,
+      "loss": 0.0832,
+      "step": 6635
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00019769773606291276,
+      "loss": 0.0476,
+      "step": 6636
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019767032538524555,
+      "loss": 0.1565,
+      "step": 6637
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019764291293676553,
+      "loss": 0.0933,
+      "step": 6638
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019761549871849104,
+      "loss": 0.2262,
+      "step": 6639
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001975880827314404,
+      "loss": 0.0698,
+      "step": 6640
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019756066497663203,
+      "loss": 0.1167,
+      "step": 6641
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019753324545508443,
+      "loss": 0.095,
+      "step": 6642
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019750582416781616,
+      "loss": 0.0433,
+      "step": 6643
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019747840111584582,
+      "loss": 0.11,
+      "step": 6644
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019745097630019211,
+      "loss": 0.1337,
+      "step": 6645
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019742354972187377,
+      "loss": 0.1275,
+      "step": 6646
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001973961213819096,
+      "loss": 0.1213,
+      "step": 6647
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019736869128131858,
+      "loss": 0.0768,
+      "step": 6648
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019734125942111953,
+      "loss": 0.0968,
+      "step": 6649
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019731382580233156,
+      "loss": 0.1606,
+      "step": 6650
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001972863904259737,
+      "loss": 0.0982,
+      "step": 6651
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019725895329306505,
+      "loss": 0.1728,
+      "step": 6652
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001972315144046249,
+      "loss": 0.1964,
+      "step": 6653
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001972040737616725,
+      "loss": 0.0194,
+      "step": 6654
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019717663136522721,
+      "loss": 0.0423,
+      "step": 6655
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019714918721630838,
+      "loss": 0.0852,
+      "step": 6656
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001971217413159355,
+      "loss": 0.0712,
+      "step": 6657
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019709429366512817,
+      "loss": 0.0652,
+      "step": 6658
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001970668442649059,
+      "loss": 0.192,
+      "step": 6659
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001970393931162884,
+      "loss": 0.0592,
+      "step": 6660
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019701194022029537,
+      "loss": 0.0797,
+      "step": 6661
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019698448557794654,
+      "loss": 0.0986,
+      "step": 6662
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019695702919026196,
+      "loss": 0.0781,
+      "step": 6663
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019692957105826143,
+      "loss": 0.132,
+      "step": 6664
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019690211118296495,
+      "loss": 0.1275,
+      "step": 6665
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001968746495653926,
+      "loss": 0.0971,
+      "step": 6666
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001968471862065644,
+      "loss": 0.128,
+      "step": 6667
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019681972110750068,
+      "loss": 0.1529,
+      "step": 6668
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00019679225426922159,
+      "loss": 0.1286,
+      "step": 6669
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019676478569274748,
+      "loss": 0.0492,
+      "step": 6670
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001967373153790987,
+      "loss": 0.0985,
+      "step": 6671
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019670984332929574,
+      "loss": 0.1038,
+      "step": 6672
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019668236954435906,
+      "loss": 0.1182,
+      "step": 6673
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019665489402530926,
+      "loss": 0.1697,
+      "step": 6674
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019662741677316692,
+      "loss": 0.1379,
+      "step": 6675
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019659993778895288,
+      "loss": 0.0494,
+      "step": 6676
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019657245707368768,
+      "loss": 0.1192,
+      "step": 6677
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019654497462839234,
+      "loss": 0.0522,
+      "step": 6678
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019651749045408764,
+      "loss": 0.1998,
+      "step": 6679
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019649000455179463,
+      "loss": 0.0286,
+      "step": 6680
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001964625169225343,
+      "loss": 0.0266,
+      "step": 6681
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019643502756732766,
+      "loss": 0.0291,
+      "step": 6682
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019640753648719598,
+      "loss": 0.1002,
+      "step": 6683
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019638004368316035,
+      "loss": 0.033,
+      "step": 6684
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001963525491562421,
+      "loss": 0.1245,
+      "step": 6685
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019632505290746258,
+      "loss": 0.0674,
+      "step": 6686
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019629755493784322,
+      "loss": 0.131,
+      "step": 6687
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019627005524840548,
+      "loss": 0.0997,
+      "step": 6688
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019624255384017084,
+      "loss": 0.103,
+      "step": 6689
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019621505071416092,
+      "loss": 0.1327,
+      "step": 6690
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001961875458713974,
+      "loss": 0.1411,
+      "step": 6691
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000196160039312902,
+      "loss": 0.0855,
+      "step": 6692
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019613253103969652,
+      "loss": 0.0136,
+      "step": 6693
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019610502105280272,
+      "loss": 0.1577,
+      "step": 6694
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019607750935324268,
+      "loss": 0.1101,
+      "step": 6695
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019604999594203826,
+      "loss": 0.0929,
+      "step": 6696
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019602248082021151,
+      "loss": 0.1223,
+      "step": 6697
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001959949639887846,
+      "loss": 0.1024,
+      "step": 6698
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019596744544877963,
+      "loss": 0.0891,
+      "step": 6699
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019593992520121884,
+      "loss": 0.0462,
+      "step": 6700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001959124032471246,
+      "loss": 0.0461,
+      "step": 6701
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00019588487958751918,
+      "loss": 0.1957,
+      "step": 6702
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.000195857354223425,
+      "loss": 0.1323,
+      "step": 6703
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019582982715586465,
+      "loss": 0.0561,
+      "step": 6704
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019580229838586056,
+      "loss": 0.1469,
+      "step": 6705
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019577476791443548,
+      "loss": 0.0741,
+      "step": 6706
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019574723574261195,
+      "loss": 0.1129,
+      "step": 6707
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019571970187141276,
+      "loss": 0.1693,
+      "step": 6708
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019569216630186073,
+      "loss": 0.0876,
+      "step": 6709
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019566462903497874,
+      "loss": 0.1046,
+      "step": 6710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019563709007178962,
+      "loss": 0.0408,
+      "step": 6711
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019560954941331648,
+      "loss": 0.1421,
+      "step": 6712
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019558200706058231,
+      "loss": 0.0488,
+      "step": 6713
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019555446301461031,
+      "loss": 0.1088,
+      "step": 6714
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019552691727642356,
+      "loss": 0.1136,
+      "step": 6715
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019549936984704534,
+      "loss": 0.0746,
+      "step": 6716
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019547182072749894,
+      "loss": 0.1083,
+      "step": 6717
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019544426991880773,
+      "loss": 0.118,
+      "step": 6718
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019541671742199517,
+      "loss": 0.1519,
+      "step": 6719
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019538916323808475,
+      "loss": 0.137,
+      "step": 6720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019536160736810003,
+      "loss": 0.0933,
+      "step": 6721
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001953340498130646,
+      "loss": 0.1319,
+      "step": 6722
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019530649057400215,
+      "loss": 0.1207,
+      "step": 6723
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019527892965193647,
+      "loss": 0.0669,
+      "step": 6724
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001952513670478913,
+      "loss": 0.0289,
+      "step": 6725
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019522380276289057,
+      "loss": 0.0891,
+      "step": 6726
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019519623679795816,
+      "loss": 0.0542,
+      "step": 6727
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019516866915411811,
+      "loss": 0.1598,
+      "step": 6728
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019514109983239443,
+      "loss": 0.0753,
+      "step": 6729
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001951135288338113,
+      "loss": 0.1523,
+      "step": 6730
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019508595615939288,
+      "loss": 0.0539,
+      "step": 6731
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019505838181016336,
+      "loss": 0.1045,
+      "step": 6732
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019503080578714708,
+      "loss": 0.1254,
+      "step": 6733
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019500322809136848,
+      "loss": 0.1324,
+      "step": 6734
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00019497564872385183,
+      "loss": 0.1417,
+      "step": 6735
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001949480676856218,
+      "loss": 0.0855,
+      "step": 6736
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019492048497770284,
+      "loss": 0.0412,
+      "step": 6737
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001948929006011196,
+      "loss": 0.0699,
+      "step": 6738
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019486531455689674,
+      "loss": 0.1477,
+      "step": 6739
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019483772684605904,
+      "loss": 0.0529,
+      "step": 6740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019481013746963123,
+      "loss": 0.1098,
+      "step": 6741
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019478254642863824,
+      "loss": 0.0578,
+      "step": 6742
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019475495372410498,
+      "loss": 0.0872,
+      "step": 6743
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001947273593570564,
+      "loss": 0.0542,
+      "step": 6744
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019469976332851758,
+      "loss": 0.1131,
+      "step": 6745
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019467216563951364,
+      "loss": 0.1369,
+      "step": 6746
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019464456629106978,
+      "loss": 0.083,
+      "step": 6747
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001946169652842112,
+      "loss": 0.0044,
+      "step": 6748
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001945893626199632,
+      "loss": 0.1087,
+      "step": 6749
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019456175829935112,
+      "loss": 0.0891,
+      "step": 6750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019453415232340037,
+      "loss": 0.1108,
+      "step": 6751
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019450654469313649,
+      "loss": 0.1154,
+      "step": 6752
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019447893540958496,
+      "loss": 0.1962,
+      "step": 6753
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019445132447377145,
+      "loss": 0.1238,
+      "step": 6754
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019442371188672155,
+      "loss": 0.0531,
+      "step": 6755
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019439609764946105,
+      "loss": 0.0556,
+      "step": 6756
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019436848176301573,
+      "loss": 0.1588,
+      "step": 6757
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001943408642284114,
+      "loss": 0.0644,
+      "step": 6758
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019431324504667397,
+      "loss": 0.125,
+      "step": 6759
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019428562421882942,
+      "loss": 0.1123,
+      "step": 6760
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019425800174590385,
+      "loss": 0.1773,
+      "step": 6761
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019423037762892332,
+      "loss": 0.1005,
+      "step": 6762
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019420275186891393,
+      "loss": 0.0391,
+      "step": 6763
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001941751244669019,
+      "loss": 0.1008,
+      "step": 6764
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001941474954239136,
+      "loss": 0.0527,
+      "step": 6765
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019411986474097523,
+      "loss": 0.0628,
+      "step": 6766
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019409223241911332,
+      "loss": 0.1305,
+      "step": 6767
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00019406459845935426,
+      "loss": 0.1731,
+      "step": 6768
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019403696286272458,
+      "loss": 0.0847,
+      "step": 6769
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019400932563025088,
+      "loss": 0.1286,
+      "step": 6770
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001939816867629598,
+      "loss": 0.1458,
+      "step": 6771
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019395404626187802,
+      "loss": 0.028,
+      "step": 6772
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001939264041280323,
+      "loss": 0.0413,
+      "step": 6773
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001938987603624495,
+      "loss": 0.2061,
+      "step": 6774
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019387111496615652,
+      "loss": 0.0726,
+      "step": 6775
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001938434679401802,
+      "loss": 0.1517,
+      "step": 6776
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019381581928554765,
+      "loss": 0.1344,
+      "step": 6777
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001937881690032859,
+      "loss": 0.1009,
+      "step": 6778
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019376051709442204,
+      "loss": 0.0835,
+      "step": 6779
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019373286355998338,
+      "loss": 0.0533,
+      "step": 6780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019370520840099702,
+      "loss": 0.1652,
+      "step": 6781
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019367755161849037,
+      "loss": 0.1197,
+      "step": 6782
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019364989321349075,
+      "loss": 0.122,
+      "step": 6783
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019362223318702555,
+      "loss": 0.1332,
+      "step": 6784
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019359457154012236,
+      "loss": 0.0863,
+      "step": 6785
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019356690827380865,
+      "loss": 0.1003,
+      "step": 6786
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019353924338911212,
+      "loss": 0.1503,
+      "step": 6787
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019351157688706028,
+      "loss": 0.1316,
+      "step": 6788
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000193483908768681,
+      "loss": 0.0951,
+      "step": 6789
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019345623903500205,
+      "loss": 0.0584,
+      "step": 6790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019342856768705122,
+      "loss": 0.1735,
+      "step": 6791
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019340089472585646,
+      "loss": 0.0847,
+      "step": 6792
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019337322015244572,
+      "loss": 0.1156,
+      "step": 6793
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019334554396784707,
+      "loss": 0.0893,
+      "step": 6794
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019331786617308858,
+      "loss": 0.077,
+      "step": 6795
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019329018676919837,
+      "loss": 0.0571,
+      "step": 6796
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019326250575720468,
+      "loss": 0.0493,
+      "step": 6797
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019323482313813575,
+      "loss": 0.0843,
+      "step": 6798
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019320713891301996,
+      "loss": 0.0991,
+      "step": 6799
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019317945308288564,
+      "loss": 0.0941,
+      "step": 6800
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00019315176564876124,
+      "loss": 0.1398,
+      "step": 6801
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019312407661167526,
+      "loss": 0.0865,
+      "step": 6802
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019309638597265637,
+      "loss": 0.0612,
+      "step": 6803
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019306869373273309,
+      "loss": 0.187,
+      "step": 6804
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019304099989293417,
+      "loss": 0.1163,
+      "step": 6805
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019301330445428827,
+      "loss": 0.097,
+      "step": 6806
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019298560741782425,
+      "loss": 0.1636,
+      "step": 6807
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019295790878457096,
+      "loss": 0.0572,
+      "step": 6808
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019293020855555733,
+      "loss": 0.0586,
+      "step": 6809
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001929025067318124,
+      "loss": 0.0597,
+      "step": 6810
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001928748033143651,
+      "loss": 0.1451,
+      "step": 6811
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001928470983042446,
+      "loss": 0.0451,
+      "step": 6812
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001928193917024801,
+      "loss": 0.1297,
+      "step": 6813
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019279168351010072,
+      "loss": 0.1347,
+      "step": 6814
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001927639737281358,
+      "loss": 0.0791,
+      "step": 6815
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001927362623576146,
+      "loss": 0.1127,
+      "step": 6816
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019270854939956662,
+      "loss": 0.0865,
+      "step": 6817
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019268083485502125,
+      "loss": 0.1423,
+      "step": 6818
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019265311872500806,
+      "loss": 0.1259,
+      "step": 6819
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001926254010105566,
+      "loss": 0.1809,
+      "step": 6820
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019259768171269646,
+      "loss": 0.1344,
+      "step": 6821
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019256996083245734,
+      "loss": 0.0587,
+      "step": 6822
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019254223837086908,
+      "loss": 0.0887,
+      "step": 6823
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019251451432896133,
+      "loss": 0.0024,
+      "step": 6824
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001924867887077641,
+      "loss": 0.0926,
+      "step": 6825
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019245906150830729,
+      "loss": 0.1559,
+      "step": 6826
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019243133273162078,
+      "loss": 0.0376,
+      "step": 6827
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019240360237873476,
+      "loss": 0.074,
+      "step": 6828
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019237587045067923,
+      "loss": 0.0881,
+      "step": 6829
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019234813694848433,
+      "loss": 0.0918,
+      "step": 6830
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019232040187318038,
+      "loss": 0.0528,
+      "step": 6831
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001922926652257976,
+      "loss": 0.0527,
+      "step": 6832
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001922649270073663,
+      "loss": 0.1008,
+      "step": 6833
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00019223718721891693,
+      "loss": 0.1022,
+      "step": 6834
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001922094458614799,
+      "loss": 0.1322,
+      "step": 6835
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019218170293608579,
+      "loss": 0.0727,
+      "step": 6836
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019215395844376506,
+      "loss": 0.0393,
+      "step": 6837
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001921262123855484,
+      "loss": 0.1284,
+      "step": 6838
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019209846476246647,
+      "loss": 0.0934,
+      "step": 6839
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019207071557555004,
+      "loss": 0.0513,
+      "step": 6840
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019204296482582987,
+      "loss": 0.0903,
+      "step": 6841
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019201521251433685,
+      "loss": 0.1626,
+      "step": 6842
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019198745864210196,
+      "loss": 0.131,
+      "step": 6843
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019195970321015605,
+      "loss": 0.0374,
+      "step": 6844
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019193194621953022,
+      "loss": 0.0494,
+      "step": 6845
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019190418767125558,
+      "loss": 0.0856,
+      "step": 6846
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019187642756636323,
+      "loss": 0.1011,
+      "step": 6847
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019184866590588439,
+      "loss": 0.0963,
+      "step": 6848
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019182090269085034,
+      "loss": 0.1409,
+      "step": 6849
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019179313792229235,
+      "loss": 0.1167,
+      "step": 6850
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019176537160124193,
+      "loss": 0.0727,
+      "step": 6851
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019173760372873038,
+      "loss": 0.09,
+      "step": 6852
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019170983430578928,
+      "loss": 0.1231,
+      "step": 6853
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001916820633334501,
+      "loss": 0.1608,
+      "step": 6854
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001916542908127445,
+      "loss": 0.0677,
+      "step": 6855
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001916265167447042,
+      "loss": 0.1052,
+      "step": 6856
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001915987411303608,
+      "loss": 0.105,
+      "step": 6857
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001915709639707462,
+      "loss": 0.1099,
+      "step": 6858
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019154318526689218,
+      "loss": 0.0617,
+      "step": 6859
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019151540501983065,
+      "loss": 0.092,
+      "step": 6860
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001914876232305936,
+      "loss": 0.0868,
+      "step": 6861
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.000191459839900213,
+      "loss": 0.0767,
+      "step": 6862
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019143205502972089,
+      "loss": 0.1354,
+      "step": 6863
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019140426862014948,
+      "loss": 0.0708,
+      "step": 6864
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019137648067253085,
+      "loss": 0.0557,
+      "step": 6865
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00019134869118789732,
+      "loss": 0.1484,
+      "step": 6866
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001913209001672812,
+      "loss": 0.0435,
+      "step": 6867
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019129310761171475,
+      "loss": 0.1121,
+      "step": 6868
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001912653135222305,
+      "loss": 0.1448,
+      "step": 6869
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019123751789986085,
+      "loss": 0.0728,
+      "step": 6870
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019120972074563835,
+      "loss": 0.0566,
+      "step": 6871
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019118192206059553,
+      "loss": 0.042,
+      "step": 6872
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019115412184576508,
+      "loss": 0.1299,
+      "step": 6873
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019112632010217972,
+      "loss": 0.0704,
+      "step": 6874
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019109851683087213,
+      "loss": 0.1165,
+      "step": 6875
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001910707120328752,
+      "loss": 0.1006,
+      "step": 6876
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019104290570922172,
+      "loss": 0.0727,
+      "step": 6877
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001910150978609447,
+      "loss": 0.1507,
+      "step": 6878
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019098728848907705,
+      "loss": 0.0571,
+      "step": 6879
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001909594775946518,
+      "loss": 0.0951,
+      "step": 6880
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001909316651787021,
+      "loss": 0.0538,
+      "step": 6881
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000190903851242261,
+      "loss": 0.0789,
+      "step": 6882
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019087603578636184,
+      "loss": 0.0806,
+      "step": 6883
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019084821881203782,
+      "loss": 0.1727,
+      "step": 6884
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019082040032032224,
+      "loss": 0.1649,
+      "step": 6885
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019079258031224852,
+      "loss": 0.1255,
+      "step": 6886
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019076475878885004,
+      "loss": 0.0931,
+      "step": 6887
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019073693575116029,
+      "loss": 0.0924,
+      "step": 6888
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019070911120021292,
+      "loss": 0.1795,
+      "step": 6889
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019068128513704134,
+      "loss": 0.1555,
+      "step": 6890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019065345756267933,
+      "loss": 0.0542,
+      "step": 6891
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019062562847816064,
+      "loss": 0.1238,
+      "step": 6892
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019059779788451898,
+      "loss": 0.0683,
+      "step": 6893
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019056996578278819,
+      "loss": 0.0838,
+      "step": 6894
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019054213217400214,
+      "loss": 0.0588,
+      "step": 6895
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019051429705919473,
+      "loss": 0.0916,
+      "step": 6896
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00019048646043940003,
+      "loss": 0.0752,
+      "step": 6897
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.000190458622315652,
+      "loss": 0.0883,
+      "step": 6898
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001904307826889848,
+      "loss": 0.1338,
+      "step": 6899
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019040294156043262,
+      "loss": 0.068,
+      "step": 6900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019037509893102963,
+      "loss": 0.1015,
+      "step": 6901
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019034725480181016,
+      "loss": 0.0631,
+      "step": 6902
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019031940917380843,
+      "loss": 0.1185,
+      "step": 6903
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001902915620480589,
+      "loss": 0.1999,
+      "step": 6904
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000190263713425596,
+      "loss": 0.0741,
+      "step": 6905
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001902358633074542,
+      "loss": 0.0719,
+      "step": 6906
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001902080116946681,
+      "loss": 0.0839,
+      "step": 6907
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019018015858827223,
+      "loss": 0.1287,
+      "step": 6908
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019015230398930136,
+      "loss": 0.0932,
+      "step": 6909
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019012444789879012,
+      "loss": 0.1029,
+      "step": 6910
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019009659031777326,
+      "loss": 0.1087,
+      "step": 6911
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001900687312472857,
+      "loss": 0.0988,
+      "step": 6912
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019004087068836225,
+      "loss": 0.0862,
+      "step": 6913
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00019001300864203788,
+      "loss": 0.071,
+      "step": 6914
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018998514510934754,
+      "loss": 0.1223,
+      "step": 6915
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018995728009132636,
+      "loss": 0.1725,
+      "step": 6916
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018992941358900943,
+      "loss": 0.054,
+      "step": 6917
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018990154560343182,
+      "loss": 0.1442,
+      "step": 6918
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018987367613562884,
+      "loss": 0.0969,
+      "step": 6919
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001898458051866357,
+      "loss": 0.0815,
+      "step": 6920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018981793275748773,
+      "loss": 0.0764,
+      "step": 6921
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018979005884922035,
+      "loss": 0.1018,
+      "step": 6922
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018976218346286892,
+      "loss": 0.1868,
+      "step": 6923
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018973430659946898,
+      "loss": 0.0777,
+      "step": 6924
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018970642826005612,
+      "loss": 0.1001,
+      "step": 6925
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018967854844566587,
+      "loss": 0.0625,
+      "step": 6926
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018965066715733394,
+      "loss": 0.0851,
+      "step": 6927
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001896227843960959,
+      "loss": 0.093,
+      "step": 6928
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001895949001629877,
+      "loss": 0.0771,
+      "step": 6929
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018956701445904504,
+      "loss": 0.1545,
+      "step": 6930
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018953912728530381,
+      "loss": 0.052,
+      "step": 6931
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00018951123864279995,
+      "loss": 0.1566,
+      "step": 6932
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018948334853256945,
+      "loss": 0.1371,
+      "step": 6933
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001894554569556483,
+      "loss": 0.0961,
+      "step": 6934
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018942756391307272,
+      "loss": 0.1009,
+      "step": 6935
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001893996694058787,
+      "loss": 0.0742,
+      "step": 6936
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018937177343510256,
+      "loss": 0.0839,
+      "step": 6937
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018934387600178038,
+      "loss": 0.128,
+      "step": 6938
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018931597710694864,
+      "loss": 0.1033,
+      "step": 6939
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018928807675164364,
+      "loss": 0.1062,
+      "step": 6940
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001892601749369018,
+      "loss": 0.0826,
+      "step": 6941
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018923227166375964,
+      "loss": 0.1013,
+      "step": 6942
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018920436693325358,
+      "loss": 0.1458,
+      "step": 6943
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018917646074642026,
+      "loss": 0.0745,
+      "step": 6944
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018914855310429632,
+      "loss": 0.1108,
+      "step": 6945
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001891206440079184,
+      "loss": 0.1596,
+      "step": 6946
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001890927334583233,
+      "loss": 0.0752,
+      "step": 6947
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001890648214565478,
+      "loss": 0.1447,
+      "step": 6948
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001890369080036287,
+      "loss": 0.0649,
+      "step": 6949
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018900899310060298,
+      "loss": 0.088,
+      "step": 6950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018898107674850757,
+      "loss": 0.1152,
+      "step": 6951
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018895315894837945,
+      "loss": 0.0679,
+      "step": 6952
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001889252397012557,
+      "loss": 0.0713,
+      "step": 6953
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018889731900817346,
+      "loss": 0.0633,
+      "step": 6954
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018886939687016985,
+      "loss": 0.1086,
+      "step": 6955
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018884147328828213,
+      "loss": 0.1631,
+      "step": 6956
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018881354826354762,
+      "loss": 0.1477,
+      "step": 6957
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018878562179700363,
+      "loss": 0.0524,
+      "step": 6958
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018875769388968747,
+      "loss": 0.0758,
+      "step": 6959
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018872976454263672,
+      "loss": 0.1016,
+      "step": 6960
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001887018337568887,
+      "loss": 0.0699,
+      "step": 6961
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018867390153348107,
+      "loss": 0.0852,
+      "step": 6962
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018864596787345145,
+      "loss": 0.1186,
+      "step": 6963
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00018861803277783735,
+      "loss": 0.1142,
+      "step": 6964
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001885900962476767,
+      "loss": 0.107,
+      "step": 6965
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018856215828400705,
+      "loss": 0.042,
+      "step": 6966
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018853421888786632,
+      "loss": 0.135,
+      "step": 6967
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018850627806029244,
+      "loss": 0.0868,
+      "step": 6968
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018847833580232318,
+      "loss": 0.0399,
+      "step": 6969
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018845039211499661,
+      "loss": 0.0396,
+      "step": 6970
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001884224469993507,
+      "loss": 0.155,
+      "step": 6971
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018839450045642355,
+      "loss": 0.1082,
+      "step": 6972
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001883665524872533,
+      "loss": 0.0772,
+      "step": 6973
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018833860309287813,
+      "loss": 0.1644,
+      "step": 6974
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018831065227433634,
+      "loss": 0.1268,
+      "step": 6975
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018828270003266612,
+      "loss": 0.121,
+      "step": 6976
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018825474636890584,
+      "loss": 0.012,
+      "step": 6977
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018822679128409396,
+      "loss": 0.1896,
+      "step": 6978
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018819883477926884,
+      "loss": 0.0605,
+      "step": 6979
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018817087685546902,
+      "loss": 0.1233,
+      "step": 6980
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018814291751373307,
+      "loss": 0.1312,
+      "step": 6981
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001881149567550996,
+      "loss": 0.072,
+      "step": 6982
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018808699458060729,
+      "loss": 0.0884,
+      "step": 6983
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018805903099129479,
+      "loss": 0.0887,
+      "step": 6984
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018803106598820087,
+      "loss": 0.1176,
+      "step": 6985
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018800309957236441,
+      "loss": 0.1996,
+      "step": 6986
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018797513174482423,
+      "loss": 0.0939,
+      "step": 6987
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018794716250661924,
+      "loss": 0.0725,
+      "step": 6988
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018791919185878845,
+      "loss": 0.0934,
+      "step": 6989
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018789121980237092,
+      "loss": 0.0813,
+      "step": 6990
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018786324633840567,
+      "loss": 0.101,
+      "step": 6991
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018783527146793184,
+      "loss": 0.168,
+      "step": 6992
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018780729519198865,
+      "loss": 0.0703,
+      "step": 6993
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001877793175116153,
+      "loss": 0.06,
+      "step": 6994
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018775133842785112,
+      "loss": 0.0935,
+      "step": 6995
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018772335794173535,
+      "loss": 0.0978,
+      "step": 6996
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001876953760543075,
+      "loss": 0.0725,
+      "step": 6997
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00018766739276660698,
+      "loss": 0.0955,
+      "step": 6998
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018763940807967324,
+      "loss": 0.1384,
+      "step": 6999
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018761142199454588,
+      "loss": 0.044,
+      "step": 7000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018758343451226453,
+      "loss": 0.1332,
+      "step": 7001
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018755544563386878,
+      "loss": 0.0971,
+      "step": 7002
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018752745536039834,
+      "loss": 0.0795,
+      "step": 7003
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.000187499463692893,
+      "loss": 0.0444,
+      "step": 7004
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001874714706323925,
+      "loss": 0.1008,
+      "step": 7005
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001874434761799368,
+      "loss": 0.1279,
+      "step": 7006
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018741548033656574,
+      "loss": 0.0717,
+      "step": 7007
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018738748310331933,
+      "loss": 0.0627,
+      "step": 7008
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018735948448123755,
+      "loss": 0.1358,
+      "step": 7009
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018733148447136046,
+      "loss": 0.059,
+      "step": 7010
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018730348307472824,
+      "loss": 0.1735,
+      "step": 7011
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018727548029238097,
+      "loss": 0.1379,
+      "step": 7012
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018724747612535892,
+      "loss": 0.0877,
+      "step": 7013
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018721947057470236,
+      "loss": 0.1367,
+      "step": 7014
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018719146364145158,
+      "loss": 0.0954,
+      "step": 7015
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018716345532664708,
+      "loss": 0.1376,
+      "step": 7016
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018713544563132912,
+      "loss": 0.0703,
+      "step": 7017
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001871074345565383,
+      "loss": 0.1185,
+      "step": 7018
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001870794221033151,
+      "loss": 0.0987,
+      "step": 7019
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018705140827270008,
+      "loss": 0.1107,
+      "step": 7020
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018702339306573388,
+      "loss": 0.1261,
+      "step": 7021
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018699537648345722,
+      "loss": 0.0757,
+      "step": 7022
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018696735852691078,
+      "loss": 0.1283,
+      "step": 7023
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018693933919713545,
+      "loss": 0.0372,
+      "step": 7024
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001869113184951719,
+      "loss": 0.101,
+      "step": 7025
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018688329642206123,
+      "loss": 0.109,
+      "step": 7026
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018685527297884416,
+      "loss": 0.0886,
+      "step": 7027
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001868272481665618,
+      "loss": 0.0949,
+      "step": 7028
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018679922198625517,
+      "loss": 0.0825,
+      "step": 7029
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00018677119443896533,
+      "loss": 0.044,
+      "step": 7030
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001867431655257335,
+      "loss": 0.1213,
+      "step": 7031
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001867151352476008,
+      "loss": 0.0999,
+      "step": 7032
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018668710360560844,
+      "loss": 0.0813,
+      "step": 7033
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018665907060079785,
+      "loss": 0.0902,
+      "step": 7034
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018663103623421022,
+      "loss": 0.1231,
+      "step": 7035
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018660300050688707,
+      "loss": 0.0439,
+      "step": 7036
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018657496341986974,
+      "loss": 0.1168,
+      "step": 7037
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018654692497419977,
+      "loss": 0.0961,
+      "step": 7038
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018651888517091873,
+      "loss": 0.0769,
+      "step": 7039
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001864908440110682,
+      "loss": 0.1399,
+      "step": 7040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018646280149568986,
+      "loss": 0.1154,
+      "step": 7041
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018643475762582533,
+      "loss": 0.1107,
+      "step": 7042
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001864067124025164,
+      "loss": 0.1108,
+      "step": 7043
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001863786658268049,
+      "loss": 0.0503,
+      "step": 7044
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001863506178997326,
+      "loss": 0.0908,
+      "step": 7045
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018632256862234143,
+      "loss": 0.1009,
+      "step": 7046
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018629451799567336,
+      "loss": 0.0668,
+      "step": 7047
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001862664660207704,
+      "loss": 0.1189,
+      "step": 7048
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018623841269867458,
+      "loss": 0.1478,
+      "step": 7049
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000186210358030428,
+      "loss": 0.0926,
+      "step": 7050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018618230201707277,
+      "loss": 0.1227,
+      "step": 7051
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018615424465965116,
+      "loss": 0.1357,
+      "step": 7052
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018612618595920536,
+      "loss": 0.1055,
+      "step": 7053
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001860981259167777,
+      "loss": 0.1229,
+      "step": 7054
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018607006453341053,
+      "loss": 0.1215,
+      "step": 7055
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001860420018101462,
+      "loss": 0.0706,
+      "step": 7056
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018601393774802727,
+      "loss": 0.0737,
+      "step": 7057
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018598587234809608,
+      "loss": 0.1559,
+      "step": 7058
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018595780561139532,
+      "loss": 0.0647,
+      "step": 7059
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001859297375389675,
+      "loss": 0.0876,
+      "step": 7060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018590166813185527,
+      "loss": 0.1107,
+      "step": 7061
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018587359739110138,
+      "loss": 0.0441,
+      "step": 7062
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018584552531774852,
+      "loss": 0.0721,
+      "step": 7063
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00018581745191283957,
+      "loss": 0.0408,
+      "step": 7064
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018578937717741726,
+      "loss": 0.1254,
+      "step": 7065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018576130111252456,
+      "loss": 0.14,
+      "step": 7066
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001857332237192044,
+      "loss": 0.1456,
+      "step": 7067
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018570514499849976,
+      "loss": 0.1317,
+      "step": 7068
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018567706495145364,
+      "loss": 0.0682,
+      "step": 7069
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018564898357910923,
+      "loss": 0.0575,
+      "step": 7070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018562090088250957,
+      "loss": 0.069,
+      "step": 7071
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000185592816862698,
+      "loss": 0.0594,
+      "step": 7072
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001855647315207176,
+      "loss": 0.1347,
+      "step": 7073
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001855366448576117,
+      "loss": 0.0844,
+      "step": 7074
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001855085568744237,
+      "loss": 0.0368,
+      "step": 7075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018548046757219685,
+      "loss": 0.0163,
+      "step": 7076
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001854523769519748,
+      "loss": 0.1097,
+      "step": 7077
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001854242850148008,
+      "loss": 0.0647,
+      "step": 7078
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018539619176171854,
+      "loss": 0.1263,
+      "step": 7079
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018536809719377158,
+      "loss": 0.1007,
+      "step": 7080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018534000131200347,
+      "loss": 0.063,
+      "step": 7081
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.000185311904117458,
+      "loss": 0.1415,
+      "step": 7082
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001852838056111788,
+      "loss": 0.0647,
+      "step": 7083
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018525570579420967,
+      "loss": 0.0617,
+      "step": 7084
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018522760466759455,
+      "loss": 0.0855,
+      "step": 7085
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018519950223237715,
+      "loss": 0.0477,
+      "step": 7086
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018517139848960146,
+      "loss": 0.1893,
+      "step": 7087
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018514329344031144,
+      "loss": 0.1569,
+      "step": 7088
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018511518708555117,
+      "loss": 0.2028,
+      "step": 7089
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001850870794263647,
+      "loss": 0.0911,
+      "step": 7090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018505897046379606,
+      "loss": 0.1217,
+      "step": 7091
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001850308601988895,
+      "loss": 0.1481,
+      "step": 7092
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001850027486326892,
+      "loss": 0.1128,
+      "step": 7093
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018497463576623945,
+      "loss": 0.0888,
+      "step": 7094
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018494652160058452,
+      "loss": 0.0565,
+      "step": 7095
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001849184061367688,
+      "loss": 0.1076,
+      "step": 7096
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00018489028937583674,
+      "loss": 0.1182,
+      "step": 7097
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018486217131883268,
+      "loss": 0.1169,
+      "step": 7098
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001848340519668012,
+      "loss": 0.0376,
+      "step": 7099
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018480593132078686,
+      "loss": 0.0615,
+      "step": 7100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018477780938183424,
+      "loss": 0.0566,
+      "step": 7101
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018474968615098797,
+      "loss": 0.1208,
+      "step": 7102
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018472156162929276,
+      "loss": 0.1887,
+      "step": 7103
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018469343581779335,
+      "loss": 0.0578,
+      "step": 7104
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018466530871753456,
+      "loss": 0.1157,
+      "step": 7105
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001846371803295612,
+      "loss": 0.0639,
+      "step": 7106
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001846090506549182,
+      "loss": 0.086,
+      "step": 7107
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001845809196946504,
+      "loss": 0.0792,
+      "step": 7108
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018455278744980286,
+      "loss": 0.0733,
+      "step": 7109
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018452465392142058,
+      "loss": 0.008,
+      "step": 7110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018449651911054863,
+      "loss": 0.1453,
+      "step": 7111
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001844683830182322,
+      "loss": 0.1071,
+      "step": 7112
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018444024564551642,
+      "loss": 0.1842,
+      "step": 7113
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018441210699344647,
+      "loss": 0.044,
+      "step": 7114
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001843839670630677,
+      "loss": 0.069,
+      "step": 7115
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018435582585542537,
+      "loss": 0.0298,
+      "step": 7116
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018432768337156485,
+      "loss": 0.101,
+      "step": 7117
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001842995396125316,
+      "loss": 0.0961,
+      "step": 7118
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018427139457937098,
+      "loss": 0.0534,
+      "step": 7119
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018424324827312854,
+      "loss": 0.1083,
+      "step": 7120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018421510069484986,
+      "loss": 0.1154,
+      "step": 7121
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018418695184558055,
+      "loss": 0.1,
+      "step": 7122
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018415880172636622,
+      "loss": 0.1681,
+      "step": 7123
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018413065033825258,
+      "loss": 0.0901,
+      "step": 7124
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001841024976822854,
+      "loss": 0.0803,
+      "step": 7125
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018407434375951039,
+      "loss": 0.0955,
+      "step": 7126
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018404618857097342,
+      "loss": 0.186,
+      "step": 7127
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018401803211772036,
+      "loss": 0.1897,
+      "step": 7128
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018398987440079723,
+      "loss": 0.0583,
+      "step": 7129
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00018396171542124992,
+      "loss": 0.0815,
+      "step": 7130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018393355518012448,
+      "loss": 0.1221,
+      "step": 7131
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018390539367846694,
+      "loss": 0.0046,
+      "step": 7132
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001838772309173235,
+      "loss": 0.1271,
+      "step": 7133
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018384906689774026,
+      "loss": 0.036,
+      "step": 7134
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018382090162076344,
+      "loss": 0.1228,
+      "step": 7135
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001837927350874393,
+      "loss": 0.0958,
+      "step": 7136
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018376456729881412,
+      "loss": 0.0894,
+      "step": 7137
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018373639825593432,
+      "loss": 0.1537,
+      "step": 7138
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018370822795984627,
+      "loss": 0.049,
+      "step": 7139
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018368005641159634,
+      "loss": 0.1177,
+      "step": 7140
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018365188361223113,
+      "loss": 0.124,
+      "step": 7141
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001836237095627971,
+      "loss": 0.0218,
+      "step": 7142
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018359553426434088,
+      "loss": 0.0648,
+      "step": 7143
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018356735771790905,
+      "loss": 0.0581,
+      "step": 7144
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018353917992454833,
+      "loss": 0.0035,
+      "step": 7145
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018351100088530547,
+      "loss": 0.1411,
+      "step": 7146
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018348282060122714,
+      "loss": 0.0324,
+      "step": 7147
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018345463907336028,
+      "loss": 0.1028,
+      "step": 7148
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018342645630275162,
+      "loss": 0.044,
+      "step": 7149
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018339827229044815,
+      "loss": 0.1379,
+      "step": 7150
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018337008703749679,
+      "loss": 0.0743,
+      "step": 7151
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018334190054494456,
+      "loss": 0.0849,
+      "step": 7152
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018331371281383855,
+      "loss": 0.0911,
+      "step": 7153
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018328552384522573,
+      "loss": 0.1262,
+      "step": 7154
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018325733364015333,
+      "loss": 0.116,
+      "step": 7155
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018322914219966855,
+      "loss": 0.0978,
+      "step": 7156
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018320094952481853,
+      "loss": 0.1026,
+      "step": 7157
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018317275561665066,
+      "loss": 0.0827,
+      "step": 7158
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018314456047621212,
+      "loss": 0.1099,
+      "step": 7159
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001831163641045504,
+      "loss": 0.1077,
+      "step": 7160
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018308816650271283,
+      "loss": 0.1194,
+      "step": 7161
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018305996767174694,
+      "loss": 0.1144,
+      "step": 7162
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00018303176761270019,
+      "loss": 0.1223,
+      "step": 7163
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018300356632662014,
+      "loss": 0.0721,
+      "step": 7164
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018297536381455435,
+      "loss": 0.0666,
+      "step": 7165
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018294716007755053,
+      "loss": 0.1483,
+      "step": 7166
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001829189551166563,
+      "loss": 0.1473,
+      "step": 7167
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018289074893291943,
+      "loss": 0.0848,
+      "step": 7168
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001828625415273877,
+      "loss": 0.1397,
+      "step": 7169
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018283433290110893,
+      "loss": 0.1118,
+      "step": 7170
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018280612305513098,
+      "loss": 0.0732,
+      "step": 7171
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018277791199050177,
+      "loss": 0.0798,
+      "step": 7172
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018274969970826922,
+      "loss": 0.0993,
+      "step": 7173
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001827214862094814,
+      "loss": 0.1225,
+      "step": 7174
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018269327149518627,
+      "loss": 0.2076,
+      "step": 7175
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000182665055566432,
+      "loss": 0.1226,
+      "step": 7176
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018263683842426672,
+      "loss": 0.1432,
+      "step": 7177
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018260862006973857,
+      "loss": 0.0661,
+      "step": 7178
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018258040050389584,
+      "loss": 0.1437,
+      "step": 7179
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018255217972778677,
+      "loss": 0.0709,
+      "step": 7180
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018252395774245967,
+      "loss": 0.0696,
+      "step": 7181
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018249573454896294,
+      "loss": 0.0792,
+      "step": 7182
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018246751014834489,
+      "loss": 0.1607,
+      "step": 7183
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001824392845416541,
+      "loss": 0.0564,
+      "step": 7184
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018241105772993902,
+      "loss": 0.1029,
+      "step": 7185
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001823828297142482,
+      "loss": 0.126,
+      "step": 7186
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018235460049563019,
+      "loss": 0.1685,
+      "step": 7187
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018232637007513367,
+      "loss": 0.0369,
+      "step": 7188
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001822981384538073,
+      "loss": 0.1065,
+      "step": 7189
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018226990563269979,
+      "loss": 0.0102,
+      "step": 7190
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018224167161285992,
+      "loss": 0.0646,
+      "step": 7191
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018221343639533653,
+      "loss": 0.1636,
+      "step": 7192
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018218519998117836,
+      "loss": 0.1037,
+      "step": 7193
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018215696237143446,
+      "loss": 0.1099,
+      "step": 7194
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018212872356715366,
+      "loss": 0.0717,
+      "step": 7195
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00018210048356938504,
+      "loss": 0.1201,
+      "step": 7196
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001820722423791776,
+      "loss": 0.0995,
+      "step": 7197
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018204399999758035,
+      "loss": 0.0299,
+      "step": 7198
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018201575642564255,
+      "loss": 0.1381,
+      "step": 7199
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018198751166441322,
+      "loss": 0.0528,
+      "step": 7200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018195926571494165,
+      "loss": 0.1124,
+      "step": 7201
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018193101857827705,
+      "loss": 0.0656,
+      "step": 7202
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018190277025546876,
+      "loss": 0.0349,
+      "step": 7203
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018187452074756614,
+      "loss": 0.1548,
+      "step": 7204
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001818462700556185,
+      "loss": 0.105,
+      "step": 7205
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018181801818067535,
+      "loss": 0.1001,
+      "step": 7206
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018178976512378612,
+      "loss": 0.129,
+      "step": 7207
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018176151088600037,
+      "loss": 0.158,
+      "step": 7208
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001817332554683676,
+      "loss": 0.1358,
+      "step": 7209
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001817049988719374,
+      "loss": 0.0791,
+      "step": 7210
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018167674109775952,
+      "loss": 0.1525,
+      "step": 7211
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018164848214688364,
+      "loss": 0.0971,
+      "step": 7212
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018162022202035939,
+      "loss": 0.098,
+      "step": 7213
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018159196071923668,
+      "loss": 0.1052,
+      "step": 7214
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018156369824456523,
+      "loss": 0.1684,
+      "step": 7215
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001815354345973949,
+      "loss": 0.0652,
+      "step": 7216
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018150716977877572,
+      "loss": 0.091,
+      "step": 7217
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018147890378975752,
+      "loss": 0.101,
+      "step": 7218
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018145063663139045,
+      "loss": 0.0476,
+      "step": 7219
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018142236830472438,
+      "loss": 0.1171,
+      "step": 7220
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018139409881080947,
+      "loss": 0.0726,
+      "step": 7221
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018136582815069592,
+      "loss": 0.1552,
+      "step": 7222
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001813375563254338,
+      "loss": 0.1043,
+      "step": 7223
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018130928333607338,
+      "loss": 0.1022,
+      "step": 7224
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001812810091836648,
+      "loss": 0.101,
+      "step": 7225
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001812527338692585,
+      "loss": 0.073,
+      "step": 7226
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018122445739390485,
+      "loss": 0.0531,
+      "step": 7227
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018119617975865415,
+      "loss": 0.0624,
+      "step": 7228
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00018116790096455685,
+      "loss": 0.0877,
+      "step": 7229
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018113962101266341,
+      "loss": 0.161,
+      "step": 7230
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018111133990402435,
+      "loss": 0.1724,
+      "step": 7231
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001810830576396903,
+      "loss": 0.1048,
+      "step": 7232
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018105477422071176,
+      "loss": 0.1724,
+      "step": 7233
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001810264896481394,
+      "loss": 0.1513,
+      "step": 7234
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018099820392302398,
+      "loss": 0.1141,
+      "step": 7235
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001809699170464162,
+      "loss": 0.0806,
+      "step": 7236
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018094162901936681,
+      "loss": 0.1499,
+      "step": 7237
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018091333984292663,
+      "loss": 0.1506,
+      "step": 7238
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018088504951814654,
+      "loss": 0.058,
+      "step": 7239
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001808567580460774,
+      "loss": 0.1401,
+      "step": 7240
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001808284654277702,
+      "loss": 0.0941,
+      "step": 7241
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018080017166427592,
+      "loss": 0.1059,
+      "step": 7242
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018077187675664556,
+      "loss": 0.0456,
+      "step": 7243
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018074358070593023,
+      "loss": 0.0682,
+      "step": 7244
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018071528351318105,
+      "loss": 0.0514,
+      "step": 7245
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018068698517944914,
+      "loss": 0.1687,
+      "step": 7246
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018065868570578574,
+      "loss": 0.0222,
+      "step": 7247
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018063038509324208,
+      "loss": 0.1302,
+      "step": 7248
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001806020833428694,
+      "loss": 0.07,
+      "step": 7249
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018057378045571905,
+      "loss": 0.0817,
+      "step": 7250
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018054547643284242,
+      "loss": 0.0443,
+      "step": 7251
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018051717127529098,
+      "loss": 0.1039,
+      "step": 7252
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018048886498411607,
+      "loss": 0.1061,
+      "step": 7253
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018046055756036923,
+      "loss": 0.1345,
+      "step": 7254
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.000180432249005102,
+      "loss": 0.0587,
+      "step": 7255
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018040393931936598,
+      "loss": 0.1545,
+      "step": 7256
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018037562850421274,
+      "loss": 0.0448,
+      "step": 7257
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.000180347316560694,
+      "loss": 0.1442,
+      "step": 7258
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018031900348986144,
+      "loss": 0.103,
+      "step": 7259
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018029068929276685,
+      "loss": 0.1051,
+      "step": 7260
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018026237397046193,
+      "loss": 0.1078,
+      "step": 7261
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00018023405752399857,
+      "loss": 0.1311,
+      "step": 7262
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001802057399544286,
+      "loss": 0.1412,
+      "step": 7263
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018017742126280402,
+      "loss": 0.0534,
+      "step": 7264
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018014910145017674,
+      "loss": 0.0994,
+      "step": 7265
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001801207805175987,
+      "loss": 0.0567,
+      "step": 7266
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018009245846612202,
+      "loss": 0.0865,
+      "step": 7267
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018006413529679876,
+      "loss": 0.0263,
+      "step": 7268
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018003581101068105,
+      "loss": 0.0217,
+      "step": 7269
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00018000748560882104,
+      "loss": 0.0307,
+      "step": 7270
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017997915909227088,
+      "loss": 0.1613,
+      "step": 7271
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017995083146208292,
+      "loss": 0.0579,
+      "step": 7272
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001799225027193094,
+      "loss": 0.1191,
+      "step": 7273
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017989417286500266,
+      "loss": 0.0603,
+      "step": 7274
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017986584190021504,
+      "loss": 0.0698,
+      "step": 7275
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017983750982599896,
+      "loss": 0.1149,
+      "step": 7276
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001798091766434069,
+      "loss": 0.0569,
+      "step": 7277
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001797808423534914,
+      "loss": 0.1592,
+      "step": 7278
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017975250695730487,
+      "loss": 0.0833,
+      "step": 7279
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001797241704559,
+      "loss": 0.0974,
+      "step": 7280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017969583285032932,
+      "loss": 0.1295,
+      "step": 7281
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017966749414164554,
+      "loss": 0.0977,
+      "step": 7282
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017963915433090135,
+      "loss": 0.1021,
+      "step": 7283
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001796108134191495,
+      "loss": 0.0822,
+      "step": 7284
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017958247140744277,
+      "loss": 0.1282,
+      "step": 7285
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017955412829683397,
+      "loss": 0.0761,
+      "step": 7286
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017952578408837595,
+      "loss": 0.1242,
+      "step": 7287
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017949743878312171,
+      "loss": 0.1494,
+      "step": 7288
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017946909238212406,
+      "loss": 0.0549,
+      "step": 7289
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017944074488643604,
+      "loss": 0.154,
+      "step": 7290
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017941239629711068,
+      "loss": 0.1157,
+      "step": 7291
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017938404661520108,
+      "loss": 0.1553,
+      "step": 7292
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017935569584176032,
+      "loss": 0.0774,
+      "step": 7293
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001793273439778415,
+      "loss": 0.0237,
+      "step": 7294
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00017929899102449791,
+      "loss": 0.1114,
+      "step": 7295
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017927063698278267,
+      "loss": 0.0466,
+      "step": 7296
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001792422818537491,
+      "loss": 0.1076,
+      "step": 7297
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017921392563845055,
+      "loss": 0.0464,
+      "step": 7298
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017918556833794032,
+      "loss": 0.0162,
+      "step": 7299
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017915720995327177,
+      "loss": 0.2108,
+      "step": 7300
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017912885048549846,
+      "loss": 0.1331,
+      "step": 7301
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001791004899356737,
+      "loss": 0.1109,
+      "step": 7302
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017907212830485113,
+      "loss": 0.24,
+      "step": 7303
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001790437655940842,
+      "loss": 0.1447,
+      "step": 7304
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017901540180442654,
+      "loss": 0.0637,
+      "step": 7305
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017898703693693184,
+      "loss": 0.228,
+      "step": 7306
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001789586709926537,
+      "loss": 0.0724,
+      "step": 7307
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017893030397264585,
+      "loss": 0.0455,
+      "step": 7308
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017890193587796204,
+      "loss": 0.0404,
+      "step": 7309
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017887356670965606,
+      "loss": 0.1133,
+      "step": 7310
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001788451964687818,
+      "loss": 0.0212,
+      "step": 7311
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000178816825156393,
+      "loss": 0.0617,
+      "step": 7312
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001787884527735437,
+      "loss": 0.1017,
+      "step": 7313
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017876007932128782,
+      "loss": 0.0565,
+      "step": 7314
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017873170480067928,
+      "loss": 0.1,
+      "step": 7315
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017870332921277214,
+      "loss": 0.1092,
+      "step": 7316
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017867495255862055,
+      "loss": 0.0849,
+      "step": 7317
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017864657483927858,
+      "loss": 0.1115,
+      "step": 7318
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001786181960558003,
+      "loss": 0.2245,
+      "step": 7319
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017858981620924,
+      "loss": 0.0198,
+      "step": 7320
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017856143530065187,
+      "loss": 0.07,
+      "step": 7321
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017853305333109015,
+      "loss": 0.0867,
+      "step": 7322
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017850467030160915,
+      "loss": 0.053,
+      "step": 7323
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017847628621326324,
+      "loss": 0.1252,
+      "step": 7324
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017844790106710681,
+      "loss": 0.0562,
+      "step": 7325
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017841951486419433,
+      "loss": 0.0243,
+      "step": 7326
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017839112760558018,
+      "loss": 0.0993,
+      "step": 7327
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017836273929231888,
+      "loss": 0.1307,
+      "step": 7328
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017833434992546503,
+      "loss": 0.0456,
+      "step": 7329
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017830595950607311,
+      "loss": 0.1082,
+      "step": 7330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017827756803519783,
+      "loss": 0.0535,
+      "step": 7331
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017824917551389382,
+      "loss": 0.1511,
+      "step": 7332
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017822078194321578,
+      "loss": 0.1501,
+      "step": 7333
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017819238732421847,
+      "loss": 0.1088,
+      "step": 7334
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001781639916579566,
+      "loss": 0.0447,
+      "step": 7335
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001781355949454851,
+      "loss": 0.1506,
+      "step": 7336
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001781071971878587,
+      "loss": 0.098,
+      "step": 7337
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017807879838613235,
+      "loss": 0.0999,
+      "step": 7338
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017805039854136098,
+      "loss": 0.0927,
+      "step": 7339
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017802199765459958,
+      "loss": 0.0463,
+      "step": 7340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017799359572690313,
+      "loss": 0.0908,
+      "step": 7341
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017796519275932666,
+      "loss": 0.0937,
+      "step": 7342
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017793678875292532,
+      "loss": 0.0563,
+      "step": 7343
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017790838370875423,
+      "loss": 0.0611,
+      "step": 7344
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017787997762786845,
+      "loss": 0.1282,
+      "step": 7345
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017785157051132333,
+      "loss": 0.1146,
+      "step": 7346
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017782316236017405,
+      "loss": 0.0835,
+      "step": 7347
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017779475317547576,
+      "loss": 0.0587,
+      "step": 7348
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000177766342958284,
+      "loss": 0.0595,
+      "step": 7349
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000177737931709654,
+      "loss": 0.0798,
+      "step": 7350
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017770951943064123,
+      "loss": 0.0668,
+      "step": 7351
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017768110612230098,
+      "loss": 0.1065,
+      "step": 7352
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017765269178568886,
+      "loss": 0.0881,
+      "step": 7353
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017762427642186034,
+      "loss": 0.077,
+      "step": 7354
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017759586003187092,
+      "loss": 0.0898,
+      "step": 7355
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017756744261677621,
+      "loss": 0.077,
+      "step": 7356
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001775390241776319,
+      "loss": 0.1228,
+      "step": 7357
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017751060471549355,
+      "loss": 0.1845,
+      "step": 7358
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017748218423141698,
+      "loss": 0.1365,
+      "step": 7359
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001774537627264578,
+      "loss": 0.2156,
+      "step": 7360
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017742534020167186,
+      "loss": 0.1478,
+      "step": 7361
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000177396916658115,
+      "loss": 0.0557,
+      "step": 7362
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017736849209684297,
+      "loss": 0.0627,
+      "step": 7363
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017734006651891172,
+      "loss": 0.1062,
+      "step": 7364
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017731163992537716,
+      "loss": 0.0772,
+      "step": 7365
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017728321231729529,
+      "loss": 0.1683,
+      "step": 7366
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001772547836957221,
+      "loss": 0.125,
+      "step": 7367
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001772263540617136,
+      "loss": 0.1285,
+      "step": 7368
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001771979234163259,
+      "loss": 0.1259,
+      "step": 7369
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001771694917606151,
+      "loss": 0.0418,
+      "step": 7370
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017714105909563734,
+      "loss": 0.0982,
+      "step": 7371
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017711262542244885,
+      "loss": 0.0833,
+      "step": 7372
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001770841907421058,
+      "loss": 0.0293,
+      "step": 7373
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017705575505566452,
+      "loss": 0.0571,
+      "step": 7374
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017702731836418127,
+      "loss": 0.1054,
+      "step": 7375
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001769988806687124,
+      "loss": 0.1429,
+      "step": 7376
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001769704419703143,
+      "loss": 0.0984,
+      "step": 7377
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017694200227004334,
+      "loss": 0.0739,
+      "step": 7378
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017691356156895603,
+      "loss": 0.1081,
+      "step": 7379
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001768851198681088,
+      "loss": 0.142,
+      "step": 7380
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017685667716855818,
+      "loss": 0.0848,
+      "step": 7381
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017682823347136086,
+      "loss": 0.1254,
+      "step": 7382
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017679978877757329,
+      "loss": 0.0541,
+      "step": 7383
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017677134308825214,
+      "loss": 0.0747,
+      "step": 7384
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001767428964044541,
+      "loss": 0.1882,
+      "step": 7385
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001767144487272359,
+      "loss": 0.0923,
+      "step": 7386
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001766860000576543,
+      "loss": 0.1428,
+      "step": 7387
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000176657550396766,
+      "loss": 0.1173,
+      "step": 7388
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017662909974562786,
+      "loss": 0.0925,
+      "step": 7389
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001766006481052968,
+      "loss": 0.0955,
+      "step": 7390
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017657219547682963,
+      "loss": 0.044,
+      "step": 7391
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017654374186128338,
+      "loss": 0.1123,
+      "step": 7392
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001765152872597149,
+      "loss": 0.1822,
+      "step": 7393
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017648683167318128,
+      "loss": 0.1611,
+      "step": 7394
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017645837510273956,
+      "loss": 0.0521,
+      "step": 7395
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017642991754944674,
+      "loss": 0.1213,
+      "step": 7396
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017640145901436,
+      "loss": 0.1177,
+      "step": 7397
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001763729994985365,
+      "loss": 0.1377,
+      "step": 7398
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017634453900303339,
+      "loss": 0.1301,
+      "step": 7399
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017631607752890795,
+      "loss": 0.1013,
+      "step": 7400
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017628761507721736,
+      "loss": 0.1025,
+      "step": 7401
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017625915164901898,
+      "loss": 0.0252,
+      "step": 7402
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001762306872453701,
+      "loss": 0.0957,
+      "step": 7403
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017620222186732812,
+      "loss": 0.1116,
+      "step": 7404
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017617375551595043,
+      "loss": 0.1074,
+      "step": 7405
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017614528819229447,
+      "loss": 0.1016,
+      "step": 7406
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017611681989741776,
+      "loss": 0.1223,
+      "step": 7407
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017608835063237773,
+      "loss": 0.1362,
+      "step": 7408
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000176059880398232,
+      "loss": 0.1538,
+      "step": 7409
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001760314091960382,
+      "loss": 0.1097,
+      "step": 7410
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001760029370268538,
+      "loss": 0.1508,
+      "step": 7411
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001759744638917365,
+      "loss": 0.1765,
+      "step": 7412
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017594598979174411,
+      "loss": 0.0703,
+      "step": 7413
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017591751472793426,
+      "loss": 0.0528,
+      "step": 7414
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017588903870136477,
+      "loss": 0.1171,
+      "step": 7415
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001758605617130934,
+      "loss": 0.056,
+      "step": 7416
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000175832083764178,
+      "loss": 0.0544,
+      "step": 7417
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017580360485567643,
+      "loss": 0.0533,
+      "step": 7418
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001757751249886466,
+      "loss": 0.0663,
+      "step": 7419
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001757466441641465,
+      "loss": 0.0241,
+      "step": 7420
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000175718162383234,
+      "loss": 0.1465,
+      "step": 7421
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001756896796469672,
+      "loss": 0.0369,
+      "step": 7422
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001756611959564042,
+      "loss": 0.1901,
+      "step": 7423
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000175632711312603,
+      "loss": 0.0848,
+      "step": 7424
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001756042257166217,
+      "loss": 0.0891,
+      "step": 7425
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00017557573916951853,
+      "loss": 0.0697,
+      "step": 7426
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017554725167235165,
+      "loss": 0.1183,
+      "step": 7427
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017551876322617933,
+      "loss": 0.0574,
+      "step": 7428
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001754902738320597,
+      "loss": 0.0811,
+      "step": 7429
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001754617834910512,
+      "loss": 0.116,
+      "step": 7430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017543329220421207,
+      "loss": 0.131,
+      "step": 7431
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017540479997260074,
+      "loss": 0.1349,
+      "step": 7432
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001753763067972756,
+      "loss": 0.2279,
+      "step": 7433
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017534781267929508,
+      "loss": 0.1612,
+      "step": 7434
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017531931761971764,
+      "loss": 0.1242,
+      "step": 7435
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017529082161960182,
+      "loss": 0.052,
+      "step": 7436
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017526232468000608,
+      "loss": 0.1706,
+      "step": 7437
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017523382680198913,
+      "loss": 0.0786,
+      "step": 7438
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017520532798660947,
+      "loss": 0.0647,
+      "step": 7439
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017517682823492582,
+      "loss": 0.0338,
+      "step": 7440
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017514832754799678,
+      "loss": 0.1519,
+      "step": 7441
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017511982592688117,
+      "loss": 0.0467,
+      "step": 7442
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001750913233726377,
+      "loss": 0.1449,
+      "step": 7443
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001750628198863251,
+      "loss": 0.1017,
+      "step": 7444
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017503431546900229,
+      "loss": 0.0633,
+      "step": 7445
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000175005810121728,
+      "loss": 0.0699,
+      "step": 7446
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017497730384556122,
+      "loss": 0.1339,
+      "step": 7447
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017494879664156087,
+      "loss": 0.0364,
+      "step": 7448
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001749202885107859,
+      "loss": 0.1136,
+      "step": 7449
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001748917794542953,
+      "loss": 0.1001,
+      "step": 7450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017486326947314802,
+      "loss": 0.1079,
+      "step": 7451
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017483475856840323,
+      "loss": 0.0829,
+      "step": 7452
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017480624674111995,
+      "loss": 0.1001,
+      "step": 7453
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017477773399235738,
+      "loss": 0.1413,
+      "step": 7454
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017474922032317467,
+      "loss": 0.1103,
+      "step": 7455
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000174720705734631,
+      "loss": 0.0615,
+      "step": 7456
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017469219022778556,
+      "loss": 0.1176,
+      "step": 7457
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001746636738036977,
+      "loss": 0.0819,
+      "step": 7458
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00017463515646342668,
+      "loss": 0.1203,
+      "step": 7459
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001746066382080318,
+      "loss": 0.0294,
+      "step": 7460
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017457811903857253,
+      "loss": 0.1225,
+      "step": 7461
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017454959895610818,
+      "loss": 0.0687,
+      "step": 7462
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017452107796169823,
+      "loss": 0.0727,
+      "step": 7463
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017449255605640212,
+      "loss": 0.1882,
+      "step": 7464
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017446403324127937,
+      "loss": 0.1319,
+      "step": 7465
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017443550951738956,
+      "loss": 0.1502,
+      "step": 7466
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017440698488579222,
+      "loss": 0.1107,
+      "step": 7467
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000174378459347547,
+      "loss": 0.1109,
+      "step": 7468
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017434993290371345,
+      "loss": 0.0049,
+      "step": 7469
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017432140555535128,
+      "loss": 0.0416,
+      "step": 7470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017429287730352024,
+      "loss": 0.0294,
+      "step": 7471
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017426434814928009,
+      "loss": 0.0999,
+      "step": 7472
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017423581809369052,
+      "loss": 0.0335,
+      "step": 7473
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017420728713781137,
+      "loss": 0.0337,
+      "step": 7474
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001741787552827025,
+      "loss": 0.1248,
+      "step": 7475
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017415022252942382,
+      "loss": 0.1514,
+      "step": 7476
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017412168887903512,
+      "loss": 0.175,
+      "step": 7477
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017409315433259645,
+      "loss": 0.1319,
+      "step": 7478
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001740646188911677,
+      "loss": 0.0921,
+      "step": 7479
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000174036082555809,
+      "loss": 0.1473,
+      "step": 7480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017400754532758032,
+      "loss": 0.0937,
+      "step": 7481
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017397900720754166,
+      "loss": 0.0685,
+      "step": 7482
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017395046819675329,
+      "loss": 0.0984,
+      "step": 7483
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001739219282962752,
+      "loss": 0.1125,
+      "step": 7484
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017389338750716763,
+      "loss": 0.1028,
+      "step": 7485
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017386484583049077,
+      "loss": 0.1104,
+      "step": 7486
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017383630326730487,
+      "loss": 0.1332,
+      "step": 7487
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017380775981867022,
+      "loss": 0.0849,
+      "step": 7488
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017377921548564714,
+      "loss": 0.0872,
+      "step": 7489
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001737506702692959,
+      "loss": 0.1089,
+      "step": 7490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017372212417067693,
+      "loss": 0.0537,
+      "step": 7491
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00017369357719085058,
+      "loss": 0.0798,
+      "step": 7492
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017366502933087734,
+      "loss": 0.1221,
+      "step": 7493
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017363648059181766,
+      "loss": 0.0825,
+      "step": 7494
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000173607930974732,
+      "loss": 0.0793,
+      "step": 7495
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017357938048068102,
+      "loss": 0.0941,
+      "step": 7496
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001735508291107251,
+      "loss": 0.2017,
+      "step": 7497
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000173522276865925,
+      "loss": 0.0548,
+      "step": 7498
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017349372374734131,
+      "loss": 0.0346,
+      "step": 7499
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017346516975603462,
+      "loss": 0.0571,
+      "step": 7500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017343661489306573,
+      "loss": 0.1006,
+      "step": 7501
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017340805915949529,
+      "loss": 0.0658,
+      "step": 7502
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001733795025563841,
+      "loss": 0.1136,
+      "step": 7503
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017335094508479296,
+      "loss": 0.1159,
+      "step": 7504
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001733223867457827,
+      "loss": 0.1198,
+      "step": 7505
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017329382754041416,
+      "loss": 0.1717,
+      "step": 7506
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017326526746974823,
+      "loss": 0.1192,
+      "step": 7507
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017323670653484583,
+      "loss": 0.036,
+      "step": 7508
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017320814473676795,
+      "loss": 0.0726,
+      "step": 7509
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017317958207657553,
+      "loss": 0.0744,
+      "step": 7510
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017315101855532958,
+      "loss": 0.085,
+      "step": 7511
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001731224541740912,
+      "loss": 0.0308,
+      "step": 7512
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017309388893392146,
+      "loss": 0.1391,
+      "step": 7513
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001730653228358815,
+      "loss": 0.0842,
+      "step": 7514
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001730367558810324,
+      "loss": 0.0574,
+      "step": 7515
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017300818807043539,
+      "loss": 0.1,
+      "step": 7516
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017297961940515168,
+      "loss": 0.1587,
+      "step": 7517
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017295104988624245,
+      "loss": 0.1708,
+      "step": 7518
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017292247951476903,
+      "loss": 0.1082,
+      "step": 7519
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017289390829179272,
+      "loss": 0.1311,
+      "step": 7520
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017286533621837483,
+      "loss": 0.0385,
+      "step": 7521
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001728367632955768,
+      "loss": 0.0793,
+      "step": 7522
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017280818952445997,
+      "loss": 0.11,
+      "step": 7523
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017277961490608581,
+      "loss": 0.0568,
+      "step": 7524
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00017275103944151571,
+      "loss": 0.1595,
+      "step": 7525
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017272246313181124,
+      "loss": 0.0789,
+      "step": 7526
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017269388597803385,
+      "loss": 0.1255,
+      "step": 7527
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001726653079812452,
+      "loss": 0.072,
+      "step": 7528
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001726367291425068,
+      "loss": 0.1523,
+      "step": 7529
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001726081494628803,
+      "loss": 0.0857,
+      "step": 7530
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017257956894342733,
+      "loss": 0.0642,
+      "step": 7531
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017255098758520962,
+      "loss": 0.1228,
+      "step": 7532
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017252240538928883,
+      "loss": 0.1246,
+      "step": 7533
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017249382235672673,
+      "loss": 0.1148,
+      "step": 7534
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001724652384885851,
+      "loss": 0.1131,
+      "step": 7535
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001724366537859257,
+      "loss": 0.0966,
+      "step": 7536
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017240806824981043,
+      "loss": 0.2437,
+      "step": 7537
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017237948188130115,
+      "loss": 0.106,
+      "step": 7538
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017235089468145974,
+      "loss": 0.1131,
+      "step": 7539
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017232230665134812,
+      "loss": 0.0369,
+      "step": 7540
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001722937177920283,
+      "loss": 0.0839,
+      "step": 7541
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017226512810456223,
+      "loss": 0.0808,
+      "step": 7542
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001722365375900119,
+      "loss": 0.1089,
+      "step": 7543
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017220794624943947,
+      "loss": 0.1371,
+      "step": 7544
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001721793540839069,
+      "loss": 0.076,
+      "step": 7545
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017215076109447638,
+      "loss": 0.1319,
+      "step": 7546
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001721221672822101,
+      "loss": 0.1345,
+      "step": 7547
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017209357264817013,
+      "loss": 0.1391,
+      "step": 7548
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001720649771934187,
+      "loss": 0.0686,
+      "step": 7549
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017203638091901812,
+      "loss": 0.0772,
+      "step": 7550
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001720077838260306,
+      "loss": 0.093,
+      "step": 7551
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017197918591551844,
+      "loss": 0.111,
+      "step": 7552
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017195058718854394,
+      "loss": 0.1441,
+      "step": 7553
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017192198764616956,
+      "loss": 0.1084,
+      "step": 7554
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017189338728945764,
+      "loss": 0.1316,
+      "step": 7555
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017186478611947052,
+      "loss": 0.1111,
+      "step": 7556
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017183618413727075,
+      "loss": 0.0683,
+      "step": 7557
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00017180758134392076,
+      "loss": 0.1169,
+      "step": 7558
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017177897774048304,
+      "loss": 0.1362,
+      "step": 7559
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001717503733280202,
+      "loss": 0.0349,
+      "step": 7560
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017172176810759476,
+      "loss": 0.1014,
+      "step": 7561
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017169316208026934,
+      "loss": 0.0727,
+      "step": 7562
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001716645552471066,
+      "loss": 0.0555,
+      "step": 7563
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001716359476091691,
+      "loss": 0.1379,
+      "step": 7564
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017160733916751964,
+      "loss": 0.0978,
+      "step": 7565
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017157872992322085,
+      "loss": 0.1203,
+      "step": 7566
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001715501198773356,
+      "loss": 0.1562,
+      "step": 7567
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017152150903092649,
+      "loss": 0.1512,
+      "step": 7568
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017149289738505649,
+      "loss": 0.0791,
+      "step": 7569
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017146428494078838,
+      "loss": 0.1704,
+      "step": 7570
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000171435671699185,
+      "loss": 0.0873,
+      "step": 7571
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017140705766130934,
+      "loss": 0.0642,
+      "step": 7572
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001713784428282242,
+      "loss": 0.1452,
+      "step": 7573
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001713498272009926,
+      "loss": 0.138,
+      "step": 7574
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017132121078067757,
+      "loss": 0.0518,
+      "step": 7575
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017129259356834207,
+      "loss": 0.0885,
+      "step": 7576
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017126397556504913,
+      "loss": 0.148,
+      "step": 7577
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017123535677186186,
+      "loss": 0.0744,
+      "step": 7578
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017120673718984335,
+      "loss": 0.0561,
+      "step": 7579
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001711781168200568,
+      "loss": 0.1633,
+      "step": 7580
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017114949566356528,
+      "loss": 0.1815,
+      "step": 7581
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017112087372143198,
+      "loss": 0.0674,
+      "step": 7582
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017109225099472023,
+      "loss": 0.0972,
+      "step": 7583
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017106362748449316,
+      "loss": 0.1035,
+      "step": 7584
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017103500319181407,
+      "loss": 0.0801,
+      "step": 7585
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001710063781177463,
+      "loss": 0.06,
+      "step": 7586
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017097775226335323,
+      "loss": 0.0665,
+      "step": 7587
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001709491256296982,
+      "loss": 0.0996,
+      "step": 7588
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017092049821784455,
+      "loss": 0.0514,
+      "step": 7589
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017089187002885574,
+      "loss": 0.0926,
+      "step": 7590
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00017086324106379522,
+      "loss": 0.1361,
+      "step": 7591
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001708346113237265,
+      "loss": 0.0539,
+      "step": 7592
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017080598080971304,
+      "loss": 0.0884,
+      "step": 7593
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001707773495228184,
+      "loss": 0.1423,
+      "step": 7594
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017074871746410616,
+      "loss": 0.1645,
+      "step": 7595
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017072008463463992,
+      "loss": 0.1069,
+      "step": 7596
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017069145103548332,
+      "loss": 0.1264,
+      "step": 7597
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017066281666769996,
+      "loss": 0.0357,
+      "step": 7598
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017063418153235354,
+      "loss": 0.114,
+      "step": 7599
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017060554563050781,
+      "loss": 0.0763,
+      "step": 7600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017057690896322647,
+      "loss": 0.1512,
+      "step": 7601
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001705482715315733,
+      "loss": 0.0574,
+      "step": 7602
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017051963333661213,
+      "loss": 0.0737,
+      "step": 7603
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017049099437940675,
+      "loss": 0.1357,
+      "step": 7604
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017046235466102103,
+      "loss": 0.1098,
+      "step": 7605
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017043371418251881,
+      "loss": 0.138,
+      "step": 7606
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017040507294496403,
+      "loss": 0.1084,
+      "step": 7607
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017037643094942065,
+      "loss": 0.0636,
+      "step": 7608
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017034778819695256,
+      "loss": 0.1318,
+      "step": 7609
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017031914468862381,
+      "loss": 0.0961,
+      "step": 7610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017029050042549848,
+      "loss": 0.0941,
+      "step": 7611
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017026185540864054,
+      "loss": 0.0835,
+      "step": 7612
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017023320963911411,
+      "loss": 0.0928,
+      "step": 7613
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017020456311798323,
+      "loss": 0.0624,
+      "step": 7614
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017017591584631212,
+      "loss": 0.119,
+      "step": 7615
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001701472678251649,
+      "loss": 0.1446,
+      "step": 7616
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017011861905560574,
+      "loss": 0.0531,
+      "step": 7617
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017008996953869887,
+      "loss": 0.0857,
+      "step": 7618
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017006131927550853,
+      "loss": 0.1239,
+      "step": 7619
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00017003266826709905,
+      "loss": 0.0299,
+      "step": 7620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001700040165145347,
+      "loss": 0.1017,
+      "step": 7621
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00016997536401887976,
+      "loss": 0.0626,
+      "step": 7622
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00016994671078119866,
+      "loss": 0.102,
+      "step": 7623
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001699180568025557,
+      "loss": 0.0723,
+      "step": 7624
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016988940208401536,
+      "loss": 0.0845,
+      "step": 7625
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016986074662664205,
+      "loss": 0.0499,
+      "step": 7626
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016983209043150023,
+      "loss": 0.0717,
+      "step": 7627
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016980343349965447,
+      "loss": 0.1012,
+      "step": 7628
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016977477583216918,
+      "loss": 0.0632,
+      "step": 7629
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016974611743010896,
+      "loss": 0.1561,
+      "step": 7630
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016971745829453842,
+      "loss": 0.1064,
+      "step": 7631
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001696887984265221,
+      "loss": 0.0583,
+      "step": 7632
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016966013782712463,
+      "loss": 0.0643,
+      "step": 7633
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016963147649741068,
+      "loss": 0.1548,
+      "step": 7634
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000169602814438445,
+      "loss": 0.0583,
+      "step": 7635
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016957415165129225,
+      "loss": 0.1223,
+      "step": 7636
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016954548813701714,
+      "loss": 0.0525,
+      "step": 7637
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016951682389668447,
+      "loss": 0.1064,
+      "step": 7638
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016948815893135904,
+      "loss": 0.1409,
+      "step": 7639
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016945949324210564,
+      "loss": 0.1284,
+      "step": 7640
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001694308268299891,
+      "loss": 0.0874,
+      "step": 7641
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016940215969607435,
+      "loss": 0.1361,
+      "step": 7642
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016937349184142626,
+      "loss": 0.0503,
+      "step": 7643
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016934482326710977,
+      "loss": 0.1085,
+      "step": 7644
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016931615397418982,
+      "loss": 0.1486,
+      "step": 7645
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016928748396373143,
+      "loss": 0.1267,
+      "step": 7646
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016925881323679953,
+      "loss": 0.1132,
+      "step": 7647
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016923014179445918,
+      "loss": 0.1704,
+      "step": 7648
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016920146963777548,
+      "loss": 0.1047,
+      "step": 7649
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016917279676781344,
+      "loss": 0.0858,
+      "step": 7650
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001691441231856383,
+      "loss": 0.1048,
+      "step": 7651
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001691154488923151,
+      "loss": 0.1159,
+      "step": 7652
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000169086773888909,
+      "loss": 0.1723,
+      "step": 7653
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016905809817648529,
+      "loss": 0.0911,
+      "step": 7654
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016902942175610905,
+      "loss": 0.0925,
+      "step": 7655
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001690007446288456,
+      "loss": 0.1334,
+      "step": 7656
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016897206679576023,
+      "loss": 0.1021,
+      "step": 7657
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001689433882579182,
+      "loss": 0.0634,
+      "step": 7658
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016891470901638487,
+      "loss": 0.1563,
+      "step": 7659
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016888602907222554,
+      "loss": 0.1088,
+      "step": 7660
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016885734842650562,
+      "loss": 0.0508,
+      "step": 7661
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001688286670802905,
+      "loss": 0.0917,
+      "step": 7662
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016879998503464561,
+      "loss": 0.1858,
+      "step": 7663
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016877130229063644,
+      "loss": 0.1519,
+      "step": 7664
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001687426188493284,
+      "loss": 0.1644,
+      "step": 7665
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016871393471178705,
+      "loss": 0.215,
+      "step": 7666
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001686852498790779,
+      "loss": 0.1966,
+      "step": 7667
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001686565643522665,
+      "loss": 0.1023,
+      "step": 7668
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001686278781324185,
+      "loss": 0.1157,
+      "step": 7669
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016859919122059938,
+      "loss": 0.1353,
+      "step": 7670
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001685705036178749,
+      "loss": 0.0952,
+      "step": 7671
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016854181532531065,
+      "loss": 0.0846,
+      "step": 7672
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001685131263439723,
+      "loss": 0.1018,
+      "step": 7673
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016848443667492562,
+      "loss": 0.0764,
+      "step": 7674
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016845574631923634,
+      "loss": 0.0552,
+      "step": 7675
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016842705527797018,
+      "loss": 0.0517,
+      "step": 7676
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016839836355219296,
+      "loss": 0.063,
+      "step": 7677
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001683696711429705,
+      "loss": 0.0684,
+      "step": 7678
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016834097805136863,
+      "loss": 0.0968,
+      "step": 7679
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016831228427845318,
+      "loss": 0.0544,
+      "step": 7680
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001682835898252901,
+      "loss": 0.0918,
+      "step": 7681
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016825489469294522,
+      "loss": 0.1135,
+      "step": 7682
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001682261988824846,
+      "loss": 0.0384,
+      "step": 7683
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016819750239497412,
+      "loss": 0.0914,
+      "step": 7684
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016816880523147975,
+      "loss": 0.1299,
+      "step": 7685
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016814010739306758,
+      "loss": 0.1225,
+      "step": 7686
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016811140888080365,
+      "loss": 0.078,
+      "step": 7687
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016808270969575397,
+      "loss": 0.1375,
+      "step": 7688
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016805400983898466,
+      "loss": 0.192,
+      "step": 7689
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001680253093115618,
+      "loss": 0.0842,
+      "step": 7690
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016799660811455148,
+      "loss": 0.1334,
+      "step": 7691
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016796790624902007,
+      "loss": 0.1208,
+      "step": 7692
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016793920371603358,
+      "loss": 0.0309,
+      "step": 7693
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016791050051665832,
+      "loss": 0.0613,
+      "step": 7694
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016788179665196045,
+      "loss": 0.03,
+      "step": 7695
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016785309212300625,
+      "loss": 0.1137,
+      "step": 7696
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001678243869308621,
+      "loss": 0.0837,
+      "step": 7697
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016779568107659417,
+      "loss": 0.1209,
+      "step": 7698
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016776697456126892,
+      "loss": 0.1347,
+      "step": 7699
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016773826738595267,
+      "loss": 0.0573,
+      "step": 7700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016770955955171176,
+      "loss": 0.1207,
+      "step": 7701
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001676808510596127,
+      "loss": 0.1415,
+      "step": 7702
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016765214191072182,
+      "loss": 0.1336,
+      "step": 7703
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016762343210610566,
+      "loss": 0.0471,
+      "step": 7704
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001675947216468307,
+      "loss": 0.1532,
+      "step": 7705
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001675660105339634,
+      "loss": 0.049,
+      "step": 7706
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016753729876857031,
+      "loss": 0.1257,
+      "step": 7707
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016750858635171802,
+      "loss": 0.1283,
+      "step": 7708
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016747987328447306,
+      "loss": 0.1654,
+      "step": 7709
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016745115956790212,
+      "loss": 0.1465,
+      "step": 7710
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016742244520307173,
+      "loss": 0.1299,
+      "step": 7711
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016739373019104866,
+      "loss": 0.0596,
+      "step": 7712
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016736501453289943,
+      "loss": 0.1311,
+      "step": 7713
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016733629822969086,
+      "loss": 0.0789,
+      "step": 7714
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016730758128248967,
+      "loss": 0.0819,
+      "step": 7715
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016727886369236256,
+      "loss": 0.1265,
+      "step": 7716
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016725014546037636,
+      "loss": 0.0417,
+      "step": 7717
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016722142658759785,
+      "loss": 0.0625,
+      "step": 7718
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016719270707509382,
+      "loss": 0.0438,
+      "step": 7719
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016716398692393115,
+      "loss": 0.0991,
+      "step": 7720
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001671352661351767,
+      "loss": 0.1432,
+      "step": 7721
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016710654470989737,
+      "loss": 0.1906,
+      "step": 7722
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00016707782264916001,
+      "loss": 0.09,
+      "step": 7723
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016704909995403166,
+      "loss": 0.0977,
+      "step": 7724
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016702037662557926,
+      "loss": 0.082,
+      "step": 7725
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016699165266486978,
+      "loss": 0.0998,
+      "step": 7726
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016696292807297025,
+      "loss": 0.0872,
+      "step": 7727
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016693420285094764,
+      "loss": 0.069,
+      "step": 7728
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016690547699986906,
+      "loss": 0.058,
+      "step": 7729
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016687675052080161,
+      "loss": 0.0935,
+      "step": 7730
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016684802341481234,
+      "loss": 0.0633,
+      "step": 7731
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001668192956829684,
+      "loss": 0.0819,
+      "step": 7732
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016679056732633697,
+      "loss": 0.1012,
+      "step": 7733
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001667618383459852,
+      "loss": 0.0829,
+      "step": 7734
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016673310874298032,
+      "loss": 0.0865,
+      "step": 7735
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016670437851838948,
+      "loss": 0.1293,
+      "step": 7736
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016667564767327992,
+      "loss": 0.0886,
+      "step": 7737
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016664691620871906,
+      "loss": 0.1338,
+      "step": 7738
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000166618184125774,
+      "loss": 0.142,
+      "step": 7739
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016658945142551212,
+      "loss": 0.1365,
+      "step": 7740
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001665607181090008,
+      "loss": 0.0597,
+      "step": 7741
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016653198417730734,
+      "loss": 0.1453,
+      "step": 7742
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016650324963149923,
+      "loss": 0.1185,
+      "step": 7743
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016647451447264372,
+      "loss": 0.0355,
+      "step": 7744
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016644577870180834,
+      "loss": 0.1043,
+      "step": 7745
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001664170423200605,
+      "loss": 0.042,
+      "step": 7746
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016638830532846765,
+      "loss": 0.099,
+      "step": 7747
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016635956772809732,
+      "loss": 0.115,
+      "step": 7748
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016633082952001704,
+      "loss": 0.0588,
+      "step": 7749
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016630209070529433,
+      "loss": 0.1148,
+      "step": 7750
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016627335128499676,
+      "loss": 0.1807,
+      "step": 7751
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001662446112601919,
+      "loss": 0.1903,
+      "step": 7752
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001662158706319474,
+      "loss": 0.0635,
+      "step": 7753
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016618712940133084,
+      "loss": 0.1124,
+      "step": 7754
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016615838756940985,
+      "loss": 0.0353,
+      "step": 7755
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016612964513725218,
+      "loss": 0.1283,
+      "step": 7756
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001661009021059255,
+      "loss": 0.0878,
+      "step": 7757
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016607215847649754,
+      "loss": 0.0805,
+      "step": 7758
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000166043414250036,
+      "loss": 0.0275,
+      "step": 7759
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001660146694276087,
+      "loss": 0.1175,
+      "step": 7760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016598592401028336,
+      "loss": 0.07,
+      "step": 7761
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016595717799912778,
+      "loss": 0.0443,
+      "step": 7762
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001659284313952099,
+      "loss": 0.0953,
+      "step": 7763
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016589968419959743,
+      "loss": 0.0995,
+      "step": 7764
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016587093641335843,
+      "loss": 0.0399,
+      "step": 7765
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016584218803756057,
+      "loss": 0.1324,
+      "step": 7766
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016581343907327194,
+      "loss": 0.1038,
+      "step": 7767
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001657846895215604,
+      "loss": 0.1385,
+      "step": 7768
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016575593938349397,
+      "loss": 0.1286,
+      "step": 7769
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016572718866014058,
+      "loss": 0.0544,
+      "step": 7770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016569843735256824,
+      "loss": 0.0684,
+      "step": 7771
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016566968546184502,
+      "loss": 0.173,
+      "step": 7772
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016564093298903886,
+      "loss": 0.0355,
+      "step": 7773
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016561217993521794,
+      "loss": 0.1156,
+      "step": 7774
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016558342630145032,
+      "loss": 0.0558,
+      "step": 7775
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016555467208880415,
+      "loss": 0.1675,
+      "step": 7776
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001655259172983475,
+      "loss": 0.1135,
+      "step": 7777
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016549716193114855,
+      "loss": 0.1174,
+      "step": 7778
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016546840598827546,
+      "loss": 0.1017,
+      "step": 7779
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016543964947079648,
+      "loss": 0.1635,
+      "step": 7780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016541089237977975,
+      "loss": 0.1405,
+      "step": 7781
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001653821347162936,
+      "loss": 0.144,
+      "step": 7782
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016535337648140627,
+      "loss": 0.0803,
+      "step": 7783
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000165324617676186,
+      "loss": 0.1096,
+      "step": 7784
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016529585830170112,
+      "loss": 0.068,
+      "step": 7785
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016526709835902,
+      "loss": 0.0862,
+      "step": 7786
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016523833784921088,
+      "loss": 0.1787,
+      "step": 7787
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016520957677334223,
+      "loss": 0.0574,
+      "step": 7788
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001651808151324824,
+      "loss": 0.0922,
+      "step": 7789
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016515205292769977,
+      "loss": 0.0951,
+      "step": 7790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016512329016006287,
+      "loss": 0.0877,
+      "step": 7791
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016509452683064006,
+      "loss": 0.1112,
+      "step": 7792
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016506576294049986,
+      "loss": 0.099,
+      "step": 7793
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001650369984907107,
+      "loss": 0.1127,
+      "step": 7794
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016500823348234117,
+      "loss": 0.0566,
+      "step": 7795
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016497946791645978,
+      "loss": 0.0411,
+      "step": 7796
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016495070179413506,
+      "loss": 0.1093,
+      "step": 7797
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016492193511643568,
+      "loss": 0.1518,
+      "step": 7798
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016489316788443012,
+      "loss": 0.0895,
+      "step": 7799
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016486440009918704,
+      "loss": 0.0604,
+      "step": 7800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001648356317617751,
+      "loss": 0.1342,
+      "step": 7801
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016480686287326297,
+      "loss": 0.0807,
+      "step": 7802
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016477809343471924,
+      "loss": 0.1763,
+      "step": 7803
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016474932344721273,
+      "loss": 0.1888,
+      "step": 7804
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001647205529118121,
+      "loss": 0.0971,
+      "step": 7805
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016469178182958608,
+      "loss": 0.0747,
+      "step": 7806
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016466301020160346,
+      "loss": 0.048,
+      "step": 7807
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016463423802893297,
+      "loss": 0.0973,
+      "step": 7808
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001646054653126435,
+      "loss": 0.0698,
+      "step": 7809
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016457669205380382,
+      "loss": 0.0494,
+      "step": 7810
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016454791825348277,
+      "loss": 0.1232,
+      "step": 7811
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001645191439127492,
+      "loss": 0.0791,
+      "step": 7812
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.000164490369032672,
+      "loss": 0.1269,
+      "step": 7813
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016446159361432013,
+      "loss": 0.0956,
+      "step": 7814
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016443281765876245,
+      "loss": 0.096,
+      "step": 7815
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001644040411670679,
+      "loss": 0.0959,
+      "step": 7816
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016437526414030546,
+      "loss": 0.0522,
+      "step": 7817
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016434648657954413,
+      "loss": 0.083,
+      "step": 7818
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016431770848585292,
+      "loss": 0.1261,
+      "step": 7819
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016428892986030075,
+      "loss": 0.0856,
+      "step": 7820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001642601507039568,
+      "loss": 0.1101,
+      "step": 7821
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016423137101789003,
+      "loss": 0.0986,
+      "step": 7822
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016420259080316958,
+      "loss": 0.1255,
+      "step": 7823
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016417381006086457,
+      "loss": 0.118,
+      "step": 7824
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000164145028792044,
+      "loss": 0.0906,
+      "step": 7825
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016411624699777717,
+      "loss": 0.0925,
+      "step": 7826
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001640874646791331,
+      "loss": 0.0607,
+      "step": 7827
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016405868183718103,
+      "loss": 0.0711,
+      "step": 7828
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016402989847299018,
+      "loss": 0.0288,
+      "step": 7829
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016400111458762974,
+      "loss": 0.0897,
+      "step": 7830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016397233018216893,
+      "loss": 0.1182,
+      "step": 7831
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001639435452576771,
+      "loss": 0.1173,
+      "step": 7832
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016391475981522336,
+      "loss": 0.1189,
+      "step": 7833
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016388597385587716,
+      "loss": 0.0899,
+      "step": 7834
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001638571873807077,
+      "loss": 0.1302,
+      "step": 7835
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016382840039078442,
+      "loss": 0.0516,
+      "step": 7836
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001637996128871766,
+      "loss": 0.0635,
+      "step": 7837
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016377082487095356,
+      "loss": 0.1744,
+      "step": 7838
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016374203634318487,
+      "loss": 0.1013,
+      "step": 7839
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016371324730493975,
+      "loss": 0.1473,
+      "step": 7840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016368445775728774,
+      "loss": 0.2416,
+      "step": 7841
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016365566770129827,
+      "loss": 0.1236,
+      "step": 7842
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016362687713804077,
+      "loss": 0.0648,
+      "step": 7843
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016359808606858473,
+      "loss": 0.0377,
+      "step": 7844
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016356929449399968,
+      "loss": 0.035,
+      "step": 7845
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001635405024153551,
+      "loss": 0.0766,
+      "step": 7846
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001635117098337206,
+      "loss": 0.0778,
+      "step": 7847
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001634829167501657,
+      "loss": 0.0691,
+      "step": 7848
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016345412316575998,
+      "loss": 0.0998,
+      "step": 7849
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016342532908157303,
+      "loss": 0.1284,
+      "step": 7850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016339653449867447,
+      "loss": 0.0764,
+      "step": 7851
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.000163367739418134,
+      "loss": 0.0856,
+      "step": 7852
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016333894384102114,
+      "loss": 0.1052,
+      "step": 7853
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016331014776840562,
+      "loss": 0.0202,
+      "step": 7854
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016328135120135715,
+      "loss": 0.1292,
+      "step": 7855
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016325255414094547,
+      "loss": 0.1442,
+      "step": 7856
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016322375658824027,
+      "loss": 0.2605,
+      "step": 7857
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016319495854431125,
+      "loss": 0.0947,
+      "step": 7858
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016316616001022828,
+      "loss": 0.0811,
+      "step": 7859
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016313736098706098,
+      "loss": 0.0264,
+      "step": 7860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016310856147587932,
+      "loss": 0.0327,
+      "step": 7861
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016307976147775298,
+      "loss": 0.0784,
+      "step": 7862
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001630509609937519,
+      "loss": 0.0931,
+      "step": 7863
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001630221600249459,
+      "loss": 0.0864,
+      "step": 7864
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016299335857240484,
+      "loss": 0.1294,
+      "step": 7865
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016296455663719858,
+      "loss": 0.0763,
+      "step": 7866
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016293575422039709,
+      "loss": 0.1448,
+      "step": 7867
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016290695132307025,
+      "loss": 0.0951,
+      "step": 7868
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016287814794628801,
+      "loss": 0.1299,
+      "step": 7869
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016284934409112033,
+      "loss": 0.0938,
+      "step": 7870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016282053975863722,
+      "loss": 0.0403,
+      "step": 7871
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016279173494990865,
+      "loss": 0.0914,
+      "step": 7872
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016276292966600463,
+      "loss": 0.1428,
+      "step": 7873
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001627341239079952,
+      "loss": 0.1533,
+      "step": 7874
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016270531767695047,
+      "loss": 0.105,
+      "step": 7875
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016267651097394035,
+      "loss": 0.048,
+      "step": 7876
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001626477038000351,
+      "loss": 0.0443,
+      "step": 7877
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001626188961563047,
+      "loss": 0.1721,
+      "step": 7878
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016259008804381928,
+      "loss": 0.1121,
+      "step": 7879
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016256127946364906,
+      "loss": 0.0605,
+      "step": 7880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001625324704168641,
+      "loss": 0.0633,
+      "step": 7881
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001625036609045347,
+      "loss": 0.1406,
+      "step": 7882
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016247485092773088,
+      "loss": 0.0791,
+      "step": 7883
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016244604048752296,
+      "loss": 0.1072,
+      "step": 7884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016241722958498117,
+      "loss": 0.1667,
+      "step": 7885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016238841822117567,
+      "loss": 0.1249,
+      "step": 7886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00016235960639717674,
+      "loss": 0.0745,
+      "step": 7887
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001623307941140547,
+      "loss": 0.1119,
+      "step": 7888
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016230198137287983,
+      "loss": 0.13,
+      "step": 7889
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016227316817472244,
+      "loss": 0.0775,
+      "step": 7890
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016224435452065283,
+      "loss": 0.1615,
+      "step": 7891
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016221554041174132,
+      "loss": 0.129,
+      "step": 7892
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016218672584905838,
+      "loss": 0.0802,
+      "step": 7893
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016215791083367423,
+      "loss": 0.1316,
+      "step": 7894
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001621290953666594,
+      "loss": 0.1284,
+      "step": 7895
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016210027944908422,
+      "loss": 0.138,
+      "step": 7896
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016207146308201917,
+      "loss": 0.17,
+      "step": 7897
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016204264626653466,
+      "loss": 0.0864,
+      "step": 7898
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016201382900370113,
+      "loss": 0.1416,
+      "step": 7899
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016198501129458909,
+      "loss": 0.0922,
+      "step": 7900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016195619314026904,
+      "loss": 0.0783,
+      "step": 7901
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016192737454181146,
+      "loss": 0.0349,
+      "step": 7902
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001618985555002869,
+      "loss": 0.0838,
+      "step": 7903
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016186973601676588,
+      "loss": 0.0461,
+      "step": 7904
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000161840916092319,
+      "loss": 0.1288,
+      "step": 7905
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016181209572801682,
+      "loss": 0.0643,
+      "step": 7906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001617832749249299,
+      "loss": 0.0892,
+      "step": 7907
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016175445368412888,
+      "loss": 0.1246,
+      "step": 7908
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001617256320066844,
+      "loss": 0.1164,
+      "step": 7909
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016169680989366704,
+      "loss": 0.094,
+      "step": 7910
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016166798734614746,
+      "loss": 0.1313,
+      "step": 7911
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016163916436519638,
+      "loss": 0.0669,
+      "step": 7912
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016161034095188453,
+      "loss": 0.1202,
+      "step": 7913
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001615815171072825,
+      "loss": 0.0674,
+      "step": 7914
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016155269283246116,
+      "loss": 0.151,
+      "step": 7915
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001615238681284911,
+      "loss": 0.1168,
+      "step": 7916
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016149504299644313,
+      "loss": 0.0864,
+      "step": 7917
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016146621743738803,
+      "loss": 0.1241,
+      "step": 7918
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016143739145239657,
+      "loss": 0.1587,
+      "step": 7919
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016140856504253954,
+      "loss": 0.0372,
+      "step": 7920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001613797382088878,
+      "loss": 0.1124,
+      "step": 7921
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016135091095251212,
+      "loss": 0.122,
+      "step": 7922
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016132208327448346,
+      "loss": 0.0994,
+      "step": 7923
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016129325517587255,
+      "loss": 0.1526,
+      "step": 7924
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016126442665775033,
+      "loss": 0.0389,
+      "step": 7925
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016123559772118773,
+      "loss": 0.1136,
+      "step": 7926
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001612067683672556,
+      "loss": 0.1143,
+      "step": 7927
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016117793859702487,
+      "loss": 0.1303,
+      "step": 7928
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001611491084115665,
+      "loss": 0.1009,
+      "step": 7929
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016112027781195142,
+      "loss": 0.0865,
+      "step": 7930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016109144679925073,
+      "loss": 0.0859,
+      "step": 7931
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016106261537453525,
+      "loss": 0.0772,
+      "step": 7932
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016103378353887608,
+      "loss": 0.045,
+      "step": 7933
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016100495129334417,
+      "loss": 0.1375,
+      "step": 7934
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001609761186390106,
+      "loss": 0.0671,
+      "step": 7935
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001609472855769464,
+      "loss": 0.0369,
+      "step": 7936
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016091845210822265,
+      "loss": 0.1557,
+      "step": 7937
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001608896182339105,
+      "loss": 0.0633,
+      "step": 7938
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016086078395508084,
+      "loss": 0.1308,
+      "step": 7939
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016083194927280495,
+      "loss": 0.1083,
+      "step": 7940
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000160803114188154,
+      "loss": 0.1635,
+      "step": 7941
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001607742787021989,
+      "loss": 0.0546,
+      "step": 7942
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000160745442816011,
+      "loss": 0.09,
+      "step": 7943
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001607166065306614,
+      "loss": 0.1225,
+      "step": 7944
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001606877698472213,
+      "loss": 0.1088,
+      "step": 7945
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016065893276676193,
+      "loss": 0.1197,
+      "step": 7946
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016063009529035442,
+      "loss": 0.1327,
+      "step": 7947
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016060125741907007,
+      "loss": 0.0368,
+      "step": 7948
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001605724191539801,
+      "loss": 0.1118,
+      "step": 7949
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016054358049615573,
+      "loss": 0.0867,
+      "step": 7950
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016051474144666828,
+      "loss": 0.0287,
+      "step": 7951
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016048590200658897,
+      "loss": 0.1053,
+      "step": 7952
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016045706217698925,
+      "loss": 0.1202,
+      "step": 7953
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016042822195894024,
+      "loss": 0.1976,
+      "step": 7954
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001603993813535134,
+      "loss": 0.1219,
+      "step": 7955
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016037054036178007,
+      "loss": 0.197,
+      "step": 7956
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001603416989848115,
+      "loss": 0.085,
+      "step": 7957
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001603128572236792,
+      "loss": 0.092,
+      "step": 7958
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016028401507945454,
+      "loss": 0.0101,
+      "step": 7959
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001602551725532088,
+      "loss": 0.1299,
+      "step": 7960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001602263296460135,
+      "loss": 0.0856,
+      "step": 7961
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016019748635894,
+      "loss": 0.0663,
+      "step": 7962
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016016864269305982,
+      "loss": 0.0641,
+      "step": 7963
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001601397986494444,
+      "loss": 0.015,
+      "step": 7964
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016011095422916518,
+      "loss": 0.0455,
+      "step": 7965
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016008210943329367,
+      "loss": 0.0972,
+      "step": 7966
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016005326426290136,
+      "loss": 0.1176,
+      "step": 7967
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016002441871905976,
+      "loss": 0.2,
+      "step": 7968
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015999557280284039,
+      "loss": 0.0932,
+      "step": 7969
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001599667265153148,
+      "loss": 0.0707,
+      "step": 7970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015993787985755454,
+      "loss": 0.1398,
+      "step": 7971
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015990903283063124,
+      "loss": 0.1552,
+      "step": 7972
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015988018543561635,
+      "loss": 0.1196,
+      "step": 7973
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001598513376735816,
+      "loss": 0.0841,
+      "step": 7974
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015982248954559853,
+      "loss": 0.0914,
+      "step": 7975
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015979364105273879,
+      "loss": 0.0578,
+      "step": 7976
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015976479219607397,
+      "loss": 0.22,
+      "step": 7977
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015973594297667576,
+      "loss": 0.1679,
+      "step": 7978
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015970709339561587,
+      "loss": 0.0462,
+      "step": 7979
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015967824345396585,
+      "loss": 0.1005,
+      "step": 7980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015964939315279758,
+      "loss": 0.0246,
+      "step": 7981
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015962054249318253,
+      "loss": 0.1712,
+      "step": 7982
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001595916914761926,
+      "loss": 0.1677,
+      "step": 7983
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015956284010289944,
+      "loss": 0.0431,
+      "step": 7984
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001595339883743748,
+      "loss": 0.2027,
+      "step": 7985
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015950513629169047,
+      "loss": 0.1112,
+      "step": 7986
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001594762838559182,
+      "loss": 0.0778,
+      "step": 7987
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001594474310681298,
+      "loss": 0.1145,
+      "step": 7988
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000159418577929397,
+      "loss": 0.1637,
+      "step": 7989
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015938972444079166,
+      "loss": 0.0586,
+      "step": 7990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001593608706033856,
+      "loss": 0.1182,
+      "step": 7991
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015933201641825067,
+      "loss": 0.1154,
+      "step": 7992
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015930316188645863,
+      "loss": 0.0738,
+      "step": 7993
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015927430700908145,
+      "loss": 0.0442,
+      "step": 7994
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015924545178719096,
+      "loss": 0.0936,
+      "step": 7995
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015921659622185902,
+      "loss": 0.1598,
+      "step": 7996
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015918774031415764,
+      "loss": 0.0541,
+      "step": 7997
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015915888406515857,
+      "loss": 0.0972,
+      "step": 7998
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015913002747593383,
+      "loss": 0.0383,
+      "step": 7999
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015910117054755536,
+      "loss": 0.1517,
+      "step": 8000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001590723132810951,
+      "loss": 0.1274,
+      "step": 8001
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000159043455677625,
+      "loss": 0.0914,
+      "step": 8002
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.000159014597738217,
+      "loss": 0.122,
+      "step": 8003
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015898573946394317,
+      "loss": 0.1151,
+      "step": 8004
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015895688085587544,
+      "loss": 0.111,
+      "step": 8005
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015892802191508587,
+      "loss": 0.0728,
+      "step": 8006
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015889916264264653,
+      "loss": 0.0595,
+      "step": 8007
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001588703030396293,
+      "loss": 0.0475,
+      "step": 8008
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015884144310710635,
+      "loss": 0.0897,
+      "step": 8009
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015881258284614972,
+      "loss": 0.0252,
+      "step": 8010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015878372225783144,
+      "loss": 0.0459,
+      "step": 8011
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015875486134322373,
+      "loss": 0.1267,
+      "step": 8012
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015872600010339853,
+      "loss": 0.0952,
+      "step": 8013
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015869713853942807,
+      "loss": 0.1075,
+      "step": 8014
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015866827665238434,
+      "loss": 0.1485,
+      "step": 8015
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015863941444333962,
+      "loss": 0.0222,
+      "step": 8016
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015861055191336592,
+      "loss": 0.1269,
+      "step": 8017
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015858168906353548,
+      "loss": 0.0245,
+      "step": 8018
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015855282589492047,
+      "loss": 0.1304,
+      "step": 8019
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015852396240859313,
+      "loss": 0.0493,
+      "step": 8020
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001584950986056255,
+      "loss": 0.0435,
+      "step": 8021
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001584662344870899,
+      "loss": 0.1846,
+      "step": 8022
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001584373700540585,
+      "loss": 0.1347,
+      "step": 8023
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015840850530760353,
+      "loss": 0.1065,
+      "step": 8024
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015837964024879728,
+      "loss": 0.093,
+      "step": 8025
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015835077487871192,
+      "loss": 0.0773,
+      "step": 8026
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015832190919841982,
+      "loss": 0.0875,
+      "step": 8027
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015829304320899318,
+      "loss": 0.0614,
+      "step": 8028
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015826417691150427,
+      "loss": 0.157,
+      "step": 8029
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001582353103070255,
+      "loss": 0.1642,
+      "step": 8030
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015820644339662905,
+      "loss": 0.1728,
+      "step": 8031
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001581775761813873,
+      "loss": 0.1033,
+      "step": 8032
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015814870866237256,
+      "loss": 0.0549,
+      "step": 8033
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015811984084065716,
+      "loss": 0.0474,
+      "step": 8034
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015809097271731357,
+      "loss": 0.0747,
+      "step": 8035
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015806210429341398,
+      "loss": 0.1344,
+      "step": 8036
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015803323557003093,
+      "loss": 0.0723,
+      "step": 8037
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001580043665482367,
+      "loss": 0.1011,
+      "step": 8038
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015797549722910373,
+      "loss": 0.195,
+      "step": 8039
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015794662761370445,
+      "loss": 0.1355,
+      "step": 8040
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001579177577031112,
+      "loss": 0.0556,
+      "step": 8041
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015788888749839651,
+      "loss": 0.1184,
+      "step": 8042
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001578600170006328,
+      "loss": 0.0624,
+      "step": 8043
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015783114621089245,
+      "loss": 0.1157,
+      "step": 8044
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015780227513024806,
+      "loss": 0.1518,
+      "step": 8045
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015777340375977198,
+      "loss": 0.0301,
+      "step": 8046
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015774453210053675,
+      "loss": 0.0899,
+      "step": 8047
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015771566015361488,
+      "loss": 0.071,
+      "step": 8048
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015768678792007883,
+      "loss": 0.0802,
+      "step": 8049
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015765791540100116,
+      "loss": 0.1099,
+      "step": 8050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001576290425974544,
+      "loss": 0.0981,
+      "step": 8051
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015760016951051108,
+      "loss": 0.0915,
+      "step": 8052
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015757129614124376,
+      "loss": 0.0814,
+      "step": 8053
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015754242249072499,
+      "loss": 0.132,
+      "step": 8054
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001575135485600273,
+      "loss": 0.1506,
+      "step": 8055
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015748467435022334,
+      "loss": 0.1243,
+      "step": 8056
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001574557998623857,
+      "loss": 0.0588,
+      "step": 8057
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015742692509758695,
+      "loss": 0.1493,
+      "step": 8058
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015739805005689968,
+      "loss": 0.2068,
+      "step": 8059
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001573691747413966,
+      "loss": 0.039,
+      "step": 8060
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015734029915215022,
+      "loss": 0.1384,
+      "step": 8061
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015731142329023328,
+      "loss": 0.0756,
+      "step": 8062
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015728254715671844,
+      "loss": 0.1259,
+      "step": 8063
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001572536707526783,
+      "loss": 0.1249,
+      "step": 8064
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001572247940791856,
+      "loss": 0.0287,
+      "step": 8065
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001571959171373129,
+      "loss": 0.1134,
+      "step": 8066
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.000157167039928133,
+      "loss": 0.1261,
+      "step": 8067
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015713816245271868,
+      "loss": 0.1371,
+      "step": 8068
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015710928471214246,
+      "loss": 0.0554,
+      "step": 8069
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001570804067074772,
+      "loss": 0.0588,
+      "step": 8070
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001570515284397956,
+      "loss": 0.082,
+      "step": 8071
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015702264991017037,
+      "loss": 0.1119,
+      "step": 8072
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015699377111967432,
+      "loss": 0.0633,
+      "step": 8073
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015696489206938017,
+      "loss": 0.0623,
+      "step": 8074
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015693601276036068,
+      "loss": 0.0916,
+      "step": 8075
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015690713319368868,
+      "loss": 0.0936,
+      "step": 8076
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001568782533704369,
+      "loss": 0.1154,
+      "step": 8077
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015684937329167825,
+      "loss": 0.0708,
+      "step": 8078
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015682049295848543,
+      "loss": 0.0547,
+      "step": 8079
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015679161237193128,
+      "loss": 0.0763,
+      "step": 8080
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001567627315330887,
+      "loss": 0.186,
+      "step": 8081
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001567338504430304,
+      "loss": 0.1369,
+      "step": 8082
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001567049691028293,
+      "loss": 0.1117,
+      "step": 8083
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015667608751355832,
+      "loss": 0.1253,
+      "step": 8084
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015664720567629023,
+      "loss": 0.1734,
+      "step": 8085
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000156618323592098,
+      "loss": 0.0799,
+      "step": 8086
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001565894412620544,
+      "loss": 0.1101,
+      "step": 8087
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015656055868723241,
+      "loss": 0.154,
+      "step": 8088
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015653167586870487,
+      "loss": 0.0561,
+      "step": 8089
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015650279280754475,
+      "loss": 0.0978,
+      "step": 8090
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015647390950482493,
+      "loss": 0.1102,
+      "step": 8091
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015644502596161836,
+      "loss": 0.1094,
+      "step": 8092
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.000156416142178998,
+      "loss": 0.1248,
+      "step": 8093
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015638725815803674,
+      "loss": 0.141,
+      "step": 8094
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001563583738998076,
+      "loss": 0.0028,
+      "step": 8095
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001563294894053835,
+      "loss": 0.0273,
+      "step": 8096
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015630060467583743,
+      "loss": 0.0233,
+      "step": 8097
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015627171971224237,
+      "loss": 0.1655,
+      "step": 8098
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015624283451567131,
+      "loss": 0.1185,
+      "step": 8099
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015621394908719725,
+      "loss": 0.0782,
+      "step": 8100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015618506342789324,
+      "loss": 0.0046,
+      "step": 8101
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015615617753883223,
+      "loss": 0.1168,
+      "step": 8102
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001561272914210873,
+      "loss": 0.1769,
+      "step": 8103
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015609840507573147,
+      "loss": 0.0709,
+      "step": 8104
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001560695185038377,
+      "loss": 0.0984,
+      "step": 8105
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001560406317064792,
+      "loss": 0.0948,
+      "step": 8106
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015601174468472887,
+      "loss": 0.0526,
+      "step": 8107
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015598285743965995,
+      "loss": 0.0777,
+      "step": 8108
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015595396997234533,
+      "loss": 0.113,
+      "step": 8109
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015592508228385823,
+      "loss": 0.0697,
+      "step": 8110
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015589619437527172,
+      "loss": 0.1112,
+      "step": 8111
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015586730624765884,
+      "loss": 0.2121,
+      "step": 8112
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015583841790209273,
+      "loss": 0.1014,
+      "step": 8113
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015580952933964655,
+      "loss": 0.1105,
+      "step": 8114
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015578064056139337,
+      "loss": 0.0531,
+      "step": 8115
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015575175156840637,
+      "loss": 0.1064,
+      "step": 8116
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015572286236175865,
+      "loss": 0.1223,
+      "step": 8117
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015569397294252333,
+      "loss": 0.1339,
+      "step": 8118
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015566508331177372,
+      "loss": 0.0695,
+      "step": 8119
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001556361934705828,
+      "loss": 0.085,
+      "step": 8120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015560730342002382,
+      "loss": 0.1259,
+      "step": 8121
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015557841316116996,
+      "loss": 0.0602,
+      "step": 8122
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015554952269509445,
+      "loss": 0.1502,
+      "step": 8123
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015552063202287043,
+      "loss": 0.103,
+      "step": 8124
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015549174114557108,
+      "loss": 0.2196,
+      "step": 8125
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015546285006426972,
+      "loss": 0.0797,
+      "step": 8126
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015543395878003943,
+      "loss": 0.0377,
+      "step": 8127
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015540506729395354,
+      "loss": 0.1079,
+      "step": 8128
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015537617560708528,
+      "loss": 0.0964,
+      "step": 8129
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001553472837205078,
+      "loss": 0.0758,
+      "step": 8130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015531839163529445,
+      "loss": 0.145,
+      "step": 8131
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001552894993525184,
+      "loss": 0.1467,
+      "step": 8132
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015526060687325299,
+      "loss": 0.1646,
+      "step": 8133
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001552317141985715,
+      "loss": 0.053,
+      "step": 8134
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015520282132954713,
+      "loss": 0.1166,
+      "step": 8135
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015517392826725323,
+      "loss": 0.043,
+      "step": 8136
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015514503501276304,
+      "loss": 0.0673,
+      "step": 8137
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015511614156714989,
+      "loss": 0.0599,
+      "step": 8138
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015508724793148707,
+      "loss": 0.1077,
+      "step": 8139
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015505835410684792,
+      "loss": 0.0552,
+      "step": 8140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001550294600943058,
+      "loss": 0.0543,
+      "step": 8141
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001550005658949339,
+      "loss": 0.0779,
+      "step": 8142
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015497167150980572,
+      "loss": 0.1014,
+      "step": 8143
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001549427769399945,
+      "loss": 0.1002,
+      "step": 8144
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001549138821865736,
+      "loss": 0.0951,
+      "step": 8145
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015488498725061637,
+      "loss": 0.0638,
+      "step": 8146
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015485609213319624,
+      "loss": 0.1346,
+      "step": 8147
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015482719683538646,
+      "loss": 0.0909,
+      "step": 8148
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001547983013582605,
+      "loss": 0.1157,
+      "step": 8149
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015476940570289171,
+      "loss": 0.0793,
+      "step": 8150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015474050987035346,
+      "loss": 0.0675,
+      "step": 8151
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015471161386171922,
+      "loss": 0.0932,
+      "step": 8152
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015468271767806233,
+      "loss": 0.0412,
+      "step": 8153
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015465382132045624,
+      "loss": 0.1175,
+      "step": 8154
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001546249247899743,
+      "loss": 0.1129,
+      "step": 8155
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015459602808768995,
+      "loss": 0.0516,
+      "step": 8156
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015456713121467664,
+      "loss": 0.0864,
+      "step": 8157
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015453823417200778,
+      "loss": 0.0772,
+      "step": 8158
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015450933696075693,
+      "loss": 0.048,
+      "step": 8159
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015448043958199732,
+      "loss": 0.1207,
+      "step": 8160
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015445154203680257,
+      "loss": 0.0328,
+      "step": 8161
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001544226443262461,
+      "loss": 0.148,
+      "step": 8162
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015439374645140135,
+      "loss": 0.102,
+      "step": 8163
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001543648484133418,
+      "loss": 0.0569,
+      "step": 8164
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015433595021314088,
+      "loss": 0.1099,
+      "step": 8165
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015430705185187215,
+      "loss": 0.1601,
+      "step": 8166
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015427815333060916,
+      "loss": 0.0925,
+      "step": 8167
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015424925465042522,
+      "loss": 0.1062,
+      "step": 8168
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000154220355812394,
+      "loss": 0.0962,
+      "step": 8169
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015419145681758888,
+      "loss": 0.0504,
+      "step": 8170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015416255766708344,
+      "loss": 0.1631,
+      "step": 8171
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001541336583619512,
+      "loss": 0.1028,
+      "step": 8172
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015410475890326568,
+      "loss": 0.0779,
+      "step": 8173
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015407585929210038,
+      "loss": 0.063,
+      "step": 8174
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001540469595295289,
+      "loss": 0.1603,
+      "step": 8175
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015401805961662472,
+      "loss": 0.1072,
+      "step": 8176
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015398915955446146,
+      "loss": 0.1471,
+      "step": 8177
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015396025934411253,
+      "loss": 0.0434,
+      "step": 8178
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015393135898665163,
+      "loss": 0.0655,
+      "step": 8179
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015390245848315227,
+      "loss": 0.0,
+      "step": 8180
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015387355783468798,
+      "loss": 0.0512,
+      "step": 8181
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015384465704233242,
+      "loss": 0.1007,
+      "step": 8182
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001538157561071591,
+      "loss": 0.1131,
+      "step": 8183
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015378685503024164,
+      "loss": 0.1194,
+      "step": 8184
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015375795381265367,
+      "loss": 0.08,
+      "step": 8185
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015372905245546867,
+      "loss": 0.1126,
+      "step": 8186
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015370015095976037,
+      "loss": 0.05,
+      "step": 8187
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015367124932660225,
+      "loss": 0.0408,
+      "step": 8188
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015364234755706798,
+      "loss": 0.115,
+      "step": 8189
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001536134456522312,
+      "loss": 0.0621,
+      "step": 8190
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001535845436131655,
+      "loss": 0.0638,
+      "step": 8191
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015355564144094457,
+      "loss": 0.106,
+      "step": 8192
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015352673913664194,
+      "loss": 0.0697,
+      "step": 8193
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015349783670133126,
+      "loss": 0.2048,
+      "step": 8194
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015346893413608628,
+      "loss": 0.1151,
+      "step": 8195
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015344003144198052,
+      "loss": 0.1467,
+      "step": 8196
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001534111286200877,
+      "loss": 0.0751,
+      "step": 8197
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015338222567148142,
+      "loss": 0.1981,
+      "step": 8198
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001533533225972354,
+      "loss": 0.0583,
+      "step": 8199
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001533244193984233,
+      "loss": 0.1014,
+      "step": 8200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015329551607611873,
+      "loss": 0.0686,
+      "step": 8201
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015326661263139545,
+      "loss": 0.0664,
+      "step": 8202
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015323770906532707,
+      "loss": 0.1848,
+      "step": 8203
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015320880537898724,
+      "loss": 0.0869,
+      "step": 8204
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015317990157344977,
+      "loss": 0.1762,
+      "step": 8205
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015315099764978823,
+      "loss": 0.0789,
+      "step": 8206
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015312209360907638,
+      "loss": 0.0753,
+      "step": 8207
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015309318945238795,
+      "loss": 0.0219,
+      "step": 8208
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015306428518079655,
+      "loss": 0.12,
+      "step": 8209
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000153035380795376,
+      "loss": 0.1021,
+      "step": 8210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015300647629719991,
+      "loss": 0.0746,
+      "step": 8211
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015297757168734205,
+      "loss": 0.0827,
+      "step": 8212
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001529486669668761,
+      "loss": 0.0608,
+      "step": 8213
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015291976213687586,
+      "loss": 0.0981,
+      "step": 8214
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015289085719841504,
+      "loss": 0.1284,
+      "step": 8215
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001528619521525673,
+      "loss": 0.0783,
+      "step": 8216
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015283304700040647,
+      "loss": 0.105,
+      "step": 8217
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015280414174300626,
+      "loss": 0.092,
+      "step": 8218
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015277523638144037,
+      "loss": 0.0673,
+      "step": 8219
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015274633091678263,
+      "loss": 0.1004,
+      "step": 8220
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001527174253501067,
+      "loss": 0.1237,
+      "step": 8221
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015268851968248636,
+      "loss": 0.0898,
+      "step": 8222
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001526596139149955,
+      "loss": 0.1175,
+      "step": 8223
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001526307080487077,
+      "loss": 0.091,
+      "step": 8224
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015260180208469688,
+      "loss": 0.0397,
+      "step": 8225
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001525728960240367,
+      "loss": 0.131,
+      "step": 8226
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015254398986780095,
+      "loss": 0.1033,
+      "step": 8227
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015251508361706348,
+      "loss": 0.0587,
+      "step": 8228
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000152486177272898,
+      "loss": 0.1142,
+      "step": 8229
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015245727083637833,
+      "loss": 0.0831,
+      "step": 8230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015242836430857823,
+      "loss": 0.1701,
+      "step": 8231
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015239945769057156,
+      "loss": 0.071,
+      "step": 8232
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015237055098343212,
+      "loss": 0.0947,
+      "step": 8233
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015234164418823356,
+      "loss": 0.1028,
+      "step": 8234
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015231273730604986,
+      "loss": 0.0705,
+      "step": 8235
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015228383033795472,
+      "loss": 0.0734,
+      "step": 8236
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015225492328502197,
+      "loss": 0.1234,
+      "step": 8237
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015222601614832545,
+      "loss": 0.1616,
+      "step": 8238
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015219710892893891,
+      "loss": 0.062,
+      "step": 8239
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015216820162793627,
+      "loss": 0.1176,
+      "step": 8240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015213929424639133,
+      "loss": 0.0817,
+      "step": 8241
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015211038678537782,
+      "loss": 0.1483,
+      "step": 8242
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015208147924596966,
+      "loss": 0.0617,
+      "step": 8243
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015205257162924065,
+      "loss": 0.1071,
+      "step": 8244
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015202366393626457,
+      "loss": 0.0829,
+      "step": 8245
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015199475616811534,
+      "loss": 0.0868,
+      "step": 8246
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001519658483258668,
+      "loss": 0.0819,
+      "step": 8247
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015193694041059274,
+      "loss": 0.0776,
+      "step": 8248
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015190803242336702,
+      "loss": 0.0961,
+      "step": 8249
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001518791243652635,
+      "loss": 0.1136,
+      "step": 8250
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015185021623735604,
+      "loss": 0.078,
+      "step": 8251
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015182130804071843,
+      "loss": 0.1015,
+      "step": 8252
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001517923997764246,
+      "loss": 0.0905,
+      "step": 8253
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001517634914455483,
+      "loss": 0.1043,
+      "step": 8254
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015173458304916355,
+      "loss": 0.1632,
+      "step": 8255
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001517056745883441,
+      "loss": 0.0605,
+      "step": 8256
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015167676606416382,
+      "loss": 0.1307,
+      "step": 8257
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015164785747769663,
+      "loss": 0.094,
+      "step": 8258
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015161894883001636,
+      "loss": 0.2,
+      "step": 8259
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015159004012219688,
+      "loss": 0.0579,
+      "step": 8260
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015156113135531208,
+      "loss": 0.1467,
+      "step": 8261
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015153222253043582,
+      "loss": 0.0896,
+      "step": 8262
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.000151503313648642,
+      "loss": 0.0873,
+      "step": 8263
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015147440471100445,
+      "loss": 0.0515,
+      "step": 8264
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001514454957185971,
+      "loss": 0.0122,
+      "step": 8265
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015141658667249384,
+      "loss": 0.144,
+      "step": 8266
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015138767757376853,
+      "loss": 0.0558,
+      "step": 8267
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001513587684234951,
+      "loss": 0.1172,
+      "step": 8268
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015132985922274738,
+      "loss": 0.0703,
+      "step": 8269
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001513009499725993,
+      "loss": 0.0765,
+      "step": 8270
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015127204067412472,
+      "loss": 0.0882,
+      "step": 8271
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015124313132839755,
+      "loss": 0.1008,
+      "step": 8272
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015121422193649172,
+      "loss": 0.1013,
+      "step": 8273
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015118531249948112,
+      "loss": 0.1157,
+      "step": 8274
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001511564030184396,
+      "loss": 0.062,
+      "step": 8275
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015112749349444114,
+      "loss": 0.1408,
+      "step": 8276
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001510985839285596,
+      "loss": 0.0495,
+      "step": 8277
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015106967432186886,
+      "loss": 0.0815,
+      "step": 8278
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015104076467544288,
+      "loss": 0.1041,
+      "step": 8279
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001510118549903555,
+      "loss": 0.1836,
+      "step": 8280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015098294526768072,
+      "loss": 0.0927,
+      "step": 8281
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015095403550849238,
+      "loss": 0.025,
+      "step": 8282
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00015092512571386444,
+      "loss": 0.1914,
+      "step": 8283
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001508962158848708,
+      "loss": 0.067,
+      "step": 8284
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015086730602258534,
+      "loss": 0.1424,
+      "step": 8285
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.000150838396128082,
+      "loss": 0.0624,
+      "step": 8286
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015080948620243474,
+      "loss": 0.0942,
+      "step": 8287
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001507805762467174,
+      "loss": 0.1383,
+      "step": 8288
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015075166626200398,
+      "loss": 0.1226,
+      "step": 8289
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015072275624936836,
+      "loss": 0.103,
+      "step": 8290
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015069384620988447,
+      "loss": 0.0967,
+      "step": 8291
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015066493614462618,
+      "loss": 0.1042,
+      "step": 8292
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015063602605466748,
+      "loss": 0.0567,
+      "step": 8293
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001506071159410823,
+      "loss": 0.0922,
+      "step": 8294
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001505782058049445,
+      "loss": 0.0463,
+      "step": 8295
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001505492956473281,
+      "loss": 0.0777,
+      "step": 8296
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015052038546930696,
+      "loss": 0.1003,
+      "step": 8297
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.000150491475271955,
+      "loss": 0.1719,
+      "step": 8298
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015046256505634622,
+      "loss": 0.1232,
+      "step": 8299
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015043365482355449,
+      "loss": 0.0561,
+      "step": 8300
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015040474457465373,
+      "loss": 0.1662,
+      "step": 8301
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015037583431071793,
+      "loss": 0.1231,
+      "step": 8302
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015034692403282098,
+      "loss": 0.1159,
+      "step": 8303
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015031801374203678,
+      "loss": 0.1568,
+      "step": 8304
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015028910343943932,
+      "loss": 0.1186,
+      "step": 8305
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015026019312610253,
+      "loss": 0.0078,
+      "step": 8306
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015023128280310035,
+      "loss": 0.07,
+      "step": 8307
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015020237247150665,
+      "loss": 0.0689,
+      "step": 8308
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015017346213239548,
+      "loss": 0.1269,
+      "step": 8309
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015014455178684062,
+      "loss": 0.1231,
+      "step": 8310
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015011564143591608,
+      "loss": 0.0534,
+      "step": 8311
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015008673108069583,
+      "loss": 0.1056,
+      "step": 8312
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001500578207222538,
+      "loss": 0.1245,
+      "step": 8313
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015002891036166387,
+      "loss": 0.1042,
+      "step": 8314
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00015,
+      "loss": 0.1169,
+      "step": 8315
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001499710896383361,
+      "loss": 0.0292,
+      "step": 8316
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014994217927774618,
+      "loss": 0.0679,
+      "step": 8317
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014991326891930417,
+      "loss": 0.0643,
+      "step": 8318
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001498843585640839,
+      "loss": 0.0341,
+      "step": 8319
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014985544821315938,
+      "loss": 0.1444,
+      "step": 8320
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014982653786760453,
+      "loss": 0.0651,
+      "step": 8321
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001497976275284933,
+      "loss": 0.0684,
+      "step": 8322
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014976871719689968,
+      "loss": 0.0601,
+      "step": 8323
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014973980687389745,
+      "loss": 0.045,
+      "step": 8324
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014971089656056066,
+      "loss": 0.081,
+      "step": 8325
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001496819862579632,
+      "loss": 0.1203,
+      "step": 8326
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014965307596717902,
+      "loss": 0.1309,
+      "step": 8327
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001496241656892821,
+      "loss": 0.0465,
+      "step": 8328
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014959525542534625,
+      "loss": 0.076,
+      "step": 8329
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014956634517644552,
+      "loss": 0.0957,
+      "step": 8330
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014953743494365375,
+      "loss": 0.0582,
+      "step": 8331
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014950852472804497,
+      "loss": 0.1264,
+      "step": 8332
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014947961453069307,
+      "loss": 0.084,
+      "step": 8333
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014945070435267192,
+      "loss": 0.0734,
+      "step": 8334
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014942179419505548,
+      "loss": 0.1489,
+      "step": 8335
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014939288405891766,
+      "loss": 0.0717,
+      "step": 8336
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014936397394533246,
+      "loss": 0.0703,
+      "step": 8337
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014933506385537385,
+      "loss": 0.1133,
+      "step": 8338
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014930615379011555,
+      "loss": 0.0953,
+      "step": 8339
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014927724375063165,
+      "loss": 0.088,
+      "step": 8340
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014924833373799602,
+      "loss": 0.0734,
+      "step": 8341
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014921942375328257,
+      "loss": 0.1421,
+      "step": 8342
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014919051379756526,
+      "loss": 0.0444,
+      "step": 8343
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014916160387191797,
+      "loss": 0.076,
+      "step": 8344
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014913269397741463,
+      "loss": 0.0793,
+      "step": 8345
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014910378411512918,
+      "loss": 0.1192,
+      "step": 8346
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014907487428613553,
+      "loss": 0.1226,
+      "step": 8347
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014904596449150754,
+      "loss": 0.1075,
+      "step": 8348
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014901705473231928,
+      "loss": 0.134,
+      "step": 8349
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014898814500964447,
+      "loss": 0.1262,
+      "step": 8350
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014895923532455714,
+      "loss": 0.11,
+      "step": 8351
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014893032567813114,
+      "loss": 0.1284,
+      "step": 8352
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001489014160714404,
+      "loss": 0.0037,
+      "step": 8353
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014887250650555886,
+      "loss": 0.091,
+      "step": 8354
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014884359698156036,
+      "loss": 0.0528,
+      "step": 8355
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014881468750051885,
+      "loss": 0.0471,
+      "step": 8356
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014878577806350825,
+      "loss": 0.0944,
+      "step": 8357
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001487568686716024,
+      "loss": 0.0844,
+      "step": 8358
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014872795932587528,
+      "loss": 0.1344,
+      "step": 8359
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001486990500274007,
+      "loss": 0.1222,
+      "step": 8360
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014867014077725262,
+      "loss": 0.0751,
+      "step": 8361
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014864123157650488,
+      "loss": 0.0966,
+      "step": 8362
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001486123224262314,
+      "loss": 0.0696,
+      "step": 8363
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014858341332750616,
+      "loss": 0.0499,
+      "step": 8364
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001485545042814029,
+      "loss": 0.0237,
+      "step": 8365
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014852559528899555,
+      "loss": 0.0744,
+      "step": 8366
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014849668635135801,
+      "loss": 0.1715,
+      "step": 8367
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014846777746956415,
+      "loss": 0.0516,
+      "step": 8368
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014843886864468795,
+      "loss": 0.0918,
+      "step": 8369
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014840995987780312,
+      "loss": 0.1,
+      "step": 8370
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014838105116998364,
+      "loss": 0.1076,
+      "step": 8371
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014835214252230334,
+      "loss": 0.066,
+      "step": 8372
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014832323393583613,
+      "loss": 0.1649,
+      "step": 8373
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014829432541165591,
+      "loss": 0.1018,
+      "step": 8374
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014826541695083648,
+      "loss": 0.0334,
+      "step": 8375
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014823650855445166,
+      "loss": 0.0614,
+      "step": 8376
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014820760022357538,
+      "loss": 0.0888,
+      "step": 8377
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014817869195928154,
+      "loss": 0.041,
+      "step": 8378
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.000148149783762644,
+      "loss": 0.1663,
+      "step": 8379
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001481208756347365,
+      "loss": 0.1499,
+      "step": 8380
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014809196757663298,
+      "loss": 0.0835,
+      "step": 8381
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014806305958940726,
+      "loss": 0.1499,
+      "step": 8382
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014803415167413317,
+      "loss": 0.0293,
+      "step": 8383
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014800524383188463,
+      "loss": 0.0978,
+      "step": 8384
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001479763360637354,
+      "loss": 0.0728,
+      "step": 8385
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014794742837075935,
+      "loss": 0.0799,
+      "step": 8386
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001479185207540303,
+      "loss": 0.0355,
+      "step": 8387
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014788961321462215,
+      "loss": 0.1045,
+      "step": 8388
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014786070575360872,
+      "loss": 0.1185,
+      "step": 8389
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001478317983720637,
+      "loss": 0.1017,
+      "step": 8390
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014780289107106106,
+      "loss": 0.0903,
+      "step": 8391
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014777398385167455,
+      "loss": 0.0807,
+      "step": 8392
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014774507671497803,
+      "loss": 0.1017,
+      "step": 8393
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001477161696620453,
+      "loss": 0.047,
+      "step": 8394
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014768726269395014,
+      "loss": 0.1294,
+      "step": 8395
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014765835581176642,
+      "loss": 0.0653,
+      "step": 8396
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001476294490165679,
+      "loss": 0.0872,
+      "step": 8397
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014760054230942839,
+      "loss": 0.1053,
+      "step": 8398
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014757163569142177,
+      "loss": 0.1168,
+      "step": 8399
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014754272916362164,
+      "loss": 0.0513,
+      "step": 8400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000147513822727102,
+      "loss": 0.0893,
+      "step": 8401
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014748491638293652,
+      "loss": 0.0362,
+      "step": 8402
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014745601013219903,
+      "loss": 0.1532,
+      "step": 8403
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014742710397596336,
+      "loss": 0.0566,
+      "step": 8404
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014739819791530315,
+      "loss": 0.1383,
+      "step": 8405
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001473692919512923,
+      "loss": 0.1079,
+      "step": 8406
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014734038608500452,
+      "loss": 0.0434,
+      "step": 8407
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014731148031751359,
+      "loss": 0.0749,
+      "step": 8408
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001472825746498933,
+      "loss": 0.1084,
+      "step": 8409
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001472536690832174,
+      "loss": 0.1076,
+      "step": 8410
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014722476361855963,
+      "loss": 0.1428,
+      "step": 8411
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014719585825699374,
+      "loss": 0.1021,
+      "step": 8412
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001471669529995935,
+      "loss": 0.0936,
+      "step": 8413
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014713804784743264,
+      "loss": 0.1185,
+      "step": 8414
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000147109142801585,
+      "loss": 0.0665,
+      "step": 8415
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014708023786312412,
+      "loss": 0.0506,
+      "step": 8416
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014705133303312388,
+      "loss": 0.1065,
+      "step": 8417
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014702242831265795,
+      "loss": 0.0666,
+      "step": 8418
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014699352370280006,
+      "loss": 0.0958,
+      "step": 8419
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014696461920462401,
+      "loss": 0.1051,
+      "step": 8420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014693571481920342,
+      "loss": 0.1399,
+      "step": 8421
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014690681054761205,
+      "loss": 0.0912,
+      "step": 8422
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001468779063909236,
+      "loss": 0.0376,
+      "step": 8423
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014684900235021172,
+      "loss": 0.1286,
+      "step": 8424
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014682009842655023,
+      "loss": 0.0433,
+      "step": 8425
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014679119462101273,
+      "loss": 0.0043,
+      "step": 8426
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014676229093467293,
+      "loss": 0.0917,
+      "step": 8427
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014673338736860453,
+      "loss": 0.103,
+      "step": 8428
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014670448392388122,
+      "loss": 0.1009,
+      "step": 8429
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014667558060157672,
+      "loss": 0.0689,
+      "step": 8430
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014664667740276458,
+      "loss": 0.0839,
+      "step": 8431
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014661777432851855,
+      "loss": 0.1157,
+      "step": 8432
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001465888713799123,
+      "loss": 0.1487,
+      "step": 8433
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014655996855801948,
+      "loss": 0.146,
+      "step": 8434
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014653106586391375,
+      "loss": 0.1695,
+      "step": 8435
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001465021632986687,
+      "loss": 0.1355,
+      "step": 8436
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014647326086335806,
+      "loss": 0.04,
+      "step": 8437
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014644435855905544,
+      "loss": 0.1231,
+      "step": 8438
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014641545638683447,
+      "loss": 0.0344,
+      "step": 8439
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014638655434776884,
+      "loss": 0.0378,
+      "step": 8440
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.000146357652442932,
+      "loss": 0.0452,
+      "step": 8441
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014632875067339775,
+      "loss": 0.0396,
+      "step": 8442
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014629984904023963,
+      "loss": 0.0724,
+      "step": 8443
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014627094754453128,
+      "loss": 0.0923,
+      "step": 8444
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014624204618734636,
+      "loss": 0.0809,
+      "step": 8445
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014621314496975833,
+      "loss": 0.1297,
+      "step": 8446
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014618424389284087,
+      "loss": 0.16,
+      "step": 8447
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014615534295766755,
+      "loss": 0.0914,
+      "step": 8448
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014612644216531196,
+      "loss": 0.07,
+      "step": 8449
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014609754151684775,
+      "loss": 0.0735,
+      "step": 8450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014606864101334837,
+      "loss": 0.1784,
+      "step": 8451
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014603974065588744,
+      "loss": 0.0821,
+      "step": 8452
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014601084044553854,
+      "loss": 0.0655,
+      "step": 8453
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014598194038337522,
+      "loss": 0.0915,
+      "step": 8454
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014595304047047112,
+      "loss": 0.0638,
+      "step": 8455
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001459241407078996,
+      "loss": 0.0643,
+      "step": 8456
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014589524109673432,
+      "loss": 0.0579,
+      "step": 8457
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014586634163804877,
+      "loss": 0.1081,
+      "step": 8458
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014583744233291653,
+      "loss": 0.0967,
+      "step": 8459
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014580854318241114,
+      "loss": 0.0671,
+      "step": 8460
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014577964418760602,
+      "loss": 0.1986,
+      "step": 8461
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014575074534957478,
+      "loss": 0.1008,
+      "step": 8462
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014572184666939084,
+      "loss": 0.1066,
+      "step": 8463
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014569294814812777,
+      "loss": 0.0775,
+      "step": 8464
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014566404978685912,
+      "loss": 0.0706,
+      "step": 8465
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014563515158665824,
+      "loss": 0.0484,
+      "step": 8466
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014560625354859865,
+      "loss": 0.1111,
+      "step": 8467
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014557735567375388,
+      "loss": 0.0731,
+      "step": 8468
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001455484579631974,
+      "loss": 0.0715,
+      "step": 8469
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001455195604180027,
+      "loss": 0.0596,
+      "step": 8470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001454906630392431,
+      "loss": 0.0823,
+      "step": 8471
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001454617658279922,
+      "loss": 0.0892,
+      "step": 8472
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014543286878532336,
+      "loss": 0.1339,
+      "step": 8473
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014540397191231005,
+      "loss": 0.0853,
+      "step": 8474
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001453750752100257,
+      "loss": 0.1211,
+      "step": 8475
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001453461786795438,
+      "loss": 0.0921,
+      "step": 8476
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014531728232193764,
+      "loss": 0.1584,
+      "step": 8477
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014528838613828076,
+      "loss": 0.1118,
+      "step": 8478
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014525949012964646,
+      "loss": 0.0533,
+      "step": 8479
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014523059429710826,
+      "loss": 0.0913,
+      "step": 8480
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014520169864173953,
+      "loss": 0.1077,
+      "step": 8481
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014517280316461354,
+      "loss": 0.1001,
+      "step": 8482
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014514390786680376,
+      "loss": 0.0946,
+      "step": 8483
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001451150127493836,
+      "loss": 0.0766,
+      "step": 8484
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014508611781342639,
+      "loss": 0.0587,
+      "step": 8485
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014505722306000553,
+      "loss": 0.1209,
+      "step": 8486
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014502832849019428,
+      "loss": 0.1464,
+      "step": 8487
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014499943410506606,
+      "loss": 0.1262,
+      "step": 8488
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001449705399056942,
+      "loss": 0.0246,
+      "step": 8489
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014494164589315203,
+      "loss": 0.0943,
+      "step": 8490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001449127520685129,
+      "loss": 0.0994,
+      "step": 8491
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001448838584328501,
+      "loss": 0.0372,
+      "step": 8492
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014485496498723696,
+      "loss": 0.164,
+      "step": 8493
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014482607173274675,
+      "loss": 0.1862,
+      "step": 8494
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001447971786704528,
+      "loss": 0.1686,
+      "step": 8495
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014476828580142854,
+      "loss": 0.0633,
+      "step": 8496
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.000144739393126747,
+      "loss": 0.0825,
+      "step": 8497
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014471050064748158,
+      "loss": 0.0872,
+      "step": 8498
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014468160836470556,
+      "loss": 0.1393,
+      "step": 8499
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001446527162794922,
+      "loss": 0.0886,
+      "step": 8500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014462382439291477,
+      "loss": 0.0663,
+      "step": 8501
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014459493270604646,
+      "loss": 0.032,
+      "step": 8502
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014456604121996055,
+      "loss": 0.1549,
+      "step": 8503
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001445371499357303,
+      "loss": 0.1069,
+      "step": 8504
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001445082588544289,
+      "loss": 0.1584,
+      "step": 8505
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001444793679771296,
+      "loss": 0.0809,
+      "step": 8506
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014445047730490555,
+      "loss": 0.0563,
+      "step": 8507
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014442158683883002,
+      "loss": 0.0591,
+      "step": 8508
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014439269657997615,
+      "loss": 0.049,
+      "step": 8509
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014436380652941717,
+      "loss": 0.0701,
+      "step": 8510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014433491668822633,
+      "loss": 0.13,
+      "step": 8511
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014430602705747664,
+      "loss": 0.1144,
+      "step": 8512
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014427713763824138,
+      "loss": 0.0553,
+      "step": 8513
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014424824843159363,
+      "loss": 0.0645,
+      "step": 8514
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001442193594386066,
+      "loss": 0.1488,
+      "step": 8515
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014419047066035345,
+      "loss": 0.0647,
+      "step": 8516
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014416158209790725,
+      "loss": 0.0839,
+      "step": 8517
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014413269375234116,
+      "loss": 0.1176,
+      "step": 8518
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014410380562472828,
+      "loss": 0.1066,
+      "step": 8519
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014407491771614172,
+      "loss": 0.0981,
+      "step": 8520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014404603002765468,
+      "loss": 0.0762,
+      "step": 8521
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001440171425603401,
+      "loss": 0.0656,
+      "step": 8522
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001439882553152711,
+      "loss": 0.1321,
+      "step": 8523
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014395936829352077,
+      "loss": 0.2076,
+      "step": 8524
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014393048149616224,
+      "loss": 0.0576,
+      "step": 8525
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014390159492426858,
+      "loss": 0.1309,
+      "step": 8526
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014387270857891272,
+      "loss": 0.082,
+      "step": 8527
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014384382246116775,
+      "loss": 0.1294,
+      "step": 8528
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014381493657210674,
+      "loss": 0.0594,
+      "step": 8529
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001437860509128027,
+      "loss": 0.0862,
+      "step": 8530
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014375716548432869,
+      "loss": 0.08,
+      "step": 8531
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001437282802877576,
+      "loss": 0.0856,
+      "step": 8532
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014369939532416255,
+      "loss": 0.069,
+      "step": 8533
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014367051059461649,
+      "loss": 0.1308,
+      "step": 8534
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014364162610019238,
+      "loss": 0.1105,
+      "step": 8535
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001436127418419632,
+      "loss": 0.1186,
+      "step": 8536
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000143583857821002,
+      "loss": 0.0499,
+      "step": 8537
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014355497403838164,
+      "loss": 0.169,
+      "step": 8538
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014352609049517507,
+      "loss": 0.1864,
+      "step": 8539
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014349720719245525,
+      "loss": 0.0678,
+      "step": 8540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001434683241312951,
+      "loss": 0.1101,
+      "step": 8541
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001434394413127676,
+      "loss": 0.0673,
+      "step": 8542
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001434105587379456,
+      "loss": 0.0972,
+      "step": 8543
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.000143381676407902,
+      "loss": 0.0439,
+      "step": 8544
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014335279432370972,
+      "loss": 0.0896,
+      "step": 8545
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014332391248644163,
+      "loss": 0.0652,
+      "step": 8546
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014329503089717066,
+      "loss": 0.0556,
+      "step": 8547
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014326614955696959,
+      "loss": 0.0956,
+      "step": 8548
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014323726846691132,
+      "loss": 0.0508,
+      "step": 8549
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014320838762806867,
+      "loss": 0.0917,
+      "step": 8550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014317950704151455,
+      "loss": 0.0706,
+      "step": 8551
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014315062670832178,
+      "loss": 0.093,
+      "step": 8552
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014312174662956307,
+      "loss": 0.1491,
+      "step": 8553
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014309286680631132,
+      "loss": 0.081,
+      "step": 8554
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014306398723963932,
+      "loss": 0.1237,
+      "step": 8555
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014303510793061983,
+      "loss": 0.2036,
+      "step": 8556
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014300622888032568,
+      "loss": 0.0852,
+      "step": 8557
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014297735008982963,
+      "loss": 0.1092,
+      "step": 8558
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001429484715602044,
+      "loss": 0.1039,
+      "step": 8559
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014291959329252278,
+      "loss": 0.0687,
+      "step": 8560
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001428907152878575,
+      "loss": 0.0595,
+      "step": 8561
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014286183754728135,
+      "loss": 0.1074,
+      "step": 8562
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014283296007186697,
+      "loss": 0.0807,
+      "step": 8563
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001428040828626871,
+      "loss": 0.0828,
+      "step": 8564
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001427752059208144,
+      "loss": 0.0596,
+      "step": 8565
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014274632924732167,
+      "loss": 0.1098,
+      "step": 8566
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014271745284328159,
+      "loss": 0.1622,
+      "step": 8567
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001426885767097667,
+      "loss": 0.0832,
+      "step": 8568
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014265970084784976,
+      "loss": 0.0907,
+      "step": 8569
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001426308252586034,
+      "loss": 0.0762,
+      "step": 8570
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001426019499431003,
+      "loss": 0.0798,
+      "step": 8571
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014257307490241305,
+      "loss": 0.1212,
+      "step": 8572
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001425442001376143,
+      "loss": 0.1501,
+      "step": 8573
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014251532564977664,
+      "loss": 0.076,
+      "step": 8574
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014248645143997266,
+      "loss": 0.1296,
+      "step": 8575
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000142457577509275,
+      "loss": 0.0518,
+      "step": 8576
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014242870385875627,
+      "loss": 0.0848,
+      "step": 8577
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014239983048948892,
+      "loss": 0.1032,
+      "step": 8578
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001423709574025456,
+      "loss": 0.0785,
+      "step": 8579
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001423420845989988,
+      "loss": 0.0567,
+      "step": 8580
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014231321207992115,
+      "loss": 0.0708,
+      "step": 8581
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014228433984638515,
+      "loss": 0.0909,
+      "step": 8582
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014225546789946325,
+      "loss": 0.0762,
+      "step": 8583
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014222659624022802,
+      "loss": 0.1056,
+      "step": 8584
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014219772486975194,
+      "loss": 0.0766,
+      "step": 8585
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001421688537891075,
+      "loss": 0.0951,
+      "step": 8586
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001421399829993672,
+      "loss": 0.1261,
+      "step": 8587
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014211111250160346,
+      "loss": 0.1123,
+      "step": 8588
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014208224229688877,
+      "loss": 0.2047,
+      "step": 8589
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014205337238629552,
+      "loss": 0.0632,
+      "step": 8590
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001420245027708962,
+      "loss": 0.0927,
+      "step": 8591
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001419956334517633,
+      "loss": 0.0762,
+      "step": 8592
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014196676442996907,
+      "loss": 0.0512,
+      "step": 8593
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000141937895706586,
+      "loss": 0.1113,
+      "step": 8594
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014190902728268643,
+      "loss": 0.1743,
+      "step": 8595
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014188015915934279,
+      "loss": 0.1089,
+      "step": 8596
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014185129133762744,
+      "loss": 0.0997,
+      "step": 8597
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014182242381861272,
+      "loss": 0.0873,
+      "step": 8598
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014179355660337095,
+      "loss": 0.1237,
+      "step": 8599
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014176468969297448,
+      "loss": 0.069,
+      "step": 8600
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014173582308849565,
+      "loss": 0.0519,
+      "step": 8601
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001417069567910068,
+      "loss": 0.1292,
+      "step": 8602
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014167809080158018,
+      "loss": 0.107,
+      "step": 8603
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014164922512128805,
+      "loss": 0.1949,
+      "step": 8604
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014162035975120272,
+      "loss": 0.049,
+      "step": 8605
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014159149469239641,
+      "loss": 0.063,
+      "step": 8606
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014156262994594146,
+      "loss": 0.0349,
+      "step": 8607
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014153376551291012,
+      "loss": 0.0513,
+      "step": 8608
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014150490139437449,
+      "loss": 0.0346,
+      "step": 8609
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001414760375914069,
+      "loss": 0.0623,
+      "step": 8610
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014144717410507948,
+      "loss": 0.1023,
+      "step": 8611
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014141831093646446,
+      "loss": 0.1036,
+      "step": 8612
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014138944808663408,
+      "loss": 0.0996,
+      "step": 8613
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001413605855566604,
+      "loss": 0.0992,
+      "step": 8614
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014133172334761563,
+      "loss": 0.1273,
+      "step": 8615
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014130286146057193,
+      "loss": 0.0806,
+      "step": 8616
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014127399989660142,
+      "loss": 0.1346,
+      "step": 8617
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001412451386567763,
+      "loss": 0.0818,
+      "step": 8618
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014121627774216854,
+      "loss": 0.1061,
+      "step": 8619
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014118741715385028,
+      "loss": 0.1236,
+      "step": 8620
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014115855689289365,
+      "loss": 0.1001,
+      "step": 8621
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001411296969603707,
+      "loss": 0.0864,
+      "step": 8622
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014110083735735353,
+      "loss": 0.0698,
+      "step": 8623
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014107197808491413,
+      "loss": 0.1336,
+      "step": 8624
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014104311914412454,
+      "loss": 0.0596,
+      "step": 8625
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014101426053605683,
+      "loss": 0.1239,
+      "step": 8626
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014098540226178296,
+      "loss": 0.144,
+      "step": 8627
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014095654432237507,
+      "loss": 0.1035,
+      "step": 8628
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014092768671890491,
+      "loss": 0.095,
+      "step": 8629
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001408988294524446,
+      "loss": 0.1145,
+      "step": 8630
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014086997252406614,
+      "loss": 0.0512,
+      "step": 8631
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001408411159348414,
+      "loss": 0.1291,
+      "step": 8632
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014081225968584242,
+      "loss": 0.0475,
+      "step": 8633
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014078340377814095,
+      "loss": 0.0866,
+      "step": 8634
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014075454821280904,
+      "loss": 0.1023,
+      "step": 8635
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014072569299091852,
+      "loss": 0.086,
+      "step": 8636
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014069683811354135,
+      "loss": 0.086,
+      "step": 8637
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014066798358174936,
+      "loss": 0.0541,
+      "step": 8638
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001406391293966144,
+      "loss": 0.022,
+      "step": 8639
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014061027555920832,
+      "loss": 0.0607,
+      "step": 8640
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000140581422070603,
+      "loss": 0.108,
+      "step": 8641
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014055256893187018,
+      "loss": 0.102,
+      "step": 8642
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014052371614408182,
+      "loss": 0.1385,
+      "step": 8643
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014049486370830953,
+      "loss": 0.0885,
+      "step": 8644
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001404660116256252,
+      "loss": 0.1129,
+      "step": 8645
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014043715989710056,
+      "loss": 0.1318,
+      "step": 8646
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014040830852380737,
+      "loss": 0.0795,
+      "step": 8647
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014037945750681747,
+      "loss": 0.0575,
+      "step": 8648
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014035060684720248,
+      "loss": 0.0949,
+      "step": 8649
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014032175654603413,
+      "loss": 0.0662,
+      "step": 8650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014029290660438413,
+      "loss": 0.1075,
+      "step": 8651
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001402640570233242,
+      "loss": 0.0676,
+      "step": 8652
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014023520780392603,
+      "loss": 0.0866,
+      "step": 8653
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014020635894726122,
+      "loss": 0.2019,
+      "step": 8654
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014017751045440144,
+      "loss": 0.1103,
+      "step": 8655
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014014866232641836,
+      "loss": 0.1983,
+      "step": 8656
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001401198145643836,
+      "loss": 0.1379,
+      "step": 8657
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001400909671693688,
+      "loss": 0.0583,
+      "step": 8658
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014006212014244546,
+      "loss": 0.0824,
+      "step": 8659
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001400332734846852,
+      "loss": 0.0829,
+      "step": 8660
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014000442719715962,
+      "loss": 0.0623,
+      "step": 8661
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013997558128094024,
+      "loss": 0.0503,
+      "step": 8662
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013994673573709864,
+      "loss": 0.1187,
+      "step": 8663
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013991789056670633,
+      "loss": 0.0511,
+      "step": 8664
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001398890457708348,
+      "loss": 0.0347,
+      "step": 8665
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001398602013505556,
+      "loss": 0.1036,
+      "step": 8666
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013983135730694012,
+      "loss": 0.068,
+      "step": 8667
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013980251364105996,
+      "loss": 0.034,
+      "step": 8668
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013977367035398653,
+      "loss": 0.0915,
+      "step": 8669
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001397448274467912,
+      "loss": 0.0584,
+      "step": 8670
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013971598492054548,
+      "loss": 0.0877,
+      "step": 8671
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013968714277632076,
+      "loss": 0.0884,
+      "step": 8672
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001396583010151884,
+      "loss": 0.0954,
+      "step": 8673
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013962945963821996,
+      "loss": 0.1172,
+      "step": 8674
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001396006186464866,
+      "loss": 0.1066,
+      "step": 8675
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013957177804105976,
+      "loss": 0.0629,
+      "step": 8676
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00013954293782301078,
+      "loss": 0.1134,
+      "step": 8677
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000139514097993411,
+      "loss": 0.1081,
+      "step": 8678
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013948525855333175,
+      "loss": 0.1106,
+      "step": 8679
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013945641950384427,
+      "loss": 0.121,
+      "step": 8680
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013942758084601993,
+      "loss": 0.0829,
+      "step": 8681
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001393987425809299,
+      "loss": 0.1077,
+      "step": 8682
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013936990470964552,
+      "loss": 0.1235,
+      "step": 8683
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001393410672332381,
+      "loss": 0.1268,
+      "step": 8684
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013931223015277867,
+      "loss": 0.1155,
+      "step": 8685
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013928339346933857,
+      "loss": 0.0868,
+      "step": 8686
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013925455718398898,
+      "loss": 0.0553,
+      "step": 8687
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001392257212978011,
+      "loss": 0.0518,
+      "step": 8688
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013919688581184606,
+      "loss": 0.1158,
+      "step": 8689
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013916805072719502,
+      "loss": 0.0812,
+      "step": 8690
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013913921604491913,
+      "loss": 0.1643,
+      "step": 8691
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013911038176608954,
+      "loss": 0.1091,
+      "step": 8692
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001390815478917773,
+      "loss": 0.1511,
+      "step": 8693
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001390527144230536,
+      "loss": 0.0913,
+      "step": 8694
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001390238813609894,
+      "loss": 0.1316,
+      "step": 8695
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001389950487066558,
+      "loss": 0.1319,
+      "step": 8696
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013896621646112392,
+      "loss": 0.0922,
+      "step": 8697
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013893738462546473,
+      "loss": 0.0519,
+      "step": 8698
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001389085532007493,
+      "loss": 0.0569,
+      "step": 8699
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013887972218804855,
+      "loss": 0.0515,
+      "step": 8700
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001388508915884335,
+      "loss": 0.1241,
+      "step": 8701
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013882206140297513,
+      "loss": 0.1432,
+      "step": 8702
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001387932316327444,
+      "loss": 0.1665,
+      "step": 8703
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001387644022788123,
+      "loss": 0.0456,
+      "step": 8704
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013873557334224965,
+      "loss": 0.0846,
+      "step": 8705
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013870674482412745,
+      "loss": 0.1107,
+      "step": 8706
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013867791672551654,
+      "loss": 0.1202,
+      "step": 8707
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013864908904748783,
+      "loss": 0.0734,
+      "step": 8708
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013862026179111222,
+      "loss": 0.0811,
+      "step": 8709
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013859143495746046,
+      "loss": 0.0966,
+      "step": 8710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00013856260854760343,
+      "loss": 0.0773,
+      "step": 8711
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013853378256261194,
+      "loss": 0.072,
+      "step": 8712
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013850495700355685,
+      "loss": 0.1112,
+      "step": 8713
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013847613187150895,
+      "loss": 0.0509,
+      "step": 8714
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013844730716753887,
+      "loss": 0.0718,
+      "step": 8715
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013841848289271747,
+      "loss": 0.0266,
+      "step": 8716
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013838965904811547,
+      "loss": 0.1069,
+      "step": 8717
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013836083563480356,
+      "loss": 0.069,
+      "step": 8718
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013833201265385254,
+      "loss": 0.094,
+      "step": 8719
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013830319010633299,
+      "loss": 0.1732,
+      "step": 8720
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013827436799331561,
+      "loss": 0.0639,
+      "step": 8721
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001382455463158711,
+      "loss": 0.0447,
+      "step": 8722
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001382167250750701,
+      "loss": 0.0903,
+      "step": 8723
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013818790427198315,
+      "loss": 0.0238,
+      "step": 8724
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013815908390768102,
+      "loss": 0.1188,
+      "step": 8725
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001381302639832341,
+      "loss": 0.0436,
+      "step": 8726
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001381014444997131,
+      "loss": 0.0947,
+      "step": 8727
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001380726254581885,
+      "loss": 0.0659,
+      "step": 8728
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013804380685973096,
+      "loss": 0.1221,
+      "step": 8729
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013801498870541092,
+      "loss": 0.1542,
+      "step": 8730
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013798617099629887,
+      "loss": 0.1684,
+      "step": 8731
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013795735373346534,
+      "loss": 0.1558,
+      "step": 8732
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001379285369179808,
+      "loss": 0.1198,
+      "step": 8733
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013789972055091573,
+      "loss": 0.0298,
+      "step": 8734
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001378709046333406,
+      "loss": 0.0411,
+      "step": 8735
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013784208916632574,
+      "loss": 0.0441,
+      "step": 8736
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013781327415094162,
+      "loss": 0.1121,
+      "step": 8737
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013778445958825863,
+      "loss": 0.0497,
+      "step": 8738
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013775564547934715,
+      "loss": 0.1522,
+      "step": 8739
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013772683182527757,
+      "loss": 0.1005,
+      "step": 8740
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013769801862712017,
+      "loss": 0.0981,
+      "step": 8741
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001376692058859453,
+      "loss": 0.0442,
+      "step": 8742
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013764039360282323,
+      "loss": 0.1026,
+      "step": 8743
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00013761158177882433,
+      "loss": 0.1086,
+      "step": 8744
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013758277041501886,
+      "loss": 0.0983,
+      "step": 8745
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013755395951247702,
+      "loss": 0.0753,
+      "step": 8746
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001375251490722691,
+      "loss": 0.1317,
+      "step": 8747
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001374963390954653,
+      "loss": 0.1709,
+      "step": 8748
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013746752958313584,
+      "loss": 0.1247,
+      "step": 8749
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013743872053635097,
+      "loss": 0.0232,
+      "step": 8750
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013740991195618072,
+      "loss": 0.0303,
+      "step": 8751
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013738110384369528,
+      "loss": 0.0863,
+      "step": 8752
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013735229619996487,
+      "loss": 0.0697,
+      "step": 8753
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001373234890260596,
+      "loss": 0.115,
+      "step": 8754
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001372946823230496,
+      "loss": 0.1255,
+      "step": 8755
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001372658760920048,
+      "loss": 0.1008,
+      "step": 8756
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013723707033399534,
+      "loss": 0.0786,
+      "step": 8757
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013720826505009132,
+      "loss": 0.0701,
+      "step": 8758
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013717946024136276,
+      "loss": 0.0975,
+      "step": 8759
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013715065590887965,
+      "loss": 0.1668,
+      "step": 8760
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000137121852053712,
+      "loss": 0.0978,
+      "step": 8761
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013709304867692975,
+      "loss": 0.1195,
+      "step": 8762
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001370642457796029,
+      "loss": 0.0579,
+      "step": 8763
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001370354433628014,
+      "loss": 0.0911,
+      "step": 8764
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001370066414275952,
+      "loss": 0.2031,
+      "step": 8765
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013697783997505413,
+      "loss": 0.0895,
+      "step": 8766
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013694903900624808,
+      "loss": 0.1192,
+      "step": 8767
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000136920238522247,
+      "loss": 0.118,
+      "step": 8768
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013689143852412068,
+      "loss": 0.1029,
+      "step": 8769
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013686263901293902,
+      "loss": 0.1951,
+      "step": 8770
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013683383998977175,
+      "loss": 0.1271,
+      "step": 8771
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013680504145568872,
+      "loss": 0.0278,
+      "step": 8772
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001367762434117597,
+      "loss": 0.0466,
+      "step": 8773
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013674744585905448,
+      "loss": 0.0413,
+      "step": 8774
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013671864879864282,
+      "loss": 0.0744,
+      "step": 8775
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013668985223159436,
+      "loss": 0.1059,
+      "step": 8776
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00013666105615897886,
+      "loss": 0.0732,
+      "step": 8777
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000136632260581866,
+      "loss": 0.0728,
+      "step": 8778
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013660346550132548,
+      "loss": 0.139,
+      "step": 8779
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000136574670918427,
+      "loss": 0.1041,
+      "step": 8780
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013654587683424002,
+      "loss": 0.1379,
+      "step": 8781
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001365170832498343,
+      "loss": 0.0458,
+      "step": 8782
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001364882901662794,
+      "loss": 0.1695,
+      "step": 8783
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013645949758464486,
+      "loss": 0.072,
+      "step": 8784
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013643070550600035,
+      "loss": 0.1458,
+      "step": 8785
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013640191393141527,
+      "loss": 0.1615,
+      "step": 8786
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013637312286195926,
+      "loss": 0.1142,
+      "step": 8787
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013634433229870173,
+      "loss": 0.1466,
+      "step": 8788
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013631554224271224,
+      "loss": 0.086,
+      "step": 8789
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001362867526950602,
+      "loss": 0.0732,
+      "step": 8790
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013625796365681516,
+      "loss": 0.1235,
+      "step": 8791
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001362291751290464,
+      "loss": 0.0694,
+      "step": 8792
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013620038711282338,
+      "loss": 0.0416,
+      "step": 8793
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013617159960921556,
+      "loss": 0.0426,
+      "step": 8794
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013614281261929224,
+      "loss": 0.0646,
+      "step": 8795
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013611402614412287,
+      "loss": 0.1241,
+      "step": 8796
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013608524018477662,
+      "loss": 0.0734,
+      "step": 8797
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013605645474232294,
+      "loss": 0.0878,
+      "step": 8798
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013602766981783102,
+      "loss": 0.0834,
+      "step": 8799
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013599888541237023,
+      "loss": 0.1011,
+      "step": 8800
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013597010152700982,
+      "loss": 0.1422,
+      "step": 8801
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013594131816281894,
+      "loss": 0.1338,
+      "step": 8802
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001359125353208669,
+      "loss": 0.1294,
+      "step": 8803
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013588375300222283,
+      "loss": 0.0965,
+      "step": 8804
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013585497120795594,
+      "loss": 0.1193,
+      "step": 8805
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001358261899391355,
+      "loss": 0.0786,
+      "step": 8806
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013579740919683042,
+      "loss": 0.159,
+      "step": 8807
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013576862898210994,
+      "loss": 0.1764,
+      "step": 8808
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001357398492960432,
+      "loss": 0.0695,
+      "step": 8809
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00013571107013969923,
+      "loss": 0.0982,
+      "step": 8810
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001356822915141471,
+      "loss": 0.0991,
+      "step": 8811
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013565351342045587,
+      "loss": 0.1356,
+      "step": 8812
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001356247358596945,
+      "loss": 0.1954,
+      "step": 8813
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013559595883293207,
+      "loss": 0.0503,
+      "step": 8814
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013556718234123753,
+      "loss": 0.0918,
+      "step": 8815
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013553840638567987,
+      "loss": 0.0628,
+      "step": 8816
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013550963096732797,
+      "loss": 0.0855,
+      "step": 8817
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013548085608725078,
+      "loss": 0.1181,
+      "step": 8818
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001354520817465172,
+      "loss": 0.1648,
+      "step": 8819
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013542330794619615,
+      "loss": 0.1647,
+      "step": 8820
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001353945346873565,
+      "loss": 0.1322,
+      "step": 8821
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013536576197106703,
+      "loss": 0.1305,
+      "step": 8822
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013533698979839654,
+      "loss": 0.1374,
+      "step": 8823
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013530821817041392,
+      "loss": 0.1566,
+      "step": 8824
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001352794470881879,
+      "loss": 0.0641,
+      "step": 8825
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013525067655278727,
+      "loss": 0.0996,
+      "step": 8826
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013522190656528073,
+      "loss": 0.1822,
+      "step": 8827
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013519313712673703,
+      "loss": 0.0881,
+      "step": 8828
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013516436823822486,
+      "loss": 0.1121,
+      "step": 8829
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013513559990081293,
+      "loss": 0.0574,
+      "step": 8830
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001351068321155699,
+      "loss": 0.0556,
+      "step": 8831
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013507806488356435,
+      "loss": 0.1771,
+      "step": 8832
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001350492982058649,
+      "loss": 0.093,
+      "step": 8833
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001350205320835402,
+      "loss": 0.0983,
+      "step": 8834
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013499176651765875,
+      "loss": 0.1663,
+      "step": 8835
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001349630015092893,
+      "loss": 0.1109,
+      "step": 8836
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013493423705950014,
+      "loss": 0.1119,
+      "step": 8837
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013490547316935995,
+      "loss": 0.0669,
+      "step": 8838
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013487670983993713,
+      "loss": 0.2115,
+      "step": 8839
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001348479470723002,
+      "loss": 0.0216,
+      "step": 8840
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013481918486751761,
+      "loss": 0.046,
+      "step": 8841
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00013479042322665777,
+      "loss": 0.0314,
+      "step": 8842
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001347616621507891,
+      "loss": 0.0871,
+      "step": 8843
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013473290164098001,
+      "loss": 0.1,
+      "step": 8844
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013470414169829885,
+      "loss": 0.0229,
+      "step": 8845
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013467538232381403,
+      "loss": 0.1427,
+      "step": 8846
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013464662351859376,
+      "loss": 0.0487,
+      "step": 8847
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001346178652837064,
+      "loss": 0.1492,
+      "step": 8848
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013458910762022023,
+      "loss": 0.1184,
+      "step": 8849
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013456035052920352,
+      "loss": 0.1045,
+      "step": 8850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001345315940117245,
+      "loss": 0.1458,
+      "step": 8851
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013450283806885145,
+      "loss": 0.1291,
+      "step": 8852
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001344740827016525,
+      "loss": 0.1213,
+      "step": 8853
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013444532791119585,
+      "loss": 0.1646,
+      "step": 8854
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013441657369854963,
+      "loss": 0.0959,
+      "step": 8855
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000134387820064782,
+      "loss": 0.1112,
+      "step": 8856
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013435906701096112,
+      "loss": 0.0133,
+      "step": 8857
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000134330314538155,
+      "loss": 0.1113,
+      "step": 8858
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013430156264743174,
+      "loss": 0.0676,
+      "step": 8859
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001342728113398594,
+      "loss": 0.0277,
+      "step": 8860
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013424406061650598,
+      "loss": 0.0512,
+      "step": 8861
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001342153104784396,
+      "loss": 0.0585,
+      "step": 8862
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013418656092672806,
+      "loss": 0.0453,
+      "step": 8863
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001341578119624394,
+      "loss": 0.0569,
+      "step": 8864
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013412906358664157,
+      "loss": 0.0841,
+      "step": 8865
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013410031580040252,
+      "loss": 0.0676,
+      "step": 8866
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001340715686047901,
+      "loss": 0.0834,
+      "step": 8867
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001340428220008722,
+      "loss": 0.0928,
+      "step": 8868
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013401407598971667,
+      "loss": 0.0867,
+      "step": 8869
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001339853305723913,
+      "loss": 0.0731,
+      "step": 8870
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013395658574996396,
+      "loss": 0.0975,
+      "step": 8871
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001339278415235025,
+      "loss": 0.128,
+      "step": 8872
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001338990978940745,
+      "loss": 0.0154,
+      "step": 8873
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001338703548627478,
+      "loss": 0.0936,
+      "step": 8874
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00013384161243059012,
+      "loss": 0.1169,
+      "step": 8875
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013381287059866914,
+      "loss": 0.0371,
+      "step": 8876
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001337841293680526,
+      "loss": 0.0755,
+      "step": 8877
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013375538873980808,
+      "loss": 0.1221,
+      "step": 8878
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013372664871500324,
+      "loss": 0.1613,
+      "step": 8879
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013369790929470567,
+      "loss": 0.0546,
+      "step": 8880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013366917047998294,
+      "loss": 0.1254,
+      "step": 8881
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013364043227190268,
+      "loss": 0.0752,
+      "step": 8882
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013361169467153235,
+      "loss": 0.048,
+      "step": 8883
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013358295767993954,
+      "loss": 0.0639,
+      "step": 8884
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013355422129819166,
+      "loss": 0.076,
+      "step": 8885
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013352548552735625,
+      "loss": 0.0563,
+      "step": 8886
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013349675036850083,
+      "loss": 0.1722,
+      "step": 8887
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013346801582269263,
+      "loss": 0.1626,
+      "step": 8888
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001334392818909992,
+      "loss": 0.1145,
+      "step": 8889
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013341054857448788,
+      "loss": 0.0842,
+      "step": 8890
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000133381815874226,
+      "loss": 0.0435,
+      "step": 8891
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000133353083791281,
+      "loss": 0.0782,
+      "step": 8892
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013332435232672006,
+      "loss": 0.1359,
+      "step": 8893
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013329562148161055,
+      "loss": 0.1474,
+      "step": 8894
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001332668912570197,
+      "loss": 0.1108,
+      "step": 8895
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013323816165401476,
+      "loss": 0.0358,
+      "step": 8896
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013320943267366306,
+      "loss": 0.1197,
+      "step": 8897
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013318070431703159,
+      "loss": 0.0523,
+      "step": 8898
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013315197658518766,
+      "loss": 0.0862,
+      "step": 8899
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001331232494791984,
+      "loss": 0.0917,
+      "step": 8900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001330945230001309,
+      "loss": 0.0729,
+      "step": 8901
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001330657971490524,
+      "loss": 0.1205,
+      "step": 8902
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013303707192702978,
+      "loss": 0.0357,
+      "step": 8903
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013300834733513022,
+      "loss": 0.1376,
+      "step": 8904
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013297962337442072,
+      "loss": 0.081,
+      "step": 8905
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001329509000459683,
+      "loss": 0.0917,
+      "step": 8906
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013292217735083999,
+      "loss": 0.1373,
+      "step": 8907
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00013289345529010263,
+      "loss": 0.1349,
+      "step": 8908
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001328647338648233,
+      "loss": 0.1087,
+      "step": 8909
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013283601307606882,
+      "loss": 0.1133,
+      "step": 8910
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013280729292490615,
+      "loss": 0.084,
+      "step": 8911
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013277857341240212,
+      "loss": 0.0928,
+      "step": 8912
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013274985453962367,
+      "loss": 0.0696,
+      "step": 8913
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013272113630763744,
+      "loss": 0.1525,
+      "step": 8914
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013269241871751035,
+      "loss": 0.0225,
+      "step": 8915
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013266370177030914,
+      "loss": 0.0264,
+      "step": 8916
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013263498546710055,
+      "loss": 0.0839,
+      "step": 8917
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001326062698089514,
+      "loss": 0.099,
+      "step": 8918
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013257755479692828,
+      "loss": 0.0428,
+      "step": 8919
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013254884043209788,
+      "loss": 0.1288,
+      "step": 8920
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001325201267155269,
+      "loss": 0.1366,
+      "step": 8921
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013249141364828196,
+      "loss": 0.0823,
+      "step": 8922
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001324627012314297,
+      "loss": 0.2425,
+      "step": 8923
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001324339894660366,
+      "loss": 0.0533,
+      "step": 8924
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013240527835316928,
+      "loss": 0.118,
+      "step": 8925
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013237656789389428,
+      "loss": 0.1303,
+      "step": 8926
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013234785808927813,
+      "loss": 0.0483,
+      "step": 8927
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013231914894038733,
+      "loss": 0.1315,
+      "step": 8928
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013229044044828824,
+      "loss": 0.1284,
+      "step": 8929
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013226173261404734,
+      "loss": 0.1373,
+      "step": 8930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013223302543873108,
+      "loss": 0.0269,
+      "step": 8931
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001322043189234058,
+      "loss": 0.1052,
+      "step": 8932
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013217561306913794,
+      "loss": 0.124,
+      "step": 8933
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013214690787699372,
+      "loss": 0.0383,
+      "step": 8934
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013211820334803955,
+      "loss": 0.1445,
+      "step": 8935
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013208949948334168,
+      "loss": 0.118,
+      "step": 8936
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013206079628396637,
+      "loss": 0.1363,
+      "step": 8937
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013203209375097993,
+      "loss": 0.1442,
+      "step": 8938
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001320033918854485,
+      "loss": 0.1118,
+      "step": 8939
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001319746906884382,
+      "loss": 0.0137,
+      "step": 8940
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00013194599016101534,
+      "loss": 0.1521,
+      "step": 8941
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.000131917290304246,
+      "loss": 0.1302,
+      "step": 8942
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013188859111919635,
+      "loss": 0.0185,
+      "step": 8943
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001318598926069324,
+      "loss": 0.1001,
+      "step": 8944
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013183119476852022,
+      "loss": 0.1153,
+      "step": 8945
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013180249760502588,
+      "loss": 0.1083,
+      "step": 8946
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013177380111751538,
+      "loss": 0.1666,
+      "step": 8947
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013174510530705475,
+      "loss": 0.0459,
+      "step": 8948
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001317164101747099,
+      "loss": 0.1576,
+      "step": 8949
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001316877157215468,
+      "loss": 0.0802,
+      "step": 8950
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013165902194863137,
+      "loss": 0.0622,
+      "step": 8951
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013163032885702948,
+      "loss": 0.1359,
+      "step": 8952
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013160163644780707,
+      "loss": 0.0752,
+      "step": 8953
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013157294472202985,
+      "loss": 0.0786,
+      "step": 8954
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001315442536807637,
+      "loss": 0.0434,
+      "step": 8955
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013151556332507438,
+      "loss": 0.0863,
+      "step": 8956
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001314868736560277,
+      "loss": 0.0365,
+      "step": 8957
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001314581846746894,
+      "loss": 0.1045,
+      "step": 8958
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013142949638212512,
+      "loss": 0.1128,
+      "step": 8959
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013140080877940063,
+      "loss": 0.0513,
+      "step": 8960
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013137212186758153,
+      "loss": 0.1034,
+      "step": 8961
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013134343564773346,
+      "loss": 0.0544,
+      "step": 8962
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013131475012092208,
+      "loss": 0.0991,
+      "step": 8963
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013128606528821292,
+      "loss": 0.0899,
+      "step": 8964
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013125738115067157,
+      "loss": 0.0622,
+      "step": 8965
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013122869770936353,
+      "loss": 0.1145,
+      "step": 8966
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013120001496535433,
+      "loss": 0.0868,
+      "step": 8967
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001311713329197095,
+      "loss": 0.0962,
+      "step": 8968
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013114265157349438,
+      "loss": 0.1802,
+      "step": 8969
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013111397092777446,
+      "loss": 0.0221,
+      "step": 8970
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013108529098361516,
+      "loss": 0.1131,
+      "step": 8971
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001310566117420818,
+      "loss": 0.0279,
+      "step": 8972
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00013102793320423977,
+      "loss": 0.1214,
+      "step": 8973
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001309992553711544,
+      "loss": 0.0669,
+      "step": 8974
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013097057824389095,
+      "loss": 0.0854,
+      "step": 8975
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013094190182351472,
+      "loss": 0.0288,
+      "step": 8976
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013091322611109098,
+      "loss": 0.1353,
+      "step": 8977
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013088455110768488,
+      "loss": 0.1041,
+      "step": 8978
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001308558768143617,
+      "loss": 0.0686,
+      "step": 8979
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013082720323218653,
+      "loss": 0.166,
+      "step": 8980
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001307985303622245,
+      "loss": 0.1382,
+      "step": 8981
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001307698582055408,
+      "loss": 0.1302,
+      "step": 8982
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013074118676320045,
+      "loss": 0.0982,
+      "step": 8983
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001307125160362686,
+      "loss": 0.1404,
+      "step": 8984
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013068384602581018,
+      "loss": 0.0557,
+      "step": 8985
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001306551767328902,
+      "loss": 0.2005,
+      "step": 8986
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001306265081585737,
+      "loss": 0.0987,
+      "step": 8987
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013059784030392563,
+      "loss": 0.1235,
+      "step": 8988
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013056917317001089,
+      "loss": 0.0538,
+      "step": 8989
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001305405067578944,
+      "loss": 0.0283,
+      "step": 8990
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013051184106864096,
+      "loss": 0.1161,
+      "step": 8991
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001304831761033155,
+      "loss": 0.1387,
+      "step": 8992
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013045451186298283,
+      "loss": 0.0916,
+      "step": 8993
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013042584834870778,
+      "loss": 0.122,
+      "step": 8994
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000130397185561555,
+      "loss": 0.1096,
+      "step": 8995
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001303685235025893,
+      "loss": 0.0695,
+      "step": 8996
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013033986217287537,
+      "loss": 0.1228,
+      "step": 8997
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013031120157347791,
+      "loss": 0.0611,
+      "step": 8998
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001302825417054616,
+      "loss": 0.0883,
+      "step": 8999
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013025388256989102,
+      "loss": 0.1008,
+      "step": 9000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001302252241678308,
+      "loss": 0.0761,
+      "step": 9001
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001301965665003455,
+      "loss": 0.0431,
+      "step": 9002
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013016790956849972,
+      "loss": 0.0685,
+      "step": 9003
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013013925337335795,
+      "loss": 0.0458,
+      "step": 9004
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013011059791598464,
+      "loss": 0.1462,
+      "step": 9005
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001300819431974443,
+      "loss": 0.1006,
+      "step": 9006
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00013005328921880134,
+      "loss": 0.0842,
+      "step": 9007
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001300246359811202,
+      "loss": 0.1054,
+      "step": 9008
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012999598348546534,
+      "loss": 0.0933,
+      "step": 9009
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012996733173290095,
+      "loss": 0.0643,
+      "step": 9010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012993868072449147,
+      "loss": 0.0596,
+      "step": 9011
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012991003046130113,
+      "loss": 0.0992,
+      "step": 9012
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001298813809443943,
+      "loss": 0.1502,
+      "step": 9013
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012985273217483513,
+      "loss": 0.0618,
+      "step": 9014
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012982408415368788,
+      "loss": 0.1753,
+      "step": 9015
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012979543688201674,
+      "loss": 0.0529,
+      "step": 9016
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001297667903608859,
+      "loss": 0.0938,
+      "step": 9017
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001297381445913594,
+      "loss": 0.1322,
+      "step": 9018
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012970949957450152,
+      "loss": 0.0705,
+      "step": 9019
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012968085531137616,
+      "loss": 0.0829,
+      "step": 9020
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012965221180304744,
+      "loss": 0.1663,
+      "step": 9021
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012962356905057935,
+      "loss": 0.1355,
+      "step": 9022
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012959492705503595,
+      "loss": 0.092,
+      "step": 9023
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012956628581748124,
+      "loss": 0.0482,
+      "step": 9024
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000129537645338979,
+      "loss": 0.0914,
+      "step": 9025
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012950900562059325,
+      "loss": 0.1156,
+      "step": 9026
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012948036666338784,
+      "loss": 0.0809,
+      "step": 9027
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012945172846842663,
+      "loss": 0.0661,
+      "step": 9028
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001294230910367735,
+      "loss": 0.0689,
+      "step": 9029
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001293944543694922,
+      "loss": 0.0713,
+      "step": 9030
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012936581846764643,
+      "loss": 0.078,
+      "step": 9031
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001293371833323,
+      "loss": 0.1209,
+      "step": 9032
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012930854896451666,
+      "loss": 0.0824,
+      "step": 9033
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012927991536536008,
+      "loss": 0.1311,
+      "step": 9034
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012925128253589384,
+      "loss": 0.1125,
+      "step": 9035
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001292226504771816,
+      "loss": 0.0886,
+      "step": 9036
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012919401919028696,
+      "loss": 0.1155,
+      "step": 9037
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001291653886762735,
+      "loss": 0.1374,
+      "step": 9038
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012913675893620475,
+      "loss": 0.0695,
+      "step": 9039
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00012910812997114426,
+      "loss": 0.0653,
+      "step": 9040
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012907950178215545,
+      "loss": 0.1138,
+      "step": 9041
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001290508743703018,
+      "loss": 0.1053,
+      "step": 9042
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012902224773664672,
+      "loss": 0.0996,
+      "step": 9043
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001289936218822536,
+      "loss": 0.0754,
+      "step": 9044
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001289649968081859,
+      "loss": 0.0137,
+      "step": 9045
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012893637251550684,
+      "loss": 0.1353,
+      "step": 9046
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012890774900527977,
+      "loss": 0.1149,
+      "step": 9047
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.000128879126278568,
+      "loss": 0.058,
+      "step": 9048
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001288505043364347,
+      "loss": 0.0818,
+      "step": 9049
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012882188317994323,
+      "loss": 0.1022,
+      "step": 9050
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012879326281015662,
+      "loss": 0.1265,
+      "step": 9051
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012876464322813814,
+      "loss": 0.0592,
+      "step": 9052
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012873602443495087,
+      "loss": 0.094,
+      "step": 9053
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012870740643165795,
+      "loss": 0.0961,
+      "step": 9054
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012867878921932245,
+      "loss": 0.0617,
+      "step": 9055
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012865017279900737,
+      "loss": 0.068,
+      "step": 9056
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001286215571717758,
+      "loss": 0.0765,
+      "step": 9057
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012859294233869069,
+      "loss": 0.2714,
+      "step": 9058
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012856432830081496,
+      "loss": 0.1024,
+      "step": 9059
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012853571505921165,
+      "loss": 0.1242,
+      "step": 9060
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012850710261494351,
+      "loss": 0.1235,
+      "step": 9061
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001284784909690735,
+      "loss": 0.1811,
+      "step": 9062
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001284498801226644,
+      "loss": 0.1707,
+      "step": 9063
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001284212700767791,
+      "loss": 0.1073,
+      "step": 9064
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012839266083248038,
+      "loss": 0.0968,
+      "step": 9065
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012836405239083088,
+      "loss": 0.1104,
+      "step": 9066
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001283354447528934,
+      "loss": 0.1347,
+      "step": 9067
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012830683791973063,
+      "loss": 0.091,
+      "step": 9068
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001282782318924052,
+      "loss": 0.0921,
+      "step": 9069
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012824962667197978,
+      "loss": 0.1061,
+      "step": 9070
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012822102225951693,
+      "loss": 0.0007,
+      "step": 9071
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012819241865607924,
+      "loss": 0.0866,
+      "step": 9072
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00012816381586272925,
+      "loss": 0.0566,
+      "step": 9073
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012813521388052945,
+      "loss": 0.1003,
+      "step": 9074
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012810661271054241,
+      "loss": 0.0948,
+      "step": 9075
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012807801235383044,
+      "loss": 0.1211,
+      "step": 9076
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012804941281145603,
+      "loss": 0.0671,
+      "step": 9077
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012802081408448156,
+      "loss": 0.1514,
+      "step": 9078
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012799221617396938,
+      "loss": 0.1942,
+      "step": 9079
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012796361908098188,
+      "loss": 0.1106,
+      "step": 9080
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012793502280658128,
+      "loss": 0.0663,
+      "step": 9081
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012790642735182987,
+      "loss": 0.0854,
+      "step": 9082
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001278778327177899,
+      "loss": 0.0929,
+      "step": 9083
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012784923890552357,
+      "loss": 0.0329,
+      "step": 9084
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012782064591609313,
+      "loss": 0.0959,
+      "step": 9085
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012779205375056056,
+      "loss": 0.0795,
+      "step": 9086
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012776346240998807,
+      "loss": 0.0926,
+      "step": 9087
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012773487189543777,
+      "loss": 0.098,
+      "step": 9088
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012770628220797167,
+      "loss": 0.1088,
+      "step": 9089
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012767769334865188,
+      "loss": 0.124,
+      "step": 9090
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012764910531854026,
+      "loss": 0.0483,
+      "step": 9091
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012762051811869885,
+      "loss": 0.0878,
+      "step": 9092
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012759193175018954,
+      "loss": 0.0634,
+      "step": 9093
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012756334621407428,
+      "loss": 0.0515,
+      "step": 9094
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012753476151141494,
+      "loss": 0.1976,
+      "step": 9095
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001275061776432733,
+      "loss": 0.1118,
+      "step": 9096
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012747759461071117,
+      "loss": 0.1185,
+      "step": 9097
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012744901241479035,
+      "loss": 0.0414,
+      "step": 9098
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012742043105657262,
+      "loss": 0.1549,
+      "step": 9099
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012739185053711968,
+      "loss": 0.0766,
+      "step": 9100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012736327085749322,
+      "loss": 0.0688,
+      "step": 9101
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012733469201875482,
+      "loss": 0.0798,
+      "step": 9102
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012730611402196612,
+      "loss": 0.085,
+      "step": 9103
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012727753686818876,
+      "loss": 0.1092,
+      "step": 9104
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012724896055848423,
+      "loss": 0.0885,
+      "step": 9105
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00012722038509391421,
+      "loss": 0.1103,
+      "step": 9106
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012719181047554,
+      "loss": 0.0764,
+      "step": 9107
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012716323670442317,
+      "loss": 0.1118,
+      "step": 9108
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001271346637816251,
+      "loss": 0.0628,
+      "step": 9109
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012710609170820726,
+      "loss": 0.0935,
+      "step": 9110
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012707752048523097,
+      "loss": 0.097,
+      "step": 9111
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012704895011375755,
+      "loss": 0.1089,
+      "step": 9112
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012702038059484832,
+      "loss": 0.078,
+      "step": 9113
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001269918119295646,
+      "loss": 0.0155,
+      "step": 9114
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012696324411896756,
+      "loss": 0.0428,
+      "step": 9115
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012693467716411852,
+      "loss": 0.0855,
+      "step": 9116
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012690611106607852,
+      "loss": 0.0983,
+      "step": 9117
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012687754582590876,
+      "loss": 0.1009,
+      "step": 9118
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001268489814446704,
+      "loss": 0.0737,
+      "step": 9119
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012682041792342447,
+      "loss": 0.1177,
+      "step": 9120
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012679185526323207,
+      "loss": 0.1459,
+      "step": 9121
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012676329346515417,
+      "loss": 0.0847,
+      "step": 9122
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012673473253025177,
+      "loss": 0.086,
+      "step": 9123
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001267061724595858,
+      "loss": 0.0875,
+      "step": 9124
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012667761325421727,
+      "loss": 0.123,
+      "step": 9125
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012664905491520707,
+      "loss": 0.0471,
+      "step": 9126
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012662049744361587,
+      "loss": 0.1202,
+      "step": 9127
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001265919408405047,
+      "loss": 0.1701,
+      "step": 9128
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012656338510693424,
+      "loss": 0.1296,
+      "step": 9129
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012653483024396533,
+      "loss": 0.0167,
+      "step": 9130
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012650627625265874,
+      "loss": 0.1528,
+      "step": 9131
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.000126477723134075,
+      "loss": 0.1134,
+      "step": 9132
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012644917088927487,
+      "loss": 0.0888,
+      "step": 9133
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.000126420619519319,
+      "loss": 0.1079,
+      "step": 9134
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012639206902526794,
+      "loss": 0.0539,
+      "step": 9135
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012636351940818234,
+      "loss": 0.0812,
+      "step": 9136
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012633497066912266,
+      "loss": 0.074,
+      "step": 9137
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001263064228091494,
+      "loss": 0.098,
+      "step": 9138
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00012627787582932304,
+      "loss": 0.1323,
+      "step": 9139
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012624932973070405,
+      "loss": 0.1001,
+      "step": 9140
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001262207845143529,
+      "loss": 0.0637,
+      "step": 9141
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012619224018132975,
+      "loss": 0.0412,
+      "step": 9142
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001261636967326951,
+      "loss": 0.2038,
+      "step": 9143
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001261351541695092,
+      "loss": 0.0345,
+      "step": 9144
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012610661249283237,
+      "loss": 0.1726,
+      "step": 9145
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012607807170372483,
+      "loss": 0.0325,
+      "step": 9146
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012604953180324674,
+      "loss": 0.1023,
+      "step": 9147
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001260209927924583,
+      "loss": 0.1066,
+      "step": 9148
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001259924546724197,
+      "loss": 0.087,
+      "step": 9149
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012596391744419097,
+      "loss": 0.1532,
+      "step": 9150
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012593538110883227,
+      "loss": 0.0787,
+      "step": 9151
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012590684566740355,
+      "loss": 0.0595,
+      "step": 9152
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012587831112096488,
+      "loss": 0.0134,
+      "step": 9153
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001258497774705762,
+      "loss": 0.1401,
+      "step": 9154
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012582124471729748,
+      "loss": 0.1362,
+      "step": 9155
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012579271286218866,
+      "loss": 0.0425,
+      "step": 9156
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001257641819063095,
+      "loss": 0.0256,
+      "step": 9157
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012573565185071994,
+      "loss": 0.0431,
+      "step": 9158
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012570712269647973,
+      "loss": 0.091,
+      "step": 9159
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001256785944446487,
+      "loss": 0.0656,
+      "step": 9160
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012565006709628658,
+      "loss": 0.062,
+      "step": 9161
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012562154065245303,
+      "loss": 0.1101,
+      "step": 9162
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012559301511420778,
+      "loss": 0.0963,
+      "step": 9163
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001255644904826104,
+      "loss": 0.0109,
+      "step": 9164
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012553596675872057,
+      "loss": 0.1028,
+      "step": 9165
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012550744394359786,
+      "loss": 0.0651,
+      "step": 9166
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001254789220383018,
+      "loss": 0.1163,
+      "step": 9167
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001254504010438918,
+      "loss": 0.1105,
+      "step": 9168
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012542188096142744,
+      "loss": 0.0747,
+      "step": 9169
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012539336179196814,
+      "loss": 0.1377,
+      "step": 9170
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001253648435365733,
+      "loss": 0.1097,
+      "step": 9171
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00012533632619630232,
+      "loss": 0.0764,
+      "step": 9172
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012530780977221444,
+      "loss": 0.0362,
+      "step": 9173
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012527929426536902,
+      "loss": 0.0676,
+      "step": 9174
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001252507796768253,
+      "loss": 0.0629,
+      "step": 9175
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001252222660076426,
+      "loss": 0.1121,
+      "step": 9176
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012519375325888003,
+      "loss": 0.1427,
+      "step": 9177
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012516524143159677,
+      "loss": 0.0701,
+      "step": 9178
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012513673052685196,
+      "loss": 0.1176,
+      "step": 9179
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012510822054570471,
+      "loss": 0.1029,
+      "step": 9180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012507971148921406,
+      "loss": 0.141,
+      "step": 9181
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012505120335843913,
+      "loss": 0.1183,
+      "step": 9182
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012502269615443878,
+      "loss": 0.1246,
+      "step": 9183
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000124994189878272,
+      "loss": 0.1172,
+      "step": 9184
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012496568453099774,
+      "loss": 0.0808,
+      "step": 9185
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012493718011367488,
+      "loss": 0.0718,
+      "step": 9186
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001249086766273623,
+      "loss": 0.0888,
+      "step": 9187
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001248801740731188,
+      "loss": 0.1393,
+      "step": 9188
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001248516724520032,
+      "loss": 0.0677,
+      "step": 9189
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012482317176507418,
+      "loss": 0.0462,
+      "step": 9190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001247946720133905,
+      "loss": 0.118,
+      "step": 9191
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012476617319801087,
+      "loss": 0.0935,
+      "step": 9192
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001247376753199939,
+      "loss": 0.047,
+      "step": 9193
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012470917838039818,
+      "loss": 0.0807,
+      "step": 9194
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012468068238028233,
+      "loss": 0.1409,
+      "step": 9195
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012465218732070487,
+      "loss": 0.0653,
+      "step": 9196
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012462369320272442,
+      "loss": 0.0752,
+      "step": 9197
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012459520002739926,
+      "loss": 0.0404,
+      "step": 9198
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001245667077957879,
+      "loss": 0.114,
+      "step": 9199
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001245382165089488,
+      "loss": 0.0483,
+      "step": 9200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012450972616794028,
+      "loss": 0.0453,
+      "step": 9201
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012448123677382073,
+      "loss": 0.0784,
+      "step": 9202
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012445274832764835,
+      "loss": 0.1049,
+      "step": 9203
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012442426083048144,
+      "loss": 0.0534,
+      "step": 9204
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00012439577428337826,
+      "loss": 0.0719,
+      "step": 9205
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012436728868739697,
+      "loss": 0.0564,
+      "step": 9206
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012433880404359582,
+      "loss": 0.0839,
+      "step": 9207
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012431032035303276,
+      "loss": 0.0834,
+      "step": 9208
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000124281837616766,
+      "loss": 0.1224,
+      "step": 9209
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001242533558358535,
+      "loss": 0.0584,
+      "step": 9210
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012422487501135337,
+      "loss": 0.0906,
+      "step": 9211
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001241963951443236,
+      "loss": 0.0343,
+      "step": 9212
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.000124167916235822,
+      "loss": 0.1074,
+      "step": 9213
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001241394382869066,
+      "loss": 0.0569,
+      "step": 9214
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001241109612986352,
+      "loss": 0.0579,
+      "step": 9215
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001240824852720657,
+      "loss": 0.1121,
+      "step": 9216
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001240540102082559,
+      "loss": 0.1244,
+      "step": 9217
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012402553610826346,
+      "loss": 0.0498,
+      "step": 9218
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012399706297314621,
+      "loss": 0.1701,
+      "step": 9219
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012396859080396184,
+      "loss": 0.1196,
+      "step": 9220
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012394011960176795,
+      "loss": 0.0661,
+      "step": 9221
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012391164936762227,
+      "loss": 0.0512,
+      "step": 9222
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012388318010258227,
+      "loss": 0.0215,
+      "step": 9223
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012385471180770553,
+      "loss": 0.222,
+      "step": 9224
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012382624448404957,
+      "loss": 0.1087,
+      "step": 9225
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012379777813267188,
+      "loss": 0.0585,
+      "step": 9226
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001237693127546299,
+      "loss": 0.1226,
+      "step": 9227
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012374084835098102,
+      "loss": 0.1265,
+      "step": 9228
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012371238492278264,
+      "loss": 0.0818,
+      "step": 9229
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012368392247109205,
+      "loss": 0.1857,
+      "step": 9230
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012365546099696656,
+      "loss": 0.1027,
+      "step": 9231
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012362700050146347,
+      "loss": 0.0937,
+      "step": 9232
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012359854098563998,
+      "loss": 0.0407,
+      "step": 9233
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012357008245055323,
+      "loss": 0.0761,
+      "step": 9234
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012354162489726044,
+      "loss": 0.1082,
+      "step": 9235
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0001235131683268187,
+      "loss": 0.0885,
+      "step": 9236
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012348471274028504,
+      "loss": 0.0356,
+      "step": 9237
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00012345625813871665,
+      "loss": 0.0708,
+      "step": 9238
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012342780452317034,
+      "loss": 0.0515,
+      "step": 9239
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012339935189470318,
+      "loss": 0.0992,
+      "step": 9240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012337090025437211,
+      "loss": 0.1097,
+      "step": 9241
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000123342449603234,
+      "loss": 0.1306,
+      "step": 9242
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012331399994234575,
+      "loss": 0.0958,
+      "step": 9243
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001232855512727641,
+      "loss": 0.1111,
+      "step": 9244
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012325710359554587,
+      "loss": 0.1846,
+      "step": 9245
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012322865691174784,
+      "loss": 0.1092,
+      "step": 9246
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001232002112224267,
+      "loss": 0.1775,
+      "step": 9247
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001231717665286392,
+      "loss": 0.1003,
+      "step": 9248
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001231433228314418,
+      "loss": 0.1704,
+      "step": 9249
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001231148801318912,
+      "loss": 0.0884,
+      "step": 9250
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012308643843104395,
+      "loss": 0.0527,
+      "step": 9251
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012305799772995663,
+      "loss": 0.0807,
+      "step": 9252
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012302955802968572,
+      "loss": 0.1003,
+      "step": 9253
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001230011193312876,
+      "loss": 0.0766,
+      "step": 9254
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012297268163581873,
+      "loss": 0.0602,
+      "step": 9255
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012294424494433545,
+      "loss": 0.1373,
+      "step": 9256
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012291580925789417,
+      "loss": 0.0768,
+      "step": 9257
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012288737457755115,
+      "loss": 0.0983,
+      "step": 9258
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012285894090436264,
+      "loss": 0.1582,
+      "step": 9259
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001228305082393849,
+      "loss": 0.1226,
+      "step": 9260
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012280207658367406,
+      "loss": 0.1072,
+      "step": 9261
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012277364593828638,
+      "loss": 0.0307,
+      "step": 9262
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001227452163042779,
+      "loss": 0.0671,
+      "step": 9263
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012271678768270471,
+      "loss": 0.056,
+      "step": 9264
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012268836007462284,
+      "loss": 0.1091,
+      "step": 9265
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012265993348108828,
+      "loss": 0.0114,
+      "step": 9266
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012263150790315703,
+      "loss": 0.1329,
+      "step": 9267
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012260308334188503,
+      "loss": 0.0859,
+      "step": 9268
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012257465979832814,
+      "loss": 0.1631,
+      "step": 9269
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00012254623727354217,
+      "loss": 0.0406,
+      "step": 9270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.000122517815768583,
+      "loss": 0.065,
+      "step": 9271
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012248939528450637,
+      "loss": 0.0964,
+      "step": 9272
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012246097582236808,
+      "loss": 0.1051,
+      "step": 9273
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012243255738322376,
+      "loss": 0.0325,
+      "step": 9274
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012240413996812905,
+      "loss": 0.168,
+      "step": 9275
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012237572357813966,
+      "loss": 0.075,
+      "step": 9276
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012234730821431111,
+      "loss": 0.0882,
+      "step": 9277
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012231889387769905,
+      "loss": 0.0857,
+      "step": 9278
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012229048056935883,
+      "loss": 0.0489,
+      "step": 9279
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000122262068290346,
+      "loss": 0.1088,
+      "step": 9280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012223365704171598,
+      "loss": 0.0963,
+      "step": 9281
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012220524682452418,
+      "loss": 0.1039,
+      "step": 9282
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.000122176837639826,
+      "loss": 0.0774,
+      "step": 9283
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012214842948867668,
+      "loss": 0.0953,
+      "step": 9284
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012212002237213152,
+      "loss": 0.0586,
+      "step": 9285
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012209161629124577,
+      "loss": 0.0919,
+      "step": 9286
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012206321124707464,
+      "loss": 0.0822,
+      "step": 9287
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012203480724067329,
+      "loss": 0.1363,
+      "step": 9288
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001220064042730969,
+      "loss": 0.1177,
+      "step": 9289
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012197800234540043,
+      "loss": 0.1516,
+      "step": 9290
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012194960145863902,
+      "loss": 0.1698,
+      "step": 9291
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012192120161386762,
+      "loss": 0.0713,
+      "step": 9292
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012189280281214126,
+      "loss": 0.0877,
+      "step": 9293
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012186440505451492,
+      "loss": 0.0692,
+      "step": 9294
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012183600834204337,
+      "loss": 0.0574,
+      "step": 9295
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012180761267578152,
+      "loss": 0.0828,
+      "step": 9296
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012177921805678419,
+      "loss": 0.0657,
+      "step": 9297
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012175082448610614,
+      "loss": 0.1741,
+      "step": 9298
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012172243196480214,
+      "loss": 0.0947,
+      "step": 9299
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012169404049392686,
+      "loss": 0.0779,
+      "step": 9300
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012166565007453497,
+      "loss": 0.1707,
+      "step": 9301
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012163726070768107,
+      "loss": 0.082,
+      "step": 9302
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012160887239441977,
+      "loss": 0.1009,
+      "step": 9303
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00012158048513580569,
+      "loss": 0.0494,
+      "step": 9304
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012155209893289316,
+      "loss": 0.1289,
+      "step": 9305
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012152371378673673,
+      "loss": 0.08,
+      "step": 9306
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012149532969839085,
+      "loss": 0.0662,
+      "step": 9307
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012146694666890987,
+      "loss": 0.1107,
+      "step": 9308
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012143856469934817,
+      "loss": 0.064,
+      "step": 9309
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012141018379076001,
+      "loss": 0.0123,
+      "step": 9310
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001213818039441997,
+      "loss": 0.0764,
+      "step": 9311
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012135342516072142,
+      "loss": 0.0912,
+      "step": 9312
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001213250474413794,
+      "loss": 0.1201,
+      "step": 9313
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012129667078722786,
+      "loss": 0.0689,
+      "step": 9314
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012126829519932071,
+      "loss": 0.118,
+      "step": 9315
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012123992067871219,
+      "loss": 0.1134,
+      "step": 9316
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012121154722645626,
+      "loss": 0.1101,
+      "step": 9317
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012118317484360694,
+      "loss": 0.0316,
+      "step": 9318
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012115480353121825,
+      "loss": 0.1492,
+      "step": 9319
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012112643329034393,
+      "loss": 0.053,
+      "step": 9320
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012109806412203796,
+      "loss": 0.0952,
+      "step": 9321
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012106969602735414,
+      "loss": 0.1715,
+      "step": 9322
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012104132900734629,
+      "loss": 0.0765,
+      "step": 9323
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012101296306306816,
+      "loss": 0.0341,
+      "step": 9324
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012098459819557342,
+      "loss": 0.1117,
+      "step": 9325
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012095623440591579,
+      "loss": 0.1268,
+      "step": 9326
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012092787169514887,
+      "loss": 0.0707,
+      "step": 9327
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012089951006432625,
+      "loss": 0.0535,
+      "step": 9328
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012087114951450158,
+      "loss": 0.1677,
+      "step": 9329
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001208427900467282,
+      "loss": 0.0995,
+      "step": 9330
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012081443166205969,
+      "loss": 0.109,
+      "step": 9331
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012078607436154943,
+      "loss": 0.0946,
+      "step": 9332
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012075771814625082,
+      "loss": 0.1141,
+      "step": 9333
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012072936301721735,
+      "loss": 0.092,
+      "step": 9334
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012070100897550211,
+      "loss": 0.0429,
+      "step": 9335
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012067265602215848,
+      "loss": 0.1219,
+      "step": 9336
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00012064430415823968,
+      "loss": 0.1044,
+      "step": 9337
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001206159533847989,
+      "loss": 0.0899,
+      "step": 9338
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001205876037028893,
+      "loss": 0.0599,
+      "step": 9339
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012055925511356395,
+      "loss": 0.0942,
+      "step": 9340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012053090761787594,
+      "loss": 0.109,
+      "step": 9341
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012050256121687829,
+      "loss": 0.0379,
+      "step": 9342
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012047421591162399,
+      "loss": 0.0593,
+      "step": 9343
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012044587170316605,
+      "loss": 0.0489,
+      "step": 9344
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012041752859255725,
+      "loss": 0.0917,
+      "step": 9345
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001203891865808505,
+      "loss": 0.0954,
+      "step": 9346
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012036084566909864,
+      "loss": 0.1041,
+      "step": 9347
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012033250585835446,
+      "loss": 0.1082,
+      "step": 9348
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012030416714967071,
+      "loss": 0.17,
+      "step": 9349
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012027582954410005,
+      "loss": 0.0137,
+      "step": 9350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012024749304269513,
+      "loss": 0.0438,
+      "step": 9351
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012021915764650861,
+      "loss": 0.0149,
+      "step": 9352
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012019082335659304,
+      "loss": 0.0802,
+      "step": 9353
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012016249017400099,
+      "loss": 0.074,
+      "step": 9354
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012013415809978498,
+      "loss": 0.0695,
+      "step": 9355
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012010582713499732,
+      "loss": 0.0674,
+      "step": 9356
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012007749728069057,
+      "loss": 0.0516,
+      "step": 9357
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012004916853791704,
+      "loss": 0.0785,
+      "step": 9358
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00012002084090772906,
+      "loss": 0.0425,
+      "step": 9359
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.000119992514391179,
+      "loss": 0.0643,
+      "step": 9360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011996418898931895,
+      "loss": 0.1169,
+      "step": 9361
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011993586470320121,
+      "loss": 0.1347,
+      "step": 9362
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011990754153387794,
+      "loss": 0.057,
+      "step": 9363
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011987921948240126,
+      "loss": 0.0855,
+      "step": 9364
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011985089854982328,
+      "loss": 0.1224,
+      "step": 9365
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011982257873719596,
+      "loss": 0.0695,
+      "step": 9366
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011979426004557135,
+      "loss": 0.0219,
+      "step": 9367
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011976594247600142,
+      "loss": 0.0655,
+      "step": 9368
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011973762602953805,
+      "loss": 0.1116,
+      "step": 9369
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011970931070723319,
+      "loss": 0.0732,
+      "step": 9370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011968099651013856,
+      "loss": 0.0886,
+      "step": 9371
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.000119652683439306,
+      "loss": 0.0768,
+      "step": 9372
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011962437149578724,
+      "loss": 0.1005,
+      "step": 9373
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011959606068063402,
+      "loss": 0.1846,
+      "step": 9374
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011956775099489801,
+      "loss": 0.0895,
+      "step": 9375
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011953944243963077,
+      "loss": 0.0733,
+      "step": 9376
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011951113501588393,
+      "loss": 0.1413,
+      "step": 9377
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011948282872470902,
+      "loss": 0.1287,
+      "step": 9378
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011945452356715751,
+      "loss": 0.1778,
+      "step": 9379
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011942621954428094,
+      "loss": 0.074,
+      "step": 9380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011939791665713061,
+      "loss": 0.074,
+      "step": 9381
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011936961490675793,
+      "loss": 0.0477,
+      "step": 9382
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011934131429421422,
+      "loss": 0.0862,
+      "step": 9383
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011931301482055081,
+      "loss": 0.0863,
+      "step": 9384
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011928471648681896,
+      "loss": 0.0987,
+      "step": 9385
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011925641929406977,
+      "loss": 0.083,
+      "step": 9386
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011922812324335442,
+      "loss": 0.1546,
+      "step": 9387
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011919982833572408,
+      "loss": 0.0231,
+      "step": 9388
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011917153457222977,
+      "loss": 0.1108,
+      "step": 9389
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001191432419539226,
+      "loss": 0.0732,
+      "step": 9390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011911495048185347,
+      "loss": 0.1212,
+      "step": 9391
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011908666015707337,
+      "loss": 0.0791,
+      "step": 9392
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011905837098063316,
+      "loss": 0.0787,
+      "step": 9393
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011903008295358377,
+      "loss": 0.1032,
+      "step": 9394
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011900179607697602,
+      "loss": 0.146,
+      "step": 9395
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011897351035186059,
+      "loss": 0.0933,
+      "step": 9396
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011894522577928823,
+      "loss": 0.1028,
+      "step": 9397
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011891694236030971,
+      "loss": 0.0816,
+      "step": 9398
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001188886600959756,
+      "loss": 0.0539,
+      "step": 9399
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011886037898733664,
+      "loss": 0.0667,
+      "step": 9400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011883209903544318,
+      "loss": 0.1306,
+      "step": 9401
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011880382024134585,
+      "loss": 0.0217,
+      "step": 9402
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011877554260609513,
+      "loss": 0.164,
+      "step": 9403
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011874726613074145,
+      "loss": 0.1174,
+      "step": 9404
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011871899081633518,
+      "loss": 0.1073,
+      "step": 9405
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011869071666392668,
+      "loss": 0.0871,
+      "step": 9406
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011866244367456621,
+      "loss": 0.0606,
+      "step": 9407
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011863417184930406,
+      "loss": 0.0067,
+      "step": 9408
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011860590118919046,
+      "loss": 0.1008,
+      "step": 9409
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011857763169527564,
+      "loss": 0.1404,
+      "step": 9410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011854936336860958,
+      "loss": 0.1211,
+      "step": 9411
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011852109621024244,
+      "loss": 0.1259,
+      "step": 9412
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011849283022122428,
+      "loss": 0.0502,
+      "step": 9413
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011846456540260506,
+      "loss": 0.1134,
+      "step": 9414
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011843630175543478,
+      "loss": 0.1232,
+      "step": 9415
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011840803928076336,
+      "loss": 0.0585,
+      "step": 9416
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011837977797964059,
+      "loss": 0.0883,
+      "step": 9417
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011835151785311635,
+      "loss": 0.0976,
+      "step": 9418
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011832325890224041,
+      "loss": 0.1419,
+      "step": 9419
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011829500112806253,
+      "loss": 0.1712,
+      "step": 9420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001182667445316324,
+      "loss": 0.1149,
+      "step": 9421
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011823848911399963,
+      "loss": 0.0648,
+      "step": 9422
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011821023487621384,
+      "loss": 0.1491,
+      "step": 9423
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011818198181932461,
+      "loss": 0.0843,
+      "step": 9424
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011815372994438145,
+      "loss": 0.0479,
+      "step": 9425
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011812547925243388,
+      "loss": 0.0907,
+      "step": 9426
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011809722974453124,
+      "loss": 0.1877,
+      "step": 9427
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011806898142172295,
+      "loss": 0.0735,
+      "step": 9428
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011804073428505836,
+      "loss": 0.1049,
+      "step": 9429
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011801248833558679,
+      "loss": 0.1432,
+      "step": 9430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011798424357435749,
+      "loss": 0.0487,
+      "step": 9431
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011795600000241965,
+      "loss": 0.076,
+      "step": 9432
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011792775762082241,
+      "loss": 0.1092,
+      "step": 9433
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011789951643061492,
+      "loss": 0.0832,
+      "step": 9434
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011787127643284628,
+      "loss": 0.0681,
+      "step": 9435
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011784303762856559,
+      "loss": 0.0866,
+      "step": 9436
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011781480001882165,
+      "loss": 0.0485,
+      "step": 9437
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011778656360466348,
+      "loss": 0.1024,
+      "step": 9438
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011775832838714006,
+      "loss": 0.0423,
+      "step": 9439
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011773009436730017,
+      "loss": 0.0778,
+      "step": 9440
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011770186154619272,
+      "loss": 0.1551,
+      "step": 9441
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011767362992486635,
+      "loss": 0.0928,
+      "step": 9442
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011764539950436981,
+      "loss": 0.1667,
+      "step": 9443
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001176171702857518,
+      "loss": 0.0608,
+      "step": 9444
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011758894227006095,
+      "loss": 0.1304,
+      "step": 9445
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001175607154583459,
+      "loss": 0.0612,
+      "step": 9446
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011753248985165509,
+      "loss": 0.0615,
+      "step": 9447
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011750426545103708,
+      "loss": 0.0903,
+      "step": 9448
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011747604225754032,
+      "loss": 0.1937,
+      "step": 9449
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001174478202722132,
+      "loss": 0.097,
+      "step": 9450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011741959949610419,
+      "loss": 0.0632,
+      "step": 9451
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011739137993026145,
+      "loss": 0.1419,
+      "step": 9452
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001173631615757333,
+      "loss": 0.0852,
+      "step": 9453
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.000117334944433568,
+      "loss": 0.1384,
+      "step": 9454
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001173067285048137,
+      "loss": 0.0618,
+      "step": 9455
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011727851379051865,
+      "loss": 0.0888,
+      "step": 9456
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011725030029173078,
+      "loss": 0.0554,
+      "step": 9457
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011722208800949823,
+      "loss": 0.0604,
+      "step": 9458
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.000117193876944869,
+      "loss": 0.146,
+      "step": 9459
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011716566709889104,
+      "loss": 0.1275,
+      "step": 9460
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011713745847261226,
+      "loss": 0.0528,
+      "step": 9461
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011710925106708053,
+      "loss": 0.1236,
+      "step": 9462
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011708104488334366,
+      "loss": 0.077,
+      "step": 9463
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011705283992244947,
+      "loss": 0.0847,
+      "step": 9464
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001170246361854456,
+      "loss": 0.1147,
+      "step": 9465
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011699643367337991,
+      "loss": 0.075,
+      "step": 9466
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011696823238729984,
+      "loss": 0.1069,
+      "step": 9467
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011694003232825307,
+      "loss": 0.0705,
+      "step": 9468
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011691183349728717,
+      "loss": 0.0376,
+      "step": 9469
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001168836358954496,
+      "loss": 0.1203,
+      "step": 9470
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001168554395237879,
+      "loss": 0.0673,
+      "step": 9471
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011682724438334938,
+      "loss": 0.1529,
+      "step": 9472
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011679905047518147,
+      "loss": 0.105,
+      "step": 9473
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011677085780033145,
+      "loss": 0.1357,
+      "step": 9474
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011674266635984662,
+      "loss": 0.0898,
+      "step": 9475
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011671447615477423,
+      "loss": 0.2044,
+      "step": 9476
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011668628718616149,
+      "loss": 0.1138,
+      "step": 9477
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011665809945505543,
+      "loss": 0.0477,
+      "step": 9478
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011662991296250319,
+      "loss": 0.1457,
+      "step": 9479
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011660172770955181,
+      "loss": 0.1411,
+      "step": 9480
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011657354369724834,
+      "loss": 0.1055,
+      "step": 9481
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011654536092663975,
+      "loss": 0.0906,
+      "step": 9482
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011651717939877282,
+      "loss": 0.064,
+      "step": 9483
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011648899911469452,
+      "loss": 0.0923,
+      "step": 9484
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011646082007545162,
+      "loss": 0.0619,
+      "step": 9485
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001164326422820909,
+      "loss": 0.1102,
+      "step": 9486
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011640446573565912,
+      "loss": 0.0438,
+      "step": 9487
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011637629043720288,
+      "loss": 0.1299,
+      "step": 9488
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011634811638776884,
+      "loss": 0.1022,
+      "step": 9489
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001163199435884036,
+      "loss": 0.1133,
+      "step": 9490
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011629177204015372,
+      "loss": 0.092,
+      "step": 9491
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011626360174406569,
+      "loss": 0.1211,
+      "step": 9492
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011623543270118588,
+      "loss": 0.0644,
+      "step": 9493
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011620726491256071,
+      "loss": 0.0433,
+      "step": 9494
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011617909837923657,
+      "loss": 0.1283,
+      "step": 9495
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011615093310225974,
+      "loss": 0.042,
+      "step": 9496
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011612276908267652,
+      "loss": 0.1738,
+      "step": 9497
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011609460632153304,
+      "loss": 0.107,
+      "step": 9498
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011606644481987552,
+      "loss": 0.0502,
+      "step": 9499
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011603828457875005,
+      "loss": 0.0547,
+      "step": 9500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011601012559920275,
+      "loss": 0.063,
+      "step": 9501
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001159819678822796,
+      "loss": 0.0678,
+      "step": 9502
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011595381142902658,
+      "loss": 0.1077,
+      "step": 9503
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011592565624048963,
+      "loss": 0.079,
+      "step": 9504
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001158975023177146,
+      "loss": 0.1749,
+      "step": 9505
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011586934966174736,
+      "loss": 0.0323,
+      "step": 9506
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011584119827363379,
+      "loss": 0.022,
+      "step": 9507
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011581304815441944,
+      "loss": 0.1005,
+      "step": 9508
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001157848993051501,
+      "loss": 0.0821,
+      "step": 9509
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011575675172687143,
+      "loss": 0.1381,
+      "step": 9510
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011572860542062902,
+      "loss": 0.095,
+      "step": 9511
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011570046038746843,
+      "loss": 0.1895,
+      "step": 9512
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011567231662843514,
+      "loss": 0.011,
+      "step": 9513
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011564417414457461,
+      "loss": 0.0576,
+      "step": 9514
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011561603293693227,
+      "loss": 0.0644,
+      "step": 9515
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011558789300655347,
+      "loss": 0.0753,
+      "step": 9516
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011555975435448359,
+      "loss": 0.013,
+      "step": 9517
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001155316169817678,
+      "loss": 0.0972,
+      "step": 9518
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011550348088945134,
+      "loss": 0.1827,
+      "step": 9519
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011547534607857942,
+      "loss": 0.0411,
+      "step": 9520
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011544721255019711,
+      "loss": 0.1347,
+      "step": 9521
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011541908030534964,
+      "loss": 0.1005,
+      "step": 9522
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011539094934508184,
+      "loss": 0.1055,
+      "step": 9523
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011536281967043878,
+      "loss": 0.0451,
+      "step": 9524
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011533469128246541,
+      "loss": 0.056,
+      "step": 9525
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001153065641822066,
+      "loss": 0.0112,
+      "step": 9526
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011527843837070724,
+      "loss": 0.1152,
+      "step": 9527
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011525031384901201,
+      "loss": 0.1322,
+      "step": 9528
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011522219061816576,
+      "loss": 0.0991,
+      "step": 9529
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011519406867921311,
+      "loss": 0.1381,
+      "step": 9530
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011516594803319874,
+      "loss": 0.0912,
+      "step": 9531
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011513782868116734,
+      "loss": 0.0675,
+      "step": 9532
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001151097106241633,
+      "loss": 0.1579,
+      "step": 9533
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011508159386323119,
+      "loss": 0.066,
+      "step": 9534
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011505347839941546,
+      "loss": 0.1491,
+      "step": 9535
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011502536423376053,
+      "loss": 0.0525,
+      "step": 9536
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001149972513673108,
+      "loss": 0.0147,
+      "step": 9537
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001149691398011105,
+      "loss": 0.083,
+      "step": 9538
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011494102953620395,
+      "loss": 0.1462,
+      "step": 9539
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001149129205736353,
+      "loss": 0.02,
+      "step": 9540
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011488481291444879,
+      "loss": 0.0413,
+      "step": 9541
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001148567065596885,
+      "loss": 0.0167,
+      "step": 9542
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011482860151039854,
+      "loss": 0.0853,
+      "step": 9543
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011480049776762285,
+      "loss": 0.1283,
+      "step": 9544
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011477239533240546,
+      "loss": 0.0837,
+      "step": 9545
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011474429420579026,
+      "loss": 0.1008,
+      "step": 9546
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011471619438882117,
+      "loss": 0.0823,
+      "step": 9547
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011468809588254204,
+      "loss": 0.0993,
+      "step": 9548
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011465999868799653,
+      "loss": 0.0442,
+      "step": 9549
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011463190280622844,
+      "loss": 0.1344,
+      "step": 9550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011460380823828145,
+      "loss": 0.1183,
+      "step": 9551
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011457571498519917,
+      "loss": 0.0829,
+      "step": 9552
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011454762304802523,
+      "loss": 0.048,
+      "step": 9553
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001145195324278031,
+      "loss": 0.1089,
+      "step": 9554
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011449144312557631,
+      "loss": 0.0408,
+      "step": 9555
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011446335514238828,
+      "loss": 0.0922,
+      "step": 9556
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011443526847928238,
+      "loss": 0.1775,
+      "step": 9557
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011440718313730204,
+      "loss": 0.0777,
+      "step": 9558
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011437909911749039,
+      "loss": 0.0699,
+      "step": 9559
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011435101642089077,
+      "loss": 0.078,
+      "step": 9560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011432293504854634,
+      "loss": 0.0654,
+      "step": 9561
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011429485500150022,
+      "loss": 0.1007,
+      "step": 9562
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011426677628079563,
+      "loss": 0.079,
+      "step": 9563
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011423869888747544,
+      "loss": 0.0753,
+      "step": 9564
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011421062282258274,
+      "loss": 0.0166,
+      "step": 9565
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011418254808716043,
+      "loss": 0.0898,
+      "step": 9566
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011415447468225144,
+      "loss": 0.1411,
+      "step": 9567
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011412640260889862,
+      "loss": 0.0936,
+      "step": 9568
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011409833186814472,
+      "loss": 0.0679,
+      "step": 9569
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011407026246103249,
+      "loss": 0.16,
+      "step": 9570
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011404219438860468,
+      "loss": 0.1453,
+      "step": 9571
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011401412765190388,
+      "loss": 0.0853,
+      "step": 9572
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011398606225197277,
+      "loss": 0.0998,
+      "step": 9573
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011395799818985378,
+      "loss": 0.084,
+      "step": 9574
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011392993546658949,
+      "loss": 0.0597,
+      "step": 9575
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001139018740832223,
+      "loss": 0.1176,
+      "step": 9576
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011387381404079464,
+      "loss": 0.0792,
+      "step": 9577
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011384575534034885,
+      "loss": 0.0466,
+      "step": 9578
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011381769798292722,
+      "loss": 0.0638,
+      "step": 9579
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011378964196957202,
+      "loss": 0.0407,
+      "step": 9580
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011376158730132541,
+      "loss": 0.069,
+      "step": 9581
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011373353397922957,
+      "loss": 0.0574,
+      "step": 9582
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011370548200432665,
+      "loss": 0.0848,
+      "step": 9583
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011367743137765857,
+      "loss": 0.1611,
+      "step": 9584
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011364938210026741,
+      "loss": 0.0654,
+      "step": 9585
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011362133417319513,
+      "loss": 0.1006,
+      "step": 9586
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011359328759748357,
+      "loss": 0.0884,
+      "step": 9587
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001135652423741747,
+      "loss": 0.0742,
+      "step": 9588
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011353719850431016,
+      "loss": 0.0885,
+      "step": 9589
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011350915598893177,
+      "loss": 0.115,
+      "step": 9590
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011348111482908123,
+      "loss": 0.1233,
+      "step": 9591
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011345307502580019,
+      "loss": 0.0638,
+      "step": 9592
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011342503658013026,
+      "loss": 0.1214,
+      "step": 9593
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011339699949311295,
+      "loss": 0.028,
+      "step": 9594
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011336896376578975,
+      "loss": 0.0874,
+      "step": 9595
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011334092939920215,
+      "loss": 0.0814,
+      "step": 9596
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011331289639439151,
+      "loss": 0.0845,
+      "step": 9597
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011328486475239925,
+      "loss": 0.0482,
+      "step": 9598
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011325683447426653,
+      "loss": 0.0911,
+      "step": 9599
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00011322880556103466,
+      "loss": 0.1003,
+      "step": 9600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011320077801374485,
+      "loss": 0.0381,
+      "step": 9601
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001131727518334382,
+      "loss": 0.1239,
+      "step": 9602
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011314472702115579,
+      "loss": 0.0947,
+      "step": 9603
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011311670357793881,
+      "loss": 0.1043,
+      "step": 9604
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011308868150482805,
+      "loss": 0.0548,
+      "step": 9605
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011306066080286455,
+      "loss": 0.1316,
+      "step": 9606
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011303264147308916,
+      "loss": 0.0699,
+      "step": 9607
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011300462351654274,
+      "loss": 0.0765,
+      "step": 9608
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011297660693426611,
+      "loss": 0.043,
+      "step": 9609
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011294859172729992,
+      "loss": 0.0794,
+      "step": 9610
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001129205778966849,
+      "loss": 0.064,
+      "step": 9611
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011289256544346168,
+      "loss": 0.0692,
+      "step": 9612
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011286455436867082,
+      "loss": 0.1108,
+      "step": 9613
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011283654467335294,
+      "loss": 0.1079,
+      "step": 9614
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011280853635854839,
+      "loss": 0.0389,
+      "step": 9615
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011278052942529764,
+      "loss": 0.0684,
+      "step": 9616
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011275252387464108,
+      "loss": 0.0444,
+      "step": 9617
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011272451970761904,
+      "loss": 0.0467,
+      "step": 9618
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001126965169252718,
+      "loss": 0.0744,
+      "step": 9619
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011266851552863954,
+      "loss": 0.1311,
+      "step": 9620
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011264051551876245,
+      "loss": 0.08,
+      "step": 9621
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011261251689668065,
+      "loss": 0.0914,
+      "step": 9622
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011258451966343422,
+      "loss": 0.1285,
+      "step": 9623
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011255652382006321,
+      "loss": 0.0893,
+      "step": 9624
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011252852936760749,
+      "loss": 0.0795,
+      "step": 9625
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.000112500536307107,
+      "loss": 0.0505,
+      "step": 9626
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011247254463960162,
+      "loss": 0.0781,
+      "step": 9627
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011244455436613119,
+      "loss": 0.1433,
+      "step": 9628
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011241656548773547,
+      "loss": 0.1688,
+      "step": 9629
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011238857800545408,
+      "loss": 0.0719,
+      "step": 9630
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011236059192032672,
+      "loss": 0.1159,
+      "step": 9631
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011233260723339301,
+      "loss": 0.0902,
+      "step": 9632
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00011230462394569246,
+      "loss": 0.0757,
+      "step": 9633
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011227664205826465,
+      "loss": 0.1169,
+      "step": 9634
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001122486615721489,
+      "loss": 0.1011,
+      "step": 9635
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011222068248838469,
+      "loss": 0.0556,
+      "step": 9636
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011219270480801132,
+      "loss": 0.0705,
+      "step": 9637
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001121647285320681,
+      "loss": 0.0735,
+      "step": 9638
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011213675366159434,
+      "loss": 0.0905,
+      "step": 9639
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001121087801976291,
+      "loss": 0.1827,
+      "step": 9640
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011208080814121151,
+      "loss": 0.0892,
+      "step": 9641
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011205283749338073,
+      "loss": 0.153,
+      "step": 9642
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011202486825517577,
+      "loss": 0.0609,
+      "step": 9643
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001119969004276356,
+      "loss": 0.0724,
+      "step": 9644
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011196893401179913,
+      "loss": 0.051,
+      "step": 9645
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011194096900870523,
+      "loss": 0.0911,
+      "step": 9646
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011191300541939272,
+      "loss": 0.0592,
+      "step": 9647
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011188504324490035,
+      "loss": 0.0819,
+      "step": 9648
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011185708248626692,
+      "loss": 0.0783,
+      "step": 9649
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011182912314453097,
+      "loss": 0.1216,
+      "step": 9650
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011180116522073115,
+      "loss": 0.1173,
+      "step": 9651
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011177320871590605,
+      "loss": 0.1917,
+      "step": 9652
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011174525363109414,
+      "loss": 0.0312,
+      "step": 9653
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011171729996733393,
+      "loss": 0.0492,
+      "step": 9654
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001116893477256637,
+      "loss": 0.1009,
+      "step": 9655
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011166139690712186,
+      "loss": 0.0646,
+      "step": 9656
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001116334475127467,
+      "loss": 0.083,
+      "step": 9657
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011160549954357647,
+      "loss": 0.1586,
+      "step": 9658
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011157755300064935,
+      "loss": 0.1298,
+      "step": 9659
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011154960788500343,
+      "loss": 0.1532,
+      "step": 9660
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011152166419767683,
+      "loss": 0.2237,
+      "step": 9661
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011149372193970757,
+      "loss": 0.0393,
+      "step": 9662
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011146578111213361,
+      "loss": 0.112,
+      "step": 9663
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001114378417159929,
+      "loss": 0.0707,
+      "step": 9664
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011140990375232335,
+      "loss": 0.144,
+      "step": 9665
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00011138196722216263,
+      "loss": 0.1197,
+      "step": 9666
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011135403212654856,
+      "loss": 0.0932,
+      "step": 9667
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001113260984665189,
+      "loss": 0.1305,
+      "step": 9668
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011129816624311127,
+      "loss": 0.1042,
+      "step": 9669
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011127023545736333,
+      "loss": 0.1095,
+      "step": 9670
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001112423061103125,
+      "loss": 0.0357,
+      "step": 9671
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011121437820299638,
+      "loss": 0.0674,
+      "step": 9672
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011118645173645234,
+      "loss": 0.1259,
+      "step": 9673
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001111585267117178,
+      "loss": 0.0387,
+      "step": 9674
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011113060312983012,
+      "loss": 0.0994,
+      "step": 9675
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011110268099182652,
+      "loss": 0.072,
+      "step": 9676
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011107476029874426,
+      "loss": 0.0301,
+      "step": 9677
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011104684105162053,
+      "loss": 0.0508,
+      "step": 9678
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001110189232514924,
+      "loss": 0.1278,
+      "step": 9679
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011099100689939702,
+      "loss": 0.1775,
+      "step": 9680
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011096309199637128,
+      "loss": 0.0389,
+      "step": 9681
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011093517854345222,
+      "loss": 0.0657,
+      "step": 9682
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001109072665416767,
+      "loss": 0.0969,
+      "step": 9683
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011087935599208158,
+      "loss": 0.1294,
+      "step": 9684
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011085144689570372,
+      "loss": 0.0434,
+      "step": 9685
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011082353925357976,
+      "loss": 0.1195,
+      "step": 9686
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011079563306674643,
+      "loss": 0.0911,
+      "step": 9687
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011076772833624038,
+      "loss": 0.0832,
+      "step": 9688
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011073982506309815,
+      "loss": 0.1815,
+      "step": 9689
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011071192324835634,
+      "loss": 0.0502,
+      "step": 9690
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011068402289305133,
+      "loss": 0.1703,
+      "step": 9691
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011065612399821959,
+      "loss": 0.1204,
+      "step": 9692
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011062822656489746,
+      "loss": 0.1478,
+      "step": 9693
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011060033059412126,
+      "loss": 0.0509,
+      "step": 9694
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001105724360869273,
+      "loss": 0.1534,
+      "step": 9695
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011054454304435165,
+      "loss": 0.069,
+      "step": 9696
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011051665146743054,
+      "loss": 0.0907,
+      "step": 9697
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011048876135720001,
+      "loss": 0.0841,
+      "step": 9698
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00011046087271469616,
+      "loss": 0.1221,
+      "step": 9699
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011043298554095497,
+      "loss": 0.0306,
+      "step": 9700
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011040509983701231,
+      "loss": 0.1196,
+      "step": 9701
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011037721560390406,
+      "loss": 0.0979,
+      "step": 9702
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011034933284266608,
+      "loss": 0.1048,
+      "step": 9703
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011032145155433409,
+      "loss": 0.0676,
+      "step": 9704
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011029357173994389,
+      "loss": 0.1245,
+      "step": 9705
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011026569340053099,
+      "loss": 0.0885,
+      "step": 9706
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011023781653713109,
+      "loss": 0.0751,
+      "step": 9707
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011020994115077964,
+      "loss": 0.0755,
+      "step": 9708
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011018206724251225,
+      "loss": 0.1219,
+      "step": 9709
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011015419481336435,
+      "loss": 0.1103,
+      "step": 9710
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011012632386437118,
+      "loss": 0.0884,
+      "step": 9711
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011009845439656819,
+      "loss": 0.1238,
+      "step": 9712
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011007058641099058,
+      "loss": 0.1147,
+      "step": 9713
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001100427199086736,
+      "loss": 0.0779,
+      "step": 9714
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00011001485489065243,
+      "loss": 0.0351,
+      "step": 9715
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010998699135796213,
+      "loss": 0.0639,
+      "step": 9716
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010995912931163774,
+      "loss": 0.0673,
+      "step": 9717
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001099312687527143,
+      "loss": 0.1311,
+      "step": 9718
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010990340968222671,
+      "loss": 0.0697,
+      "step": 9719
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010987555210120992,
+      "loss": 0.0646,
+      "step": 9720
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010984769601069867,
+      "loss": 0.0901,
+      "step": 9721
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010981984141172775,
+      "loss": 0.0732,
+      "step": 9722
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010979198830533191,
+      "loss": 0.1415,
+      "step": 9723
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010976413669254577,
+      "loss": 0.1281,
+      "step": 9724
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010973628657440401,
+      "loss": 0.0648,
+      "step": 9725
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001097084379519411,
+      "loss": 0.1024,
+      "step": 9726
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010968059082619157,
+      "loss": 0.1819,
+      "step": 9727
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010965274519818984,
+      "loss": 0.0855,
+      "step": 9728
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010962490106897033,
+      "loss": 0.0972,
+      "step": 9729
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010959705843956733,
+      "loss": 0.1275,
+      "step": 9730
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010956921731101515,
+      "loss": 0.1003,
+      "step": 9731
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010954137768434799,
+      "loss": 0.1364,
+      "step": 9732
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010951353956059997,
+      "loss": 0.1367,
+      "step": 9733
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010948570294080524,
+      "loss": 0.1025,
+      "step": 9734
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010945786782599784,
+      "loss": 0.1022,
+      "step": 9735
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010943003421721183,
+      "loss": 0.0491,
+      "step": 9736
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.000109402202115481,
+      "loss": 0.0996,
+      "step": 9737
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010937437152183932,
+      "loss": 0.0649,
+      "step": 9738
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001093465424373206,
+      "loss": 0.1654,
+      "step": 9739
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010931871486295862,
+      "loss": 0.0742,
+      "step": 9740
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010929088879978711,
+      "loss": 0.1431,
+      "step": 9741
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010926306424883969,
+      "loss": 0.0687,
+      "step": 9742
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010923524121114994,
+      "loss": 0.0261,
+      "step": 9743
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010920741968775146,
+      "loss": 0.0994,
+      "step": 9744
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010917959967967772,
+      "loss": 0.0922,
+      "step": 9745
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010915178118796219,
+      "loss": 0.1192,
+      "step": 9746
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010912396421363815,
+      "loss": 0.0728,
+      "step": 9747
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010909614875773897,
+      "loss": 0.0715,
+      "step": 9748
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010906833482129789,
+      "loss": 0.0648,
+      "step": 9749
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010904052240534817,
+      "loss": 0.0678,
+      "step": 9750
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010901271151092298,
+      "loss": 0.0301,
+      "step": 9751
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010898490213905531,
+      "loss": 0.1694,
+      "step": 9752
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010895709429077825,
+      "loss": 0.1831,
+      "step": 9753
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001089292879671248,
+      "loss": 0.0848,
+      "step": 9754
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010890148316912783,
+      "loss": 0.1132,
+      "step": 9755
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001088736798978203,
+      "loss": 0.1191,
+      "step": 9756
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001088458781542349,
+      "loss": 0.1269,
+      "step": 9757
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010881807793940444,
+      "loss": 0.1215,
+      "step": 9758
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010879027925436164,
+      "loss": 0.0798,
+      "step": 9759
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010876248210013912,
+      "loss": 0.1149,
+      "step": 9760
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001087346864777695,
+      "loss": 0.0607,
+      "step": 9761
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010870689238828526,
+      "loss": 0.0719,
+      "step": 9762
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010867909983271883,
+      "loss": 0.0216,
+      "step": 9763
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010865130881210268,
+      "loss": 0.0364,
+      "step": 9764
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010862351932746914,
+      "loss": 0.0731,
+      "step": 9765
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010859573137985058,
+      "loss": 0.1182,
+      "step": 9766
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010856794497027913,
+      "loss": 0.0348,
+      "step": 9767
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010854016009978703,
+      "loss": 0.0558,
+      "step": 9768
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001085123767694064,
+      "loss": 0.0438,
+      "step": 9769
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010848459498016931,
+      "loss": 0.074,
+      "step": 9770
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010845681473310785,
+      "loss": 0.1336,
+      "step": 9771
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010842903602925377,
+      "loss": 0.1011,
+      "step": 9772
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010840125886963918,
+      "loss": 0.1173,
+      "step": 9773
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010837348325529581,
+      "loss": 0.1409,
+      "step": 9774
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010834570918725546,
+      "loss": 0.0363,
+      "step": 9775
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010831793666654993,
+      "loss": 0.0923,
+      "step": 9776
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010829016569421077,
+      "loss": 0.1557,
+      "step": 9777
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010826239627126964,
+      "loss": 0.1162,
+      "step": 9778
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010823462839875809,
+      "loss": 0.1079,
+      "step": 9779
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001082068620777076,
+      "loss": 0.2043,
+      "step": 9780
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010817909730914968,
+      "loss": 0.0908,
+      "step": 9781
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010815133409411562,
+      "loss": 0.1446,
+      "step": 9782
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010812357243363678,
+      "loss": 0.1188,
+      "step": 9783
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010809581232874442,
+      "loss": 0.0745,
+      "step": 9784
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010806805378046974,
+      "loss": 0.1104,
+      "step": 9785
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010804029678984398,
+      "loss": 0.0795,
+      "step": 9786
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010801254135789808,
+      "loss": 0.0891,
+      "step": 9787
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010798478748566313,
+      "loss": 0.0743,
+      "step": 9788
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010795703517417012,
+      "loss": 0.0499,
+      "step": 9789
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010792928442444996,
+      "loss": 0.0792,
+      "step": 9790
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001079015352375335,
+      "loss": 0.0659,
+      "step": 9791
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010787378761445163,
+      "loss": 0.0341,
+      "step": 9792
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010784604155623496,
+      "loss": 0.1517,
+      "step": 9793
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010781829706391423,
+      "loss": 0.0776,
+      "step": 9794
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010779055413852005,
+      "loss": 0.1678,
+      "step": 9795
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010776281278108303,
+      "loss": 0.1096,
+      "step": 9796
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010773507299263369,
+      "loss": 0.128,
+      "step": 9797
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001077073347742024,
+      "loss": 0.1069,
+      "step": 9798
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001076795981268196,
+      "loss": 0.047,
+      "step": 9799
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010765186305151562,
+      "loss": 0.1281,
+      "step": 9800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010762412954932076,
+      "loss": 0.1576,
+      "step": 9801
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010759639762126529,
+      "loss": 0.1416,
+      "step": 9802
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001075686672683792,
+      "loss": 0.1058,
+      "step": 9803
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010754093849169273,
+      "loss": 0.1348,
+      "step": 9804
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010751321129223586,
+      "loss": 0.136,
+      "step": 9805
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010748548567103863,
+      "loss": 0.1399,
+      "step": 9806
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010745776162913095,
+      "loss": 0.0538,
+      "step": 9807
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010743003916754263,
+      "loss": 0.06,
+      "step": 9808
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010740231828730354,
+      "loss": 0.0439,
+      "step": 9809
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010737459898944339,
+      "loss": 0.053,
+      "step": 9810
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010734688127499188,
+      "loss": 0.1314,
+      "step": 9811
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010731916514497875,
+      "loss": 0.1487,
+      "step": 9812
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010729145060043335,
+      "loss": 0.1744,
+      "step": 9813
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010726373764238536,
+      "loss": 0.1271,
+      "step": 9814
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010723602627186419,
+      "loss": 0.1208,
+      "step": 9815
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010720831648989925,
+      "loss": 0.0713,
+      "step": 9816
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010718060829751993,
+      "loss": 0.0524,
+      "step": 9817
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010715290169575535,
+      "loss": 0.1148,
+      "step": 9818
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010712519668563486,
+      "loss": 0.1067,
+      "step": 9819
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010709749326818758,
+      "loss": 0.0487,
+      "step": 9820
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001070697914444426,
+      "loss": 0.1306,
+      "step": 9821
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010704209121542902,
+      "loss": 0.1062,
+      "step": 9822
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010701439258217574,
+      "loss": 0.0861,
+      "step": 9823
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010698669554571173,
+      "loss": 0.1051,
+      "step": 9824
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010695900010706584,
+      "loss": 0.1471,
+      "step": 9825
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010693130626726686,
+      "loss": 0.0585,
+      "step": 9826
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010690361402734365,
+      "loss": 0.1421,
+      "step": 9827
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010687592338832472,
+      "loss": 0.0974,
+      "step": 9828
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010684823435123878,
+      "loss": 0.0686,
+      "step": 9829
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010682054691711438,
+      "loss": 0.1139,
+      "step": 9830
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010679286108698006,
+      "loss": 0.0127,
+      "step": 9831
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010676517686186426,
+      "loss": 0.0814,
+      "step": 9832
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010673749424279534,
+      "loss": 0.125,
+      "step": 9833
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010670981323080163,
+      "loss": 0.0979,
+      "step": 9834
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010668213382691139,
+      "loss": 0.0858,
+      "step": 9835
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010665445603215289,
+      "loss": 0.1403,
+      "step": 9836
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010662677984755424,
+      "loss": 0.1151,
+      "step": 9837
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010659910527414352,
+      "loss": 0.1019,
+      "step": 9838
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010657143231294875,
+      "loss": 0.1006,
+      "step": 9839
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010654376096499794,
+      "loss": 0.0527,
+      "step": 9840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010651609123131896,
+      "loss": 0.1036,
+      "step": 9841
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010648842311293972,
+      "loss": 0.0484,
+      "step": 9842
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010646075661088794,
+      "loss": 0.0366,
+      "step": 9843
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010643309172619134,
+      "loss": 0.0849,
+      "step": 9844
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010640542845987764,
+      "loss": 0.1308,
+      "step": 9845
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010637776681297443,
+      "loss": 0.0648,
+      "step": 9846
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010635010678650929,
+      "loss": 0.0457,
+      "step": 9847
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010632244838150964,
+      "loss": 0.0454,
+      "step": 9848
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010629479159900298,
+      "loss": 0.0691,
+      "step": 9849
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010626713644001661,
+      "loss": 0.1018,
+      "step": 9850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001062394829055779,
+      "loss": 0.0834,
+      "step": 9851
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010621183099671407,
+      "loss": 0.0651,
+      "step": 9852
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010618418071445238,
+      "loss": 0.1154,
+      "step": 9853
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010615653205981978,
+      "loss": 0.09,
+      "step": 9854
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010612888503384348,
+      "loss": 0.0387,
+      "step": 9855
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010610123963755044,
+      "loss": 0.0741,
+      "step": 9856
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010607359587196764,
+      "loss": 0.0764,
+      "step": 9857
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010604595373812199,
+      "loss": 0.1064,
+      "step": 9858
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001060183132370402,
+      "loss": 0.104,
+      "step": 9859
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010599067436974908,
+      "loss": 0.0469,
+      "step": 9860
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010596303713727538,
+      "loss": 0.0536,
+      "step": 9861
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010593540154064571,
+      "loss": 0.1591,
+      "step": 9862
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010590776758088668,
+      "loss": 0.0582,
+      "step": 9863
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010588013525902473,
+      "loss": 0.0677,
+      "step": 9864
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010585250457608641,
+      "loss": 0.1071,
+      "step": 9865
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010582487553309807,
+      "loss": 0.035,
+      "step": 9866
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010579724813108605,
+      "loss": 0.0788,
+      "step": 9867
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010576962237107674,
+      "loss": 0.1164,
+      "step": 9868
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010574199825409614,
+      "loss": 0.0507,
+      "step": 9869
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010571437578117054,
+      "loss": 0.1154,
+      "step": 9870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010568675495332604,
+      "loss": 0.0903,
+      "step": 9871
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001056591357715886,
+      "loss": 0.1689,
+      "step": 9872
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010563151823698431,
+      "loss": 0.0416,
+      "step": 9873
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010560390235053896,
+      "loss": 0.0869,
+      "step": 9874
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010557628811327843,
+      "loss": 0.1278,
+      "step": 9875
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010554867552622854,
+      "loss": 0.0748,
+      "step": 9876
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010552106459041502,
+      "loss": 0.0948,
+      "step": 9877
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010549345530686353,
+      "loss": 0.0832,
+      "step": 9878
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010546584767659962,
+      "loss": 0.1374,
+      "step": 9879
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010543824170064888,
+      "loss": 0.0728,
+      "step": 9880
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010541063738003679,
+      "loss": 0.111,
+      "step": 9881
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010538303471578877,
+      "loss": 0.0364,
+      "step": 9882
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010535543370893022,
+      "loss": 0.0594,
+      "step": 9883
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010532783436048633,
+      "loss": 0.0658,
+      "step": 9884
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010530023667148239,
+      "loss": 0.079,
+      "step": 9885
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010527264064294359,
+      "loss": 0.0157,
+      "step": 9886
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010524504627589503,
+      "loss": 0.02,
+      "step": 9887
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010521745357136178,
+      "loss": 0.1038,
+      "step": 9888
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010518986253036878,
+      "loss": 0.0819,
+      "step": 9889
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010516227315394097,
+      "loss": 0.087,
+      "step": 9890
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010513468544310324,
+      "loss": 0.0555,
+      "step": 9891
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010510709939888038,
+      "loss": 0.0266,
+      "step": 9892
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010507951502229718,
+      "loss": 0.0139,
+      "step": 9893
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010505193231437821,
+      "loss": 0.0246,
+      "step": 9894
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010502435127614815,
+      "loss": 0.0345,
+      "step": 9895
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010499677190863152,
+      "loss": 0.0588,
+      "step": 9896
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010496919421285288,
+      "loss": 0.0475,
+      "step": 9897
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010494161818983667,
+      "loss": 0.0604,
+      "step": 9898
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010491404384060716,
+      "loss": 0.0469,
+      "step": 9899
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001048864711661887,
+      "loss": 0.0385,
+      "step": 9900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010485890016760554,
+      "loss": 0.0455,
+      "step": 9901
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010483133084588187,
+      "loss": 0.0968,
+      "step": 9902
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010480376320204184,
+      "loss": 0.0316,
+      "step": 9903
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010477619723710943,
+      "loss": 0.0619,
+      "step": 9904
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010474863295210867,
+      "loss": 0.0084,
+      "step": 9905
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010472107034806352,
+      "loss": 0.0429,
+      "step": 9906
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010469350942599781,
+      "loss": 0.0202,
+      "step": 9907
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010466595018693541,
+      "loss": 0.0182,
+      "step": 9908
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010463839263189998,
+      "loss": 0.0663,
+      "step": 9909
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010461083676191525,
+      "loss": 0.0354,
+      "step": 9910
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010458328257800482,
+      "loss": 0.0132,
+      "step": 9911
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010455573008119227,
+      "loss": 0.023,
+      "step": 9912
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001045281792725011,
+      "loss": 0.0411,
+      "step": 9913
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010450063015295468,
+      "loss": 0.059,
+      "step": 9914
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010447308272357646,
+      "loss": 0.0596,
+      "step": 9915
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010444553698538969,
+      "loss": 0.0656,
+      "step": 9916
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010441799293941762,
+      "loss": 0.0514,
+      "step": 9917
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010439045058668347,
+      "loss": 0.0471,
+      "step": 9918
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010436290992821035,
+      "loss": 0.0562,
+      "step": 9919
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010433537096502126,
+      "loss": 0.0308,
+      "step": 9920
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010430783369813923,
+      "loss": 0.0366,
+      "step": 9921
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001042802981285872,
+      "loss": 0.0294,
+      "step": 9922
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010425276425738802,
+      "loss": 0.0456,
+      "step": 9923
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010422523208556455,
+      "loss": 0.037,
+      "step": 9924
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001041977016141394,
+      "loss": 0.0706,
+      "step": 9925
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010417017284413534,
+      "loss": 0.0646,
+      "step": 9926
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010414264577657498,
+      "loss": 0.0563,
+      "step": 9927
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010411512041248084,
+      "loss": 0.0561,
+      "step": 9928
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010408759675287544,
+      "loss": 0.0397,
+      "step": 9929
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010406007479878116,
+      "loss": 0.0636,
+      "step": 9930
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010403255455122037,
+      "loss": 0.0172,
+      "step": 9931
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010400503601121539,
+      "loss": 0.0397,
+      "step": 9932
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010397751917978843,
+      "loss": 0.0482,
+      "step": 9933
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010395000405796176,
+      "loss": 0.0266,
+      "step": 9934
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010392249064675734,
+      "loss": 0.0234,
+      "step": 9935
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010389497894719724,
+      "loss": 0.0366,
+      "step": 9936
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010386746896030347,
+      "loss": 0.0503,
+      "step": 9937
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010383996068709794,
+      "loss": 0.0258,
+      "step": 9938
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010381245412860262,
+      "loss": 0.0374,
+      "step": 9939
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010378494928583908,
+      "loss": 0.0235,
+      "step": 9940
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010375744615982916,
+      "loss": 0.046,
+      "step": 9941
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010372994475159452,
+      "loss": 0.049,
+      "step": 9942
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010370244506215673,
+      "loss": 0.1063,
+      "step": 9943
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010367494709253738,
+      "loss": 0.0537,
+      "step": 9944
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001036474508437579,
+      "loss": 0.0539,
+      "step": 9945
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010361995631683965,
+      "loss": 0.0579,
+      "step": 9946
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010359246351280402,
+      "loss": 0.0288,
+      "step": 9947
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010356497243267228,
+      "loss": 0.0364,
+      "step": 9948
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010353748307746572,
+      "loss": 0.0579,
+      "step": 9949
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010350999544820535,
+      "loss": 0.0378,
+      "step": 9950
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010348250954591232,
+      "loss": 0.057,
+      "step": 9951
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010345502537160765,
+      "loss": 0.0235,
+      "step": 9952
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010342754292631228,
+      "loss": 0.0394,
+      "step": 9953
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010340006221104718,
+      "loss": 0.0487,
+      "step": 9954
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010337258322683305,
+      "loss": 0.0213,
+      "step": 9955
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010334510597469073,
+      "loss": 0.0655,
+      "step": 9956
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010331763045564091,
+      "loss": 0.0506,
+      "step": 9957
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010329015667070424,
+      "loss": 0.0472,
+      "step": 9958
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001032626846209013,
+      "loss": 0.0731,
+      "step": 9959
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010323521430725251,
+      "loss": 0.0355,
+      "step": 9960
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010320774573077839,
+      "loss": 0.0671,
+      "step": 9961
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001031802788924993,
+      "loss": 0.0461,
+      "step": 9962
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010315281379343554,
+      "loss": 0.0507,
+      "step": 9963
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010312535043460745,
+      "loss": 0.0551,
+      "step": 9964
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010309788881703507,
+      "loss": 0.0377,
+      "step": 9965
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010307042894173857,
+      "loss": 0.03,
+      "step": 9966
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010304297080973801,
+      "loss": 0.0462,
+      "step": 9967
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001030155144220534,
+      "loss": 0.0359,
+      "step": 9968
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010298805977970468,
+      "loss": 0.0314,
+      "step": 9969
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010296060688371165,
+      "loss": 0.0599,
+      "step": 9970
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010293315573509412,
+      "loss": 0.0605,
+      "step": 9971
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010290570633487182,
+      "loss": 0.0275,
+      "step": 9972
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010287825868406444,
+      "loss": 0.0987,
+      "step": 9973
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010285081278369163,
+      "loss": 0.0286,
+      "step": 9974
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010282336863477281,
+      "loss": 0.0165,
+      "step": 9975
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001027959262383275,
+      "loss": 0.0138,
+      "step": 9976
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001027684855953751,
+      "loss": 0.0561,
+      "step": 9977
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001027410467069349,
+      "loss": 0.0522,
+      "step": 9978
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010271360957402628,
+      "loss": 0.0679,
+      "step": 9979
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010268617419766847,
+      "loss": 0.0525,
+      "step": 9980
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010265874057888045,
+      "loss": 0.0612,
+      "step": 9981
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010263130871868142,
+      "loss": 0.0304,
+      "step": 9982
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010260387861809034,
+      "loss": 0.0369,
+      "step": 9983
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010257645027812622,
+      "loss": 0.0508,
+      "step": 9984
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001025490236998079,
+      "loss": 0.0187,
+      "step": 9985
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010252159888415417,
+      "loss": 0.0398,
+      "step": 9986
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010249417583218384,
+      "loss": 0.0249,
+      "step": 9987
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010246675454491554,
+      "loss": 0.0441,
+      "step": 9988
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010243933502336792,
+      "loss": 0.0255,
+      "step": 9989
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010241191726855962,
+      "loss": 0.0259,
+      "step": 9990
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010238450128150895,
+      "loss": 0.0358,
+      "step": 9991
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010235708706323444,
+      "loss": 0.0317,
+      "step": 9992
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010232967461475443,
+      "loss": 0.0206,
+      "step": 9993
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010230226393708722,
+      "loss": 0.0351,
+      "step": 9994
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010227485503125107,
+      "loss": 0.0539,
+      "step": 9995
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010224744789826406,
+      "loss": 0.0538,
+      "step": 9996
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010222004253914433,
+      "loss": 0.0565,
+      "step": 9997
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010219263895490992,
+      "loss": 0.0585,
+      "step": 9998
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010216523714657878,
+      "loss": 0.071,
+      "step": 9999
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010213783711516882,
+      "loss": 0.083,
+      "step": 10000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010211043886169782,
+      "loss": 0.0698,
+      "step": 10001
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010208304238718358,
+      "loss": 0.0224,
+      "step": 10002
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010205564769264383,
+      "loss": 0.0354,
+      "step": 10003
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010202825477909616,
+      "loss": 0.0607,
+      "step": 10004
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010200086364755819,
+      "loss": 0.0477,
+      "step": 10005
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010197347429904734,
+      "loss": 0.039,
+      "step": 10006
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010194608673458107,
+      "loss": 0.0508,
+      "step": 10007
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010191870095517676,
+      "loss": 0.0688,
+      "step": 10008
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010189131696185172,
+      "loss": 0.0405,
+      "step": 10009
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001018639347556232,
+      "loss": 0.0525,
+      "step": 10010
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010183655433750834,
+      "loss": 0.035,
+      "step": 10011
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010180917570852423,
+      "loss": 0.0406,
+      "step": 10012
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010178179886968791,
+      "loss": 0.0278,
+      "step": 10013
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010175442382201638,
+      "loss": 0.041,
+      "step": 10014
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010172705056652659,
+      "loss": 0.0217,
+      "step": 10015
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010169967910423524,
+      "loss": 0.0481,
+      "step": 10016
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010167230943615918,
+      "loss": 0.0504,
+      "step": 10017
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001016449415633151,
+      "loss": 0.0625,
+      "step": 10018
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010161757548671961,
+      "loss": 0.0399,
+      "step": 10019
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010159021120738941,
+      "loss": 0.0155,
+      "step": 10020
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010156284872634082,
+      "loss": 0.0124,
+      "step": 10021
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010153548804459037,
+      "loss": 0.0555,
+      "step": 10022
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010150812916315443,
+      "loss": 0.0373,
+      "step": 10023
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001014807720830493,
+      "loss": 0.0357,
+      "step": 10024
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010145341680529121,
+      "loss": 0.0465,
+      "step": 10025
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010142606333089631,
+      "loss": 0.0417,
+      "step": 10026
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010139871166088069,
+      "loss": 0.0194,
+      "step": 10027
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010137136179626045,
+      "loss": 0.0392,
+      "step": 10028
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010134401373805148,
+      "loss": 0.0127,
+      "step": 10029
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001013166674872698,
+      "loss": 0.0215,
+      "step": 10030
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001012893230449311,
+      "loss": 0.0366,
+      "step": 10031
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010126198041205118,
+      "loss": 0.0295,
+      "step": 10032
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001012346395896458,
+      "loss": 0.0336,
+      "step": 10033
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010120730057873053,
+      "loss": 0.017,
+      "step": 10034
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010117996338032102,
+      "loss": 0.0349,
+      "step": 10035
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010115262799543264,
+      "loss": 0.0289,
+      "step": 10036
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010112529442508089,
+      "loss": 0.0543,
+      "step": 10037
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010109796267028115,
+      "loss": 0.0723,
+      "step": 10038
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010107063273204869,
+      "loss": 0.0158,
+      "step": 10039
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010104330461139873,
+      "loss": 0.0326,
+      "step": 10040
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001010159783093465,
+      "loss": 0.0537,
+      "step": 10041
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010098865382690695,
+      "loss": 0.0606,
+      "step": 10042
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010096133116509522,
+      "loss": 0.0554,
+      "step": 10043
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010093401032492621,
+      "loss": 0.0733,
+      "step": 10044
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010090669130741485,
+      "loss": 0.0458,
+      "step": 10045
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010087937411357598,
+      "loss": 0.0437,
+      "step": 10046
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010085205874442429,
+      "loss": 0.0625,
+      "step": 10047
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010082474520097446,
+      "loss": 0.0599,
+      "step": 10048
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010079743348424117,
+      "loss": 0.0331,
+      "step": 10049
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010077012359523893,
+      "loss": 0.051,
+      "step": 10050
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010074281553498224,
+      "loss": 0.0669,
+      "step": 10051
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010071550930448551,
+      "loss": 0.0517,
+      "step": 10052
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010068820490476307,
+      "loss": 0.0938,
+      "step": 10053
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010066090233682921,
+      "loss": 0.033,
+      "step": 10054
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010063360160169814,
+      "loss": 0.0208,
+      "step": 10055
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001006063027003841,
+      "loss": 0.0426,
+      "step": 10056
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010057900563390098,
+      "loss": 0.0461,
+      "step": 10057
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010055171040326286,
+      "loss": 0.0243,
+      "step": 10058
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010052441700948371,
+      "loss": 0.0342,
+      "step": 10059
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010049712545357739,
+      "loss": 0.0653,
+      "step": 10060
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001004698357365577,
+      "loss": 0.087,
+      "step": 10061
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010044254785943836,
+      "loss": 0.066,
+      "step": 10062
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010041526182323302,
+      "loss": 0.0426,
+      "step": 10063
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010038797762895531,
+      "loss": 0.0641,
+      "step": 10064
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010036069527761874,
+      "loss": 0.0245,
+      "step": 10065
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010033341477023681,
+      "loss": 0.054,
+      "step": 10066
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010030613610782283,
+      "loss": 0.0432,
+      "step": 10067
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001002788592913902,
+      "loss": 0.0601,
+      "step": 10068
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010025158432195211,
+      "loss": 0.0472,
+      "step": 10069
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001002243112005218,
+      "loss": 0.0247,
+      "step": 10070
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010019703992811241,
+      "loss": 0.0477,
+      "step": 10071
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010016977050573687,
+      "loss": 0.0321,
+      "step": 10072
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010014250293440825,
+      "loss": 0.0552,
+      "step": 10073
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010011523721513942,
+      "loss": 0.0681,
+      "step": 10074
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010008797334894325,
+      "loss": 0.0408,
+      "step": 10075
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010006071133683254,
+      "loss": 0.0463,
+      "step": 10076
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010003345117981993,
+      "loss": 0.0579,
+      "step": 10077
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010000619287891808,
+      "loss": 0.0453,
+      "step": 10078
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.997893643513955e-05,
+      "loss": 0.0478,
+      "step": 10079
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.995168184949685e-05,
+      "loss": 0.057,
+      "step": 10080
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.992442912300247e-05,
+      "loss": 0.0639,
+      "step": 10081
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.989717825666866e-05,
+      "loss": 0.0386,
+      "step": 10082
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.986992925150768e-05,
+      "loss": 0.0461,
+      "step": 10083
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.984268210853187e-05,
+      "loss": 0.037,
+      "step": 10084
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.981543682875334e-05,
+      "loss": 0.0309,
+      "step": 10085
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.978819341318423e-05,
+      "loss": 0.0375,
+      "step": 10086
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.976095186283643e-05,
+      "loss": 0.0406,
+      "step": 10087
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.973371217872195e-05,
+      "loss": 0.0082,
+      "step": 10088
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.970647436185264e-05,
+      "loss": 0.014,
+      "step": 10089
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.967923841324031e-05,
+      "loss": 0.0349,
+      "step": 10090
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.965200433389678e-05,
+      "loss": 0.017,
+      "step": 10091
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.962477212483359e-05,
+      "loss": 0.0735,
+      "step": 10092
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.959754178706239e-05,
+      "loss": 0.0704,
+      "step": 10093
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.957031332159471e-05,
+      "loss": 0.053,
+      "step": 10094
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.954308672944202e-05,
+      "loss": 0.0383,
+      "step": 10095
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.951586201161575e-05,
+      "loss": 0.0629,
+      "step": 10096
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.948863916912708e-05,
+      "loss": 0.0261,
+      "step": 10097
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.946141820298738e-05,
+      "loss": 0.0469,
+      "step": 10098
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.943419911420774e-05,
+      "loss": 0.0508,
+      "step": 10099
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.940698190379936e-05,
+      "loss": 0.0247,
+      "step": 10100
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.937976657277326e-05,
+      "loss": 0.0573,
+      "step": 10101
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.935255312214035e-05,
+      "loss": 0.025,
+      "step": 10102
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.932534155291159e-05,
+      "loss": 0.039,
+      "step": 10103
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.929813186609778e-05,
+      "loss": 0.0617,
+      "step": 10104
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.927092406270967e-05,
+      "loss": 0.0354,
+      "step": 10105
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.9243718143758e-05,
+      "loss": 0.048,
+      "step": 10106
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.921651411025336e-05,
+      "loss": 0.0476,
+      "step": 10107
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.918931196320629e-05,
+      "loss": 0.0418,
+      "step": 10108
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.916211170362727e-05,
+      "loss": 0.0406,
+      "step": 10109
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.913491333252673e-05,
+      "loss": 0.0439,
+      "step": 10110
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.910771685091497e-05,
+      "loss": 0.0405,
+      "step": 10111
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.908052225980237e-05,
+      "loss": 0.0494,
+      "step": 10112
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.9053329560199e-05,
+      "loss": 0.0833,
+      "step": 10113
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.902613875311502e-05,
+      "loss": 0.0459,
+      "step": 10114
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.899894983956051e-05,
+      "loss": 0.0534,
+      "step": 10115
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.897176282054546e-05,
+      "loss": 0.0473,
+      "step": 10116
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.894457769707982e-05,
+      "loss": 0.0169,
+      "step": 10117
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.891739447017334e-05,
+      "loss": 0.0383,
+      "step": 10118
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.889021314083587e-05,
+      "loss": 0.0525,
+      "step": 10119
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.886303371007711e-05,
+      "loss": 0.0546,
+      "step": 10120
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.883585617890669e-05,
+      "loss": 0.0238,
+      "step": 10121
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.880868054833425e-05,
+      "loss": 0.0715,
+      "step": 10122
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.878150681936912e-05,
+      "loss": 0.0634,
+      "step": 10123
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.875433499302079e-05,
+      "loss": 0.0312,
+      "step": 10124
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.872716507029869e-05,
+      "loss": 0.0412,
+      "step": 10125
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.869999705221203e-05,
+      "loss": 0.0705,
+      "step": 10126
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.867283093977011e-05,
+      "loss": 0.0533,
+      "step": 10127
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.864566673398194e-05,
+      "loss": 0.0947,
+      "step": 10128
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.861850443585666e-05,
+      "loss": 0.0598,
+      "step": 10129
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.859134404640327e-05,
+      "loss": 0.019,
+      "step": 10130
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.856418556663068e-05,
+      "loss": 0.0447,
+      "step": 10131
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.853702899754781e-05,
+      "loss": 0.0604,
+      "step": 10132
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.850987434016337e-05,
+      "loss": 0.0814,
+      "step": 10133
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.848272159548608e-05,
+      "loss": 0.0703,
+      "step": 10134
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.845557076452461e-05,
+      "loss": 0.0376,
+      "step": 10135
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.842842184828757e-05,
+      "loss": 0.042,
+      "step": 10136
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.840127484778346e-05,
+      "loss": 0.0805,
+      "step": 10137
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.837412976402062e-05,
+      "loss": 0.0405,
+      "step": 10138
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.834698659800748e-05,
+      "loss": 0.01,
+      "step": 10139
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.83198453507523e-05,
+      "loss": 0.034,
+      "step": 10140
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.829270602326332e-05,
+      "loss": 0.0681,
+      "step": 10141
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.826556861654871e-05,
+      "loss": 0.0906,
+      "step": 10142
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.823843313161649e-05,
+      "loss": 0.027,
+      "step": 10143
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.821129956947468e-05,
+      "loss": 0.0163,
+      "step": 10144
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.818416793113123e-05,
+      "loss": 0.0349,
+      "step": 10145
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.8157038217594e-05,
+      "loss": 0.0794,
+      "step": 10146
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.81299104298708e-05,
+      "loss": 0.0497,
+      "step": 10147
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.810278456896927e-05,
+      "loss": 0.0315,
+      "step": 10148
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.807566063589711e-05,
+      "loss": 0.0208,
+      "step": 10149
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.80485386316619e-05,
+      "loss": 0.0455,
+      "step": 10150
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.802141855727113e-05,
+      "loss": 0.0293,
+      "step": 10151
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.799430041373231e-05,
+      "loss": 0.0485,
+      "step": 10152
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.796718420205265e-05,
+      "loss": 0.0448,
+      "step": 10153
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.794006992323951e-05,
+      "loss": 0.0539,
+      "step": 10154
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.791295757830012e-05,
+      "loss": 0.0302,
+      "step": 10155
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.788584716824159e-05,
+      "loss": 0.0509,
+      "step": 10156
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.785873869407106e-05,
+      "loss": 0.0385,
+      "step": 10157
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.783163215679546e-05,
+      "loss": 0.0681,
+      "step": 10158
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.780452755742174e-05,
+      "loss": 0.0439,
+      "step": 10159
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.777742489695675e-05,
+      "loss": 0.0608,
+      "step": 10160
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.775032417640728e-05,
+      "loss": 0.0499,
+      "step": 10161
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.772322539678011e-05,
+      "loss": 0.0603,
+      "step": 10162
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.769612855908178e-05,
+      "loss": 0.0201,
+      "step": 10163
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.766903366431885e-05,
+      "loss": 0.061,
+      "step": 10164
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.76419407134979e-05,
+      "loss": 0.0393,
+      "step": 10165
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.761484970762526e-05,
+      "loss": 0.0191,
+      "step": 10166
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.758776064770736e-05,
+      "loss": 0.0342,
+      "step": 10167
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.756067353475055e-05,
+      "loss": 0.0322,
+      "step": 10168
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.753358836976086e-05,
+      "loss": 0.0609,
+      "step": 10169
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.75065051537445e-05,
+      "loss": 0.046,
+      "step": 10170
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.747942388770755e-05,
+      "loss": 0.0232,
+      "step": 10171
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.745234457265597e-05,
+      "loss": 0.0667,
+      "step": 10172
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.742526720959574e-05,
+      "loss": 0.043,
+      "step": 10173
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.739819179953261e-05,
+      "loss": 0.0174,
+      "step": 10174
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.737111834347242e-05,
+      "loss": 0.0533,
+      "step": 10175
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.734404684242085e-05,
+      "loss": 0.0434,
+      "step": 10176
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.731697729738352e-05,
+      "loss": 0.0211,
+      "step": 10177
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.728990970936606e-05,
+      "loss": 0.0496,
+      "step": 10178
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.726284407937381e-05,
+      "loss": 0.0369,
+      "step": 10179
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.723578040841226e-05,
+      "loss": 0.0508,
+      "step": 10180
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.720871869748673e-05,
+      "loss": 0.0477,
+      "step": 10181
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.71816589476025e-05,
+      "loss": 0.0311,
+      "step": 10182
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.715460115976478e-05,
+      "loss": 0.065,
+      "step": 10183
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.712754533497864e-05,
+      "loss": 0.0272,
+      "step": 10184
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.710049147424912e-05,
+      "loss": 0.0317,
+      "step": 10185
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.707343957858123e-05,
+      "loss": 0.0201,
+      "step": 10186
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.704638964897985e-05,
+      "loss": 0.0551,
+      "step": 10187
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.701934168644985e-05,
+      "loss": 0.0552,
+      "step": 10188
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.699229569199591e-05,
+      "loss": 0.0287,
+      "step": 10189
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.696525166662271e-05,
+      "loss": 0.0309,
+      "step": 10190
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.693820961133492e-05,
+      "loss": 0.0484,
+      "step": 10191
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.691116952713705e-05,
+      "loss": 0.0458,
+      "step": 10192
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.688413141503358e-05,
+      "loss": 0.0191,
+      "step": 10193
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.685709527602883e-05,
+      "loss": 0.0507,
+      "step": 10194
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.683006111112712e-05,
+      "loss": 0.0274,
+      "step": 10195
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.680302892133274e-05,
+      "loss": 0.0455,
+      "step": 10196
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.677599870764983e-05,
+      "loss": 0.0422,
+      "step": 10197
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.674897047108254e-05,
+      "loss": 0.0655,
+      "step": 10198
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.67219442126348e-05,
+      "loss": 0.0298,
+      "step": 10199
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.669491993331059e-05,
+      "loss": 0.0564,
+      "step": 10200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.666789763411378e-05,
+      "loss": 0.041,
+      "step": 10201
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.66408773160482e-05,
+      "loss": 0.0557,
+      "step": 10202
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.66138589801176e-05,
+      "loss": 0.1123,
+      "step": 10203
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.658684262732554e-05,
+      "loss": 0.0598,
+      "step": 10204
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.655982825867562e-05,
+      "loss": 0.0354,
+      "step": 10205
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.653281587517138e-05,
+      "loss": 0.0363,
+      "step": 10206
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.65058054778162e-05,
+      "loss": 0.0262,
+      "step": 10207
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.647879706761359e-05,
+      "loss": 0.0692,
+      "step": 10208
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.645179064556664e-05,
+      "loss": 0.0671,
+      "step": 10209
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.642478621267863e-05,
+      "loss": 0.0331,
+      "step": 10210
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.639778376995271e-05,
+      "loss": 0.065,
+      "step": 10211
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.637078331839195e-05,
+      "loss": 0.0308,
+      "step": 10212
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.634378485899935e-05,
+      "loss": 0.066,
+      "step": 10213
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.631678839277776e-05,
+      "loss": 0.0774,
+      "step": 10214
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.628979392073006e-05,
+      "loss": 0.0337,
+      "step": 10215
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.626280144385901e-05,
+      "loss": 0.0371,
+      "step": 10216
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.623581096316731e-05,
+      "loss": 0.0485,
+      "step": 10217
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.620882247965762e-05,
+      "loss": 0.0714,
+      "step": 10218
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.618183599433237e-05,
+      "loss": 0.0364,
+      "step": 10219
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.61548515081941e-05,
+      "loss": 0.0437,
+      "step": 10220
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.61278690222452e-05,
+      "loss": 0.0273,
+      "step": 10221
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.610088853748799e-05,
+      "loss": 0.0308,
+      "step": 10222
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.607391005492474e-05,
+      "loss": 0.0534,
+      "step": 10223
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.604693357555755e-05,
+      "loss": 0.0304,
+      "step": 10224
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.601995910038858e-05,
+      "loss": 0.0448,
+      "step": 10225
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.599298663041982e-05,
+      "loss": 0.0446,
+      "step": 10226
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.596601616665323e-05,
+      "loss": 0.024,
+      "step": 10227
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.593904771009072e-05,
+      "loss": 0.1044,
+      "step": 10228
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.591208126173406e-05,
+      "loss": 0.0765,
+      "step": 10229
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.588511682258493e-05,
+      "loss": 0.034,
+      "step": 10230
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.585815439364505e-05,
+      "loss": 0.0383,
+      "step": 10231
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.583119397591593e-05,
+      "loss": 0.0559,
+      "step": 10232
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.580423557039912e-05,
+      "loss": 0.0259,
+      "step": 10233
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.577727917809609e-05,
+      "loss": 0.0345,
+      "step": 10234
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.575032480000808e-05,
+      "loss": 0.0291,
+      "step": 10235
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.57233724371364e-05,
+      "loss": 0.0351,
+      "step": 10236
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.569642209048229e-05,
+      "loss": 0.0532,
+      "step": 10237
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.566947376104683e-05,
+      "loss": 0.0385,
+      "step": 10238
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.564252744983116e-05,
+      "loss": 0.0778,
+      "step": 10239
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.561558315783615e-05,
+      "loss": 0.0485,
+      "step": 10240
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.558864088606273e-05,
+      "loss": 0.0492,
+      "step": 10241
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.556170063551174e-05,
+      "loss": 0.0241,
+      "step": 10242
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.553476240718396e-05,
+      "loss": 0.0493,
+      "step": 10243
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.550782620208008e-05,
+      "loss": 0.0302,
+      "step": 10244
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.548089202120061e-05,
+      "loss": 0.0764,
+      "step": 10245
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.54539598655461e-05,
+      "loss": 0.0398,
+      "step": 10246
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.542702973611706e-05,
+      "loss": 0.0698,
+      "step": 10247
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.540010163391376e-05,
+      "loss": 0.015,
+      "step": 10248
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.537317555993669e-05,
+      "loss": 0.0534,
+      "step": 10249
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.53462515151859e-05,
+      "loss": 0.0204,
+      "step": 10250
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.53193295006616e-05,
+      "loss": 0.0735,
+      "step": 10251
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.529240951736384e-05,
+      "loss": 0.0465,
+      "step": 10252
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.526549156629262e-05,
+      "loss": 0.0253,
+      "step": 10253
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.523857564844796e-05,
+      "loss": 0.055,
+      "step": 10254
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.521166176482957e-05,
+      "loss": 0.0279,
+      "step": 10255
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.518474991643727e-05,
+      "loss": 0.0312,
+      "step": 10256
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.51578401042708e-05,
+      "loss": 0.0557,
+      "step": 10257
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.513093232932971e-05,
+      "loss": 0.0361,
+      "step": 10258
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.510402659261367e-05,
+      "loss": 0.0372,
+      "step": 10259
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.507712289512199e-05,
+      "loss": 0.0256,
+      "step": 10260
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.505022123785412e-05,
+      "loss": 0.0184,
+      "step": 10261
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.502332162180943e-05,
+      "loss": 0.0706,
+      "step": 10262
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.499642404798708e-05,
+      "loss": 0.0647,
+      "step": 10263
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.496952851738634e-05,
+      "loss": 0.0384,
+      "step": 10264
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.494263503100619e-05,
+      "loss": 0.046,
+      "step": 10265
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.491574358984571e-05,
+      "loss": 0.0483,
+      "step": 10266
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.488885419490382e-05,
+      "loss": 0.0116,
+      "step": 10267
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.486196684717938e-05,
+      "loss": 0.0546,
+      "step": 10268
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.483508154767124e-05,
+      "loss": 0.0557,
+      "step": 10269
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.480819829737798e-05,
+      "loss": 0.0127,
+      "step": 10270
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.47813170972983e-05,
+      "loss": 0.0677,
+      "step": 10271
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.475443794843078e-05,
+      "loss": 0.0248,
+      "step": 10272
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.47275608517739e-05,
+      "loss": 0.0336,
+      "step": 10273
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.470068580832609e-05,
+      "loss": 0.0495,
+      "step": 10274
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.467381281908556e-05,
+      "loss": 0.0439,
+      "step": 10275
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.464694188505065e-05,
+      "loss": 0.0092,
+      "step": 10276
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.462007300721957e-05,
+      "loss": 0.06,
+      "step": 10277
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.459320618659032e-05,
+      "loss": 0.043,
+      "step": 10278
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.456634142416105e-05,
+      "loss": 0.0481,
+      "step": 10279
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.45394787209296e-05,
+      "loss": 0.0455,
+      "step": 10280
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.451261807789387e-05,
+      "loss": 0.071,
+      "step": 10281
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.448575949605166e-05,
+      "loss": 0.0354,
+      "step": 10282
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.445890297640072e-05,
+      "loss": 0.0414,
+      "step": 10283
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.44320485199387e-05,
+      "loss": 0.0444,
+      "step": 10284
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.440519612766306e-05,
+      "loss": 0.0285,
+      "step": 10285
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.437834580057135e-05,
+      "loss": 0.0785,
+      "step": 10286
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.4351497539661e-05,
+      "loss": 0.0298,
+      "step": 10287
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.432465134592931e-05,
+      "loss": 0.0554,
+      "step": 10288
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.42978072203736e-05,
+      "loss": 0.0396,
+      "step": 10289
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.427096516399098e-05,
+      "loss": 0.0141,
+      "step": 10290
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.424412517777857e-05,
+      "loss": 0.044,
+      "step": 10291
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.421728726273343e-05,
+      "loss": 0.0254,
+      "step": 10292
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.419045141985246e-05,
+      "loss": 0.0398,
+      "step": 10293
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.416361765013258e-05,
+      "loss": 0.0297,
+      "step": 10294
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.413678595457056e-05,
+      "loss": 0.0722,
+      "step": 10295
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.410995633416312e-05,
+      "loss": 0.0404,
+      "step": 10296
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.408312878990691e-05,
+      "loss": 0.069,
+      "step": 10297
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.405630332279847e-05,
+      "loss": 0.0513,
+      "step": 10298
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.402947993383432e-05,
+      "loss": 0.054,
+      "step": 10299
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.400265862401095e-05,
+      "loss": 0.0283,
+      "step": 10300
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.39758393943245e-05,
+      "loss": 0.0626,
+      "step": 10301
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.394902224577134e-05,
+      "loss": 0.0704,
+      "step": 10302
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.392220717934762e-05,
+      "loss": 0.072,
+      "step": 10303
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.389539419604948e-05,
+      "loss": 0.092,
+      "step": 10304
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.386858329687294e-05,
+      "loss": 0.0266,
+      "step": 10305
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.38417744828139e-05,
+      "loss": 0.0417,
+      "step": 10306
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.381496775486826e-05,
+      "loss": 0.0448,
+      "step": 10307
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.37881631140318e-05,
+      "loss": 0.0001,
+      "step": 10308
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.376136056130024e-05,
+      "loss": 0.0217,
+      "step": 10309
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.373456009766926e-05,
+      "loss": 0.0354,
+      "step": 10310
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.370776172413435e-05,
+      "loss": 0.0301,
+      "step": 10311
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.368096544169096e-05,
+      "loss": 0.0354,
+      "step": 10312
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.365417125133459e-05,
+      "loss": 0.0309,
+      "step": 10313
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.362737915406053e-05,
+      "loss": 0.027,
+      "step": 10314
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.360058915086407e-05,
+      "loss": 0.0328,
+      "step": 10315
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.357380124274028e-05,
+      "loss": 0.0879,
+      "step": 10316
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.354701543068431e-05,
+      "loss": 0.0076,
+      "step": 10317
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.352023171569115e-05,
+      "loss": 0.0245,
+      "step": 10318
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.349345009875577e-05,
+      "loss": 0.0255,
+      "step": 10319
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.346667058087306e-05,
+      "loss": 0.0476,
+      "step": 10320
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.343989316303772e-05,
+      "loss": 0.013,
+      "step": 10321
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.341311784624448e-05,
+      "loss": 0.0583,
+      "step": 10322
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.338634463148796e-05,
+      "loss": 0.0381,
+      "step": 10323
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.335957351976273e-05,
+      "loss": 0.0529,
+      "step": 10324
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.33328045120633e-05,
+      "loss": 0.0457,
+      "step": 10325
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.330603760938395e-05,
+      "loss": 0.0385,
+      "step": 10326
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.327927281271906e-05,
+      "loss": 0.0352,
+      "step": 10327
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.325251012306284e-05,
+      "loss": 0.0674,
+      "step": 10328
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.322574954140948e-05,
+      "loss": 0.0684,
+      "step": 10329
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.319899106875304e-05,
+      "loss": 0.0423,
+      "step": 10330
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.31722347060875e-05,
+      "loss": 0.0366,
+      "step": 10331
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.314548045440679e-05,
+      "loss": 0.0173,
+      "step": 10332
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.311872831470475e-05,
+      "loss": 0.0198,
+      "step": 10333
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.309197828797517e-05,
+      "loss": 0.0315,
+      "step": 10334
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.306523037521171e-05,
+      "loss": 0.0711,
+      "step": 10335
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.303848457740798e-05,
+      "loss": 0.036,
+      "step": 10336
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.30117408955575e-05,
+      "loss": 0.0323,
+      "step": 10337
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.298499933065373e-05,
+      "loss": 0.0159,
+      "step": 10338
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.295825988369003e-05,
+      "loss": 0.0568,
+      "step": 10339
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.29315225556598e-05,
+      "loss": 0.1114,
+      "step": 10340
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.290478734755607e-05,
+      "loss": 0.0166,
+      "step": 10341
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.287805426037203e-05,
+      "loss": 0.0655,
+      "step": 10342
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.285132329510081e-05,
+      "loss": 0.0791,
+      "step": 10343
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.282459445273532e-05,
+      "loss": 0.0235,
+      "step": 10344
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.279786773426851e-05,
+      "loss": 0.056,
+      "step": 10345
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.277114314069316e-05,
+      "loss": 0.0226,
+      "step": 10346
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.2744420673002e-05,
+      "loss": 0.0519,
+      "step": 10347
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.271770033218772e-05,
+      "loss": 0.0556,
+      "step": 10348
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.269098211924289e-05,
+      "loss": 0.0167,
+      "step": 10349
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.266426603516007e-05,
+      "loss": 0.0725,
+      "step": 10350
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.26375520809316e-05,
+      "loss": 0.0306,
+      "step": 10351
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.261084025754983e-05,
+      "loss": 0.0281,
+      "step": 10352
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.258413056600701e-05,
+      "loss": 0.0482,
+      "step": 10353
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.255742300729543e-05,
+      "loss": 0.0587,
+      "step": 10354
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.253071758240712e-05,
+      "loss": 0.0965,
+      "step": 10355
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.250401429233418e-05,
+      "loss": 0.0625,
+      "step": 10356
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.247731313806846e-05,
+      "loss": 0.0727,
+      "step": 10357
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.245061412060188e-05,
+      "loss": 0.0481,
+      "step": 10358
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.24239172409262e-05,
+      "loss": 0.0334,
+      "step": 10359
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.239722250003316e-05,
+      "loss": 0.0396,
+      "step": 10360
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.23705298989144e-05,
+      "loss": 0.0193,
+      "step": 10361
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.234383943856146e-05,
+      "loss": 0.0501,
+      "step": 10362
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.231715111996579e-05,
+      "loss": 0.0469,
+      "step": 10363
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.22904649441188e-05,
+      "loss": 0.044,
+      "step": 10364
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.226378091201182e-05,
+      "loss": 0.0619,
+      "step": 10365
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.223709902463611e-05,
+      "loss": 0.0304,
+      "step": 10366
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.221041928298272e-05,
+      "loss": 0.061,
+      "step": 10367
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.218374168804282e-05,
+      "loss": 0.0319,
+      "step": 10368
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.215706624080736e-05,
+      "loss": 0.0319,
+      "step": 10369
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.213039294226724e-05,
+      "loss": 0.0423,
+      "step": 10370
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.210372179341338e-05,
+      "loss": 0.059,
+      "step": 10371
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.207705279523644e-05,
+      "loss": 0.052,
+      "step": 10372
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.205038594872712e-05,
+      "loss": 0.0338,
+      "step": 10373
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.202372125487602e-05,
+      "loss": 0.031,
+      "step": 10374
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.199705871467369e-05,
+      "loss": 0.0449,
+      "step": 10375
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.197039832911056e-05,
+      "loss": 0.0317,
+      "step": 10376
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.19437400991769e-05,
+      "loss": 0.0099,
+      "step": 10377
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.191708402586307e-05,
+      "loss": 0.0257,
+      "step": 10378
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.189043011015926e-05,
+      "loss": 0.0155,
+      "step": 10379
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.186377835305552e-05,
+      "loss": 0.0552,
+      "step": 10380
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.183712875554203e-05,
+      "loss": 0.0558,
+      "step": 10381
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.181048131860858e-05,
+      "loss": 0.0169,
+      "step": 10382
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.178383604324509e-05,
+      "loss": 0.0118,
+      "step": 10383
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.175719293044138e-05,
+      "loss": 0.0509,
+      "step": 10384
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.173055198118718e-05,
+      "loss": 0.0132,
+      "step": 10385
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.170391319647211e-05,
+      "loss": 0.0345,
+      "step": 10386
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.167727657728567e-05,
+      "loss": 0.0511,
+      "step": 10387
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.16506421246174e-05,
+      "loss": 0.0899,
+      "step": 10388
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.162400983945667e-05,
+      "loss": 0.0155,
+      "step": 10389
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.159737972279278e-05,
+      "loss": 0.0321,
+      "step": 10390
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.157075177561502e-05,
+      "loss": 0.0222,
+      "step": 10391
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.154412599891244e-05,
+      "loss": 0.029,
+      "step": 10392
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.151750239367415e-05,
+      "loss": 0.044,
+      "step": 10393
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.149088096088911e-05,
+      "loss": 0.042,
+      "step": 10394
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.146426170154632e-05,
+      "loss": 0.0618,
+      "step": 10395
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.143764461663459e-05,
+      "loss": 0.0648,
+      "step": 10396
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.141102970714259e-05,
+      "loss": 0.0529,
+      "step": 10397
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.138441697405903e-05,
+      "loss": 0.02,
+      "step": 10398
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.135780641837247e-05,
+      "loss": 0.0286,
+      "step": 10399
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.133119804107144e-05,
+      "loss": 0.0397,
+      "step": 10400
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.130459184314438e-05,
+      "loss": 0.0427,
+      "step": 10401
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.12779878255796e-05,
+      "loss": 0.0212,
+      "step": 10402
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.125138598936535e-05,
+      "loss": 0.0277,
+      "step": 10403
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.122478633548985e-05,
+      "loss": 0.0431,
+      "step": 10404
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.119818886494117e-05,
+      "loss": 0.0832,
+      "step": 10405
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.11715935787074e-05,
+      "loss": 0.0631,
+      "step": 10406
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.114500047777637e-05,
+      "loss": 0.0199,
+      "step": 10407
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.111840956313596e-05,
+      "loss": 0.036,
+      "step": 10408
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.109182083577397e-05,
+      "loss": 0.0655,
+      "step": 10409
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.10652342966781e-05,
+      "loss": 0.0118,
+      "step": 10410
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.103864994683597e-05,
+      "loss": 0.067,
+      "step": 10411
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.101206778723507e-05,
+      "loss": 0.0305,
+      "step": 10412
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.098548781886287e-05,
+      "loss": 0.0532,
+      "step": 10413
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.095891004270674e-05,
+      "loss": 0.0574,
+      "step": 10414
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.093233445975396e-05,
+      "loss": 0.0384,
+      "step": 10415
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.090576107099174e-05,
+      "loss": 0.0534,
+      "step": 10416
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.087918987740724e-05,
+      "loss": 0.0478,
+      "step": 10417
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.085262087998743e-05,
+      "loss": 0.0193,
+      "step": 10418
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.08260540797193e-05,
+      "loss": 0.0217,
+      "step": 10419
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.079948947758974e-05,
+      "loss": 0.038,
+      "step": 10420
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.077292707458555e-05,
+      "loss": 0.0173,
+      "step": 10421
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.07463668716935e-05,
+      "loss": 0.0203,
+      "step": 10422
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.071980886990008e-05,
+      "loss": 0.0203,
+      "step": 10423
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.069325307019195e-05,
+      "loss": 0.0589,
+      "step": 10424
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.066669947355554e-05,
+      "loss": 0.0455,
+      "step": 10425
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.064014808097725e-05,
+      "loss": 0.073,
+      "step": 10426
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.061359889344342e-05,
+      "loss": 0.0205,
+      "step": 10427
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.05870519119402e-05,
+      "loss": 0.0526,
+      "step": 10428
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.056050713745378e-05,
+      "loss": 0.0267,
+      "step": 10429
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.053396457097022e-05,
+      "loss": 0.0418,
+      "step": 10430
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.05074242134755e-05,
+      "loss": 0.0533,
+      "step": 10431
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.048088606595555e-05,
+      "loss": 0.0608,
+      "step": 10432
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.04543501293961e-05,
+      "loss": 0.0828,
+      "step": 10433
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.042781640478291e-05,
+      "loss": 0.056,
+      "step": 10434
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.040128489310164e-05,
+      "loss": 0.0396,
+      "step": 10435
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.037475559533788e-05,
+      "loss": 0.0424,
+      "step": 10436
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.034822851247717e-05,
+      "loss": 0.0337,
+      "step": 10437
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.032170364550477e-05,
+      "loss": 0.0176,
+      "step": 10438
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.029518099540608e-05,
+      "loss": 0.0127,
+      "step": 10439
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.026866056316635e-05,
+      "loss": 0.0156,
+      "step": 10440
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.024214234977069e-05,
+      "loss": 0.044,
+      "step": 10441
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.021562635620425e-05,
+      "loss": 0.0364,
+      "step": 10442
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.018911258345193e-05,
+      "loss": 0.0359,
+      "step": 10443
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.016260103249871e-05,
+      "loss": 0.0227,
+      "step": 10444
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.013609170432937e-05,
+      "loss": 0.0388,
+      "step": 10445
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.010958459992869e-05,
+      "loss": 0.0301,
+      "step": 10446
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.008307972028136e-05,
+      "loss": 0.0467,
+      "step": 10447
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.005657706637187e-05,
+      "loss": 0.0411,
+      "step": 10448
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.003007663918473e-05,
+      "loss": 0.0605,
+      "step": 10449
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.000357843970439e-05,
+      "loss": 0.0968,
+      "step": 10450
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.997708246891518e-05,
+      "loss": 0.0561,
+      "step": 10451
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.995058872780139e-05,
+      "loss": 0.0712,
+      "step": 10452
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.992409721734708e-05,
+      "loss": 0.0589,
+      "step": 10453
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.989760793853641e-05,
+      "loss": 0.0462,
+      "step": 10454
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.987112089235333e-05,
+      "loss": 0.0498,
+      "step": 10455
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 8.98446360797818e-05,
+      "loss": 0.0486,
+      "step": 10456
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.981815350180568e-05,
+      "loss": 0.0586,
+      "step": 10457
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.97916731594086e-05,
+      "loss": 0.0556,
+      "step": 10458
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.976519505357432e-05,
+      "loss": 0.0449,
+      "step": 10459
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.973871918528642e-05,
+      "loss": 0.0518,
+      "step": 10460
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.97122455555284e-05,
+      "loss": 0.0768,
+      "step": 10461
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.968577416528368e-05,
+      "loss": 0.0497,
+      "step": 10462
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.965930501553556e-05,
+      "loss": 0.0599,
+      "step": 10463
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.963283810726731e-05,
+      "loss": 0.0379,
+      "step": 10464
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.960637344146208e-05,
+      "loss": 0.0606,
+      "step": 10465
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.957991101910297e-05,
+      "loss": 0.0552,
+      "step": 10466
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.955345084117303e-05,
+      "loss": 0.0431,
+      "step": 10467
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.95269929086551e-05,
+      "loss": 0.0554,
+      "step": 10468
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.950053722253205e-05,
+      "loss": 0.0224,
+      "step": 10469
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.947408378378663e-05,
+      "loss": 0.0491,
+      "step": 10470
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.94476325934015e-05,
+      "loss": 0.0499,
+      "step": 10471
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.942118365235931e-05,
+      "loss": 0.0376,
+      "step": 10472
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.939473696164245e-05,
+      "loss": 0.0506,
+      "step": 10473
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.936829252223338e-05,
+      "loss": 0.0184,
+      "step": 10474
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.934185033511444e-05,
+      "loss": 0.0206,
+      "step": 10475
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.931541040126788e-05,
+      "loss": 0.0462,
+      "step": 10476
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.928897272167592e-05,
+      "loss": 0.0268,
+      "step": 10477
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.926253729732054e-05,
+      "loss": 0.0706,
+      "step": 10478
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.92361041291838e-05,
+      "loss": 0.0006,
+      "step": 10479
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.920967321824761e-05,
+      "loss": 0.0174,
+      "step": 10480
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.918324456549377e-05,
+      "loss": 0.0362,
+      "step": 10481
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.915681817190404e-05,
+      "loss": 0.0365,
+      "step": 10482
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.913039403846017e-05,
+      "loss": 0.0784,
+      "step": 10483
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.910397216614358e-05,
+      "loss": 0.0652,
+      "step": 10484
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.907755255593589e-05,
+      "loss": 0.0414,
+      "step": 10485
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.905113520881846e-05,
+      "loss": 0.0381,
+      "step": 10486
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.90247201257726e-05,
+      "loss": 0.0596,
+      "step": 10487
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.899830730777966e-05,
+      "loss": 0.0304,
+      "step": 10488
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 8.897189675582066e-05,
+      "loss": 0.055,
+      "step": 10489
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.894548847087674e-05,
+      "loss": 0.0341,
+      "step": 10490
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.891908245392886e-05,
+      "loss": 0.0715,
+      "step": 10491
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.889267870595796e-05,
+      "loss": 0.0291,
+      "step": 10492
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.886627722794488e-05,
+      "loss": 0.0247,
+      "step": 10493
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.88398780208703e-05,
+      "loss": 0.0544,
+      "step": 10494
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.881348108571489e-05,
+      "loss": 0.0484,
+      "step": 10495
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.878708642345924e-05,
+      "loss": 0.0424,
+      "step": 10496
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.876069403508381e-05,
+      "loss": 0.0666,
+      "step": 10497
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.873430392156909e-05,
+      "loss": 0.0467,
+      "step": 10498
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.870791608389524e-05,
+      "loss": 0.0624,
+      "step": 10499
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.868153052304257e-05,
+      "loss": 0.0294,
+      "step": 10500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.865514723999125e-05,
+      "loss": 0.0529,
+      "step": 10501
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.86287662357213e-05,
+      "loss": 0.0475,
+      "step": 10502
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.860238751121279e-05,
+      "loss": 0.044,
+      "step": 10503
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.857601106744546e-05,
+      "loss": 0.0438,
+      "step": 10504
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.854963690539921e-05,
+      "loss": 0.0299,
+      "step": 10505
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.852326502605373e-05,
+      "loss": 0.0491,
+      "step": 10506
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.849689543038869e-05,
+      "loss": 0.0298,
+      "step": 10507
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.847052811938367e-05,
+      "loss": 0.0727,
+      "step": 10508
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.844416309401805e-05,
+      "loss": 0.0394,
+      "step": 10509
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.841780035527125e-05,
+      "loss": 0.0281,
+      "step": 10510
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.839143990412258e-05,
+      "loss": 0.0374,
+      "step": 10511
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.836508174155128e-05,
+      "loss": 0.0979,
+      "step": 10512
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.833872586853648e-05,
+      "loss": 0.0379,
+      "step": 10513
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.831237228605716e-05,
+      "loss": 0.0807,
+      "step": 10514
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.82860209950923e-05,
+      "loss": 0.0693,
+      "step": 10515
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.825967199662077e-05,
+      "loss": 0.0398,
+      "step": 10516
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.823332529162139e-05,
+      "loss": 0.0633,
+      "step": 10517
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.820698088107287e-05,
+      "loss": 0.0644,
+      "step": 10518
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.818063876595379e-05,
+      "loss": 0.0254,
+      "step": 10519
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.815429894724268e-05,
+      "loss": 0.0335,
+      "step": 10520
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.812796142591801e-05,
+      "loss": 0.02,
+      "step": 10521
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 8.810162620295814e-05,
+      "loss": 0.0252,
+      "step": 10522
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.807529327934136e-05,
+      "loss": 0.0422,
+      "step": 10523
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.804896265604581e-05,
+      "loss": 0.0441,
+      "step": 10524
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.802263433404962e-05,
+      "loss": 0.0496,
+      "step": 10525
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.799630831433083e-05,
+      "loss": 0.0554,
+      "step": 10526
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.796998459786736e-05,
+      "loss": 0.0267,
+      "step": 10527
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.794366318563712e-05,
+      "loss": 0.0469,
+      "step": 10528
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.791734407861776e-05,
+      "loss": 0.1201,
+      "step": 10529
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.7891027277787e-05,
+      "loss": 0.0722,
+      "step": 10530
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.786471278412245e-05,
+      "loss": 0.0192,
+      "step": 10531
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.783840059860162e-05,
+      "loss": 0.0331,
+      "step": 10532
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.781209072220195e-05,
+      "loss": 0.0373,
+      "step": 10533
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.778578315590071e-05,
+      "loss": 0.0252,
+      "step": 10534
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.775947790067518e-05,
+      "loss": 0.0439,
+      "step": 10535
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.773317495750253e-05,
+      "loss": 0.0661,
+      "step": 10536
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.770687432735981e-05,
+      "loss": 0.0139,
+      "step": 10537
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.768057601122412e-05,
+      "loss": 0.0424,
+      "step": 10538
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.765428001007222e-05,
+      "loss": 0.0387,
+      "step": 10539
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.762798632488096e-05,
+      "loss": 0.0217,
+      "step": 10540
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.760169495662713e-05,
+      "loss": 0.0669,
+      "step": 10541
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.757540590628736e-05,
+      "loss": 0.0344,
+      "step": 10542
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.754911917483824e-05,
+      "loss": 0.0319,
+      "step": 10543
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.752283476325618e-05,
+      "loss": 0.0623,
+      "step": 10544
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.749655267251754e-05,
+      "loss": 0.0663,
+      "step": 10545
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.747027290359875e-05,
+      "loss": 0.0232,
+      "step": 10546
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.744399545747589e-05,
+      "loss": 0.0555,
+      "step": 10547
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.741772033512521e-05,
+      "loss": 0.0397,
+      "step": 10548
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.739144753752271e-05,
+      "loss": 0.072,
+      "step": 10549
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.736517706564435e-05,
+      "loss": 0.0396,
+      "step": 10550
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.733890892046589e-05,
+      "loss": 0.046,
+      "step": 10551
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.731264310296329e-05,
+      "loss": 0.0128,
+      "step": 10552
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.728637961411211e-05,
+      "loss": 0.0402,
+      "step": 10553
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.726011845488807e-05,
+      "loss": 0.0624,
+      "step": 10554
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 8.723385962626666e-05,
+      "loss": 0.0925,
+      "step": 10555
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.720760312922325e-05,
+      "loss": 0.0363,
+      "step": 10556
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.71813489647333e-05,
+      "loss": 0.0429,
+      "step": 10557
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.715509713377195e-05,
+      "loss": 0.0231,
+      "step": 10558
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.712884763731458e-05,
+      "loss": 0.051,
+      "step": 10559
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.710260047633602e-05,
+      "loss": 0.0188,
+      "step": 10560
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.707635565181143e-05,
+      "loss": 0.0449,
+      "step": 10561
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.705011316471576e-05,
+      "loss": 0.0392,
+      "step": 10562
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.702387301602371e-05,
+      "loss": 0.11,
+      "step": 10563
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.699763520671024e-05,
+      "loss": 0.0557,
+      "step": 10564
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.697139973774974e-05,
+      "loss": 0.0373,
+      "step": 10565
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.694516661011698e-05,
+      "loss": 0.0545,
+      "step": 10566
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.69189358247863e-05,
+      "loss": 0.0173,
+      "step": 10567
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.689270738273222e-05,
+      "loss": 0.0638,
+      "step": 10568
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.686648128492903e-05,
+      "loss": 0.0602,
+      "step": 10569
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.684025753235083e-05,
+      "loss": 0.0274,
+      "step": 10570
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.681403612597194e-05,
+      "loss": 0.0227,
+      "step": 10571
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.678781706676625e-05,
+      "loss": 0.0628,
+      "step": 10572
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.676160035570783e-05,
+      "loss": 0.0631,
+      "step": 10573
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.673538599377052e-05,
+      "loss": 0.0261,
+      "step": 10574
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.67091739819281e-05,
+      "loss": 0.0564,
+      "step": 10575
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.668296432115421e-05,
+      "loss": 0.0457,
+      "step": 10576
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.665675701242257e-05,
+      "loss": 0.0362,
+      "step": 10577
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.663055205670663e-05,
+      "loss": 0.0618,
+      "step": 10578
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.660434945497986e-05,
+      "loss": 0.0628,
+      "step": 10579
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.657814920821563e-05,
+      "loss": 0.0303,
+      "step": 10580
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.655195131738712e-05,
+      "loss": 0.0456,
+      "step": 10581
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.65257557834676e-05,
+      "loss": 0.0157,
+      "step": 10582
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.649956260743006e-05,
+      "loss": 0.0353,
+      "step": 10583
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.647337179024763e-05,
+      "loss": 0.0457,
+      "step": 10584
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.644718333289315e-05,
+      "loss": 0.0403,
+      "step": 10585
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.642099723633935e-05,
+      "loss": 0.0475,
+      "step": 10586
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.639481350155915e-05,
+      "loss": 0.0445,
+      "step": 10587
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 8.636863212952503e-05,
+      "loss": 0.0503,
+      "step": 10588
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.634245312120976e-05,
+      "loss": 0.0372,
+      "step": 10589
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.631627647758555e-05,
+      "loss": 0.0485,
+      "step": 10590
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.6290102199625e-05,
+      "loss": 0.0668,
+      "step": 10591
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.626393028830025e-05,
+      "loss": 0.0541,
+      "step": 10592
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.623776074458364e-05,
+      "loss": 0.0339,
+      "step": 10593
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.621159356944726e-05,
+      "loss": 0.0169,
+      "step": 10594
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.618542876386304e-05,
+      "loss": 0.073,
+      "step": 10595
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.615926632880307e-05,
+      "loss": 0.0376,
+      "step": 10596
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.613310626523909e-05,
+      "loss": 0.0238,
+      "step": 10597
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.610694857414297e-05,
+      "loss": 0.0455,
+      "step": 10598
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.608079325648637e-05,
+      "loss": 0.0982,
+      "step": 10599
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.605464031324082e-05,
+      "loss": 0.0791,
+      "step": 10600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.602848974537785e-05,
+      "loss": 0.039,
+      "step": 10601
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.600234155386887e-05,
+      "loss": 0.0569,
+      "step": 10602
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.597619573968529e-05,
+      "loss": 0.0297,
+      "step": 10603
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.595005230379822e-05,
+      "loss": 0.0513,
+      "step": 10604
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.592391124717902e-05,
+      "loss": 0.0685,
+      "step": 10605
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.589777257079847e-05,
+      "loss": 0.0614,
+      "step": 10606
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.587163627562779e-05,
+      "loss": 0.0704,
+      "step": 10607
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.584550236263768e-05,
+      "loss": 0.0532,
+      "step": 10608
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.581937083279909e-05,
+      "loss": 0.026,
+      "step": 10609
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.579324168708267e-05,
+      "loss": 0.0315,
+      "step": 10610
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.576711492645897e-05,
+      "loss": 0.051,
+      "step": 10611
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.574099055189865e-05,
+      "loss": 0.0826,
+      "step": 10612
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.571486856437204e-05,
+      "loss": 0.0538,
+      "step": 10613
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.568874896484961e-05,
+      "loss": 0.0358,
+      "step": 10614
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.566263175430155e-05,
+      "loss": 0.0584,
+      "step": 10615
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.563651693369805e-05,
+      "loss": 0.041,
+      "step": 10616
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.561040450400916e-05,
+      "loss": 0.0498,
+      "step": 10617
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.558429446620497e-05,
+      "loss": 0.0639,
+      "step": 10618
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.55581868212553e-05,
+      "loss": 0.04,
+      "step": 10619
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.553208157013007e-05,
+      "loss": 0.0075,
+      "step": 10620
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 8.550597871379894e-05,
+      "loss": 0.0562,
+      "step": 10621
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.547987825323153e-05,
+      "loss": 0.0325,
+      "step": 10622
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.545378018939749e-05,
+      "loss": 0.0273,
+      "step": 10623
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.542768452326619e-05,
+      "loss": 0.0397,
+      "step": 10624
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.54015912558071e-05,
+      "loss": 0.0611,
+      "step": 10625
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.537550038798945e-05,
+      "loss": 0.0381,
+      "step": 10626
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.534941192078244e-05,
+      "loss": 0.0687,
+      "step": 10627
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.53233258551552e-05,
+      "loss": 0.0186,
+      "step": 10628
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.52972421920767e-05,
+      "loss": 0.0628,
+      "step": 10629
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.527116093251603e-05,
+      "loss": 0.0795,
+      "step": 10630
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.524508207744181e-05,
+      "loss": 0.0237,
+      "step": 10631
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.521900562782294e-05,
+      "loss": 0.0435,
+      "step": 10632
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.519293158462798e-05,
+      "loss": 0.0182,
+      "step": 10633
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.516685994882563e-05,
+      "loss": 0.0181,
+      "step": 10634
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.514079072138431e-05,
+      "loss": 0.0297,
+      "step": 10635
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.511472390327237e-05,
+      "loss": 0.0893,
+      "step": 10636
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.508865949545821e-05,
+      "loss": 0.0511,
+      "step": 10637
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.506259749890995e-05,
+      "loss": 0.0405,
+      "step": 10638
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.503653791459583e-05,
+      "loss": 0.0889,
+      "step": 10639
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.501048074348381e-05,
+      "loss": 0.0533,
+      "step": 10640
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.498442598654186e-05,
+      "loss": 0.0494,
+      "step": 10641
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.495837364473777e-05,
+      "loss": 0.0571,
+      "step": 10642
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.49323237190394e-05,
+      "loss": 0.0235,
+      "step": 10643
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.490627621041435e-05,
+      "loss": 0.0311,
+      "step": 10644
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.48802311198303e-05,
+      "loss": 0.0322,
+      "step": 10645
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.485418844825473e-05,
+      "loss": 0.0199,
+      "step": 10646
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.482814819665495e-05,
+      "loss": 0.0094,
+      "step": 10647
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.48021103659984e-05,
+      "loss": 0.0407,
+      "step": 10648
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.477607495725221e-05,
+      "loss": 0.0525,
+      "step": 10649
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.475004197138362e-05,
+      "loss": 0.0557,
+      "step": 10650
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.472401140935963e-05,
+      "loss": 0.0346,
+      "step": 10651
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.469798327214714e-05,
+      "loss": 0.0362,
+      "step": 10652
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.467195756071312e-05,
+      "loss": 0.0201,
+      "step": 10653
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 8.464593427602427e-05,
+      "loss": 0.0571,
+      "step": 10654
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.461991341904741e-05,
+      "loss": 0.0432,
+      "step": 10655
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.459389499074893e-05,
+      "loss": 0.0282,
+      "step": 10656
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.456787899209549e-05,
+      "loss": 0.007,
+      "step": 10657
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.454186542405344e-05,
+      "loss": 0.0247,
+      "step": 10658
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.451585428758918e-05,
+      "loss": 0.034,
+      "step": 10659
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.448984558366889e-05,
+      "loss": 0.0494,
+      "step": 10660
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.44638393132587e-05,
+      "loss": 0.0469,
+      "step": 10661
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.443783547732473e-05,
+      "loss": 0.0377,
+      "step": 10662
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.441183407683286e-05,
+      "loss": 0.047,
+      "step": 10663
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.438583511274908e-05,
+      "loss": 0.0251,
+      "step": 10664
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.435983858603912e-05,
+      "loss": 0.0182,
+      "step": 10665
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.433384449766859e-05,
+      "loss": 0.0847,
+      "step": 10666
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.430785284860325e-05,
+      "loss": 0.0321,
+      "step": 10667
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.428186363980847e-05,
+      "loss": 0.0207,
+      "step": 10668
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.42558768722498e-05,
+      "loss": 0.026,
+      "step": 10669
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.422989254689244e-05,
+      "loss": 0.0557,
+      "step": 10670
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.420391066470184e-05,
+      "loss": 0.0144,
+      "step": 10671
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.41779312266429e-05,
+      "loss": 0.0586,
+      "step": 10672
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.415195423368083e-05,
+      "loss": 0.0366,
+      "step": 10673
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.412597968678051e-05,
+      "loss": 0.0012,
+      "step": 10674
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.410000758690693e-05,
+      "loss": 0.0312,
+      "step": 10675
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.407403793502481e-05,
+      "loss": 0.0261,
+      "step": 10676
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.404807073209881e-05,
+      "loss": 0.0486,
+      "step": 10677
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.402210597909364e-05,
+      "loss": 0.0476,
+      "step": 10678
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.399614367697369e-05,
+      "loss": 0.0313,
+      "step": 10679
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.397018382670349e-05,
+      "loss": 0.0484,
+      "step": 10680
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.394422642924735e-05,
+      "loss": 0.0428,
+      "step": 10681
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.391827148556949e-05,
+      "loss": 0.0706,
+      "step": 10682
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.3892318996634e-05,
+      "loss": 0.01,
+      "step": 10683
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.386636896340505e-05,
+      "loss": 0.0314,
+      "step": 10684
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.384042138684652e-05,
+      "loss": 0.0487,
+      "step": 10685
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.381447626792237e-05,
+      "loss": 0.0299,
+      "step": 10686
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 8.378853360759633e-05,
+      "loss": 0.0608,
+      "step": 10687
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.376259340683206e-05,
+      "loss": 0.0423,
+      "step": 10688
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.373665566659326e-05,
+      "loss": 0.0211,
+      "step": 10689
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.371072038784333e-05,
+      "loss": 0.0318,
+      "step": 10690
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.368478757154579e-05,
+      "loss": 0.035,
+      "step": 10691
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.365885721866395e-05,
+      "loss": 0.04,
+      "step": 10692
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.363292933016095e-05,
+      "loss": 0.0516,
+      "step": 10693
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.360700390700005e-05,
+      "loss": 0.0431,
+      "step": 10694
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.358108095014421e-05,
+      "loss": 0.0574,
+      "step": 10695
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.355516046055655e-05,
+      "loss": 0.0454,
+      "step": 10696
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.352924243919973e-05,
+      "loss": 0.0526,
+      "step": 10697
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.350332688703668e-05,
+      "loss": 0.0642,
+      "step": 10698
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.347741380502998e-05,
+      "loss": 0.0373,
+      "step": 10699
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.345150319414232e-05,
+      "loss": 0.0406,
+      "step": 10700
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.34255950553362e-05,
+      "loss": 0.0335,
+      "step": 10701
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.339968938957392e-05,
+      "loss": 0.0357,
+      "step": 10702
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.337378619781794e-05,
+      "loss": 0.0313,
+      "step": 10703
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.334788548103035e-05,
+      "loss": 0.0637,
+      "step": 10704
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.332198724017342e-05,
+      "loss": 0.0421,
+      "step": 10705
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.329609147620915e-05,
+      "loss": 0.0597,
+      "step": 10706
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.327019819009939e-05,
+      "loss": 0.036,
+      "step": 10707
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.324430738280616e-05,
+      "loss": 0.0574,
+      "step": 10708
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.321841905529108e-05,
+      "loss": 0.038,
+      "step": 10709
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.319253320851598e-05,
+      "loss": 0.0974,
+      "step": 10710
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.316664984344232e-05,
+      "loss": 0.0527,
+      "step": 10711
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.314076896103168e-05,
+      "loss": 0.0337,
+      "step": 10712
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.31148905622453e-05,
+      "loss": 0.0656,
+      "step": 10713
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.30890146480447e-05,
+      "loss": 0.0261,
+      "step": 10714
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.30631412193909e-05,
+      "loss": 0.0522,
+      "step": 10715
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.30372702772452e-05,
+      "loss": 0.0522,
+      "step": 10716
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.301140182256854e-05,
+      "loss": 0.0481,
+      "step": 10717
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.298553585632176e-05,
+      "loss": 0.0239,
+      "step": 10718
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.295967237946591e-05,
+      "loss": 0.0606,
+      "step": 10719
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 8.293381139296155e-05,
+      "loss": 0.0478,
+      "step": 10720
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.290795289776956e-05,
+      "loss": 0.0317,
+      "step": 10721
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.288209689485025e-05,
+      "loss": 0.0495,
+      "step": 10722
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.285624338516428e-05,
+      "loss": 0.0347,
+      "step": 10723
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.283039236967191e-05,
+      "loss": 0.0425,
+      "step": 10724
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.280454384933356e-05,
+      "loss": 0.0574,
+      "step": 10725
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.277869782510936e-05,
+      "loss": 0.0067,
+      "step": 10726
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.275285429795933e-05,
+      "loss": 0.06,
+      "step": 10727
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.272701326884361e-05,
+      "loss": 0.052,
+      "step": 10728
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.270117473872205e-05,
+      "loss": 0.0654,
+      "step": 10729
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.267533870855453e-05,
+      "loss": 0.0806,
+      "step": 10730
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.26495051793007e-05,
+      "loss": 0.0769,
+      "step": 10731
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.26236741519203e-05,
+      "loss": 0.0118,
+      "step": 10732
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.259784562737282e-05,
+      "loss": 0.043,
+      "step": 10733
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.257201960661769e-05,
+      "loss": 0.0316,
+      "step": 10734
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.254619609061435e-05,
+      "loss": 0.0689,
+      "step": 10735
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.252037508032193e-05,
+      "loss": 0.0307,
+      "step": 10736
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.249455657669984e-05,
+      "loss": 0.0003,
+      "step": 10737
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.246874058070688e-05,
+      "loss": 0.0236,
+      "step": 10738
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.244292709330224e-05,
+      "loss": 0.0511,
+      "step": 10739
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.241711611544471e-05,
+      "loss": 0.0293,
+      "step": 10740
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.239130764809316e-05,
+      "loss": 0.0605,
+      "step": 10741
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.23655016922063e-05,
+      "loss": 0.067,
+      "step": 10742
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.233969824874264e-05,
+      "loss": 0.0497,
+      "step": 10743
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.231389731866082e-05,
+      "loss": 0.0199,
+      "step": 10744
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.22880989029192e-05,
+      "loss": 0.0307,
+      "step": 10745
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.226230300247618e-05,
+      "loss": 0.0606,
+      "step": 10746
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.223650961828997e-05,
+      "loss": 0.0258,
+      "step": 10747
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.221071875131872e-05,
+      "loss": 0.0382,
+      "step": 10748
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.218493040252039e-05,
+      "loss": 0.0213,
+      "step": 10749
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.215914457285305e-05,
+      "loss": 0.0111,
+      "step": 10750
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.213336126327462e-05,
+      "loss": 0.0318,
+      "step": 10751
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.21075804747428e-05,
+      "loss": 0.053,
+      "step": 10752
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 8.208180220821524e-05,
+      "loss": 0.06,
+      "step": 10753
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.205602646464952e-05,
+      "loss": 0.0427,
+      "step": 10754
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.203025324500325e-05,
+      "loss": 0.0284,
+      "step": 10755
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.200448255023367e-05,
+      "loss": 0.0721,
+      "step": 10756
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.197871438129822e-05,
+      "loss": 0.0721,
+      "step": 10757
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.195294873915408e-05,
+      "loss": 0.0412,
+      "step": 10758
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.192718562475828e-05,
+      "loss": 0.0349,
+      "step": 10759
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.190142503906798e-05,
+      "loss": 0.0435,
+      "step": 10760
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.187566698303998e-05,
+      "loss": 0.0115,
+      "step": 10761
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.18499114576313e-05,
+      "loss": 0.0439,
+      "step": 10762
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.182415846379843e-05,
+      "loss": 0.0336,
+      "step": 10763
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.179840800249822e-05,
+      "loss": 0.0632,
+      "step": 10764
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.177266007468709e-05,
+      "loss": 0.0411,
+      "step": 10765
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.174691468132162e-05,
+      "loss": 0.0263,
+      "step": 10766
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.172117182335811e-05,
+      "loss": 0.0477,
+      "step": 10767
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.16954315017528e-05,
+      "loss": 0.0463,
+      "step": 10768
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.166969371746195e-05,
+      "loss": 0.045,
+      "step": 10769
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.164395847144156e-05,
+      "loss": 0.0666,
+      "step": 10770
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.161822576464773e-05,
+      "loss": 0.1385,
+      "step": 10771
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.159249559803626e-05,
+      "loss": 0.0237,
+      "step": 10772
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.156676797256291e-05,
+      "loss": 0.0604,
+      "step": 10773
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.154104288918352e-05,
+      "loss": 0.018,
+      "step": 10774
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.151532034885359e-05,
+      "loss": 0.0864,
+      "step": 10775
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.148960035252872e-05,
+      "loss": 0.0634,
+      "step": 10776
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.146388290116429e-05,
+      "loss": 0.0823,
+      "step": 10777
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.143816799571565e-05,
+      "loss": 0.0291,
+      "step": 10778
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.141245563713794e-05,
+      "loss": 0.0368,
+      "step": 10779
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.138674582638642e-05,
+      "loss": 0.0603,
+      "step": 10780
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.136103856441605e-05,
+      "loss": 0.0457,
+      "step": 10781
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.133533385218185e-05,
+      "loss": 0.0883,
+      "step": 10782
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.130963169063866e-05,
+      "loss": 0.0149,
+      "step": 10783
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.128393208074115e-05,
+      "loss": 0.029,
+      "step": 10784
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.12582350234441e-05,
+      "loss": 0.019,
+      "step": 10785
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 8.123254051970202e-05,
+      "loss": 0.0498,
+      "step": 10786
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.120684857046947e-05,
+      "loss": 0.0779,
+      "step": 10787
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.118115917670068e-05,
+      "loss": 0.0145,
+      "step": 10788
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.115547233935005e-05,
+      "loss": 0.0358,
+      "step": 10789
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.112978805937171e-05,
+      "loss": 0.033,
+      "step": 10790
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.110410633771975e-05,
+      "loss": 0.065,
+      "step": 10791
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.107842717534827e-05,
+      "loss": 0.0469,
+      "step": 10792
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.105275057321112e-05,
+      "loss": 0.0504,
+      "step": 10793
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.10270765322621e-05,
+      "loss": 0.0436,
+      "step": 10794
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.100140505345485e-05,
+      "loss": 0.0392,
+      "step": 10795
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.097573613774315e-05,
+      "loss": 0.016,
+      "step": 10796
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.095006978608037e-05,
+      "loss": 0.1005,
+      "step": 10797
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.092440599942005e-05,
+      "loss": 0.0135,
+      "step": 10798
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.089874477871552e-05,
+      "loss": 0.0424,
+      "step": 10799
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.08730861249199e-05,
+      "loss": 0.0178,
+      "step": 10800
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.084743003898648e-05,
+      "loss": 0.0348,
+      "step": 10801
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.08217765218682e-05,
+      "loss": 0.0354,
+      "step": 10802
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.079612557451817e-05,
+      "loss": 0.0477,
+      "step": 10803
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.0770477197889e-05,
+      "loss": 0.0469,
+      "step": 10804
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.074483139293366e-05,
+      "loss": 0.0626,
+      "step": 10805
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.071918816060469e-05,
+      "loss": 0.0272,
+      "step": 10806
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.069354750185477e-05,
+      "loss": 0.0319,
+      "step": 10807
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.066790941763635e-05,
+      "loss": 0.0153,
+      "step": 10808
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.064227390890168e-05,
+      "loss": 0.0635,
+      "step": 10809
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.06166409766032e-05,
+      "loss": 0.0438,
+      "step": 10810
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.059101062169301e-05,
+      "loss": 0.0422,
+      "step": 10811
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.05653828451233e-05,
+      "loss": 0.0189,
+      "step": 10812
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.053975764784599e-05,
+      "loss": 0.0638,
+      "step": 10813
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.051413503081293e-05,
+      "loss": 0.0859,
+      "step": 10814
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.048851499497605e-05,
+      "loss": 0.0265,
+      "step": 10815
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.046289754128693e-05,
+      "loss": 0.0351,
+      "step": 10816
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.043728267069732e-05,
+      "loss": 0.0447,
+      "step": 10817
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.041167038415865e-05,
+      "loss": 0.0367,
+      "step": 10818
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.03860606826224e-05,
+      "loss": 0.0474,
+      "step": 10819
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.036045356703976e-05,
+      "loss": 0.0442,
+      "step": 10820
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.033484903836211e-05,
+      "loss": 0.047,
+      "step": 10821
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.03092470975405e-05,
+      "loss": 0.0612,
+      "step": 10822
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.028364774552602e-05,
+      "loss": 0.0369,
+      "step": 10823
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.025805098326961e-05,
+      "loss": 0.0556,
+      "step": 10824
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.023245681172201e-05,
+      "loss": 0.0474,
+      "step": 10825
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.020686523183411e-05,
+      "loss": 0.0501,
+      "step": 10826
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.018127624455646e-05,
+      "loss": 0.0538,
+      "step": 10827
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.015568985083975e-05,
+      "loss": 0.0062,
+      "step": 10828
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.013010605163424e-05,
+      "loss": 0.0399,
+      "step": 10829
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.010452484789044e-05,
+      "loss": 0.0386,
+      "step": 10830
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.007894624055854e-05,
+      "loss": 0.0389,
+      "step": 10831
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.005337023058874e-05,
+      "loss": 0.0558,
+      "step": 10832
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.002779681893123e-05,
+      "loss": 0.024,
+      "step": 10833
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.000222600653576e-05,
+      "loss": 0.0203,
+      "step": 10834
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.997665779435239e-05,
+      "loss": 0.0655,
+      "step": 10835
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.995109218333079e-05,
+      "loss": 0.0246,
+      "step": 10836
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.992552917442073e-05,
+      "loss": 0.0665,
+      "step": 10837
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.98999687685718e-05,
+      "loss": 0.0398,
+      "step": 10838
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.98744109667334e-05,
+      "loss": 0.0261,
+      "step": 10839
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.984885576985503e-05,
+      "loss": 0.0442,
+      "step": 10840
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.98233031788859e-05,
+      "loss": 0.0348,
+      "step": 10841
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.979775319477533e-05,
+      "loss": 0.036,
+      "step": 10842
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.977220581847234e-05,
+      "loss": 0.008,
+      "step": 10843
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.974666105092601e-05,
+      "loss": 0.0522,
+      "step": 10844
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.972111889308512e-05,
+      "loss": 0.0338,
+      "step": 10845
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.969557934589864e-05,
+      "loss": 0.0537,
+      "step": 10846
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.967004241031514e-05,
+      "loss": 0.0352,
+      "step": 10847
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.964450808728339e-05,
+      "loss": 0.0558,
+      "step": 10848
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.961897637775185e-05,
+      "loss": 0.0129,
+      "step": 10849
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.95934472826689e-05,
+      "loss": 0.0516,
+      "step": 10850
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.956792080298297e-05,
+      "loss": 0.0127,
+      "step": 10851
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 7.954239693964217e-05,
+      "loss": 0.0555,
+      "step": 10852
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.951687569359479e-05,
+      "loss": 0.0192,
+      "step": 10853
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.949135706578878e-05,
+      "loss": 0.0406,
+      "step": 10854
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.946584105717203e-05,
+      "loss": 0.0334,
+      "step": 10855
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.94403276686925e-05,
+      "loss": 0.0491,
+      "step": 10856
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.941481690129784e-05,
+      "loss": 0.0451,
+      "step": 10857
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.938930875593578e-05,
+      "loss": 0.0183,
+      "step": 10858
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.936380323355384e-05,
+      "loss": 0.0548,
+      "step": 10859
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.933830033509947e-05,
+      "loss": 0.0196,
+      "step": 10860
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.931280006151998e-05,
+      "loss": 0.0726,
+      "step": 10861
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.928730241376273e-05,
+      "loss": 0.0355,
+      "step": 10862
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.926180739277476e-05,
+      "loss": 0.0624,
+      "step": 10863
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.923631499950331e-05,
+      "loss": 0.0498,
+      "step": 10864
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.921082523489521e-05,
+      "loss": 0.0481,
+      "step": 10865
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.918533809989732e-05,
+      "loss": 0.046,
+      "step": 10866
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.915985359545652e-05,
+      "loss": 0.0604,
+      "step": 10867
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.913437172251933e-05,
+      "loss": 0.0584,
+      "step": 10868
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.910889248203257e-05,
+      "loss": 0.0253,
+      "step": 10869
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.908341587494244e-05,
+      "loss": 0.0153,
+      "step": 10870
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.905794190219549e-05,
+      "loss": 0.0422,
+      "step": 10871
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.903247056473794e-05,
+      "loss": 0.0335,
+      "step": 10872
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.900700186351597e-05,
+      "loss": 0.0236,
+      "step": 10873
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.898153579947581e-05,
+      "loss": 0.0256,
+      "step": 10874
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.895607237356325e-05,
+      "loss": 0.0383,
+      "step": 10875
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.893061158672429e-05,
+      "loss": 0.0477,
+      "step": 10876
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.890515343990467e-05,
+      "loss": 0.0228,
+      "step": 10877
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.887969793405014e-05,
+      "loss": 0.0041,
+      "step": 10878
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.88542450701063e-05,
+      "loss": 0.0312,
+      "step": 10879
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.882879484901857e-05,
+      "loss": 0.0512,
+      "step": 10880
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.880334727173245e-05,
+      "loss": 0.0786,
+      "step": 10881
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.877790233919313e-05,
+      "loss": 0.0626,
+      "step": 10882
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.875246005234594e-05,
+      "loss": 0.0097,
+      "step": 10883
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.872702041213594e-05,
+      "loss": 0.0492,
+      "step": 10884
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 7.870158341950813e-05,
+      "loss": 0.0446,
+      "step": 10885
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.867614907540735e-05,
+      "loss": 0.0471,
+      "step": 10886
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.865071738077853e-05,
+      "loss": 0.0412,
+      "step": 10887
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.862528833656628e-05,
+      "loss": 0.0586,
+      "step": 10888
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.85998619437153e-05,
+      "loss": 0.0628,
+      "step": 10889
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.857443820317009e-05,
+      "loss": 0.0613,
+      "step": 10890
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.854901711587498e-05,
+      "loss": 0.0772,
+      "step": 10891
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.852359868277441e-05,
+      "loss": 0.0128,
+      "step": 10892
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.84981829048125e-05,
+      "loss": 0.0174,
+      "step": 10893
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.847276978293354e-05,
+      "loss": 0.0425,
+      "step": 10894
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.84473593180813e-05,
+      "loss": 0.0305,
+      "step": 10895
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.842195151119984e-05,
+      "loss": 0.059,
+      "step": 10896
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.839654636323303e-05,
+      "loss": 0.0398,
+      "step": 10897
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.837114387512451e-05,
+      "loss": 0.0506,
+      "step": 10898
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.834574404781805e-05,
+      "loss": 0.0462,
+      "step": 10899
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.832034688225695e-05,
+      "loss": 0.0768,
+      "step": 10900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.829495237938481e-05,
+      "loss": 0.0506,
+      "step": 10901
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.826956054014489e-05,
+      "loss": 0.0483,
+      "step": 10902
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.82441713654805e-05,
+      "loss": 0.047,
+      "step": 10903
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.821878485633471e-05,
+      "loss": 0.0455,
+      "step": 10904
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.819340101365051e-05,
+      "loss": 0.0447,
+      "step": 10905
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.816801983837094e-05,
+      "loss": 0.0347,
+      "step": 10906
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.814264133143875e-05,
+      "loss": 0.0711,
+      "step": 10907
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.811726549379676e-05,
+      "loss": 0.051,
+      "step": 10908
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.809189232638757e-05,
+      "loss": 0.0346,
+      "step": 10909
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.806652183015372e-05,
+      "loss": 0.0384,
+      "step": 10910
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.804115400603756e-05,
+      "loss": 0.0416,
+      "step": 10911
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.801578885498156e-05,
+      "loss": 0.0613,
+      "step": 10912
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.79904263779279e-05,
+      "loss": 0.0529,
+      "step": 10913
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.796506657581875e-05,
+      "loss": 0.0363,
+      "step": 10914
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.793970944959614e-05,
+      "loss": 0.0299,
+      "step": 10915
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.791435500020196e-05,
+      "loss": 0.0428,
+      "step": 10916
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 7.788900322857817e-05,
+      "loss": 0.0259,
+      "step": 10917
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.786365413566639e-05,
+      "loss": 0.0208,
+      "step": 10918
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.783830772240836e-05,
+      "loss": 0.044,
+      "step": 10919
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.781296398974559e-05,
+      "loss": 0.0138,
+      "step": 10920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.778762293861949e-05,
+      "loss": 0.0434,
+      "step": 10921
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.776228456997146e-05,
+      "loss": 0.0545,
+      "step": 10922
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.773694888474267e-05,
+      "loss": 0.0284,
+      "step": 10923
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.77116158838744e-05,
+      "loss": 0.0681,
+      "step": 10924
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.768628556830761e-05,
+      "loss": 0.0455,
+      "step": 10925
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.766095793898324e-05,
+      "loss": 0.0539,
+      "step": 10926
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.763563299684207e-05,
+      "loss": 0.023,
+      "step": 10927
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.761031074282502e-05,
+      "loss": 0.0448,
+      "step": 10928
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.758499117787257e-05,
+      "loss": 0.0355,
+      "step": 10929
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.755967430292542e-05,
+      "loss": 0.0698,
+      "step": 10930
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.753436011892392e-05,
+      "loss": 0.0739,
+      "step": 10931
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.75090486268084e-05,
+      "loss": 0.0573,
+      "step": 10932
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.748373982751918e-05,
+      "loss": 0.0852,
+      "step": 10933
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.745843372199632e-05,
+      "loss": 0.0297,
+      "step": 10934
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.743313031118006e-05,
+      "loss": 0.0435,
+      "step": 10935
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.740782959601008e-05,
+      "loss": 0.0706,
+      "step": 10936
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.738253157742635e-05,
+      "loss": 0.0163,
+      "step": 10937
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.735723625636869e-05,
+      "loss": 0.0323,
+      "step": 10938
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.733194363377661e-05,
+      "loss": 0.0199,
+      "step": 10939
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.730665371058985e-05,
+      "loss": 0.0464,
+      "step": 10940
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.728136648774761e-05,
+      "loss": 0.0395,
+      "step": 10941
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.725608196618942e-05,
+      "loss": 0.0242,
+      "step": 10942
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.723080014685442e-05,
+      "loss": 0.0339,
+      "step": 10943
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.720552103068185e-05,
+      "loss": 0.0389,
+      "step": 10944
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.718024461861069e-05,
+      "loss": 0.0494,
+      "step": 10945
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.715497091157987e-05,
+      "loss": 0.0512,
+      "step": 10946
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.71296999105283e-05,
+      "loss": 0.0157,
+      "step": 10947
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.710443161639464e-05,
+      "loss": 0.026,
+      "step": 10948
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.707916603011764e-05,
+      "loss": 0.0373,
+      "step": 10949
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 7.705390315263578e-05,
+      "loss": 0.0503,
+      "step": 10950
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.702864298488751e-05,
+      "loss": 0.0562,
+      "step": 10951
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.70033855278111e-05,
+      "loss": 0.0518,
+      "step": 10952
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.697813078234492e-05,
+      "loss": 0.0295,
+      "step": 10953
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.695287874942699e-05,
+      "loss": 0.0125,
+      "step": 10954
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.692762942999545e-05,
+      "loss": 0.0088,
+      "step": 10955
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.690238282498821e-05,
+      "loss": 0.0513,
+      "step": 10956
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.687713893534302e-05,
+      "loss": 0.0404,
+      "step": 10957
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.685189776199777e-05,
+      "loss": 0.0682,
+      "step": 10958
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.682665930588993e-05,
+      "loss": 0.0485,
+      "step": 10959
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.680142356795718e-05,
+      "loss": 0.0662,
+      "step": 10960
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.67761905491369e-05,
+      "loss": 0.0225,
+      "step": 10961
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.675096025036636e-05,
+      "loss": 0.0437,
+      "step": 10962
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.672573267258289e-05,
+      "loss": 0.0483,
+      "step": 10963
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.670050781672352e-05,
+      "loss": 0.0334,
+      "step": 10964
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.667528568372545e-05,
+      "loss": 0.0201,
+      "step": 10965
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.66500662745254e-05,
+      "loss": 0.0263,
+      "step": 10966
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.662484959006032e-05,
+      "loss": 0.0488,
+      "step": 10967
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.659963563126686e-05,
+      "loss": 0.0609,
+      "step": 10968
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.657442439908176e-05,
+      "loss": 0.018,
+      "step": 10969
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.654921589444147e-05,
+      "loss": 0.07,
+      "step": 10970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.652401011828235e-05,
+      "loss": 0.0805,
+      "step": 10971
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.649880707154085e-05,
+      "loss": 0.0297,
+      "step": 10972
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.647360675515308e-05,
+      "loss": 0.0375,
+      "step": 10973
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.644840917005524e-05,
+      "loss": 0.0364,
+      "step": 10974
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.642321431718333e-05,
+      "loss": 0.0446,
+      "step": 10975
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.639802219747324e-05,
+      "loss": 0.0627,
+      "step": 10976
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.637283281186073e-05,
+      "loss": 0.0499,
+      "step": 10977
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.634764616128158e-05,
+      "loss": 0.0262,
+      "step": 10978
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.632246224667143e-05,
+      "loss": 0.013,
+      "step": 10979
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.62972810689657e-05,
+      "loss": 0.0406,
+      "step": 10980
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.627210262909997e-05,
+      "loss": 0.031,
+      "step": 10981
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.62469269280093e-05,
+      "loss": 0.0553,
+      "step": 10982
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 7.622175396662907e-05,
+      "loss": 0.034,
+      "step": 10983
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.619658374589427e-05,
+      "loss": 0.0441,
+      "step": 10984
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.617141626674002e-05,
+      "loss": 0.0708,
+      "step": 10985
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.614625153010115e-05,
+      "loss": 0.0607,
+      "step": 10986
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.612108953691243e-05,
+      "loss": 0.0271,
+      "step": 10987
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.609593028810861e-05,
+      "loss": 0.0542,
+      "step": 10988
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.607077378462422e-05,
+      "loss": 0.0198,
+      "step": 10989
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.604562002739384e-05,
+      "loss": 0.0428,
+      "step": 10990
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.60204690173518e-05,
+      "loss": 0.0472,
+      "step": 10991
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.59953207554324e-05,
+      "loss": 0.0314,
+      "step": 10992
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.597017524256974e-05,
+      "loss": 0.044,
+      "step": 10993
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.594503247969805e-05,
+      "loss": 0.0326,
+      "step": 10994
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.59198924677512e-05,
+      "loss": 0.0488,
+      "step": 10995
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.589475520766314e-05,
+      "loss": 0.022,
+      "step": 10996
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.586962070036762e-05,
+      "loss": 0.0331,
+      "step": 10997
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.584448894679824e-05,
+      "loss": 0.0642,
+      "step": 10998
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.581935994788873e-05,
+      "loss": 0.0274,
+      "step": 10999
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.579423370457236e-05,
+      "loss": 0.0339,
+      "step": 11000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.576911021778268e-05,
+      "loss": 0.0419,
+      "step": 11001
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.574398948845287e-05,
+      "loss": 0.0685,
+      "step": 11002
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.571887151751605e-05,
+      "loss": 0.0391,
+      "step": 11003
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.569375630590538e-05,
+      "loss": 0.0475,
+      "step": 11004
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.566864385455369e-05,
+      "loss": 0.0366,
+      "step": 11005
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.564353416439404e-05,
+      "loss": 0.0148,
+      "step": 11006
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.561842723635891e-05,
+      "loss": 0.0499,
+      "step": 11007
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.559332307138116e-05,
+      "loss": 0.0349,
+      "step": 11008
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.556822167039323e-05,
+      "loss": 0.0349,
+      "step": 11009
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.554312303432762e-05,
+      "loss": 0.0431,
+      "step": 11010
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.551802716411665e-05,
+      "loss": 0.0439,
+      "step": 11011
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.549293406069248e-05,
+      "loss": 0.0222,
+      "step": 11012
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.546784372498739e-05,
+      "loss": 0.0439,
+      "step": 11013
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.544275615793329e-05,
+      "loss": 0.0431,
+      "step": 11014
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.541767136046223e-05,
+      "loss": 0.055,
+      "step": 11015
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 7.539258933350596e-05,
+      "loss": 0.0453,
+      "step": 11016
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.536751007799621e-05,
+      "loss": 0.0824,
+      "step": 11017
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.534243359486455e-05,
+      "loss": 0.0353,
+      "step": 11018
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.531735988504254e-05,
+      "loss": 0.0348,
+      "step": 11019
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.529228894946167e-05,
+      "loss": 0.0127,
+      "step": 11020
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.526722078905319e-05,
+      "loss": 0.0446,
+      "step": 11021
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.524215540474828e-05,
+      "loss": 0.0603,
+      "step": 11022
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.521709279747803e-05,
+      "loss": 0.0633,
+      "step": 11023
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.519203296817354e-05,
+      "loss": 0.061,
+      "step": 11024
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.516697591776562e-05,
+      "loss": 0.065,
+      "step": 11025
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.514192164718512e-05,
+      "loss": 0.0676,
+      "step": 11026
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.511687015736274e-05,
+      "loss": 0.035,
+      "step": 11027
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.509182144922897e-05,
+      "loss": 0.0567,
+      "step": 11028
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.506677552371441e-05,
+      "loss": 0.0388,
+      "step": 11029
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.504173238174936e-05,
+      "loss": 0.0626,
+      "step": 11030
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.501669202426426e-05,
+      "loss": 0.0573,
+      "step": 11031
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.499165445218903e-05,
+      "loss": 0.0258,
+      "step": 11032
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.496661966645393e-05,
+      "loss": 0.0418,
+      "step": 11033
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.494158766798884e-05,
+      "loss": 0.0344,
+      "step": 11034
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.491655845772372e-05,
+      "loss": 0.0514,
+      "step": 11035
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.489153203658826e-05,
+      "loss": 0.0545,
+      "step": 11036
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.486650840551208e-05,
+      "loss": 0.0584,
+      "step": 11037
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.484148756542484e-05,
+      "loss": 0.0302,
+      "step": 11038
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.481646951725588e-05,
+      "loss": 0.0312,
+      "step": 11039
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.479145426193466e-05,
+      "loss": 0.0448,
+      "step": 11040
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.476644180039036e-05,
+      "loss": 0.0216,
+      "step": 11041
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.474143213355208e-05,
+      "loss": 0.0533,
+      "step": 11042
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.471642526234897e-05,
+      "loss": 0.0236,
+      "step": 11043
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.469142118770981e-05,
+      "loss": 0.0653,
+      "step": 11044
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.466641991056359e-05,
+      "loss": 0.0317,
+      "step": 11045
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.464142143183888e-05,
+      "loss": 0.082,
+      "step": 11046
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.461642575246451e-05,
+      "loss": 0.068,
+      "step": 11047
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.459143287336875e-05,
+      "loss": 0.0295,
+      "step": 11048
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 7.456644279548016e-05,
+      "loss": 0.0348,
+      "step": 11049
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.454145551972698e-05,
+      "loss": 0.0713,
+      "step": 11050
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.451647104703748e-05,
+      "loss": 0.0706,
+      "step": 11051
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.449148937833975e-05,
+      "loss": 0.0191,
+      "step": 11052
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.44665105145617e-05,
+      "loss": 0.0334,
+      "step": 11053
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.444153445663132e-05,
+      "loss": 0.0269,
+      "step": 11054
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.441656120547634e-05,
+      "loss": 0.0682,
+      "step": 11055
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.43915907620245e-05,
+      "loss": 0.0535,
+      "step": 11056
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.436662312720335e-05,
+      "loss": 0.0368,
+      "step": 11057
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.434165830194034e-05,
+      "loss": 0.0472,
+      "step": 11058
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.431669628716281e-05,
+      "loss": 0.0681,
+      "step": 11059
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.429173708379809e-05,
+      "loss": 0.0023,
+      "step": 11060
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.426678069277338e-05,
+      "loss": 0.0706,
+      "step": 11061
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.424182711501566e-05,
+      "loss": 0.0199,
+      "step": 11062
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.421687635145193e-05,
+      "loss": 0.0392,
+      "step": 11063
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.419192840300893e-05,
+      "loss": 0.0598,
+      "step": 11064
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.416698327061355e-05,
+      "loss": 0.0557,
+      "step": 11065
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.41420409551923e-05,
+      "loss": 0.0362,
+      "step": 11066
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.411710145767185e-05,
+      "loss": 0.0455,
+      "step": 11067
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.409216477897856e-05,
+      "loss": 0.0386,
+      "step": 11068
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.406723092003867e-05,
+      "loss": 0.0365,
+      "step": 11069
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.404229988177856e-05,
+      "loss": 0.031,
+      "step": 11070
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.40173716651242e-05,
+      "loss": 0.0301,
+      "step": 11071
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.399244627100177e-05,
+      "loss": 0.0141,
+      "step": 11072
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.396752370033697e-05,
+      "loss": 0.032,
+      "step": 11073
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.394260395405574e-05,
+      "loss": 0.0255,
+      "step": 11074
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.39176870330837e-05,
+      "loss": 0.0231,
+      "step": 11075
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.389277293834652e-05,
+      "loss": 0.0387,
+      "step": 11076
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.386786167076965e-05,
+      "loss": 0.0067,
+      "step": 11077
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.384295323127839e-05,
+      "loss": 0.0287,
+      "step": 11078
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.381804762079816e-05,
+      "loss": 0.0427,
+      "step": 11079
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.3793144840254e-05,
+      "loss": 0.0189,
+      "step": 11080
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.376824489057111e-05,
+      "loss": 0.0357,
+      "step": 11081
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 7.374334777267436e-05,
+      "loss": 0.0149,
+      "step": 11082
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.371845348748854e-05,
+      "loss": 0.0706,
+      "step": 11083
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.369356203593857e-05,
+      "loss": 0.0245,
+      "step": 11084
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.366867341894893e-05,
+      "loss": 0.0273,
+      "step": 11085
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.364378763744429e-05,
+      "loss": 0.0695,
+      "step": 11086
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.361890469234903e-05,
+      "loss": 0.0418,
+      "step": 11087
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.359402458458747e-05,
+      "loss": 0.0334,
+      "step": 11088
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.35691473150838e-05,
+      "loss": 0.0215,
+      "step": 11089
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.354427288476222e-05,
+      "loss": 0.0449,
+      "step": 11090
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.351940129454664e-05,
+      "loss": 0.0354,
+      "step": 11091
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.349453254536108e-05,
+      "loss": 0.0422,
+      "step": 11092
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.346966663812929e-05,
+      "loss": 0.0444,
+      "step": 11093
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.34448035737749e-05,
+      "loss": 0.0261,
+      "step": 11094
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.341994335322161e-05,
+      "loss": 0.112,
+      "step": 11095
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.33950859773928e-05,
+      "loss": 0.0301,
+      "step": 11096
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.337023144721203e-05,
+      "loss": 0.0491,
+      "step": 11097
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.334537976360233e-05,
+      "loss": 0.0634,
+      "step": 11098
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.332053092748704e-05,
+      "loss": 0.0796,
+      "step": 11099
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.32956849397891e-05,
+      "loss": 0.0681,
+      "step": 11100
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.327084180143155e-05,
+      "loss": 0.0485,
+      "step": 11101
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.324600151333732e-05,
+      "loss": 0.0312,
+      "step": 11102
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.322116407642894e-05,
+      "loss": 0.0759,
+      "step": 11103
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.319632949162921e-05,
+      "loss": 0.0352,
+      "step": 11104
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.317149775986059e-05,
+      "loss": 0.051,
+      "step": 11105
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.314666888204556e-05,
+      "loss": 0.0361,
+      "step": 11106
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.312184285910636e-05,
+      "loss": 0.0565,
+      "step": 11107
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.309701969196531e-05,
+      "loss": 0.0357,
+      "step": 11108
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.307219938154446e-05,
+      "loss": 0.0652,
+      "step": 11109
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.304738192876575e-05,
+      "loss": 0.0415,
+      "step": 11110
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.302256733455121e-05,
+      "loss": 0.0375,
+      "step": 11111
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.29977555998225e-05,
+      "loss": 0.0309,
+      "step": 11112
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.297294672550149e-05,
+      "loss": 0.0569,
+      "step": 11113
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.29481407125095e-05,
+      "loss": 0.0389,
+      "step": 11114
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 7.292333756176818e-05,
+      "loss": 0.0427,
+      "step": 11115
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.28985372741988e-05,
+      "loss": 0.034,
+      "step": 11116
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.287373985072272e-05,
+      "loss": 0.0494,
+      "step": 11117
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.284894529226102e-05,
+      "loss": 0.0347,
+      "step": 11118
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.28241535997347e-05,
+      "loss": 0.0223,
+      "step": 11119
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.279936477406483e-05,
+      "loss": 0.067,
+      "step": 11120
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.277457881617211e-05,
+      "loss": 0.0592,
+      "step": 11121
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.274979572697738e-05,
+      "loss": 0.0548,
+      "step": 11122
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.272501550740121e-05,
+      "loss": 0.015,
+      "step": 11123
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.270023815836405e-05,
+      "loss": 0.0519,
+      "step": 11124
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.26754636807864e-05,
+      "loss": 0.0279,
+      "step": 11125
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.265069207558849e-05,
+      "loss": 0.0256,
+      "step": 11126
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.262592334369057e-05,
+      "loss": 0.0253,
+      "step": 11127
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.260115748601273e-05,
+      "loss": 0.0598,
+      "step": 11128
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.257639450347491e-05,
+      "loss": 0.0779,
+      "step": 11129
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.255163439699693e-05,
+      "loss": 0.034,
+      "step": 11130
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.252687716749867e-05,
+      "loss": 0.0602,
+      "step": 11131
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.25021228158997e-05,
+      "loss": 0.0296,
+      "step": 11132
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.247737134311966e-05,
+      "loss": 0.0231,
+      "step": 11133
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.245262275007792e-05,
+      "loss": 0.0447,
+      "step": 11134
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.242787703769379e-05,
+      "loss": 0.0299,
+      "step": 11135
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.24031342068866e-05,
+      "loss": 0.0643,
+      "step": 11136
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.237839425857539e-05,
+      "loss": 0.0262,
+      "step": 11137
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.23536571936793e-05,
+      "loss": 0.056,
+      "step": 11138
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.232892301311704e-05,
+      "loss": 0.0334,
+      "step": 11139
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.23041917178076e-05,
+      "loss": 0.0495,
+      "step": 11140
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.227946330866952e-05,
+      "loss": 0.0877,
+      "step": 11141
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.225473778662147e-05,
+      "loss": 0.0515,
+      "step": 11142
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.223001515258203e-05,
+      "loss": 0.0876,
+      "step": 11143
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.220529540746935e-05,
+      "loss": 0.0225,
+      "step": 11144
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.21805785522019e-05,
+      "loss": 0.0318,
+      "step": 11145
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.215586458769768e-05,
+      "loss": 0.0553,
+      "step": 11146
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.213115351487486e-05,
+      "loss": 0.0293,
+      "step": 11147
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 7.210644533465136e-05,
+      "loss": 0.0618,
+      "step": 11148
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.208174004794491e-05,
+      "loss": 0.021,
+      "step": 11149
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.205703765567341e-05,
+      "loss": 0.0564,
+      "step": 11150
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.203233815875431e-05,
+      "loss": 0.0237,
+      "step": 11151
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.20076415581053e-05,
+      "loss": 0.0461,
+      "step": 11152
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.198294785464367e-05,
+      "loss": 0.0511,
+      "step": 11153
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.195825704928675e-05,
+      "loss": 0.0482,
+      "step": 11154
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.193356914295167e-05,
+      "loss": 0.0439,
+      "step": 11155
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.190888413655562e-05,
+      "loss": 0.0393,
+      "step": 11156
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.188420203101548e-05,
+      "loss": 0.0616,
+      "step": 11157
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.185952282724821e-05,
+      "loss": 0.038,
+      "step": 11158
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.183484652617051e-05,
+      "loss": 0.0986,
+      "step": 11159
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.1810173128699e-05,
+      "loss": 0.0352,
+      "step": 11160
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.178550263575034e-05,
+      "loss": 0.011,
+      "step": 11161
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.176083504824082e-05,
+      "loss": 0.0743,
+      "step": 11162
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.173617036708697e-05,
+      "loss": 0.0196,
+      "step": 11163
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.171150859320475e-05,
+      "loss": 0.0481,
+      "step": 11164
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.168684972751042e-05,
+      "loss": 0.0502,
+      "step": 11165
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.166219377092002e-05,
+      "loss": 0.0724,
+      "step": 11166
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.163754072434936e-05,
+      "loss": 0.0763,
+      "step": 11167
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.161289058871431e-05,
+      "loss": 0.0662,
+      "step": 11168
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.15882433649305e-05,
+      "loss": 0.0388,
+      "step": 11169
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.156359905391352e-05,
+      "loss": 0.0367,
+      "step": 11170
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.153895765657874e-05,
+      "loss": 0.0377,
+      "step": 11171
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.151431917384167e-05,
+      "loss": 0.0313,
+      "step": 11172
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.148968360661746e-05,
+      "loss": 0.0283,
+      "step": 11173
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.146505095582129e-05,
+      "loss": 0.0191,
+      "step": 11174
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.144042122236819e-05,
+      "loss": 0.0457,
+      "step": 11175
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.141579440717301e-05,
+      "loss": 0.0432,
+      "step": 11176
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.139117051115069e-05,
+      "loss": 0.0547,
+      "step": 11177
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.136654953521579e-05,
+      "loss": 0.0472,
+      "step": 11178
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.13419314802831e-05,
+      "loss": 0.0422,
+      "step": 11179
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.13173163472669e-05,
+      "loss": 0.0475,
+      "step": 11180
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 7.129270413708168e-05,
+      "loss": 0.0403,
+      "step": 11181
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.126809485064169e-05,
+      "loss": 0.0095,
+      "step": 11182
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.124348848886105e-05,
+      "loss": 0.0453,
+      "step": 11183
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.121888505265399e-05,
+      "loss": 0.0095,
+      "step": 11184
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.119428454293423e-05,
+      "loss": 0.033,
+      "step": 11185
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.116968696061573e-05,
+      "loss": 0.0551,
+      "step": 11186
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.114509230661215e-05,
+      "loss": 0.0644,
+      "step": 11187
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.112050058183722e-05,
+      "loss": 0.0283,
+      "step": 11188
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.109591178720438e-05,
+      "loss": 0.0513,
+      "step": 11189
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.107132592362695e-05,
+      "loss": 0.0561,
+      "step": 11190
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.104674299201838e-05,
+      "loss": 0.0558,
+      "step": 11191
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.102216299329173e-05,
+      "loss": 0.0682,
+      "step": 11192
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.099758592836016e-05,
+      "loss": 0.0382,
+      "step": 11193
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.097301179813663e-05,
+      "loss": 0.0458,
+      "step": 11194
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.094844060353394e-05,
+      "loss": 0.048,
+      "step": 11195
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.092387234546481e-05,
+      "loss": 0.0494,
+      "step": 11196
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.0899307024842e-05,
+      "loss": 0.034,
+      "step": 11197
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.087474464257792e-05,
+      "loss": 0.0292,
+      "step": 11198
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.085018519958508e-05,
+      "loss": 0.0546,
+      "step": 11199
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.082562869677578e-05,
+      "loss": 0.018,
+      "step": 11200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.080107513506214e-05,
+      "loss": 0.0583,
+      "step": 11201
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.077652451535635e-05,
+      "loss": 0.0591,
+      "step": 11202
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.07519768385703e-05,
+      "loss": 0.0198,
+      "step": 11203
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.072743210561603e-05,
+      "loss": 0.0657,
+      "step": 11204
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.070289031740507e-05,
+      "loss": 0.017,
+      "step": 11205
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.067835147484923e-05,
+      "loss": 0.0622,
+      "step": 11206
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.065381557886005e-05,
+      "loss": 0.0253,
+      "step": 11207
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.062928263034891e-05,
+      "loss": 0.0749,
+      "step": 11208
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.060475263022727e-05,
+      "loss": 0.0173,
+      "step": 11209
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.058022557940613e-05,
+      "loss": 0.0226,
+      "step": 11210
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.055570147879675e-05,
+      "loss": 0.0764,
+      "step": 11211
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.053118032931006e-05,
+      "loss": 0.0353,
+      "step": 11212
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.050666213185703e-05,
+      "loss": 0.0869,
+      "step": 11213
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 7.048214688734839e-05,
+      "loss": 0.0319,
+      "step": 11214
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.045763459669475e-05,
+      "loss": 0.0461,
+      "step": 11215
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.043312526080678e-05,
+      "loss": 0.0431,
+      "step": 11216
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.040861888059482e-05,
+      "loss": 0.0576,
+      "step": 11217
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.038411545696935e-05,
+      "loss": 0.03,
+      "step": 11218
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.03596149908405e-05,
+      "loss": 0.037,
+      "step": 11219
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.03351174831184e-05,
+      "loss": 0.0607,
+      "step": 11220
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.031062293471305e-05,
+      "loss": 0.0204,
+      "step": 11221
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.028613134653438e-05,
+      "loss": 0.0566,
+      "step": 11222
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.026164271949215e-05,
+      "loss": 0.048,
+      "step": 11223
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.02371570544961e-05,
+      "loss": 0.0184,
+      "step": 11224
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.021267435245578e-05,
+      "loss": 0.0391,
+      "step": 11225
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.018819461428057e-05,
+      "loss": 0.1524,
+      "step": 11226
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.016371784087993e-05,
+      "loss": 0.0282,
+      "step": 11227
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.0139244033163e-05,
+      "loss": 0.0322,
+      "step": 11228
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.011477319203904e-05,
+      "loss": 0.0181,
+      "step": 11229
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.009030531841697e-05,
+      "loss": 0.0105,
+      "step": 11230
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.006584041320568e-05,
+      "loss": 0.0246,
+      "step": 11231
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.004137847731407e-05,
+      "loss": 0.064,
+      "step": 11232
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 7.001691951165072e-05,
+      "loss": 0.0506,
+      "step": 11233
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.99924635171243e-05,
+      "loss": 0.0477,
+      "step": 11234
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.996801049464326e-05,
+      "loss": 0.0353,
+      "step": 11235
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.994356044511594e-05,
+      "loss": 0.0786,
+      "step": 11236
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.991911336945051e-05,
+      "loss": 0.0183,
+      "step": 11237
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.989466926855525e-05,
+      "loss": 0.0485,
+      "step": 11238
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.987022814333806e-05,
+      "loss": 0.0367,
+      "step": 11239
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.984578999470699e-05,
+      "loss": 0.0158,
+      "step": 11240
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.982135482356974e-05,
+      "loss": 0.0517,
+      "step": 11241
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.9796922630834e-05,
+      "loss": 0.0437,
+      "step": 11242
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.977249341740745e-05,
+      "loss": 0.053,
+      "step": 11243
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.974806718419745e-05,
+      "loss": 0.0438,
+      "step": 11244
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.972364393211151e-05,
+      "loss": 0.053,
+      "step": 11245
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.96992236620567e-05,
+      "loss": 0.0366,
+      "step": 11246
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.967480637494023e-05,
+      "loss": 0.0415,
+      "step": 11247
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.965039207166924e-05,
+      "loss": 0.0265,
+      "step": 11248
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.962598075315046e-05,
+      "loss": 0.0538,
+      "step": 11249
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.960157242029095e-05,
+      "loss": 0.0813,
+      "step": 11250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.957716707399712e-05,
+      "loss": 0.062,
+      "step": 11251
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.955276471517577e-05,
+      "loss": 0.0314,
+      "step": 11252
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.952836534473323e-05,
+      "loss": 0.0123,
+      "step": 11253
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.950396896357602e-05,
+      "loss": 0.0518,
+      "step": 11254
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.947957557261027e-05,
+      "loss": 0.0398,
+      "step": 11255
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.945518517274212e-05,
+      "loss": 0.0509,
+      "step": 11256
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.943079776487772e-05,
+      "loss": 0.0433,
+      "step": 11257
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.940641334992286e-05,
+      "loss": 0.085,
+      "step": 11258
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.938203192878345e-05,
+      "loss": 0.0419,
+      "step": 11259
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.935765350236513e-05,
+      "loss": 0.092,
+      "step": 11260
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.933327807157352e-05,
+      "loss": 0.0274,
+      "step": 11261
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.930890563731401e-05,
+      "loss": 0.0381,
+      "step": 11262
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.92845362004921e-05,
+      "loss": 0.0319,
+      "step": 11263
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.92601697620129e-05,
+      "loss": 0.0378,
+      "step": 11264
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.923580632278169e-05,
+      "loss": 0.0648,
+      "step": 11265
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.921144588370345e-05,
+      "loss": 0.0564,
+      "step": 11266
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.918708844568302e-05,
+      "loss": 0.025,
+      "step": 11267
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.916273400962531e-05,
+      "loss": 0.0302,
+      "step": 11268
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.913838257643494e-05,
+      "loss": 0.0218,
+      "step": 11269
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.911403414701658e-05,
+      "loss": 0.0518,
+      "step": 11270
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.908968872227466e-05,
+      "loss": 0.0664,
+      "step": 11271
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.906534630311348e-05,
+      "loss": 0.0392,
+      "step": 11272
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.904100689043742e-05,
+      "loss": 0.0327,
+      "step": 11273
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.901667048515046e-05,
+      "loss": 0.048,
+      "step": 11274
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.899233708815685e-05,
+      "loss": 0.039,
+      "step": 11275
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.896800670036023e-05,
+      "loss": 0.0365,
+      "step": 11276
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.894367932266459e-05,
+      "loss": 0.0292,
+      "step": 11277
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.891935495597352e-05,
+      "loss": 0.064,
+      "step": 11278
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.88950336011907e-05,
+      "loss": 0.036,
+      "step": 11279
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.887071525921954e-05,
+      "loss": 0.0411,
+      "step": 11280
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.884639993096336e-05,
+      "loss": 0.0321,
+      "step": 11281
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.882208761732548e-05,
+      "loss": 0.0483,
+      "step": 11282
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.879777831920894e-05,
+      "loss": 0.0457,
+      "step": 11283
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.877347203751688e-05,
+      "loss": 0.0496,
+      "step": 11284
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.874916877315211e-05,
+      "loss": 0.029,
+      "step": 11285
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.872486852701748e-05,
+      "loss": 0.0675,
+      "step": 11286
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.870057130001559e-05,
+      "loss": 0.0459,
+      "step": 11287
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.867627709304906e-05,
+      "loss": 0.0186,
+      "step": 11288
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.86519859070204e-05,
+      "loss": 0.052,
+      "step": 11289
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.862769774283194e-05,
+      "loss": 0.0465,
+      "step": 11290
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.860341260138586e-05,
+      "loss": 0.072,
+      "step": 11291
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.857913048358426e-05,
+      "loss": 0.0151,
+      "step": 11292
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.855485139032927e-05,
+      "loss": 0.0314,
+      "step": 11293
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.853057532252265e-05,
+      "loss": 0.029,
+      "step": 11294
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.850630228106629e-05,
+      "loss": 0.0385,
+      "step": 11295
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.848203226686184e-05,
+      "loss": 0.049,
+      "step": 11296
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.845776528081078e-05,
+      "loss": 0.0487,
+      "step": 11297
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.843350132381468e-05,
+      "loss": 0.0366,
+      "step": 11298
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.840924039677475e-05,
+      "loss": 0.0236,
+      "step": 11299
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.838498250059235e-05,
+      "loss": 0.0481,
+      "step": 11300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.83607276361685e-05,
+      "loss": 0.0251,
+      "step": 11301
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.833647580440421e-05,
+      "loss": 0.0233,
+      "step": 11302
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.831222700620032e-05,
+      "loss": 0.0343,
+      "step": 11303
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.82879812424577e-05,
+      "loss": 0.0313,
+      "step": 11304
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.82637385140769e-05,
+      "loss": 0.0144,
+      "step": 11305
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.823949882195859e-05,
+      "loss": 0.0143,
+      "step": 11306
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.821526216700314e-05,
+      "loss": 0.0531,
+      "step": 11307
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.819102855011079e-05,
+      "loss": 0.051,
+      "step": 11308
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.81667979721819e-05,
+      "loss": 0.0092,
+      "step": 11309
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.814257043411645e-05,
+      "loss": 0.0847,
+      "step": 11310
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.81183459368145e-05,
+      "loss": 0.0526,
+      "step": 11311
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.809412448117588e-05,
+      "loss": 0.067,
+      "step": 11312
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.80699060681003e-05,
+      "loss": 0.0328,
+      "step": 11313
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.804569069848752e-05,
+      "loss": 0.0323,
+      "step": 11314
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.802147837323693e-05,
+      "loss": 0.0257,
+      "step": 11315
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.799726909324812e-05,
+      "loss": 0.0477,
+      "step": 11316
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.797306285942019e-05,
+      "loss": 0.045,
+      "step": 11317
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.79488596726525e-05,
+      "loss": 0.0392,
+      "step": 11318
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.792465953384398e-05,
+      "loss": 0.0446,
+      "step": 11319
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.790046244389375e-05,
+      "loss": 0.03,
+      "step": 11320
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.787626840370059e-05,
+      "loss": 0.032,
+      "step": 11321
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.785207741416317e-05,
+      "loss": 0.0775,
+      "step": 11322
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.782788947618022e-05,
+      "loss": 0.0638,
+      "step": 11323
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.780370459065016e-05,
+      "loss": 0.0588,
+      "step": 11324
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.777952275847151e-05,
+      "loss": 0.0443,
+      "step": 11325
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.775534398054246e-05,
+      "loss": 0.0366,
+      "step": 11326
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.77311682577612e-05,
+      "loss": 0.0405,
+      "step": 11327
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.770699559102574e-05,
+      "loss": 0.0727,
+      "step": 11328
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.768282598123414e-05,
+      "loss": 0.0395,
+      "step": 11329
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.765865942928408e-05,
+      "loss": 0.0559,
+      "step": 11330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.763449593607343e-05,
+      "loss": 0.0309,
+      "step": 11331
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.761033550249972e-05,
+      "loss": 0.043,
+      "step": 11332
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.758617812946039e-05,
+      "loss": 0.0267,
+      "step": 11333
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.75620238178529e-05,
+      "loss": 0.0354,
+      "step": 11334
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.753787256857444e-05,
+      "loss": 0.0445,
+      "step": 11335
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.751372438252224e-05,
+      "loss": 0.0497,
+      "step": 11336
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.748957926059329e-05,
+      "loss": 0.0312,
+      "step": 11337
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.746543720368446e-05,
+      "loss": 0.051,
+      "step": 11338
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.744129821269268e-05,
+      "loss": 0.0686,
+      "step": 11339
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.741716228851449e-05,
+      "loss": 0.0304,
+      "step": 11340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.739302943204667e-05,
+      "loss": 0.0513,
+      "step": 11341
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.736889964418545e-05,
+      "loss": 0.0258,
+      "step": 11342
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.734477292582736e-05,
+      "loss": 0.0478,
+      "step": 11343
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.732064927786852e-05,
+      "loss": 0.0478,
+      "step": 11344
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.729652870120513e-05,
+      "loss": 0.0529,
+      "step": 11345
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.727241119673319e-05,
+      "loss": 0.0622,
+      "step": 11346
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.724829676534853e-05,
+      "loss": 0.0543,
+      "step": 11347
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.722418540794702e-05,
+      "loss": 0.0405,
+      "step": 11348
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.720007712542423e-05,
+      "loss": 0.0212,
+      "step": 11349
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.717597191867584e-05,
+      "loss": 0.0475,
+      "step": 11350
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.715186978859718e-05,
+      "loss": 0.0663,
+      "step": 11351
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.712777073608358e-05,
+      "loss": 0.0265,
+      "step": 11352
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.710367476203032e-05,
+      "loss": 0.0272,
+      "step": 11353
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.707958186733239e-05,
+      "loss": 0.0449,
+      "step": 11354
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.70554920528849e-05,
+      "loss": 0.0489,
+      "step": 11355
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.703140531958259e-05,
+      "loss": 0.0468,
+      "step": 11356
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.700732166832039e-05,
+      "loss": 0.0385,
+      "step": 11357
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.698324109999269e-05,
+      "loss": 0.007,
+      "step": 11358
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.69591636154942e-05,
+      "loss": 0.0265,
+      "step": 11359
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.69350892157192e-05,
+      "loss": 0.0188,
+      "step": 11360
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.69110179015621e-05,
+      "loss": 0.0773,
+      "step": 11361
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.688694967391706e-05,
+      "loss": 0.0371,
+      "step": 11362
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.686288453367802e-05,
+      "loss": 0.046,
+      "step": 11363
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.68388224817391e-05,
+      "loss": 0.0583,
+      "step": 11364
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.6814763518994e-05,
+      "loss": 0.0541,
+      "step": 11365
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.679070764633655e-05,
+      "loss": 0.0645,
+      "step": 11366
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.676665486466029e-05,
+      "loss": 0.0471,
+      "step": 11367
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.674260517485874e-05,
+      "loss": 0.0356,
+      "step": 11368
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.671855857782518e-05,
+      "loss": 0.0411,
+      "step": 11369
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.669451507445303e-05,
+      "loss": 0.043,
+      "step": 11370
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.66704746656353e-05,
+      "loss": 0.0498,
+      "step": 11371
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.66464373522651e-05,
+      "loss": 0.0526,
+      "step": 11372
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.662240313523534e-05,
+      "loss": 0.037,
+      "step": 11373
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.659837201543875e-05,
+      "loss": 0.0376,
+      "step": 11374
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.65743439937681e-05,
+      "loss": 0.0073,
+      "step": 11375
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.655031907111589e-05,
+      "loss": 0.0343,
+      "step": 11376
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.652629724837467e-05,
+      "loss": 0.0615,
+      "step": 11377
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.65022785264367e-05,
+      "loss": 0.0574,
+      "step": 11378
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.647826290619421e-05,
+      "loss": 0.0151,
+      "step": 11379
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.645425038853935e-05,
+      "loss": 0.0228,
+      "step": 11380
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.643024097436404e-05,
+      "loss": 0.0932,
+      "step": 11381
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.640623466456033e-05,
+      "loss": 0.0496,
+      "step": 11382
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.638223146001976e-05,
+      "loss": 0.0307,
+      "step": 11383
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.635823136163413e-05,
+      "loss": 0.034,
+      "step": 11384
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.633423437029488e-05,
+      "loss": 0.0373,
+      "step": 11385
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.631024048689352e-05,
+      "loss": 0.0461,
+      "step": 11386
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.628624971232132e-05,
+      "loss": 0.0239,
+      "step": 11387
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.62622620474694e-05,
+      "loss": 0.07,
+      "step": 11388
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.623827749322893e-05,
+      "loss": 0.0578,
+      "step": 11389
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.621429605049076e-05,
+      "loss": 0.0622,
+      "step": 11390
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.619031772014588e-05,
+      "loss": 0.0593,
+      "step": 11391
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.61663425030849e-05,
+      "loss": 0.0123,
+      "step": 11392
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.614237040019839e-05,
+      "loss": 0.0245,
+      "step": 11393
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.6118401412377e-05,
+      "loss": 0.0576,
+      "step": 11394
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.609443554051095e-05,
+      "loss": 0.0415,
+      "step": 11395
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.60704727854906e-05,
+      "loss": 0.0398,
+      "step": 11396
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.604651314820608e-05,
+      "loss": 0.0352,
+      "step": 11397
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.60225566295474e-05,
+      "loss": 0.0673,
+      "step": 11398
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.599860323040443e-05,
+      "loss": 0.0705,
+      "step": 11399
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.597465295166709e-05,
+      "loss": 0.0432,
+      "step": 11400
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.595070579422491e-05,
+      "loss": 0.0127,
+      "step": 11401
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.592676175896761e-05,
+      "loss": 0.0575,
+      "step": 11402
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.590282084678455e-05,
+      "loss": 0.064,
+      "step": 11403
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.587888305856504e-05,
+      "loss": 0.0225,
+      "step": 11404
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.585494839519839e-05,
+      "loss": 0.0677,
+      "step": 11405
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.583101685757359e-05,
+      "loss": 0.0515,
+      "step": 11406
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.580708844657983e-05,
+      "loss": 0.0565,
+      "step": 11407
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.578316316310571e-05,
+      "loss": 0.0446,
+      "step": 11408
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.575924100804016e-05,
+      "loss": 0.0141,
+      "step": 11409
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.573532198227174e-05,
+      "loss": 0.0501,
+      "step": 11410
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.571140608668905e-05,
+      "loss": 0.0472,
+      "step": 11411
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.568749332218044e-05,
+      "loss": 0.0291,
+      "step": 11412
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.566358368963417e-05,
+      "loss": 0.0627,
+      "step": 11413
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.56396771899385e-05,
+      "loss": 0.0378,
+      "step": 11414
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.561577382398139e-05,
+      "loss": 0.019,
+      "step": 11415
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.559187359265087e-05,
+      "loss": 0.039,
+      "step": 11416
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.556797649683472e-05,
+      "loss": 0.022,
+      "step": 11417
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.554408253742059e-05,
+      "loss": 0.0644,
+      "step": 11418
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.55201917152962e-05,
+      "loss": 0.0271,
+      "step": 11419
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.54963040313489e-05,
+      "loss": 0.0145,
+      "step": 11420
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.547241948646614e-05,
+      "loss": 0.0455,
+      "step": 11421
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.544853808153506e-05,
+      "loss": 0.0356,
+      "step": 11422
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.542465981744298e-05,
+      "loss": 0.0515,
+      "step": 11423
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.540078469507665e-05,
+      "loss": 0.0236,
+      "step": 11424
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.537691271532313e-05,
+      "loss": 0.0555,
+      "step": 11425
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.53530438790691e-05,
+      "loss": 0.0422,
+      "step": 11426
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.532917818720133e-05,
+      "loss": 0.0235,
+      "step": 11427
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.530531564060628e-05,
+      "loss": 0.0318,
+      "step": 11428
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.528145624017033e-05,
+      "loss": 0.0425,
+      "step": 11429
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.525759998677992e-05,
+      "loss": 0.0115,
+      "step": 11430
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.52337468813211e-05,
+      "loss": 0.0612,
+      "step": 11431
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.520989692468005e-05,
+      "loss": 0.0269,
+      "step": 11432
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.51860501177427e-05,
+      "loss": 0.0059,
+      "step": 11433
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.51622064613949e-05,
+      "loss": 0.0348,
+      "step": 11434
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.513836595652226e-05,
+      "loss": 0.0164,
+      "step": 11435
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.511452860401046e-05,
+      "loss": 0.0331,
+      "step": 11436
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.509069440474507e-05,
+      "loss": 0.0573,
+      "step": 11437
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.506686335961139e-05,
+      "loss": 0.0464,
+      "step": 11438
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.504303546949466e-05,
+      "loss": 0.0275,
+      "step": 11439
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.501921073527997e-05,
+      "loss": 0.0116,
+      "step": 11440
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.499538915785245e-05,
+      "loss": 0.0164,
+      "step": 11441
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.49715707380969e-05,
+      "loss": 0.0914,
+      "step": 11442
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.494775547689822e-05,
+      "loss": 0.0332,
+      "step": 11443
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.4923943375141e-05,
+      "loss": 0.0862,
+      "step": 11444
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.490013443370976e-05,
+      "loss": 0.0777,
+      "step": 11445
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.4876328653489e-05,
+      "loss": 0.0502,
+      "step": 11446
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.485252603536298e-05,
+      "loss": 0.038,
+      "step": 11447
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.4828726580216e-05,
+      "loss": 0.0353,
+      "step": 11448
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.4804930288932e-05,
+      "loss": 0.0381,
+      "step": 11449
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.478113716239504e-05,
+      "loss": 0.0593,
+      "step": 11450
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.475734720148886e-05,
+      "loss": 0.0261,
+      "step": 11451
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.473356040709735e-05,
+      "loss": 0.0339,
+      "step": 11452
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.470977678010399e-05,
+      "loss": 0.0519,
+      "step": 11453
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.468599632139229e-05,
+      "loss": 0.0269,
+      "step": 11454
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.466221903184569e-05,
+      "loss": 0.053,
+      "step": 11455
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.463844491234733e-05,
+      "loss": 0.0599,
+      "step": 11456
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.46146739637805e-05,
+      "loss": 0.0274,
+      "step": 11457
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.459090618702813e-05,
+      "loss": 0.0447,
+      "step": 11458
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.456714158297308e-05,
+      "loss": 0.0221,
+      "step": 11459
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.454338015249825e-05,
+      "loss": 0.0366,
+      "step": 11460
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.451962189648618e-05,
+      "loss": 0.0652,
+      "step": 11461
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.449586681581956e-05,
+      "loss": 0.0387,
+      "step": 11462
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.447211491138075e-05,
+      "loss": 0.0622,
+      "step": 11463
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.444836618405204e-05,
+      "loss": 0.0335,
+      "step": 11464
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.442462063471561e-05,
+      "loss": 0.0494,
+      "step": 11465
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.440087826425363e-05,
+      "loss": 0.0574,
+      "step": 11466
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.437713907354794e-05,
+      "loss": 0.0521,
+      "step": 11467
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.435340306348051e-05,
+      "loss": 0.0404,
+      "step": 11468
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.4329670234933e-05,
+      "loss": 0.0309,
+      "step": 11469
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.430594058878697e-05,
+      "loss": 0.0382,
+      "step": 11470
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.428221412592397e-05,
+      "loss": 0.047,
+      "step": 11471
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.425849084722533e-05,
+      "loss": 0.0429,
+      "step": 11472
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.423477075357242e-05,
+      "loss": 0.0513,
+      "step": 11473
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.421105384584617e-05,
+      "loss": 0.0726,
+      "step": 11474
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.418734012492773e-05,
+      "loss": 0.0355,
+      "step": 11475
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.416362959169794e-05,
+      "loss": 0.0503,
+      "step": 11476
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.413992224703757e-05,
+      "loss": 0.0377,
+      "step": 11477
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.411621809182741e-05,
+      "loss": 0.0496,
+      "step": 11478
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.409251712694781e-05,
+      "loss": 0.0604,
+      "step": 11479
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.406881935327931e-05,
+      "loss": 0.0422,
+      "step": 11480
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.404512477170214e-05,
+      "loss": 0.0439,
+      "step": 11481
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.402143338309657e-05,
+      "loss": 0.0228,
+      "step": 11482
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.399774518834257e-05,
+      "loss": 0.0466,
+      "step": 11483
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.39740601883202e-05,
+      "loss": 0.0623,
+      "step": 11484
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.39503783839092e-05,
+      "loss": 0.0704,
+      "step": 11485
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.392669977598928e-05,
+      "loss": 0.0669,
+      "step": 11486
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.390302436544009e-05,
+      "loss": 0.0368,
+      "step": 11487
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.387935215314101e-05,
+      "loss": 0.0342,
+      "step": 11488
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.385568313997158e-05,
+      "loss": 0.0529,
+      "step": 11489
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.383201732681077e-05,
+      "loss": 0.0428,
+      "step": 11490
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.38083547145379e-05,
+      "loss": 0.0706,
+      "step": 11491
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.378469530403186e-05,
+      "loss": 0.0875,
+      "step": 11492
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.376103909617159e-05,
+      "loss": 0.0013,
+      "step": 11493
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.373738609183583e-05,
+      "loss": 0.0562,
+      "step": 11494
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.371373629190317e-05,
+      "loss": 0.0159,
+      "step": 11495
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.369008969725224e-05,
+      "loss": 0.0447,
+      "step": 11496
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.36664463087613e-05,
+      "loss": 0.0212,
+      "step": 11497
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.364280612730877e-05,
+      "loss": 0.0609,
+      "step": 11498
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.361916915377275e-05,
+      "loss": 0.0349,
+      "step": 11499
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.359553538903122e-05,
+      "loss": 0.0422,
+      "step": 11500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.357190483396225e-05,
+      "loss": 0.0681,
+      "step": 11501
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.35482774894435e-05,
+      "loss": 0.0546,
+      "step": 11502
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.352465335635277e-05,
+      "loss": 0.0345,
+      "step": 11503
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.35010324355676e-05,
+      "loss": 0.0626,
+      "step": 11504
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.347741472796542e-05,
+      "loss": 0.0464,
+      "step": 11505
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.345380023442348e-05,
+      "loss": 0.0454,
+      "step": 11506
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.343018895581913e-05,
+      "loss": 0.028,
+      "step": 11507
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.340658089302934e-05,
+      "loss": 0.0601,
+      "step": 11508
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.33829760469312e-05,
+      "loss": 0.0193,
+      "step": 11509
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.33593744184015e-05,
+      "loss": 0.034,
+      "step": 11510
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.33357760083169e-05,
+      "loss": 0.0474,
+      "step": 11511
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.331218081755415e-05,
+      "loss": 0.0304,
+      "step": 11512
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.328858884698961e-05,
+      "loss": 0.0247,
+      "step": 11513
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.326500009749983e-05,
+      "loss": 0.0791,
+      "step": 11514
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.324141456996083e-05,
+      "loss": 0.0326,
+      "step": 11515
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.321783226524892e-05,
+      "loss": 0.0389,
+      "step": 11516
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.319425318423999e-05,
+      "loss": 0.0666,
+      "step": 11517
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.317067732781002e-05,
+      "loss": 0.0654,
+      "step": 11518
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.314710469683487e-05,
+      "loss": 0.0333,
+      "step": 11519
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.312353529218997e-05,
+      "loss": 0.0525,
+      "step": 11520
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.309996911475101e-05,
+      "loss": 0.0847,
+      "step": 11521
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.307640616539333e-05,
+      "loss": 0.0229,
+      "step": 11522
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.305284644499231e-05,
+      "loss": 0.0591,
+      "step": 11523
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.302928995442308e-05,
+      "loss": 0.0268,
+      "step": 11524
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.300573669456062e-05,
+      "loss": 0.063,
+      "step": 11525
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.298218666628e-05,
+      "loss": 0.0561,
+      "step": 11526
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.29586398704559e-05,
+      "loss": 0.0581,
+      "step": 11527
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.293509630796314e-05,
+      "loss": 0.0364,
+      "step": 11528
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.291155597967625e-05,
+      "loss": 0.0702,
+      "step": 11529
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.288801888646966e-05,
+      "loss": 0.0547,
+      "step": 11530
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.286448502921766e-05,
+      "loss": 0.0636,
+      "step": 11531
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.284095440879458e-05,
+      "loss": 0.0149,
+      "step": 11532
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.281742702607439e-05,
+      "loss": 0.0265,
+      "step": 11533
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.279390288193117e-05,
+      "loss": 0.0355,
+      "step": 11534
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.277038197723875e-05,
+      "loss": 0.0557,
+      "step": 11535
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.274686431287077e-05,
+      "loss": 0.0451,
+      "step": 11536
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.272334988970098e-05,
+      "loss": 0.0338,
+      "step": 11537
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.269983870860275e-05,
+      "loss": 0.0283,
+      "step": 11538
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.267633077044961e-05,
+      "loss": 0.0249,
+      "step": 11539
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.265282607611459e-05,
+      "loss": 0.0337,
+      "step": 11540
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.262932462647094e-05,
+      "loss": 0.0443,
+      "step": 11541
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.260582642239173e-05,
+      "loss": 0.0867,
+      "step": 11542
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.258233146474972e-05,
+      "loss": 0.0443,
+      "step": 11543
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.255883975441782e-05,
+      "loss": 0.0429,
+      "step": 11544
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.253535129226859e-05,
+      "loss": 0.0546,
+      "step": 11545
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.251186607917459e-05,
+      "loss": 0.0258,
+      "step": 11546
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.248838411600813e-05,
+      "loss": 0.0283,
+      "step": 11547
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.246490540364164e-05,
+      "loss": 0.0284,
+      "step": 11548
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.244142994294716e-05,
+      "loss": 0.0402,
+      "step": 11549
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.241795773479686e-05,
+      "loss": 0.0521,
+      "step": 11550
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.239448878006258e-05,
+      "loss": 0.018,
+      "step": 11551
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.237102307961612e-05,
+      "loss": 0.0489,
+      "step": 11552
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.23475606343292e-05,
+      "loss": 0.0486,
+      "step": 11553
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.232410144507332e-05,
+      "loss": 0.0617,
+      "step": 11554
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.230064551272008e-05,
+      "loss": 0.0583,
+      "step": 11555
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.227719283814057e-05,
+      "loss": 0.0302,
+      "step": 11556
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.225374342220616e-05,
+      "loss": 0.0248,
+      "step": 11557
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.223029726578781e-05,
+      "loss": 0.0496,
+      "step": 11558
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.220685436975654e-05,
+      "loss": 0.0587,
+      "step": 11559
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.218341473498328e-05,
+      "loss": 0.0512,
+      "step": 11560
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.215997836233853e-05,
+      "loss": 0.0915,
+      "step": 11561
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.213654525269304e-05,
+      "loss": 0.0534,
+      "step": 11562
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.211311540691721e-05,
+      "loss": 0.0257,
+      "step": 11563
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.208968882588146e-05,
+      "loss": 0.0724,
+      "step": 11564
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.206626551045599e-05,
+      "loss": 0.0485,
+      "step": 11565
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.20428454615108e-05,
+      "loss": 0.0533,
+      "step": 11566
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.201942867991604e-05,
+      "loss": 0.0306,
+      "step": 11567
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.199601516654145e-05,
+      "loss": 0.0317,
+      "step": 11568
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.19726049222569e-05,
+      "loss": 0.0509,
+      "step": 11569
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.194919794793192e-05,
+      "loss": 0.0312,
+      "step": 11570
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.192579424443604e-05,
+      "loss": 0.0542,
+      "step": 11571
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.190239381263856e-05,
+      "loss": 0.0306,
+      "step": 11572
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.187899665340888e-05,
+      "loss": 0.0204,
+      "step": 11573
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.1855602767616e-05,
+      "loss": 0.034,
+      "step": 11574
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.183221215612904e-05,
+      "loss": 0.0188,
+      "step": 11575
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.180882481981685e-05,
+      "loss": 0.0225,
+      "step": 11576
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.178544075954814e-05,
+      "loss": 0.038,
+      "step": 11577
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.176205997619169e-05,
+      "loss": 0.0521,
+      "step": 11578
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.173868247061589e-05,
+      "loss": 0.052,
+      "step": 11579
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.171530824368934e-05,
+      "loss": 0.0558,
+      "step": 11580
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.169193729628006e-05,
+      "loss": 0.0332,
+      "step": 11581
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.166856962925637e-05,
+      "loss": 0.0393,
+      "step": 11582
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.164520524348632e-05,
+      "loss": 0.0538,
+      "step": 11583
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.162184413983775e-05,
+      "loss": 0.042,
+      "step": 11584
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.159848631917864e-05,
+      "loss": 0.0248,
+      "step": 11585
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.157513178237639e-05,
+      "loss": 0.0461,
+      "step": 11586
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.155178053029874e-05,
+      "loss": 0.0113,
+      "step": 11587
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.152843256381304e-05,
+      "loss": 0.0301,
+      "step": 11588
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.150508788378666e-05,
+      "loss": 0.0305,
+      "step": 11589
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.148174649108679e-05,
+      "loss": 0.0519,
+      "step": 11590
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.145840838658038e-05,
+      "loss": 0.046,
+      "step": 11591
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.14350735711345e-05,
+      "loss": 0.0274,
+      "step": 11592
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.141174204561588e-05,
+      "loss": 0.0779,
+      "step": 11593
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.138841381089132e-05,
+      "loss": 0.0176,
+      "step": 11594
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.136508886782733e-05,
+      "loss": 0.059,
+      "step": 11595
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.134176721729037e-05,
+      "loss": 0.0239,
+      "step": 11596
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.131844886014671e-05,
+      "loss": 0.0335,
+      "step": 11597
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.129513379726267e-05,
+      "loss": 0.0216,
+      "step": 11598
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.127182202950424e-05,
+      "loss": 0.0719,
+      "step": 11599
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.124851355773748e-05,
+      "loss": 0.0556,
+      "step": 11600
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.122520838282818e-05,
+      "loss": 0.0576,
+      "step": 11601
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.120190650564201e-05,
+      "loss": 0.0588,
+      "step": 11602
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.117860792704466e-05,
+      "loss": 0.0364,
+      "step": 11603
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.115531264790151e-05,
+      "loss": 0.0323,
+      "step": 11604
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.1132020669078e-05,
+      "loss": 0.0419,
+      "step": 11605
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.110873199143934e-05,
+      "loss": 0.0301,
+      "step": 11606
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.108544661585055e-05,
+      "loss": 0.0027,
+      "step": 11607
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.106216454317672e-05,
+      "loss": 0.0567,
+      "step": 11608
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.103888577428264e-05,
+      "loss": 0.0272,
+      "step": 11609
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.101561031003311e-05,
+      "loss": 0.0725,
+      "step": 11610
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0992338151292735e-05,
+      "loss": 0.0411,
+      "step": 11611
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0969069298925964e-05,
+      "loss": 0.0112,
+      "step": 11612
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.094580375379715e-05,
+      "loss": 0.0141,
+      "step": 11613
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.092254151677062e-05,
+      "loss": 0.0453,
+      "step": 11614
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.089928258871041e-05,
+      "loss": 0.0806,
+      "step": 11615
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.087602697048061e-05,
+      "loss": 0.0245,
+      "step": 11616
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.085277466294506e-05,
+      "loss": 0.0128,
+      "step": 11617
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.082952566696746e-05,
+      "loss": 0.0751,
+      "step": 11618
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.080627998341154e-05,
+      "loss": 0.0001,
+      "step": 11619
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.07830376131407e-05,
+      "loss": 0.0215,
+      "step": 11620
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0759798557018504e-05,
+      "loss": 0.0462,
+      "step": 11621
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.073656281590798e-05,
+      "loss": 0.0378,
+      "step": 11622
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0713330390672394e-05,
+      "loss": 0.0255,
+      "step": 11623
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.06901012821748e-05,
+      "loss": 0.0316,
+      "step": 11624
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.066687549127799e-05,
+      "loss": 0.0605,
+      "step": 11625
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.064365301884489e-05,
+      "loss": 0.0134,
+      "step": 11626
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.062043386573795e-05,
+      "loss": 0.0473,
+      "step": 11627
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.059721803281984e-05,
+      "loss": 0.0265,
+      "step": 11628
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0574005520952856e-05,
+      "loss": 0.0169,
+      "step": 11629
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.055079633099938e-05,
+      "loss": 0.0432,
+      "step": 11630
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0527590463821505e-05,
+      "loss": 0.0687,
+      "step": 11631
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.050438792028122e-05,
+      "loss": 0.056,
+      "step": 11632
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.048118870124053e-05,
+      "loss": 0.0803,
+      "step": 11633
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0457992807561125e-05,
+      "loss": 0.0213,
+      "step": 11634
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0434800240104756e-05,
+      "loss": 0.0503,
+      "step": 11635
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.041161099973292e-05,
+      "loss": 0.0359,
+      "step": 11636
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0388425087306994e-05,
+      "loss": 0.0179,
+      "step": 11637
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0365242503688255e-05,
+      "loss": 0.0232,
+      "step": 11638
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.034206324973795e-05,
+      "loss": 0.038,
+      "step": 11639
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.031888732631704e-05,
+      "loss": 0.0495,
+      "step": 11640
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0295714734286524e-05,
+      "loss": 0.0344,
+      "step": 11641
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.0272545474507145e-05,
+      "loss": 0.0344,
+      "step": 11642
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.024937954783952e-05,
+      "loss": 0.0379,
+      "step": 11643
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0226216955144316e-05,
+      "loss": 0.0663,
+      "step": 11644
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0203057697281834e-05,
+      "loss": 0.037,
+      "step": 11645
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.017990177511249e-05,
+      "loss": 0.0365,
+      "step": 11646
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0156749189496395e-05,
+      "loss": 0.032,
+      "step": 11647
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.0133599941293566e-05,
+      "loss": 0.0277,
+      "step": 11648
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.011045403136399e-05,
+      "loss": 0.0926,
+      "step": 11649
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.008731146056742e-05,
+      "loss": 0.0411,
+      "step": 11650
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.006417222976367e-05,
+      "loss": 0.0253,
+      "step": 11651
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.004103633981206e-05,
+      "loss": 0.0346,
+      "step": 11652
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.001790379157223e-05,
+      "loss": 0.0492,
+      "step": 11653
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.999477458590333e-05,
+      "loss": 0.0109,
+      "step": 11654
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9971648723664663e-05,
+      "loss": 0.0749,
+      "step": 11655
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.994852620571526e-05,
+      "loss": 0.0248,
+      "step": 11656
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9925407032913965e-05,
+      "loss": 0.0416,
+      "step": 11657
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.990229120611972e-05,
+      "loss": 0.0473,
+      "step": 11658
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.987917872619109e-05,
+      "loss": 0.0137,
+      "step": 11659
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9856069593986746e-05,
+      "loss": 0.0249,
+      "step": 11660
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.983296381036507e-05,
+      "loss": 0.0578,
+      "step": 11661
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.980986137618438e-05,
+      "loss": 0.0426,
+      "step": 11662
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.978676229230281e-05,
+      "loss": 0.0376,
+      "step": 11663
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.976366655957846e-05,
+      "loss": 0.0383,
+      "step": 11664
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.974057417886934e-05,
+      "loss": 0.0127,
+      "step": 11665
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.971748515103322e-05,
+      "loss": 0.0541,
+      "step": 11666
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.969439947692776e-05,
+      "loss": 0.054,
+      "step": 11667
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9671317157410494e-05,
+      "loss": 0.0402,
+      "step": 11668
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9648238193338956e-05,
+      "loss": 0.0629,
+      "step": 11669
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9625162585570365e-05,
+      "loss": 0.0285,
+      "step": 11670
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9602090334962014e-05,
+      "loss": 0.0544,
+      "step": 11671
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9579021442370916e-05,
+      "loss": 0.0487,
+      "step": 11672
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9555955908653965e-05,
+      "loss": 0.0365,
+      "step": 11673
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.9532893734668085e-05,
+      "loss": 0.0385,
+      "step": 11674
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.950983492126985e-05,
+      "loss": 0.0486,
+      "step": 11675
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9486779469315935e-05,
+      "loss": 0.0023,
+      "step": 11676
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9463727379662744e-05,
+      "loss": 0.0473,
+      "step": 11677
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9440678653166575e-05,
+      "loss": 0.0287,
+      "step": 11678
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.941763329068358e-05,
+      "loss": 0.0538,
+      "step": 11679
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.939459129306995e-05,
+      "loss": 0.0516,
+      "step": 11680
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.937155266118148e-05,
+      "loss": 0.0739,
+      "step": 11681
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.934851739587412e-05,
+      "loss": 0.0405,
+      "step": 11682
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.932548549800352e-05,
+      "loss": 0.0351,
+      "step": 11683
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.930245696842517e-05,
+      "loss": 0.0468,
+      "step": 11684
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.927943180799461e-05,
+      "loss": 0.0635,
+      "step": 11685
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.92564100175671e-05,
+      "loss": 0.0375,
+      "step": 11686
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.92333915979979e-05,
+      "loss": 0.0525,
+      "step": 11687
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9210376550142017e-05,
+      "loss": 0.0392,
+      "step": 11688
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9187364874854364e-05,
+      "loss": 0.0147,
+      "step": 11689
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.916435657298985e-05,
+      "loss": 0.037,
+      "step": 11690
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9141351645403065e-05,
+      "loss": 0.0552,
+      "step": 11691
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.911835009294873e-05,
+      "loss": 0.032,
+      "step": 11692
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9095351916481086e-05,
+      "loss": 0.0065,
+      "step": 11693
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.907235711685459e-05,
+      "loss": 0.0407,
+      "step": 11694
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.904936569492332e-05,
+      "loss": 0.026,
+      "step": 11695
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.902637765154147e-05,
+      "loss": 0.0301,
+      "step": 11696
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.9003392987562906e-05,
+      "loss": 0.0164,
+      "step": 11697
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.898041170384138e-05,
+      "loss": 0.053,
+      "step": 11698
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8957433801230714e-05,
+      "loss": 0.0607,
+      "step": 11699
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.893445928058436e-05,
+      "loss": 0.043,
+      "step": 11700
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8911488142755843e-05,
+      "loss": 0.0397,
+      "step": 11701
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.888852038859842e-05,
+      "loss": 0.0308,
+      "step": 11702
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.886555601896528e-05,
+      "loss": 0.0517,
+      "step": 11703
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.884259503470943e-05,
+      "loss": 0.0424,
+      "step": 11704
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.881963743668386e-05,
+      "loss": 0.038,
+      "step": 11705
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8796683225741435e-05,
+      "loss": 0.0176,
+      "step": 11706
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.877373240273478e-05,
+      "loss": 0.0328,
+      "step": 11707
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.875078496851644e-05,
+      "loss": 0.0385,
+      "step": 11708
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.87278409239388e-05,
+      "loss": 0.0189,
+      "step": 11709
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.870490026985426e-05,
+      "loss": 0.0506,
+      "step": 11710
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.868196300711492e-05,
+      "loss": 0.025,
+      "step": 11711
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.865902913657291e-05,
+      "loss": 0.0376,
+      "step": 11712
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.863609865908011e-05,
+      "loss": 0.0385,
+      "step": 11713
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.861317157548827e-05,
+      "loss": 0.0238,
+      "step": 11714
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.859024788664916e-05,
+      "loss": 0.0195,
+      "step": 11715
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8567327593414245e-05,
+      "loss": 0.0392,
+      "step": 11716
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.854441069663509e-05,
+      "loss": 0.0301,
+      "step": 11717
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.852149719716278e-05,
+      "loss": 0.0285,
+      "step": 11718
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8498587095848636e-05,
+      "loss": 0.0451,
+      "step": 11719
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.847568039354362e-05,
+      "loss": 0.0517,
+      "step": 11720
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.845277709109872e-05,
+      "loss": 0.04,
+      "step": 11721
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.842987718936471e-05,
+      "loss": 0.0316,
+      "step": 11722
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.840698068919217e-05,
+      "loss": 0.0346,
+      "step": 11723
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.838408759143174e-05,
+      "loss": 0.0375,
+      "step": 11724
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.836119789693376e-05,
+      "loss": 0.0352,
+      "step": 11725
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.83383116065486e-05,
+      "loss": 0.0531,
+      "step": 11726
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.831542872112637e-05,
+      "loss": 0.0233,
+      "step": 11727
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.829254924151705e-05,
+      "loss": 0.0376,
+      "step": 11728
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.826967316857065e-05,
+      "loss": 0.0629,
+      "step": 11729
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8246800503136825e-05,
+      "loss": 0.0174,
+      "step": 11730
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8223931246065377e-05,
+      "loss": 0.0581,
+      "step": 11731
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.820106539820567e-05,
+      "loss": 0.0612,
+      "step": 11732
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8178202960407315e-05,
+      "loss": 0.0346,
+      "step": 11733
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.8155343933519334e-05,
+      "loss": 0.0639,
+      "step": 11734
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.813248831839106e-05,
+      "loss": 0.049,
+      "step": 11735
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.810963611587138e-05,
+      "loss": 0.0127,
+      "step": 11736
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.80867873268093e-05,
+      "loss": 0.0052,
+      "step": 11737
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.806394195205356e-05,
+      "loss": 0.0454,
+      "step": 11738
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.804109999245271e-05,
+      "loss": 0.0312,
+      "step": 11739
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.801826144885537e-05,
+      "loss": 0.0407,
+      "step": 11740
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.799542632210985e-05,
+      "loss": 0.0557,
+      "step": 11741
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.797259461306449e-05,
+      "loss": 0.0314,
+      "step": 11742
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.794976632256735e-05,
+      "loss": 0.0803,
+      "step": 11743
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.792694145146648e-05,
+      "loss": 0.043,
+      "step": 11744
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.790412000060967e-05,
+      "loss": 0.0578,
+      "step": 11745
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7881301970844736e-05,
+      "loss": 0.0298,
+      "step": 11746
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.785848736301936e-05,
+      "loss": 0.0409,
+      "step": 11747
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.783567617798096e-05,
+      "loss": 0.0369,
+      "step": 11748
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.781286841657693e-05,
+      "loss": 0.0359,
+      "step": 11749
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.779006407965444e-05,
+      "loss": 0.0158,
+      "step": 11750
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.776726316806073e-05,
+      "loss": 0.0041,
+      "step": 11751
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.774446568264267e-05,
+      "loss": 0.0469,
+      "step": 11752
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.772167162424722e-05,
+      "loss": 0.0476,
+      "step": 11753
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7698880993721066e-05,
+      "loss": 0.0726,
+      "step": 11754
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.767609379191075e-05,
+      "loss": 0.0344,
+      "step": 11755
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.765331001966289e-05,
+      "loss": 0.0111,
+      "step": 11756
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.763052967782368e-05,
+      "loss": 0.0697,
+      "step": 11757
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.760775276723953e-05,
+      "loss": 0.0462,
+      "step": 11758
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7584979288756325e-05,
+      "loss": 0.0586,
+      "step": 11759
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.75622092432202e-05,
+      "loss": 0.0309,
+      "step": 11760
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.753944263147687e-05,
+      "loss": 0.021,
+      "step": 11761
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.751667945437216e-05,
+      "loss": 0.0334,
+      "step": 11762
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.74939197127516e-05,
+      "loss": 0.0694,
+      "step": 11763
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7471163407460597e-05,
+      "loss": 0.0118,
+      "step": 11764
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.744841053934459e-05,
+      "loss": 0.0239,
+      "step": 11765
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.742566110924868e-05,
+      "loss": 0.0634,
+      "step": 11766
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.740291511801803e-05,
+      "loss": 0.0346,
+      "step": 11767
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7380172566497546e-05,
+      "loss": 0.0548,
+      "step": 11768
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7357433455532006e-05,
+      "loss": 0.0922,
+      "step": 11769
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.733469778596618e-05,
+      "loss": 0.0395,
+      "step": 11770
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.731196555864454e-05,
+      "loss": 0.0852,
+      "step": 11771
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7289236774411634e-05,
+      "loss": 0.0354,
+      "step": 11772
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.72665114341117e-05,
+      "loss": 0.0451,
+      "step": 11773
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.724378953858893e-05,
+      "loss": 0.0524,
+      "step": 11774
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.722107108868733e-05,
+      "loss": 0.031,
+      "step": 11775
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7198356085250916e-05,
+      "loss": 0.0498,
+      "step": 11776
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7175644529123385e-05,
+      "loss": 0.0362,
+      "step": 11777
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.71529364211485e-05,
+      "loss": 0.0575,
+      "step": 11778
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7130231762169755e-05,
+      "loss": 0.0655,
+      "step": 11779
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7107530553030505e-05,
+      "loss": 0.0281,
+      "step": 11780
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.708483279457415e-05,
+      "loss": 0.0395,
+      "step": 11781
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.706213848764374e-05,
+      "loss": 0.0345,
+      "step": 11782
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.703944763308244e-05,
+      "loss": 0.0386,
+      "step": 11783
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.701676023173295e-05,
+      "loss": 0.0347,
+      "step": 11784
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.699407628443821e-05,
+      "loss": 0.041,
+      "step": 11785
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.697139579204073e-05,
+      "loss": 0.0786,
+      "step": 11786
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6948718755383096e-05,
+      "loss": 0.0287,
+      "step": 11787
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.692604517530781e-05,
+      "loss": 0.0543,
+      "step": 11788
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6903375052656894e-05,
+      "loss": 0.0261,
+      "step": 11789
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6880708388272647e-05,
+      "loss": 0.0527,
+      "step": 11790
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.685804518299696e-05,
+      "loss": 0.0634,
+      "step": 11791
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.683538543767181e-05,
+      "loss": 0.05,
+      "step": 11792
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.681272915313888e-05,
+      "loss": 0.0653,
+      "step": 11793
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6790076330239747e-05,
+      "loss": 0.0162,
+      "step": 11794
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.676742696981599e-05,
+      "loss": 0.0506,
+      "step": 11795
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6744781072708875e-05,
+      "loss": 0.0558,
+      "step": 11796
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.672213863975971e-05,
+      "loss": 0.0401,
+      "step": 11797
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.669949967180954e-05,
+      "loss": 0.0069,
+      "step": 11798
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.667686416969944e-05,
+      "loss": 0.0539,
+      "step": 11799
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.665423213427005e-05,
+      "loss": 0.0501,
+      "step": 11800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.663160356636227e-05,
+      "loss": 0.0265,
+      "step": 11801
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.660897846681656e-05,
+      "loss": 0.0472,
+      "step": 11802
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6586356836473486e-05,
+      "loss": 0.047,
+      "step": 11803
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.656373867617332e-05,
+      "loss": 0.0629,
+      "step": 11804
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.654112398675623e-05,
+      "loss": 0.0527,
+      "step": 11805
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.651851276906235e-05,
+      "loss": 0.0497,
+      "step": 11806
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.649590502393153e-05,
+      "loss": 0.0106,
+      "step": 11807
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.647330075220371e-05,
+      "loss": 0.0702,
+      "step": 11808
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.645069995471849e-05,
+      "loss": 0.0694,
+      "step": 11809
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.642810263231539e-05,
+      "loss": 0.0301,
+      "step": 11810
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.640550878583394e-05,
+      "loss": 0.0747,
+      "step": 11811
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6382918416113305e-05,
+      "loss": 0.0473,
+      "step": 11812
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.636033152399279e-05,
+      "loss": 0.0272,
+      "step": 11813
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.633774811031135e-05,
+      "loss": 0.0428,
+      "step": 11814
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.631516817590792e-05,
+      "loss": 0.0559,
+      "step": 11815
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.629259172162119e-05,
+      "loss": 0.0285,
+      "step": 11816
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6270018748289943e-05,
+      "loss": 0.0452,
+      "step": 11817
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.624744925675259e-05,
+      "loss": 0.0435,
+      "step": 11818
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.622488324784763e-05,
+      "loss": 0.0091,
+      "step": 11819
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6202320722413256e-05,
+      "loss": 0.0378,
+      "step": 11820
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.617976168128755e-05,
+      "loss": 0.04,
+      "step": 11821
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.615720612530864e-05,
+      "loss": 0.0162,
+      "step": 11822
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.613465405531426e-05,
+      "loss": 0.023,
+      "step": 11823
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6112105472142346e-05,
+      "loss": 0.0256,
+      "step": 11824
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.608956037663027e-05,
+      "loss": 0.0279,
+      "step": 11825
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.606701876961571e-05,
+      "loss": 0.0338,
+      "step": 11826
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.604448065193586e-05,
+      "loss": 0.0556,
+      "step": 11827
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6021946024428056e-05,
+      "loss": 0.0493,
+      "step": 11828
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5999414887929476e-05,
+      "loss": 0.0894,
+      "step": 11829
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.597688724327686e-05,
+      "loss": 0.0295,
+      "step": 11830
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5954363091307224e-05,
+      "loss": 0.0108,
+      "step": 11831
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.593184243285716e-05,
+      "loss": 0.0409,
+      "step": 11832
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5909325268763336e-05,
+      "loss": 0.0618,
+      "step": 11833
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.588681159986217e-05,
+      "loss": 0.0386,
+      "step": 11834
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5864301426989914e-05,
+      "loss": 0.0173,
+      "step": 11835
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.584179475098287e-05,
+      "loss": 0.0558,
+      "step": 11836
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5819291572676975e-05,
+      "loss": 0.0934,
+      "step": 11837
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.579679189290827e-05,
+      "loss": 0.0286,
+      "step": 11838
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.57742957125125e-05,
+      "loss": 0.063,
+      "step": 11839
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5751803032325324e-05,
+      "loss": 0.019,
+      "step": 11840
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.572931385318224e-05,
+      "loss": 0.0576,
+      "step": 11841
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.570682817591874e-05,
+      "loss": 0.0491,
+      "step": 11842
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.568434600137001e-05,
+      "loss": 0.0452,
+      "step": 11843
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.566186733037131e-05,
+      "loss": 0.0625,
+      "step": 11844
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.563939216375759e-05,
+      "loss": 0.0379,
+      "step": 11845
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.561692050236369e-05,
+      "loss": 0.0244,
+      "step": 11846
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5594452347024465e-05,
+      "loss": 0.0207,
+      "step": 11847
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.557198769857444e-05,
+      "loss": 0.0375,
+      "step": 11848
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5549526557848274e-05,
+      "loss": 0.0221,
+      "step": 11849
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.552706892568012e-05,
+      "loss": 0.0232,
+      "step": 11850
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.550461480290431e-05,
+      "loss": 0.036,
+      "step": 11851
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5482164190355005e-05,
+      "loss": 0.0276,
+      "step": 11852
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.545971708886608e-05,
+      "loss": 0.0423,
+      "step": 11853
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.543727349927153e-05,
+      "loss": 0.0504,
+      "step": 11854
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.541483342240487e-05,
+      "loss": 0.0321,
+      "step": 11855
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.539239685909982e-05,
+      "loss": 0.0368,
+      "step": 11856
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.536996381018976e-05,
+      "loss": 0.014,
+      "step": 11857
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5347534276508076e-05,
+      "loss": 0.0195,
+      "step": 11858
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.532510825888789e-05,
+      "loss": 0.0444,
+      "step": 11859
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.530268575816233e-05,
+      "loss": 0.0287,
+      "step": 11860
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.528026677516433e-05,
+      "loss": 0.0443,
+      "step": 11861
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.525785131072658e-05,
+      "loss": 0.0382,
+      "step": 11862
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.523543936568188e-05,
+      "loss": 0.0521,
+      "step": 11863
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5213030940862666e-05,
+      "loss": 0.0511,
+      "step": 11864
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.519062603710149e-05,
+      "loss": 0.0442,
+      "step": 11865
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.516822465523044e-05,
+      "loss": 0.0626,
+      "step": 11866
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.514582679608178e-05,
+      "loss": 0.0324,
+      "step": 11867
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.512343246048746e-05,
+      "loss": 0.0488,
+      "step": 11868
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.51010416492794e-05,
+      "loss": 0.037,
+      "step": 11869
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.507865436328946e-05,
+      "loss": 0.0817,
+      "step": 11870
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5056270603349064e-05,
+      "loss": 0.0118,
+      "step": 11871
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.503389037028983e-05,
+      "loss": 0.0286,
+      "step": 11872
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.501151366494304e-05,
+      "loss": 0.0515,
+      "step": 11873
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.498914048814e-05,
+      "loss": 0.0393,
+      "step": 11874
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4966770840711786e-05,
+      "loss": 0.0323,
+      "step": 11875
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.49444047234893e-05,
+      "loss": 0.0558,
+      "step": 11876
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.492204213730349e-05,
+      "loss": 0.0216,
+      "step": 11877
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.489968308298493e-05,
+      "loss": 0.0416,
+      "step": 11878
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.487732756136433e-05,
+      "loss": 0.0343,
+      "step": 11879
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4854975573272055e-05,
+      "loss": 0.0423,
+      "step": 11880
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.483262711953845e-05,
+      "loss": 0.057,
+      "step": 11881
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.48102822009936e-05,
+      "loss": 0.0336,
+      "step": 11882
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.478794081846768e-05,
+      "loss": 0.0211,
+      "step": 11883
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.476560297279051e-05,
+      "loss": 0.0298,
+      "step": 11884
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.474326866479196e-05,
+      "loss": 0.0516,
+      "step": 11885
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4720937895301636e-05,
+      "loss": 0.0457,
+      "step": 11886
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.469861066514902e-05,
+      "loss": 0.0147,
+      "step": 11887
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4676286975163606e-05,
+      "loss": 0.0303,
+      "step": 11888
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4653966826174526e-05,
+      "loss": 0.0386,
+      "step": 11889
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.46316502190111e-05,
+      "loss": 0.0497,
+      "step": 11890
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.460933715450209e-05,
+      "loss": 0.0528,
+      "step": 11891
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4587027633476485e-05,
+      "loss": 0.0675,
+      "step": 11892
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4564721656763056e-05,
+      "loss": 0.0457,
+      "step": 11893
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4542419225190296e-05,
+      "loss": 0.0456,
+      "step": 11894
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.452012033958683e-05,
+      "loss": 0.036,
+      "step": 11895
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.449782500078083e-05,
+      "loss": 0.0003,
+      "step": 11896
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4475533209600594e-05,
+      "loss": 0.0382,
+      "step": 11897
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.445324496687413e-05,
+      "loss": 0.0397,
+      "step": 11898
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4430960273429484e-05,
+      "loss": 0.0562,
+      "step": 11899
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.440867913009441e-05,
+      "loss": 0.0498,
+      "step": 11900
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4386401537696536e-05,
+      "loss": 0.0443,
+      "step": 11901
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.436412749706352e-05,
+      "loss": 0.0704,
+      "step": 11902
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.434185700902264e-05,
+      "loss": 0.0259,
+      "step": 11903
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.431959007440134e-05,
+      "loss": 0.0425,
+      "step": 11904
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4297326694026675e-05,
+      "loss": 0.0479,
+      "step": 11905
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.427506686872567e-05,
+      "loss": 0.0482,
+      "step": 11906
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.425281059932517e-05,
+      "loss": 0.0676,
+      "step": 11907
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.4230557886652044e-05,
+      "loss": 0.0302,
+      "step": 11908
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.42083087315328e-05,
+      "loss": 0.0456,
+      "step": 11909
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.4186063134794e-05,
+      "loss": 0.0315,
+      "step": 11910
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.416382109726201e-05,
+      "loss": 0.0082,
+      "step": 11911
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.4141582619762983e-05,
+      "loss": 0.0237,
+      "step": 11912
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.41193477031231e-05,
+      "loss": 0.0426,
+      "step": 11913
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.409711634816823e-05,
+      "loss": 0.0625,
+      "step": 11914
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.407488855572431e-05,
+      "loss": 0.0316,
+      "step": 11915
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.405266432661698e-05,
+      "loss": 0.0456,
+      "step": 11916
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.4030443661671755e-05,
+      "loss": 0.0377,
+      "step": 11917
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.400822656171417e-05,
+      "loss": 0.0565,
+      "step": 11918
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.398601302756942e-05,
+      "loss": 0.0226,
+      "step": 11919
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.396380306006278e-05,
+      "loss": 0.0488,
+      "step": 11920
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.394159666001925e-05,
+      "loss": 0.0447,
+      "step": 11921
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.39193938282637e-05,
+      "loss": 0.0243,
+      "step": 11922
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3897194565620846e-05,
+      "loss": 0.0458,
+      "step": 11923
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.387499887291548e-05,
+      "loss": 0.019,
+      "step": 11924
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.385280675097195e-05,
+      "loss": 0.0617,
+      "step": 11925
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.383061820061476e-05,
+      "loss": 0.063,
+      "step": 11926
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.380843322266809e-05,
+      "loss": 0.0409,
+      "step": 11927
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3786251817955976e-05,
+      "loss": 0.0706,
+      "step": 11928
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3764073987302534e-05,
+      "loss": 0.0432,
+      "step": 11929
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.374189973153147e-05,
+      "loss": 0.0614,
+      "step": 11930
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3719729051466683e-05,
+      "loss": 0.0283,
+      "step": 11931
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.36975619479315e-05,
+      "loss": 0.0495,
+      "step": 11932
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.367539842174949e-05,
+      "loss": 0.0377,
+      "step": 11933
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3653238473744e-05,
+      "loss": 0.0913,
+      "step": 11934
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.363108210473813e-05,
+      "loss": 0.0417,
+      "step": 11935
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.360892931555505e-05,
+      "loss": 0.0174,
+      "step": 11936
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3586780107017495e-05,
+      "loss": 0.0532,
+      "step": 11937
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.356463447994837e-05,
+      "loss": 0.0529,
+      "step": 11938
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.354249243517023e-05,
+      "loss": 0.0428,
+      "step": 11939
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.352035397350569e-05,
+      "loss": 0.0457,
+      "step": 11940
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.349821909577707e-05,
+      "loss": 0.0272,
+      "step": 11941
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3476087802806564e-05,
+      "loss": 0.0591,
+      "step": 11942
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.345396009541641e-05,
+      "loss": 0.0517,
+      "step": 11943
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.343183597442846e-05,
+      "loss": 0.0704,
+      "step": 11944
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.340971544066468e-05,
+      "loss": 0.039,
+      "step": 11945
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.338759849494671e-05,
+      "loss": 0.0365,
+      "step": 11946
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.336548513809616e-05,
+      "loss": 0.045,
+      "step": 11947
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.334337537093439e-05,
+      "loss": 0.0385,
+      "step": 11948
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.332126919428284e-05,
+      "loss": 0.0239,
+      "step": 11949
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3299166608962574e-05,
+      "loss": 0.0355,
+      "step": 11950
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.327706761579475e-05,
+      "loss": 0.0454,
+      "step": 11951
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.325497221560023e-05,
+      "loss": 0.0463,
+      "step": 11952
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3232880409199735e-05,
+      "loss": 0.0372,
+      "step": 11953
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.321079219741401e-05,
+      "loss": 0.0475,
+      "step": 11954
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.31887075810635e-05,
+      "loss": 0.0161,
+      "step": 11955
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3166626560968635e-05,
+      "loss": 0.0582,
+      "step": 11956
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.314454913794963e-05,
+      "loss": 0.0424,
+      "step": 11957
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.312247531282655e-05,
+      "loss": 0.0307,
+      "step": 11958
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.310040508641947e-05,
+      "loss": 0.0548,
+      "step": 11959
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.307833845954814e-05,
+      "loss": 0.0273,
+      "step": 11960
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.305627543303242e-05,
+      "loss": 0.0075,
+      "step": 11961
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3034216007691674e-05,
+      "loss": 0.0758,
+      "step": 11962
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.3012160184345504e-05,
+      "loss": 0.0431,
+      "step": 11963
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.299010796381313e-05,
+      "loss": 0.0641,
+      "step": 11964
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.2968059346913795e-05,
+      "loss": 0.0548,
+      "step": 11965
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.294601433446653e-05,
+      "loss": 0.051,
+      "step": 11966
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.2923972927290166e-05,
+      "loss": 0.0378,
+      "step": 11967
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.290193512620359e-05,
+      "loss": 0.056,
+      "step": 11968
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.2879900932025326e-05,
+      "loss": 0.0225,
+      "step": 11969
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.285787034557399e-05,
+      "loss": 0.0398,
+      "step": 11970
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.283584336766792e-05,
+      "loss": 0.0336,
+      "step": 11971
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.281381999912533e-05,
+      "loss": 0.0535,
+      "step": 11972
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.279180024076427e-05,
+      "loss": 0.0562,
+      "step": 11973
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.276978409340278e-05,
+      "loss": 0.0521,
+      "step": 11974
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.274777155785875e-05,
+      "loss": 0.0635,
+      "step": 11975
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2725762634949794e-05,
+      "loss": 0.0694,
+      "step": 11976
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.270375732549351e-05,
+      "loss": 0.0596,
+      "step": 11977
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2681755630307273e-05,
+      "loss": 0.0206,
+      "step": 11978
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.265975755020848e-05,
+      "loss": 0.0873,
+      "step": 11979
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.26377630860142e-05,
+      "loss": 0.0192,
+      "step": 11980
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.261577223854154e-05,
+      "loss": 0.0482,
+      "step": 11981
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.259378500860739e-05,
+      "loss": 0.0051,
+      "step": 11982
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.257180139702843e-05,
+      "loss": 0.0399,
+      "step": 11983
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2549821404621385e-05,
+      "loss": 0.0223,
+      "step": 11984
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2527845032202646e-05,
+      "loss": 0.0719,
+      "step": 11985
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.25058722805887e-05,
+      "loss": 0.0264,
+      "step": 11986
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.248390315059569e-05,
+      "loss": 0.0681,
+      "step": 11987
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.246193764303971e-05,
+      "loss": 0.0425,
+      "step": 11988
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.243997575873667e-05,
+      "loss": 0.0562,
+      "step": 11989
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.24180174985025e-05,
+      "loss": 0.0694,
+      "step": 11990
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.239606286315276e-05,
+      "loss": 0.0407,
+      "step": 11991
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.237411185350311e-05,
+      "loss": 0.0439,
+      "step": 11992
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.235216447036893e-05,
+      "loss": 0.0568,
+      "step": 11993
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2330220714565435e-05,
+      "loss": 0.0496,
+      "step": 11994
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2308280586907876e-05,
+      "loss": 0.0478,
+      "step": 11995
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.228634408821118e-05,
+      "loss": 0.0446,
+      "step": 11996
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.226441121929028e-05,
+      "loss": 0.0531,
+      "step": 11997
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.224248198095991e-05,
+      "loss": 0.0776,
+      "step": 11998
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.222055637403462e-05,
+      "loss": 0.0509,
+      "step": 11999
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.2198634399328965e-05,
+      "loss": 0.0195,
+      "step": 12000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.21767160576572e-05,
+      "loss": 0.0451,
+      "step": 12001
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.215480134983366e-05,
+      "loss": 0.0609,
+      "step": 12002
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.213289027667223e-05,
+      "loss": 0.0356,
+      "step": 12003
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.211098283898698e-05,
+      "loss": 0.0627,
+      "step": 12004
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.2089079037591605e-05,
+      "loss": 0.0367,
+      "step": 12005
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.2067178873299874e-05,
+      "loss": 0.0429,
+      "step": 12006
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.2045282346925266e-05,
+      "loss": 0.0383,
+      "step": 12007
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.202338945928112e-05,
+      "loss": 0.0262,
+      "step": 12008
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.20015002111808e-05,
+      "loss": 0.0606,
+      "step": 12009
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.19796146034373e-05,
+      "loss": 0.0134,
+      "step": 12010
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1957732636863724e-05,
+      "loss": 0.041,
+      "step": 12011
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.19358543122729e-05,
+      "loss": 0.0376,
+      "step": 12012
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1913979630477496e-05,
+      "loss": 0.0417,
+      "step": 12013
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1892108592290064e-05,
+      "loss": 0.0094,
+      "step": 12014
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1870241198523146e-05,
+      "loss": 0.0347,
+      "step": 12015
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1848377449988964e-05,
+      "loss": 0.0531,
+      "step": 12016
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.182651734749978e-05,
+      "loss": 0.009,
+      "step": 12017
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.180466089186759e-05,
+      "loss": 0.0676,
+      "step": 12018
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.178280808390424e-05,
+      "loss": 0.0521,
+      "step": 12019
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.176095892442159e-05,
+      "loss": 0.0646,
+      "step": 12020
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.173911341423119e-05,
+      "loss": 0.0413,
+      "step": 12021
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.171727155414465e-05,
+      "loss": 0.0374,
+      "step": 12022
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.169543334497323e-05,
+      "loss": 0.0329,
+      "step": 12023
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1673598787528145e-05,
+      "loss": 0.0411,
+      "step": 12024
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1651767882620585e-05,
+      "loss": 0.0161,
+      "step": 12025
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.162994063106137e-05,
+      "loss": 0.0266,
+      "step": 12026
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.160811703366151e-05,
+      "loss": 0.051,
+      "step": 12027
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.158629709123147e-05,
+      "loss": 0.0271,
+      "step": 12028
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1564480804581935e-05,
+      "loss": 0.039,
+      "step": 12029
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1542668174523254e-05,
+      "loss": 0.0181,
+      "step": 12030
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.152085920186576e-05,
+      "loss": 0.0338,
+      "step": 12031
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1499053887419554e-05,
+      "loss": 0.034,
+      "step": 12032
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.147725223199461e-05,
+      "loss": 0.0555,
+      "step": 12033
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.145545423640087e-05,
+      "loss": 0.0441,
+      "step": 12034
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.1433659901447966e-05,
+      "loss": 0.0145,
+      "step": 12035
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.141186922794561e-05,
+      "loss": 0.0243,
+      "step": 12036
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.139008221670321e-05,
+      "loss": 0.0419,
+      "step": 12037
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.136829886853003e-05,
+      "loss": 0.0275,
+      "step": 12038
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1346519184235364e-05,
+      "loss": 0.0108,
+      "step": 12039
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.132474316462817e-05,
+      "loss": 0.028,
+      "step": 12040
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.130297081051745e-05,
+      "loss": 0.0089,
+      "step": 12041
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.128120212271193e-05,
+      "loss": 0.0578,
+      "step": 12042
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.125943710202027e-05,
+      "loss": 0.0262,
+      "step": 12043
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.12376757492509e-05,
+      "loss": 0.0531,
+      "step": 12044
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1215918065212336e-05,
+      "loss": 0.0476,
+      "step": 12045
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.119416405071266e-05,
+      "loss": 0.0089,
+      "step": 12046
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1172413706560096e-05,
+      "loss": 0.0082,
+      "step": 12047
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.115066703356256e-05,
+      "loss": 0.0278,
+      "step": 12048
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1128924032527824e-05,
+      "loss": 0.0389,
+      "step": 12049
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1107184704263674e-05,
+      "loss": 0.0231,
+      "step": 12050
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1085449049577555e-05,
+      "loss": 0.048,
+      "step": 12051
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.106371706927699e-05,
+      "loss": 0.0549,
+      "step": 12052
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.10419887641692e-05,
+      "loss": 0.0349,
+      "step": 12053
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.1020264135061346e-05,
+      "loss": 0.0151,
+      "step": 12054
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.099854318276038e-05,
+      "loss": 0.0429,
+      "step": 12055
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.097682590807325e-05,
+      "loss": 0.0582,
+      "step": 12056
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.095511231180662e-05,
+      "loss": 0.0578,
+      "step": 12057
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.093340239476716e-05,
+      "loss": 0.0086,
+      "step": 12058
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0911696157761294e-05,
+      "loss": 0.0453,
+      "step": 12059
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.088999360159528e-05,
+      "loss": 0.0689,
+      "step": 12060
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0868294727075436e-05,
+      "loss": 0.0312,
+      "step": 12061
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.084659953500767e-05,
+      "loss": 0.0491,
+      "step": 12062
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0824908026198035e-05,
+      "loss": 0.037,
+      "step": 12063
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.080322020145224e-05,
+      "loss": 0.0564,
+      "step": 12064
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.078153606157585e-05,
+      "loss": 0.0001,
+      "step": 12065
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.0759855607374514e-05,
+      "loss": 0.0261,
+      "step": 12066
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.073817883965346e-05,
+      "loss": 0.0462,
+      "step": 12067
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.071650575921808e-05,
+      "loss": 0.0429,
+      "step": 12068
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.069483636687327e-05,
+      "loss": 0.0227,
+      "step": 12069
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.067317066342412e-05,
+      "loss": 0.029,
+      "step": 12070
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0651508649675365e-05,
+      "loss": 0.0667,
+      "step": 12071
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.062985032643178e-05,
+      "loss": 0.0437,
+      "step": 12072
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0608195694497845e-05,
+      "loss": 0.0361,
+      "step": 12073
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0586544754677924e-05,
+      "loss": 0.0398,
+      "step": 12074
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.05648975077764e-05,
+      "loss": 0.025,
+      "step": 12075
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.054325395459728e-05,
+      "loss": 0.0588,
+      "step": 12076
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.052161409594468e-05,
+      "loss": 0.0284,
+      "step": 12077
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0499977932622386e-05,
+      "loss": 0.0061,
+      "step": 12078
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.047834546543409e-05,
+      "loss": 0.0252,
+      "step": 12079
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.045671669518347e-05,
+      "loss": 0.0209,
+      "step": 12080
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0435091622673865e-05,
+      "loss": 0.0249,
+      "step": 12081
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.041347024870868e-05,
+      "loss": 0.0495,
+      "step": 12082
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0391852574091044e-05,
+      "loss": 0.0407,
+      "step": 12083
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0370238599623986e-05,
+      "loss": 0.0735,
+      "step": 12084
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.034862832611034e-05,
+      "loss": 0.0264,
+      "step": 12085
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.032702175435298e-05,
+      "loss": 0.034,
+      "step": 12086
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.030541888515443e-05,
+      "loss": 0.0171,
+      "step": 12087
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.028381971931728e-05,
+      "loss": 0.0571,
+      "step": 12088
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.026222425764379e-05,
+      "loss": 0.0428,
+      "step": 12089
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.024063250093615e-05,
+      "loss": 0.0685,
+      "step": 12090
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.021904444999652e-05,
+      "loss": 0.0287,
+      "step": 12091
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.019746010562672e-05,
+      "loss": 0.0556,
+      "step": 12092
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.017587946862874e-05,
+      "loss": 0.06,
+      "step": 12093
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.015430253980397e-05,
+      "loss": 0.0779,
+      "step": 12094
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.013272931995415e-05,
+      "loss": 0.0197,
+      "step": 12095
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.011115980988049e-05,
+      "loss": 0.0522,
+      "step": 12096
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.00895940103844e-05,
+      "loss": 0.0299,
+      "step": 12097
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.0068031922266895e-05,
+      "loss": 0.0542,
+      "step": 12098
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.004647354632892e-05,
+      "loss": 0.0245,
+      "step": 12099
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.002491888337137e-05,
+      "loss": 0.0468,
+      "step": 12100
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.000336793419488e-05,
+      "loss": 0.0473,
+      "step": 12101
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.998182069960007e-05,
+      "loss": 0.033,
+      "step": 12102
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.9960277180387315e-05,
+      "loss": 0.0456,
+      "step": 12103
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.993873737735687e-05,
+      "loss": 0.0556,
+      "step": 12104
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.991720129130894e-05,
+      "loss": 0.0625,
+      "step": 12105
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.989566892304344e-05,
+      "loss": 0.065,
+      "step": 12106
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9874140273360334e-05,
+      "loss": 0.0501,
+      "step": 12107
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.985261534305925e-05,
+      "loss": 0.0525,
+      "step": 12108
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.983109413293994e-05,
+      "loss": 0.0434,
+      "step": 12109
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.980957664380164e-05,
+      "loss": 0.0252,
+      "step": 12110
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9788062876443796e-05,
+      "loss": 0.0557,
+      "step": 12111
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9766552831665514e-05,
+      "loss": 0.022,
+      "step": 12112
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.97450465102659e-05,
+      "loss": 0.0248,
+      "step": 12113
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.972354391304381e-05,
+      "loss": 0.0504,
+      "step": 12114
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9702045040797975e-05,
+      "loss": 0.0372,
+      "step": 12115
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.968054989432707e-05,
+      "loss": 0.0463,
+      "step": 12116
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9659058474429515e-05,
+      "loss": 0.0378,
+      "step": 12117
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.963757078190375e-05,
+      "loss": 0.0234,
+      "step": 12118
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.96160868175479e-05,
+      "loss": 0.0389,
+      "step": 12119
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.959460658216006e-05,
+      "loss": 0.015,
+      "step": 12120
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.957313007653809e-05,
+      "loss": 0.0459,
+      "step": 12121
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9551657301479854e-05,
+      "loss": 0.0458,
+      "step": 12122
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9530188257783035e-05,
+      "loss": 0.0503,
+      "step": 12123
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.950872294624509e-05,
+      "loss": 0.0204,
+      "step": 12124
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.94872613676634e-05,
+      "loss": 0.0444,
+      "step": 12125
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9465803522835144e-05,
+      "loss": 0.0451,
+      "step": 12126
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9444349412557545e-05,
+      "loss": 0.0283,
+      "step": 12127
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.942289903762742e-05,
+      "loss": 0.0216,
+      "step": 12128
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.940145239884171e-05,
+      "loss": 0.0495,
+      "step": 12129
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.938000949699704e-05,
+      "loss": 0.0137,
+      "step": 12130
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.935857033288989e-05,
+      "loss": 0.052,
+      "step": 12131
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.933713490731679e-05,
+      "loss": 0.0035,
+      "step": 12132
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.931570322107388e-05,
+      "loss": 0.0761,
+      "step": 12133
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.929427527495745e-05,
+      "loss": 0.0395,
+      "step": 12134
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.927285106976326e-05,
+      "loss": 0.0491,
+      "step": 12135
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.9251430606287335e-05,
+      "loss": 0.0928,
+      "step": 12136
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.923001388532528e-05,
+      "loss": 0.0354,
+      "step": 12137
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.920860090767273e-05,
+      "loss": 0.0213,
+      "step": 12138
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.9187191674125094e-05,
+      "loss": 0.0747,
+      "step": 12139
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.9165786185477614e-05,
+      "loss": 0.0433,
+      "step": 12140
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.914438444252554e-05,
+      "loss": 0.0286,
+      "step": 12141
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.9122986446063764e-05,
+      "loss": 0.0389,
+      "step": 12142
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.910159219688727e-05,
+      "loss": 0.0183,
+      "step": 12143
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.9080201695790757e-05,
+      "loss": 0.046,
+      "step": 12144
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.9058814943568754e-05,
+      "loss": 0.0372,
+      "step": 12145
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.903743194101581e-05,
+      "loss": 0.0364,
+      "step": 12146
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.901605268892615e-05,
+      "loss": 0.024,
+      "step": 12147
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.899467718809406e-05,
+      "loss": 0.0457,
+      "step": 12148
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.897330543931349e-05,
+      "loss": 0.0095,
+      "step": 12149
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8951937443378396e-05,
+      "loss": 0.0777,
+      "step": 12150
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.893057320108243e-05,
+      "loss": 0.0297,
+      "step": 12151
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8909212713219355e-05,
+      "loss": 0.0572,
+      "step": 12152
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.888785598058252e-05,
+      "loss": 0.0351,
+      "step": 12153
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.886650300396538e-05,
+      "loss": 0.045,
+      "step": 12154
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.884515378416108e-05,
+      "loss": 0.0389,
+      "step": 12155
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.882380832196265e-05,
+      "loss": 0.0665,
+      "step": 12156
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.880246661816307e-05,
+      "loss": 0.0381,
+      "step": 12157
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.878112867355505e-05,
+      "loss": 0.0521,
+      "step": 12158
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.875979448893139e-05,
+      "loss": 0.0543,
+      "step": 12159
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.873846406508437e-05,
+      "loss": 0.0254,
+      "step": 12160
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.871713740280652e-05,
+      "loss": 0.0122,
+      "step": 12161
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.869581450288996e-05,
+      "loss": 0.0383,
+      "step": 12162
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.867449536612681e-05,
+      "loss": 0.0383,
+      "step": 12163
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8653179993309126e-05,
+      "loss": 0.0145,
+      "step": 12164
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.86318683852285e-05,
+      "loss": 0.0129,
+      "step": 12165
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.8610560542676766e-05,
+      "loss": 0.0581,
+      "step": 12166
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.858925646644533e-05,
+      "loss": 0.0194,
+      "step": 12167
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.856795615732569e-05,
+      "loss": 0.076,
+      "step": 12168
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.854665961610902e-05,
+      "loss": 0.0414,
+      "step": 12169
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8525366843586385e-05,
+      "loss": 0.0118,
+      "step": 12170
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.850407784054885e-05,
+      "loss": 0.048,
+      "step": 12171
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.848279260778715e-05,
+      "loss": 0.041,
+      "step": 12172
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.846151114609205e-05,
+      "loss": 0.0114,
+      "step": 12173
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8440233456254e-05,
+      "loss": 0.0265,
+      "step": 12174
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8418959539063565e-05,
+      "loss": 0.0489,
+      "step": 12175
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8397689395310794e-05,
+      "loss": 0.0243,
+      "step": 12176
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8376423025785956e-05,
+      "loss": 0.0437,
+      "step": 12177
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.835516043127896e-05,
+      "loss": 0.0625,
+      "step": 12178
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.833390161257972e-05,
+      "loss": 0.0075,
+      "step": 12179
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.831264657047791e-05,
+      "loss": 0.0671,
+      "step": 12180
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8291395305763034e-05,
+      "loss": 0.043,
+      "step": 12181
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8270147819224604e-05,
+      "loss": 0.0638,
+      "step": 12182
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8248904111651827e-05,
+      "loss": 0.061,
+      "step": 12183
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8227664183833926e-05,
+      "loss": 0.0533,
+      "step": 12184
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.820642803655985e-05,
+      "loss": 0.0539,
+      "step": 12185
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.818519567061841e-05,
+      "loss": 0.0299,
+      "step": 12186
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.816396708679845e-05,
+      "loss": 0.0835,
+      "step": 12187
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.814274228588841e-05,
+      "loss": 0.0431,
+      "step": 12188
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.812152126867685e-05,
+      "loss": 0.0555,
+      "step": 12189
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8100304035952044e-05,
+      "loss": 0.0852,
+      "step": 12190
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8079090588502105e-05,
+      "loss": 0.0635,
+      "step": 12191
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.805788092711503e-05,
+      "loss": 0.0454,
+      "step": 12192
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.8036675052578776e-05,
+      "loss": 0.0464,
+      "step": 12193
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.801547296568099e-05,
+      "loss": 0.0345,
+      "step": 12194
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7994274667209375e-05,
+      "loss": 0.0444,
+      "step": 12195
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7973080157951324e-05,
+      "loss": 0.009,
+      "step": 12196
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.79518894386941e-05,
+      "loss": 0.0499,
+      "step": 12197
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.793070251022497e-05,
+      "loss": 0.0409,
+      "step": 12198
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.79095193733309e-05,
+      "loss": 0.0812,
+      "step": 12199
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7888340028798886e-05,
+      "loss": 0.03,
+      "step": 12200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.78671644774155e-05,
+      "loss": 0.0508,
+      "step": 12201
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.7845992719967516e-05,
+      "loss": 0.0161,
+      "step": 12202
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7824824757241284e-05,
+      "loss": 0.0443,
+      "step": 12203
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7803660590023176e-05,
+      "loss": 0.0351,
+      "step": 12204
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7782500219099504e-05,
+      "loss": 0.0236,
+      "step": 12205
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.776134364525609e-05,
+      "loss": 0.032,
+      "step": 12206
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7740190869279e-05,
+      "loss": 0.0492,
+      "step": 12207
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7719041891953904e-05,
+      "loss": 0.0424,
+      "step": 12208
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.76978967140665e-05,
+      "loss": 0.0324,
+      "step": 12209
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.767675533640224e-05,
+      "loss": 0.0512,
+      "step": 12210
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.76556177597464e-05,
+      "loss": 0.0193,
+      "step": 12211
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.763448398488429e-05,
+      "loss": 0.0203,
+      "step": 12212
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7613354012600884e-05,
+      "loss": 0.0638,
+      "step": 12213
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.759222784368117e-05,
+      "loss": 0.0181,
+      "step": 12214
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.757110547890986e-05,
+      "loss": 0.0419,
+      "step": 12215
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.754998691907165e-05,
+      "loss": 0.0527,
+      "step": 12216
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.752887216495092e-05,
+      "loss": 0.0219,
+      "step": 12217
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.750776121733216e-05,
+      "loss": 0.0393,
+      "step": 12218
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.748665407699945e-05,
+      "loss": 0.0256,
+      "step": 12219
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.746555074473698e-05,
+      "loss": 0.0242,
+      "step": 12220
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7444451221328614e-05,
+      "loss": 0.0534,
+      "step": 12221
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7423355507558094e-05,
+      "loss": 0.0368,
+      "step": 12222
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.740226360420918e-05,
+      "loss": 0.0812,
+      "step": 12223
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.738117551206524e-05,
+      "loss": 0.0648,
+      "step": 12224
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7360091231909804e-05,
+      "loss": 0.0568,
+      "step": 12225
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.733901076452589e-05,
+      "loss": 0.0641,
+      "step": 12226
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7317934110696685e-05,
+      "loss": 0.0549,
+      "step": 12227
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.729686127120516e-05,
+      "loss": 0.0473,
+      "step": 12228
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7275792246834024e-05,
+      "loss": 0.0321,
+      "step": 12229
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.725472703836604e-05,
+      "loss": 0.0493,
+      "step": 12230
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.723366564658358e-05,
+      "loss": 0.024,
+      "step": 12231
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.7212608072269106e-05,
+      "loss": 0.0274,
+      "step": 12232
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.719155431620479e-05,
+      "loss": 0.0421,
+      "step": 12233
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.717050437917279e-05,
+      "loss": 0.0134,
+      "step": 12234
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.714945826195496e-05,
+      "loss": 0.0659,
+      "step": 12235
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.71284159653332e-05,
+      "loss": 0.0462,
+      "step": 12236
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.710737749008912e-05,
+      "loss": 0.0401,
+      "step": 12237
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.7086342837004184e-05,
+      "loss": 0.039,
+      "step": 12238
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.706531200685986e-05,
+      "loss": 0.048,
+      "step": 12239
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.7044285000437305e-05,
+      "loss": 0.0099,
+      "step": 12240
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.702326181851773e-05,
+      "loss": 0.0604,
+      "step": 12241
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.700224246188192e-05,
+      "loss": 0.0419,
+      "step": 12242
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.69812269313108e-05,
+      "loss": 0.0374,
+      "step": 12243
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.696021522758497e-05,
+      "loss": 0.0257,
+      "step": 12244
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6939207351484967e-05,
+      "loss": 0.0507,
+      "step": 12245
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.691820330379128e-05,
+      "loss": 0.0362,
+      "step": 12246
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.689720308528398e-05,
+      "loss": 0.0377,
+      "step": 12247
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6876206696743275e-05,
+      "loss": 0.029,
+      "step": 12248
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.685521413894902e-05,
+      "loss": 0.0291,
+      "step": 12249
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.683422541268116e-05,
+      "loss": 0.0508,
+      "step": 12250
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.681324051871928e-05,
+      "loss": 0.0186,
+      "step": 12251
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.679225945784288e-05,
+      "loss": 0.0398,
+      "step": 12252
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6771282230831435e-05,
+      "loss": 0.0415,
+      "step": 12253
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.67503088384641e-05,
+      "loss": 0.0613,
+      "step": 12254
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.672933928152005e-05,
+      "loss": 0.0558,
+      "step": 12255
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.670837356077821e-05,
+      "loss": 0.0961,
+      "step": 12256
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.668741167701741e-05,
+      "loss": 0.0408,
+      "step": 12257
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6666453631016224e-05,
+      "loss": 0.0734,
+      "step": 12258
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.664549942355333e-05,
+      "loss": 0.0321,
+      "step": 12259
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.662454905540701e-05,
+      "loss": 0.0315,
+      "step": 12260
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.660360252735557e-05,
+      "loss": 0.0703,
+      "step": 12261
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.658265984017711e-05,
+      "loss": 0.0306,
+      "step": 12262
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6561720994649506e-05,
+      "loss": 0.0258,
+      "step": 12263
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.654078599155068e-05,
+      "loss": 0.0786,
+      "step": 12264
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6519854831658224e-05,
+      "loss": 0.0412,
+      "step": 12265
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6498927515749813e-05,
+      "loss": 0.0453,
+      "step": 12266
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6478004044602616e-05,
+      "loss": 0.0611,
+      "step": 12267
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.6457084418994e-05,
+      "loss": 0.0207,
+      "step": 12268
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.643616863970112e-05,
+      "loss": 0.0307,
+      "step": 12269
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.641525670750082e-05,
+      "loss": 0.0246,
+      "step": 12270
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6394348623170076e-05,
+      "loss": 0.026,
+      "step": 12271
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6373444387485345e-05,
+      "loss": 0.0229,
+      "step": 12272
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.635254400122333e-05,
+      "loss": 0.0329,
+      "step": 12273
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.633164746516032e-05,
+      "loss": 0.0222,
+      "step": 12274
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.631075478007265e-05,
+      "loss": 0.0526,
+      "step": 12275
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.628986594673637e-05,
+      "loss": 0.0312,
+      "step": 12276
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.626898096592739e-05,
+      "loss": 0.0166,
+      "step": 12277
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.624809983842163e-05,
+      "loss": 0.0393,
+      "step": 12278
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.622722256499466e-05,
+      "loss": 0.0281,
+      "step": 12279
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.620634914642212e-05,
+      "loss": 0.0125,
+      "step": 12280
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.618547958347932e-05,
+      "loss": 0.0396,
+      "step": 12281
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.616461387694153e-05,
+      "loss": 0.0268,
+      "step": 12282
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.61437520275838e-05,
+      "loss": 0.0345,
+      "step": 12283
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6122894036181164e-05,
+      "loss": 0.0318,
+      "step": 12284
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.610203990350834e-05,
+      "loss": 0.0422,
+      "step": 12285
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.608118963034012e-05,
+      "loss": 0.0684,
+      "step": 12286
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.606034321745096e-05,
+      "loss": 0.0623,
+      "step": 12287
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.60395006656152e-05,
+      "loss": 0.0479,
+      "step": 12288
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.6018661975607175e-05,
+      "loss": 0.0355,
+      "step": 12289
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.599782714820089e-05,
+      "loss": 0.0241,
+      "step": 12290
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.597699618417039e-05,
+      "loss": 0.0755,
+      "step": 12291
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.595616908428944e-05,
+      "loss": 0.054,
+      "step": 12292
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.5935345849331656e-05,
+      "loss": 0.0361,
+      "step": 12293
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.5914526480070634e-05,
+      "loss": 0.0279,
+      "step": 12294
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.5893710977279704e-05,
+      "loss": 0.0307,
+      "step": 12295
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.587289934173215e-05,
+      "loss": 0.0186,
+      "step": 12296
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.585209157420104e-05,
+      "loss": 0.0286,
+      "step": 12297
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.5831287675459336e-05,
+      "loss": 0.0852,
+      "step": 12298
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.581048764627975e-05,
+      "loss": 0.0428,
+      "step": 12299
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.578969148743507e-05,
+      "loss": 0.0749,
+      "step": 12300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 4.576889919969772e-05,
+      "loss": 0.0497,
+      "step": 12301
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.574811078384014e-05,
+      "loss": 0.0459,
+      "step": 12302
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.572732624063454e-05,
+      "loss": 0.0544,
+      "step": 12303
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5706545570852945e-05,
+      "loss": 0.0169,
+      "step": 12304
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.568576877526739e-05,
+      "loss": 0.0513,
+      "step": 12305
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5664995854649586e-05,
+      "loss": 0.0309,
+      "step": 12306
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.564422680977131e-05,
+      "loss": 0.0611,
+      "step": 12307
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.562346164140389e-05,
+      "loss": 0.0066,
+      "step": 12308
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.560270035031879e-05,
+      "loss": 0.0691,
+      "step": 12309
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.558194293728728e-05,
+      "loss": 0.0238,
+      "step": 12310
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.556118940308035e-05,
+      "loss": 0.0473,
+      "step": 12311
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.554043974846905e-05,
+      "loss": 0.0512,
+      "step": 12312
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.551969397422399e-05,
+      "loss": 0.0595,
+      "step": 12313
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.549895208111596e-05,
+      "loss": 0.042,
+      "step": 12314
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.547821406991538e-05,
+      "loss": 0.0169,
+      "step": 12315
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5457479941392674e-05,
+      "loss": 0.0551,
+      "step": 12316
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5436749696318036e-05,
+      "loss": 0.0331,
+      "step": 12317
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.541602333546146e-05,
+      "loss": 0.0219,
+      "step": 12318
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5395300859592984e-05,
+      "loss": 0.0347,
+      "step": 12319
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.537458226948227e-05,
+      "loss": 0.0647,
+      "step": 12320
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5353867565899056e-05,
+      "loss": 0.0414,
+      "step": 12321
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.533315674961281e-05,
+      "loss": 0.0376,
+      "step": 12322
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.531244982139283e-05,
+      "loss": 0.0264,
+      "step": 12323
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5291746782008326e-05,
+      "loss": 0.0373,
+      "step": 12324
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.527104763222841e-05,
+      "loss": 0.0573,
+      "step": 12325
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5250352372821916e-05,
+      "loss": 0.079,
+      "step": 12326
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.522966100455769e-05,
+      "loss": 0.0753,
+      "step": 12327
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.520897352820433e-05,
+      "loss": 0.0544,
+      "step": 12328
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5188289944530253e-05,
+      "loss": 0.0328,
+      "step": 12329
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.516761025430391e-05,
+      "loss": 0.058,
+      "step": 12330
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.514693445829337e-05,
+      "loss": 0.0421,
+      "step": 12331
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5126262557266775e-05,
+      "loss": 0.0427,
+      "step": 12332
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.5105594551992e-05,
+      "loss": 0.029,
+      "step": 12333
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.508493044323673e-05,
+      "loss": 0.0108,
+      "step": 12334
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.5064270231768685e-05,
+      "loss": 0.022,
+      "step": 12335
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.5043613918355235e-05,
+      "loss": 0.0518,
+      "step": 12336
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.502296150376384e-05,
+      "loss": 0.0425,
+      "step": 12337
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.500231298876149e-05,
+      "loss": 0.0897,
+      "step": 12338
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4981668374115346e-05,
+      "loss": 0.0719,
+      "step": 12339
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.496102766059221e-05,
+      "loss": 0.0518,
+      "step": 12340
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.494039084895894e-05,
+      "loss": 0.0258,
+      "step": 12341
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4919757939982046e-05,
+      "loss": 0.0524,
+      "step": 12342
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.489912893442796e-05,
+      "loss": 0.0312,
+      "step": 12343
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.487850383306307e-05,
+      "loss": 0.0221,
+      "step": 12344
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4857882636653464e-05,
+      "loss": 0.0548,
+      "step": 12345
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4837265345965216e-05,
+      "loss": 0.0606,
+      "step": 12346
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.481665196176419e-05,
+      "loss": 0.0603,
+      "step": 12347
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.479604248481608e-05,
+      "loss": 0.054,
+      "step": 12348
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.477543691588645e-05,
+      "loss": 0.0551,
+      "step": 12349
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.475483525574078e-05,
+      "loss": 0.0315,
+      "step": 12350
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.473423750514438e-05,
+      "loss": 0.0715,
+      "step": 12351
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.471364366486237e-05,
+      "loss": 0.0453,
+      "step": 12352
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.469305373565975e-05,
+      "loss": 0.0177,
+      "step": 12353
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4672467718301327e-05,
+      "loss": 0.0419,
+      "step": 12354
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.465188561355191e-05,
+      "loss": 0.0566,
+      "step": 12355
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.463130742217595e-05,
+      "loss": 0.0194,
+      "step": 12356
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.461073314493799e-05,
+      "loss": 0.0175,
+      "step": 12357
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.459016278260224e-05,
+      "loss": 0.0474,
+      "step": 12358
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4569596335932776e-05,
+      "loss": 0.0355,
+      "step": 12359
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.454903380569369e-05,
+      "loss": 0.0739,
+      "step": 12360
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4528475192648724e-05,
+      "loss": 0.0335,
+      "step": 12361
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4507920497561643e-05,
+      "loss": 0.0355,
+      "step": 12362
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.448736972119598e-05,
+      "loss": 0.046,
+      "step": 12363
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.44668228643151e-05,
+      "loss": 0.0289,
+      "step": 12364
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.444627992768225e-05,
+      "loss": 0.0267,
+      "step": 12365
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.44257409120606e-05,
+      "loss": 0.0695,
+      "step": 12366
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.4405205818213044e-05,
+      "loss": 0.0346,
+      "step": 12367
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.438467464690247e-05,
+      "loss": 0.0379,
+      "step": 12368
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4364147398891525e-05,
+      "loss": 0.0445,
+      "step": 12369
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.434362407494268e-05,
+      "loss": 0.0561,
+      "step": 12370
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.432310467581843e-05,
+      "loss": 0.046,
+      "step": 12371
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.430258920228089e-05,
+      "loss": 0.0323,
+      "step": 12372
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.428207765509225e-05,
+      "loss": 0.0417,
+      "step": 12373
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.426157003501441e-05,
+      "loss": 0.0246,
+      "step": 12374
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.424106634280912e-05,
+      "loss": 0.0535,
+      "step": 12375
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.422056657923813e-05,
+      "loss": 0.019,
+      "step": 12376
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.420007074506285e-05,
+      "loss": 0.0439,
+      "step": 12377
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.417957884104476e-05,
+      "loss": 0.0189,
+      "step": 12378
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.415909086794494e-05,
+      "loss": 0.065,
+      "step": 12379
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4138606826524546e-05,
+      "loss": 0.021,
+      "step": 12380
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.411812671754443e-05,
+      "loss": 0.0509,
+      "step": 12381
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.409765054176546e-05,
+      "loss": 0.0441,
+      "step": 12382
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.407717829994821e-05,
+      "loss": 0.0741,
+      "step": 12383
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4056709992853124e-05,
+      "loss": 0.0548,
+      "step": 12384
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.403624562124063e-05,
+      "loss": 0.0343,
+      "step": 12385
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.4015785185870835e-05,
+      "loss": 0.0429,
+      "step": 12386
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.399532868750386e-05,
+      "loss": 0.0278,
+      "step": 12387
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3974876126899575e-05,
+      "loss": 0.0535,
+      "step": 12388
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3954427504817726e-05,
+      "loss": 0.0362,
+      "step": 12389
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3933982822017876e-05,
+      "loss": 0.0325,
+      "step": 12390
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.391354207925952e-05,
+      "loss": 0.0181,
+      "step": 12391
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3893105277302034e-05,
+      "loss": 0.0162,
+      "step": 12392
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.387267241690453e-05,
+      "loss": 0.0409,
+      "step": 12393
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.385224349882603e-05,
+      "loss": 0.0369,
+      "step": 12394
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.383181852382536e-05,
+      "loss": 0.0576,
+      "step": 12395
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.381139749266134e-05,
+      "loss": 0.0644,
+      "step": 12396
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3790980406092464e-05,
+      "loss": 0.0247,
+      "step": 12397
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.377056726487726e-05,
+      "loss": 0.0644,
+      "step": 12398
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.375015806977396e-05,
+      "loss": 0.0405,
+      "step": 12399
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.3729752821540676e-05,
+      "loss": 0.0909,
+      "step": 12400
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.370935152093548e-05,
+      "loss": 0.0597,
+      "step": 12401
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.368895416871614e-05,
+      "loss": 0.0252,
+      "step": 12402
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3668560765640475e-05,
+      "loss": 0.0684,
+      "step": 12403
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.364817131246589e-05,
+      "loss": 0.0542,
+      "step": 12404
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3627785809949894e-05,
+      "loss": 0.0676,
+      "step": 12405
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.36074042588497e-05,
+      "loss": 0.0606,
+      "step": 12406
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3587026659922474e-05,
+      "loss": 0.026,
+      "step": 12407
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.356665301392515e-05,
+      "loss": 0.0275,
+      "step": 12408
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.354628332161451e-05,
+      "loss": 0.0226,
+      "step": 12409
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3525917583747314e-05,
+      "loss": 0.0246,
+      "step": 12410
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.350555580108001e-05,
+      "loss": 0.043,
+      "step": 12411
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.348519797436907e-05,
+      "loss": 0.0517,
+      "step": 12412
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.346484410437065e-05,
+      "loss": 0.0674,
+      "step": 12413
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.344449419184084e-05,
+      "loss": 0.0339,
+      "step": 12414
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.342414823753562e-05,
+      "loss": 0.046,
+      "step": 12415
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.340380624221073e-05,
+      "loss": 0.0311,
+      "step": 12416
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.338346820662189e-05,
+      "loss": 0.0462,
+      "step": 12417
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.336313413152456e-05,
+      "loss": 0.063,
+      "step": 12418
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.334280401767407e-05,
+      "loss": 0.0458,
+      "step": 12419
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3322477865825614e-05,
+      "loss": 0.0465,
+      "step": 12420
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.330215567673432e-05,
+      "loss": 0.0221,
+      "step": 12421
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.328183745115501e-05,
+      "loss": 0.028,
+      "step": 12422
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3261523189842515e-05,
+      "loss": 0.0687,
+      "step": 12423
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.324121289355144e-05,
+      "loss": 0.0354,
+      "step": 12424
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3220906563036205e-05,
+      "loss": 0.0456,
+      "step": 12425
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.320060419905119e-05,
+      "loss": 0.0554,
+      "step": 12426
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3180305802350513e-05,
+      "loss": 0.0574,
+      "step": 12427
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3160011373688267e-05,
+      "loss": 0.0505,
+      "step": 12428
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3139720913818294e-05,
+      "loss": 0.04,
+      "step": 12429
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3119434423494314e-05,
+      "loss": 0.0318,
+      "step": 12430
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.3099151903469894e-05,
+      "loss": 0.0669,
+      "step": 12431
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.307887335449847e-05,
+      "loss": 0.0549,
+      "step": 12432
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.305859877733342e-05,
+      "loss": 0.0367,
+      "step": 12433
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3038328172727834e-05,
+      "loss": 0.0501,
+      "step": 12434
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.3018061541434675e-05,
+      "loss": 0.0489,
+      "step": 12435
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2997798884206775e-05,
+      "loss": 0.0321,
+      "step": 12436
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.29775402017969e-05,
+      "loss": 0.0012,
+      "step": 12437
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.295728549495752e-05,
+      "loss": 0.0323,
+      "step": 12438
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2937034764441135e-05,
+      "loss": 0.018,
+      "step": 12439
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.291678801099994e-05,
+      "loss": 0.0106,
+      "step": 12440
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.289654523538601e-05,
+      "loss": 0.0493,
+      "step": 12441
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2876306438351406e-05,
+      "loss": 0.0479,
+      "step": 12442
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2856071620647826e-05,
+      "loss": 0.0545,
+      "step": 12443
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2835840783027076e-05,
+      "loss": 0.0667,
+      "step": 12444
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.281561392624049e-05,
+      "loss": 0.0715,
+      "step": 12445
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.279539105103958e-05,
+      "loss": 0.0493,
+      "step": 12446
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.277517215817548e-05,
+      "loss": 0.045,
+      "step": 12447
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.275495724839934e-05,
+      "loss": 0.0523,
+      "step": 12448
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.273474632246204e-05,
+      "loss": 0.0411,
+      "step": 12449
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.271453938111432e-05,
+      "loss": 0.0222,
+      "step": 12450
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.269433642510689e-05,
+      "loss": 0.0425,
+      "step": 12451
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.267413745519015e-05,
+      "loss": 0.0158,
+      "step": 12452
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.265394247211452e-05,
+      "loss": 0.0338,
+      "step": 12453
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.263375147663014e-05,
+      "loss": 0.0451,
+      "step": 12454
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.261356446948699e-05,
+      "loss": 0.0693,
+      "step": 12455
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.259338145143506e-05,
+      "loss": 0.0852,
+      "step": 12456
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.257320242322401e-05,
+      "loss": 0.0722,
+      "step": 12457
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2553027385603494e-05,
+      "loss": 0.0528,
+      "step": 12458
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2532856339322924e-05,
+      "loss": 0.0417,
+      "step": 12459
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.25126892851316e-05,
+      "loss": 0.0262,
+      "step": 12460
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.249252622377863e-05,
+      "loss": 0.0702,
+      "step": 12461
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2472367156013074e-05,
+      "loss": 0.0419,
+      "step": 12462
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2452212082583726e-05,
+      "loss": 0.0385,
+      "step": 12463
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.243206100423934e-05,
+      "loss": 0.0294,
+      "step": 12464
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.2411913921728475e-05,
+      "loss": 0.0331,
+      "step": 12465
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2391770835799455e-05,
+      "loss": 0.0295,
+      "step": 12466
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.237163174720063e-05,
+      "loss": 0.029,
+      "step": 12467
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.235149665668002e-05,
+      "loss": 0.0073,
+      "step": 12468
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.233136556498574e-05,
+      "loss": 0.0601,
+      "step": 12469
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.231123847286538e-05,
+      "loss": 0.0292,
+      "step": 12470
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2291115381066774e-05,
+      "loss": 0.069,
+      "step": 12471
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.227099629033733e-05,
+      "loss": 0.045,
+      "step": 12472
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.225088120142447e-05,
+      "loss": 0.0584,
+      "step": 12473
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2230770115075484e-05,
+      "loss": 0.0587,
+      "step": 12474
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.221066303203727e-05,
+      "loss": 0.018,
+      "step": 12475
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.219055995305687e-05,
+      "loss": 0.0086,
+      "step": 12476
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.217046087888099e-05,
+      "loss": 0.0696,
+      "step": 12477
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.215036581025632e-05,
+      "loss": 0.0675,
+      "step": 12478
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2130274747929274e-05,
+      "loss": 0.0249,
+      "step": 12479
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.211018769264618e-05,
+      "loss": 0.0415,
+      "step": 12480
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.2090104645153266e-05,
+      "loss": 0.0408,
+      "step": 12481
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.207002560619647e-05,
+      "loss": 0.0212,
+      "step": 12482
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.204995057652177e-05,
+      "loss": 0.0074,
+      "step": 12483
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.202987955687484e-05,
+      "loss": 0.033,
+      "step": 12484
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.200981254800126e-05,
+      "loss": 0.0342,
+      "step": 12485
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.198974955064643e-05,
+      "loss": 0.0161,
+      "step": 12486
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.196969056555572e-05,
+      "loss": 0.0378,
+      "step": 12487
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.1949635593474154e-05,
+      "loss": 0.0401,
+      "step": 12488
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.192958463514681e-05,
+      "loss": 0.0104,
+      "step": 12489
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.190953769131849e-05,
+      "loss": 0.0269,
+      "step": 12490
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.188949476273382e-05,
+      "loss": 0.0308,
+      "step": 12491
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.1869455850137445e-05,
+      "loss": 0.0605,
+      "step": 12492
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.184942095427365e-05,
+      "loss": 0.0471,
+      "step": 12493
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.1829390075886755e-05,
+      "loss": 0.043,
+      "step": 12494
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.180936321572081e-05,
+      "loss": 0.0423,
+      "step": 12495
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.17893403745197e-05,
+      "loss": 0.0186,
+      "step": 12496
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.176932155302734e-05,
+      "loss": 0.0148,
+      "step": 12497
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.174930675198725e-05,
+      "loss": 0.0378,
+      "step": 12498
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1729295972143e-05,
+      "loss": 0.0253,
+      "step": 12499
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1709289214237916e-05,
+      "loss": 0.0454,
+      "step": 12500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.168928647901517e-05,
+      "loss": 0.0788,
+      "step": 12501
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.166928776721779e-05,
+      "loss": 0.0159,
+      "step": 12502
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1649293079588715e-05,
+      "loss": 0.028,
+      "step": 12503
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1629302416870616e-05,
+      "loss": 0.0052,
+      "step": 12504
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.160931577980619e-05,
+      "loss": 0.0507,
+      "step": 12505
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.158933316913783e-05,
+      "loss": 0.0161,
+      "step": 12506
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.15693545856078e-05,
+      "loss": 0.0149,
+      "step": 12507
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.154938002995831e-05,
+      "loss": 0.0407,
+      "step": 12508
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1529409502931275e-05,
+      "loss": 0.0531,
+      "step": 12509
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1509443005268684e-05,
+      "loss": 0.0413,
+      "step": 12510
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1489480537712055e-05,
+      "loss": 0.0568,
+      "step": 12511
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.146952210100305e-05,
+      "loss": 0.0329,
+      "step": 12512
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.144956769588301e-05,
+      "loss": 0.0556,
+      "step": 12513
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.14296173230932e-05,
+      "loss": 0.0654,
+      "step": 12514
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.14096709833748e-05,
+      "loss": 0.044,
+      "step": 12515
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1389728677468624e-05,
+      "loss": 0.0354,
+      "step": 12516
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1369790406115555e-05,
+      "loss": 0.021,
+      "step": 12517
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1349856170056184e-05,
+      "loss": 0.0542,
+      "step": 12518
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.132992597003107e-05,
+      "loss": 0.0288,
+      "step": 12519
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.130999980678055e-05,
+      "loss": 0.0281,
+      "step": 12520
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1290077681044756e-05,
+      "loss": 0.049,
+      "step": 12521
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.127015959356383e-05,
+      "loss": 0.0419,
+      "step": 12522
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1250245545077584e-05,
+      "loss": 0.0387,
+      "step": 12523
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.123033553632586e-05,
+      "loss": 0.0386,
+      "step": 12524
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1210429568048206e-05,
+      "loss": 0.0662,
+      "step": 12525
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1190527640984063e-05,
+      "loss": 0.038,
+      "step": 12526
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1170629755872694e-05,
+      "loss": 0.0187,
+      "step": 12527
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1150735913453335e-05,
+      "loss": 0.0303,
+      "step": 12528
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1130846114464904e-05,
+      "loss": 0.0826,
+      "step": 12529
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.1110960359646324e-05,
+      "loss": 0.0095,
+      "step": 12530
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.109107864973625e-05,
+      "loss": 0.012,
+      "step": 12531
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.107120098547317e-05,
+      "loss": 0.0518,
+      "step": 12532
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.105132736759561e-05,
+      "loss": 0.0455,
+      "step": 12533
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.10314577968417e-05,
+      "loss": 0.0004,
+      "step": 12534
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.101159227394967e-05,
+      "loss": 0.0411,
+      "step": 12535
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.099173079965729e-05,
+      "loss": 0.0276,
+      "step": 12536
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0971873374702445e-05,
+      "loss": 0.0818,
+      "step": 12537
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0952019999822846e-05,
+      "loss": 0.0108,
+      "step": 12538
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.093217067575586e-05,
+      "loss": 0.0441,
+      "step": 12539
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0912325403238986e-05,
+      "loss": 0.0261,
+      "step": 12540
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.089248418300924e-05,
+      "loss": 0.0693,
+      "step": 12541
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0872647015803786e-05,
+      "loss": 0.0471,
+      "step": 12542
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.085281390235945e-05,
+      "loss": 0.0135,
+      "step": 12543
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.083298484341305e-05,
+      "loss": 0.0554,
+      "step": 12544
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0813159839701144e-05,
+      "loss": 0.0608,
+      "step": 12545
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0793338891960095e-05,
+      "loss": 0.0198,
+      "step": 12546
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.077352200092631e-05,
+      "loss": 0.0372,
+      "step": 12547
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0753709167335845e-05,
+      "loss": 0.0527,
+      "step": 12548
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.073390039192476e-05,
+      "loss": 0.0516,
+      "step": 12549
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.071409567542881e-05,
+      "loss": 0.0349,
+      "step": 12550
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0694295018583815e-05,
+      "loss": 0.053,
+      "step": 12551
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0674498422125144e-05,
+      "loss": 0.0337,
+      "step": 12552
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.06547058867883e-05,
+      "loss": 0.0404,
+      "step": 12553
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.063491741330843e-05,
+      "loss": 0.0544,
+      "step": 12554
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.061513300242068e-05,
+      "loss": 0.0323,
+      "step": 12555
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.059535265486005e-05,
+      "loss": 0.0635,
+      "step": 12556
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0575576371361144e-05,
+      "loss": 0.0159,
+      "step": 12557
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0555804152658756e-05,
+      "loss": 0.0134,
+      "step": 12558
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0536035999487245e-05,
+      "loss": 0.0371,
+      "step": 12559
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.051627191258106e-05,
+      "loss": 0.0285,
+      "step": 12560
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0496511892674285e-05,
+      "loss": 0.063,
+      "step": 12561
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0476755940500955e-05,
+      "loss": 0.03,
+      "step": 12562
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0457004056795e-05,
+      "loss": 0.0126,
+      "step": 12563
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.0437256242290074e-05,
+      "loss": 0.0479,
+      "step": 12564
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0417512497719836e-05,
+      "loss": 0.0405,
+      "step": 12565
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.039777282381765e-05,
+      "loss": 0.0267,
+      "step": 12566
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0378037221316793e-05,
+      "loss": 0.0288,
+      "step": 12567
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.035830569095035e-05,
+      "loss": 0.032,
+      "step": 12568
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.033857823345139e-05,
+      "loss": 0.033,
+      "step": 12569
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.031885484955262e-05,
+      "loss": 0.0493,
+      "step": 12570
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.029913553998678e-05,
+      "loss": 0.0345,
+      "step": 12571
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0279420305486384e-05,
+      "loss": 0.0275,
+      "step": 12572
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0259709146783716e-05,
+      "loss": 0.0197,
+      "step": 12573
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.024000206461108e-05,
+      "loss": 0.0513,
+      "step": 12574
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.022029905970045e-05,
+      "loss": 0.0337,
+      "step": 12575
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.020060013278388e-05,
+      "loss": 0.0229,
+      "step": 12576
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0180905284592954e-05,
+      "loss": 0.0239,
+      "step": 12577
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.016121451585934e-05,
+      "loss": 0.0521,
+      "step": 12578
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0141527827314565e-05,
+      "loss": 0.0219,
+      "step": 12579
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.012184521968982e-05,
+      "loss": 0.0599,
+      "step": 12580
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.01021666937164e-05,
+      "loss": 0.0597,
+      "step": 12581
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.008249225012512e-05,
+      "loss": 0.0315,
+      "step": 12582
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.006282188964698e-05,
+      "loss": 0.0395,
+      "step": 12583
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.0043155613012564e-05,
+      "loss": 0.0529,
+      "step": 12584
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.002349342095252e-05,
+      "loss": 0.0339,
+      "step": 12585
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.000383531419719e-05,
+      "loss": 0.0327,
+      "step": 12586
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.998418129347678e-05,
+      "loss": 0.035,
+      "step": 12587
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.996453135952145e-05,
+      "loss": 0.0417,
+      "step": 12588
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9944885513061066e-05,
+      "loss": 0.0388,
+      "step": 12589
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.992524375482549e-05,
+      "loss": 0.0334,
+      "step": 12590
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.990560608554432e-05,
+      "loss": 0.0399,
+      "step": 12591
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.988597250594704e-05,
+      "loss": 0.0216,
+      "step": 12592
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.986634301676292e-05,
+      "loss": 0.0302,
+      "step": 12593
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.984671761872124e-05,
+      "loss": 0.0243,
+      "step": 12594
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.9827096312550935e-05,
+      "loss": 0.0179,
+      "step": 12595
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.980747909898096e-05,
+      "loss": 0.0616,
+      "step": 12596
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.978786597874001e-05,
+      "loss": 0.0128,
+      "step": 12597
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.97682569525566e-05,
+      "loss": 0.0345,
+      "step": 12598
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.974865202115923e-05,
+      "loss": 0.053,
+      "step": 12599
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9729051185276095e-05,
+      "loss": 0.0451,
+      "step": 12600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.970945444563539e-05,
+      "loss": 0.0325,
+      "step": 12601
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.968986180296501e-05,
+      "loss": 0.0229,
+      "step": 12602
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.967027325799274e-05,
+      "loss": 0.0634,
+      "step": 12603
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9650688811446325e-05,
+      "loss": 0.054,
+      "step": 12604
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.963110846405319e-05,
+      "loss": 0.0574,
+      "step": 12605
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.961153221654081e-05,
+      "loss": 0.0456,
+      "step": 12606
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.95919600696362e-05,
+      "loss": 0.0666,
+      "step": 12607
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.957239202406654e-05,
+      "loss": 0.0519,
+      "step": 12608
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9552828080558656e-05,
+      "loss": 0.0511,
+      "step": 12609
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.953326823983937e-05,
+      "loss": 0.0428,
+      "step": 12610
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9513712502635186e-05,
+      "loss": 0.0129,
+      "step": 12611
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.949416086967262e-05,
+      "loss": 0.0257,
+      "step": 12612
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.947461334167794e-05,
+      "loss": 0.0495,
+      "step": 12613
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.94550699193772e-05,
+      "loss": 0.0094,
+      "step": 12614
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.943553060349649e-05,
+      "loss": 0.0608,
+      "step": 12615
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.941599539476155e-05,
+      "loss": 0.0455,
+      "step": 12616
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9396464293898176e-05,
+      "loss": 0.0091,
+      "step": 12617
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.937693730163174e-05,
+      "loss": 0.0117,
+      "step": 12618
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.935741441868767e-05,
+      "loss": 0.0749,
+      "step": 12619
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.933789564579124e-05,
+      "loss": 0.0564,
+      "step": 12620
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.931838098366743e-05,
+      "loss": 0.0241,
+      "step": 12621
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9298870433041293e-05,
+      "loss": 0.0416,
+      "step": 12622
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.927936399463739e-05,
+      "loss": 0.0264,
+      "step": 12623
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9259861669180484e-05,
+      "loss": 0.0626,
+      "step": 12624
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.924036345739492e-05,
+      "loss": 0.0062,
+      "step": 12625
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.922086936000511e-05,
+      "loss": 0.0356,
+      "step": 12626
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.9201379377735145e-05,
+      "loss": 0.0687,
+      "step": 12627
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.918189351130896e-05,
+      "loss": 0.0571,
+      "step": 12628
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.916241176145051e-05,
+      "loss": 0.0321,
+      "step": 12629
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.914293412888341e-05,
+      "loss": 0.0234,
+      "step": 12630
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.912346061433125e-05,
+      "loss": 0.0601,
+      "step": 12631
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.910399121851739e-05,
+      "loss": 0.0157,
+      "step": 12632
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.908452594216505e-05,
+      "loss": 0.0506,
+      "step": 12633
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.906506478599727e-05,
+      "loss": 0.0419,
+      "step": 12634
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.904560775073708e-05,
+      "loss": 0.0676,
+      "step": 12635
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.902615483710712e-05,
+      "loss": 0.0484,
+      "step": 12636
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.900670604583014e-05,
+      "loss": 0.0207,
+      "step": 12637
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.898726137762855e-05,
+      "loss": 0.0425,
+      "step": 12638
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.896782083322461e-05,
+      "loss": 0.0533,
+      "step": 12639
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.894838441334057e-05,
+      "loss": 0.0117,
+      "step": 12640
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.892895211869834e-05,
+      "loss": 0.0299,
+      "step": 12641
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.890952395001987e-05,
+      "loss": 0.0161,
+      "step": 12642
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.889009990802683e-05,
+      "loss": 0.0478,
+      "step": 12643
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.88706799934407e-05,
+      "loss": 0.0456,
+      "step": 12644
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.885126420698296e-05,
+      "loss": 0.0246,
+      "step": 12645
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.883185254937479e-05,
+      "loss": 0.0237,
+      "step": 12646
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.881244502133736e-05,
+      "loss": 0.0439,
+      "step": 12647
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.879304162359148e-05,
+      "loss": 0.0434,
+      "step": 12648
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8773642356858016e-05,
+      "loss": 0.0279,
+      "step": 12649
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.875424722185754e-05,
+      "loss": 0.0176,
+      "step": 12650
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8734856219310586e-05,
+      "loss": 0.0243,
+      "step": 12651
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.871546934993746e-05,
+      "loss": 0.0524,
+      "step": 12652
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.8696086614458244e-05,
+      "loss": 0.0191,
+      "step": 12653
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.867670801359308e-05,
+      "loss": 0.047,
+      "step": 12654
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.86573335480617e-05,
+      "loss": 0.0386,
+      "step": 12655
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.863796321858391e-05,
+      "loss": 0.0538,
+      "step": 12656
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.861859702587923e-05,
+      "loss": 0.0107,
+      "step": 12657
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.859923497066704e-05,
+      "loss": 0.0599,
+      "step": 12658
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.857987705366656e-05,
+      "loss": 0.0309,
+      "step": 12659
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.85605232755969e-05,
+      "loss": 0.049,
+      "step": 12660
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.854117363717704e-05,
+      "loss": 0.0157,
+      "step": 12661
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.852182813912575e-05,
+      "loss": 0.0358,
+      "step": 12662
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.850248678216163e-05,
+      "loss": 0.0678,
+      "step": 12663
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.848314956700312e-05,
+      "loss": 0.0468,
+      "step": 12664
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8463816494368624e-05,
+      "loss": 0.0577,
+      "step": 12665
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8444487564976226e-05,
+      "loss": 0.069,
+      "step": 12666
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.842516277954402e-05,
+      "loss": 0.0322,
+      "step": 12667
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.840584213878983e-05,
+      "loss": 0.0305,
+      "step": 12668
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8386525643431324e-05,
+      "loss": 0.058,
+      "step": 12669
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.836721329418611e-05,
+      "loss": 0.0557,
+      "step": 12670
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8347905091771534e-05,
+      "loss": 0.0077,
+      "step": 12671
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.832860103690494e-05,
+      "loss": 0.0317,
+      "step": 12672
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.830930113030326e-05,
+      "loss": 0.0461,
+      "step": 12673
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8290005372683554e-05,
+      "loss": 0.0588,
+      "step": 12674
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.827071376476254e-05,
+      "loss": 0.0681,
+      "step": 12675
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.825142630725688e-05,
+      "loss": 0.0479,
+      "step": 12676
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8232143000883015e-05,
+      "loss": 0.0574,
+      "step": 12677
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.821286384635731e-05,
+      "loss": 0.0491,
+      "step": 12678
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8193588844395904e-05,
+      "loss": 0.0552,
+      "step": 12679
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.817431799571476e-05,
+      "loss": 0.0246,
+      "step": 12680
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.815505130102981e-05,
+      "loss": 0.0209,
+      "step": 12681
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.813578876105669e-05,
+      "loss": 0.0322,
+      "step": 12682
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8116530376511026e-05,
+      "loss": 0.0286,
+      "step": 12683
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8097276148108156e-05,
+      "loss": 0.0455,
+      "step": 12684
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8078026076563276e-05,
+      "loss": 0.0213,
+      "step": 12685
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.8058780162591567e-05,
+      "loss": 0.0582,
+      "step": 12686
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.803953840690787e-05,
+      "loss": 0.0409,
+      "step": 12687
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.802030081022708e-05,
+      "loss": 0.0272,
+      "step": 12688
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.800106737326365e-05,
+      "loss": 0.0457,
+      "step": 12689
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.798183809673218e-05,
+      "loss": 0.0218,
+      "step": 12690
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.796261298134689e-05,
+      "loss": 0.0242,
+      "step": 12691
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.794339202782204e-05,
+      "loss": 0.061,
+      "step": 12692
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7924175236871554e-05,
+      "loss": 0.0287,
+      "step": 12693
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.790496260920926e-05,
+      "loss": 0.0323,
+      "step": 12694
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.788575414554893e-05,
+      "loss": 0.0552,
+      "step": 12695
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.786654984660401e-05,
+      "loss": 0.043,
+      "step": 12696
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7847349713087995e-05,
+      "loss": 0.0168,
+      "step": 12697
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.782815374571404e-05,
+      "loss": 0.0449,
+      "step": 12698
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7808961945195247e-05,
+      "loss": 0.0315,
+      "step": 12699
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.778977431224446e-05,
+      "loss": 0.0447,
+      "step": 12700
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.777059084757455e-05,
+      "loss": 0.0505,
+      "step": 12701
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7751411551898055e-05,
+      "loss": 0.0192,
+      "step": 12702
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.773223642592747e-05,
+      "loss": 0.0135,
+      "step": 12703
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7713065470375096e-05,
+      "loss": 0.0307,
+      "step": 12704
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.769389868595302e-05,
+      "loss": 0.0608,
+      "step": 12705
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7674736073373306e-05,
+      "loss": 0.0185,
+      "step": 12706
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.765557763334771e-05,
+      "loss": 0.0752,
+      "step": 12707
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.763642336658803e-05,
+      "loss": 0.0413,
+      "step": 12708
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.761727327380568e-05,
+      "loss": 0.047,
+      "step": 12709
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.759812735571205e-05,
+      "loss": 0.0298,
+      "step": 12710
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.757898561301841e-05,
+      "loss": 0.0389,
+      "step": 12711
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.755984804643574e-05,
+      "loss": 0.0454,
+      "step": 12712
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.754071465667506e-05,
+      "loss": 0.0518,
+      "step": 12713
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.752158544444697e-05,
+      "loss": 0.0148,
+      "step": 12714
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.75024604104622e-05,
+      "loss": 0.0552,
+      "step": 12715
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7483339555431055e-05,
+      "loss": 0.039,
+      "step": 12716
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7464222880063966e-05,
+      "loss": 0.0452,
+      "step": 12717
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.744511038507097e-05,
+      "loss": 0.0541,
+      "step": 12718
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.742600207116202e-05,
+      "loss": 0.061,
+      "step": 12719
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.740689793904701e-05,
+      "loss": 0.0471,
+      "step": 12720
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.738779798943554e-05,
+      "loss": 0.05,
+      "step": 12721
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7368702223037164e-05,
+      "loss": 0.0545,
+      "step": 12722
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7349610640561226e-05,
+      "loss": 0.0088,
+      "step": 12723
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.733052324271687e-05,
+      "loss": 0.0266,
+      "step": 12724
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.731144003021321e-05,
+      "loss": 0.0236,
+      "step": 12725
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.729236100375903e-05,
+      "loss": 0.0316,
+      "step": 12726
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7273286164063203e-05,
+      "loss": 0.037,
+      "step": 12727
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7254215511834215e-05,
+      "loss": 0.0451,
+      "step": 12728
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.723514904778049e-05,
+      "loss": 0.0319,
+      "step": 12729
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.721608677261027e-05,
+      "loss": 0.0122,
+      "step": 12730
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.719702868703173e-05,
+      "loss": 0.0381,
+      "step": 12731
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.717797479175273e-05,
+      "loss": 0.037,
+      "step": 12732
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.715892508748116e-05,
+      "loss": 0.0312,
+      "step": 12733
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7139879574924625e-05,
+      "loss": 0.0388,
+      "step": 12734
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.712083825479057e-05,
+      "loss": 0.064,
+      "step": 12735
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.710180112778639e-05,
+      "loss": 0.0435,
+      "step": 12736
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.708276819461919e-05,
+      "loss": 0.0613,
+      "step": 12737
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.706373945599605e-05,
+      "loss": 0.0295,
+      "step": 12738
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7044714912623824e-05,
+      "loss": 0.0453,
+      "step": 12739
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.70256945652092e-05,
+      "loss": 0.0357,
+      "step": 12740
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.700667841445868e-05,
+      "loss": 0.0573,
+      "step": 12741
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.698766646107876e-05,
+      "loss": 0.0634,
+      "step": 12742
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.696865870577557e-05,
+      "loss": 0.0314,
+      "step": 12743
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6949655149255296e-05,
+      "loss": 0.0565,
+      "step": 12744
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.69306557922238e-05,
+      "loss": 0.0831,
+      "step": 12745
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.691166063538685e-05,
+      "loss": 0.0576,
+      "step": 12746
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.68926696794501e-05,
+      "loss": 0.0275,
+      "step": 12747
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.687368292511896e-05,
+      "loss": 0.0505,
+      "step": 12748
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6854700373098794e-05,
+      "loss": 0.0637,
+      "step": 12749
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.683572202409472e-05,
+      "loss": 0.011,
+      "step": 12750
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6816747878811677e-05,
+      "loss": 0.0211,
+      "step": 12751
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6797777937954594e-05,
+      "loss": 0.0204,
+      "step": 12752
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.677881220222804e-05,
+      "loss": 0.0604,
+      "step": 12753
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6759850672336674e-05,
+      "loss": 0.0143,
+      "step": 12754
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6740893348984717e-05,
+      "loss": 0.0463,
+      "step": 12755
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.672194023287646e-05,
+      "loss": 0.0571,
+      "step": 12756
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6702991324715914e-05,
+      "loss": 0.0537,
+      "step": 12757
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.668404662520704e-05,
+      "loss": 0.0247,
+      "step": 12758
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.6665106135053534e-05,
+      "loss": 0.0217,
+      "step": 12759
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.664616985495894e-05,
+      "loss": 0.0222,
+      "step": 12760
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.662723778562678e-05,
+      "loss": 0.0589,
+      "step": 12761
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.660830992776022e-05,
+      "loss": 0.0187,
+      "step": 12762
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.658938628206248e-05,
+      "loss": 0.0337,
+      "step": 12763
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.657046684923647e-05,
+      "loss": 0.0489,
+      "step": 12764
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.655155162998493e-05,
+      "loss": 0.071,
+      "step": 12765
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.653264062501061e-05,
+      "loss": 0.0742,
+      "step": 12766
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6513733835015914e-05,
+      "loss": 0.0412,
+      "step": 12767
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6494831260703236e-05,
+      "loss": 0.0247,
+      "step": 12768
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.647593290277474e-05,
+      "loss": 0.0767,
+      "step": 12769
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6457038761932425e-05,
+      "loss": 0.0615,
+      "step": 12770
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6438148838878125e-05,
+      "loss": 0.0749,
+      "step": 12771
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.641926313431361e-05,
+      "loss": 0.0537,
+      "step": 12772
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.640038164894035e-05,
+      "loss": 0.0451,
+      "step": 12773
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.638150438345984e-05,
+      "loss": 0.0137,
+      "step": 12774
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.636263133857325e-05,
+      "loss": 0.018,
+      "step": 12775
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.634376251498163e-05,
+      "loss": 0.0517,
+      "step": 12776
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.632489791338598e-05,
+      "loss": 0.0181,
+      "step": 12777
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.630603753448697e-05,
+      "loss": 0.0438,
+      "step": 12778
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.628718137898536e-05,
+      "loss": 0.0629,
+      "step": 12779
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6268329447581404e-05,
+      "loss": 0.0607,
+      "step": 12780
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.624948174097554e-05,
+      "loss": 0.0361,
+      "step": 12781
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.623063825986783e-05,
+      "loss": 0.026,
+      "step": 12782
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.621179900495831e-05,
+      "loss": 0.0444,
+      "step": 12783
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.619296397694677e-05,
+      "loss": 0.0244,
+      "step": 12784
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6174133176532854e-05,
+      "loss": 0.0375,
+      "step": 12785
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6155306604416134e-05,
+      "loss": 0.0178,
+      "step": 12786
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.61364842612959e-05,
+      "loss": 0.0335,
+      "step": 12787
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6117666147871395e-05,
+      "loss": 0.0558,
+      "step": 12788
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.609885226484164e-05,
+      "loss": 0.0524,
+      "step": 12789
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.608004261290547e-05,
+      "loss": 0.0514,
+      "step": 12790
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6061237192761694e-05,
+      "loss": 0.0466,
+      "step": 12791
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6042436005108806e-05,
+      "loss": 0.0286,
+      "step": 12792
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6023639050645264e-05,
+      "loss": 0.0597,
+      "step": 12793
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.6004846330069295e-05,
+      "loss": 0.0186,
+      "step": 12794
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.598605784407901e-05,
+      "loss": 0.0481,
+      "step": 12795
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5967273593372285e-05,
+      "loss": 0.0604,
+      "step": 12796
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.594849357864699e-05,
+      "loss": 0.0257,
+      "step": 12797
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.592971780060067e-05,
+      "loss": 0.011,
+      "step": 12798
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5910946259930865e-05,
+      "loss": 0.0285,
+      "step": 12799
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.589217895733484e-05,
+      "loss": 0.0234,
+      "step": 12800
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.587341589350971e-05,
+      "loss": 0.0097,
+      "step": 12801
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5854657069152546e-05,
+      "loss": 0.054,
+      "step": 12802
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5835902484960094e-05,
+      "loss": 0.022,
+      "step": 12803
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.581715214162914e-05,
+      "loss": 0.0207,
+      "step": 12804
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.579840603985613e-05,
+      "loss": 0.0182,
+      "step": 12805
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5779664180337455e-05,
+      "loss": 0.0468,
+      "step": 12806
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.576092656376925e-05,
+      "loss": 0.0405,
+      "step": 12807
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.574219319084762e-05,
+      "loss": 0.0324,
+      "step": 12808
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.572346406226851e-05,
+      "loss": 0.0405,
+      "step": 12809
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.57047391787276e-05,
+      "loss": 0.0727,
+      "step": 12810
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.568601854092045e-05,
+      "loss": 0.0341,
+      "step": 12811
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.566730214954245e-05,
+      "loss": 0.0362,
+      "step": 12812
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5648590005288944e-05,
+      "loss": 0.0473,
+      "step": 12813
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.562988210885494e-05,
+      "loss": 0.0455,
+      "step": 12814
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.561117846093546e-05,
+      "loss": 0.0819,
+      "step": 12815
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5592479062225274e-05,
+      "loss": 0.036,
+      "step": 12816
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.557378391341894e-05,
+      "loss": 0.0293,
+      "step": 12817
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5555093015211026e-05,
+      "loss": 0.0502,
+      "step": 12818
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5536406368295764e-05,
+      "loss": 0.0121,
+      "step": 12819
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.551772397336742e-05,
+      "loss": 0.0502,
+      "step": 12820
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5499045831119824e-05,
+      "loss": 0.0285,
+      "step": 12821
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.548037194224697e-05,
+      "loss": 0.0491,
+      "step": 12822
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.546170230744241e-05,
+      "loss": 0.0442,
+      "step": 12823
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5443036927399783e-05,
+      "loss": 0.036,
+      "step": 12824
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.542437580281239e-05,
+      "loss": 0.0472,
+      "step": 12825
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5405718934373396e-05,
+      "loss": 0.0502,
+      "step": 12826
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.5387066322775964e-05,
+      "loss": 0.0563,
+      "step": 12827
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.536841796871286e-05,
+      "loss": 0.0891,
+      "step": 12828
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.534977387287691e-05,
+      "loss": 0.0287,
+      "step": 12829
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.533113403596066e-05,
+      "loss": 0.0406,
+      "step": 12830
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.531249845865649e-05,
+      "loss": 0.0515,
+      "step": 12831
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.529386714165671e-05,
+      "loss": 0.0385,
+      "step": 12832
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5275240085653334e-05,
+      "loss": 0.0001,
+      "step": 12833
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.525661729133843e-05,
+      "loss": 0.0425,
+      "step": 12834
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.523799875940371e-05,
+      "loss": 0.0298,
+      "step": 12835
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.521938449054078e-05,
+      "loss": 0.0438,
+      "step": 12836
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.52007744854411e-05,
+      "loss": 0.0306,
+      "step": 12837
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.518216874479603e-05,
+      "loss": 0.0061,
+      "step": 12838
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.516356726929666e-05,
+      "loss": 0.024,
+      "step": 12839
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5144970059634055e-05,
+      "loss": 0.0506,
+      "step": 12840
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5126377116498994e-05,
+      "loss": 0.0295,
+      "step": 12841
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.510778844058213e-05,
+      "loss": 0.0589,
+      "step": 12842
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.508920403257404e-05,
+      "loss": 0.08,
+      "step": 12843
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5070623893164996e-05,
+      "loss": 0.04,
+      "step": 12844
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5052048023045334e-05,
+      "loss": 0.0544,
+      "step": 12845
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.503347642290493e-05,
+      "loss": 0.0357,
+      "step": 12846
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.501490909343378e-05,
+      "loss": 0.0569,
+      "step": 12847
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4996346035321526e-05,
+      "loss": 0.0361,
+      "step": 12848
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4977787249257775e-05,
+      "loss": 0.0514,
+      "step": 12849
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4959232735932016e-05,
+      "loss": 0.0347,
+      "step": 12850
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.49406824960333e-05,
+      "loss": 0.0695,
+      "step": 12851
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4922136530250886e-05,
+      "loss": 0.0188,
+      "step": 12852
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4903594839273576e-05,
+      "loss": 0.0291,
+      "step": 12853
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.488505742379026e-05,
+      "loss": 0.0958,
+      "step": 12854
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.486652428448947e-05,
+      "loss": 0.0233,
+      "step": 12855
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.484799542205965e-05,
+      "loss": 0.0667,
+      "step": 12856
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.482947083718916e-05,
+      "loss": 0.018,
+      "step": 12857
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4810950530566035e-05,
+      "loss": 0.0633,
+      "step": 12858
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.479243450287837e-05,
+      "loss": 0.0288,
+      "step": 12859
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.477392275481391e-05,
+      "loss": 0.0372,
+      "step": 12860
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.475541528706032e-05,
+      "loss": 0.0164,
+      "step": 12861
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4736912100305075e-05,
+      "loss": 0.0539,
+      "step": 12862
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.471841319523557e-05,
+      "loss": 0.0409,
+      "step": 12863
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4699918572538896e-05,
+      "loss": 0.039,
+      "step": 12864
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4681428232902205e-05,
+      "loss": 0.0321,
+      "step": 12865
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4662942177012266e-05,
+      "loss": 0.0048,
+      "step": 12866
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.464446040555577e-05,
+      "loss": 0.0352,
+      "step": 12867
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.462598291921933e-05,
+      "loss": 0.0237,
+      "step": 12868
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.460750971868925e-05,
+      "loss": 0.032,
+      "step": 12869
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4589040804651836e-05,
+      "loss": 0.0257,
+      "step": 12870
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.457057617779313e-05,
+      "loss": 0.0105,
+      "step": 12871
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.455211583879897e-05,
+      "loss": 0.0484,
+      "step": 12872
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.453365978835521e-05,
+      "loss": 0.023,
+      "step": 12873
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.451520802714733e-05,
+      "loss": 0.0315,
+      "step": 12874
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4496760555860874e-05,
+      "loss": 0.0707,
+      "step": 12875
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.447831737518104e-05,
+      "loss": 0.0316,
+      "step": 12876
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.445987848579295e-05,
+      "loss": 0.008,
+      "step": 12877
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.444144388838153e-05,
+      "loss": 0.0881,
+      "step": 12878
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.442301358363163e-05,
+      "loss": 0.0053,
+      "step": 12879
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.44045875722278e-05,
+      "loss": 0.0359,
+      "step": 12880
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4386165854854616e-05,
+      "loss": 0.0407,
+      "step": 12881
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.436774843219632e-05,
+      "loss": 0.0326,
+      "step": 12882
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4349335304937054e-05,
+      "loss": 0.0598,
+      "step": 12883
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4330926473760874e-05,
+      "loss": 0.025,
+      "step": 12884
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4312521939351534e-05,
+      "loss": 0.0388,
+      "step": 12885
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.429412170239284e-05,
+      "loss": 0.0092,
+      "step": 12886
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4275725763568146e-05,
+      "loss": 0.035,
+      "step": 12887
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.425733412356092e-05,
+      "loss": 0.036,
+      "step": 12888
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4238946783054266e-05,
+      "loss": 0.0131,
+      "step": 12889
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.422056374273127e-05,
+      "loss": 0.0274,
+      "step": 12890
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.420218500327491e-05,
+      "loss": 0.056,
+      "step": 12891
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.418381056536771e-05,
+      "loss": 0.008,
+      "step": 12892
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.4165440429692366e-05,
+      "loss": 0.0493,
+      "step": 12893
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.414707459693118e-05,
+      "loss": 0.0685,
+      "step": 12894
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.412871306776647e-05,
+      "loss": 0.0174,
+      "step": 12895
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.41103558428803e-05,
+      "loss": 0.0532,
+      "step": 12896
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.409200292295451e-05,
+      "loss": 0.0195,
+      "step": 12897
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.4073654308670976e-05,
+      "loss": 0.0392,
+      "step": 12898
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.405531000071116e-05,
+      "loss": 0.0604,
+      "step": 12899
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.4036969999756635e-05,
+      "loss": 0.0484,
+      "step": 12900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.401863430648862e-05,
+      "loss": 0.0335,
+      "step": 12901
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.400030292158821e-05,
+      "loss": 0.052,
+      "step": 12902
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3981975845736356e-05,
+      "loss": 0.0438,
+      "step": 12903
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3963653079613914e-05,
+      "loss": 0.038,
+      "step": 12904
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3945334623901445e-05,
+      "loss": 0.0519,
+      "step": 12905
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.392702047927951e-05,
+      "loss": 0.0516,
+      "step": 12906
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.390871064642837e-05,
+      "loss": 0.0152,
+      "step": 12907
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.389040512602817e-05,
+      "loss": 0.0526,
+      "step": 12908
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.387210391875896e-05,
+      "loss": 0.0496,
+      "step": 12909
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3853807025300525e-05,
+      "loss": 0.0462,
+      "step": 12910
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.383551444633262e-05,
+      "loss": 0.0645,
+      "step": 12911
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3817226182534634e-05,
+      "loss": 0.0269,
+      "step": 12912
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3798942234586e-05,
+      "loss": 0.0536,
+      "step": 12913
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3780662603165944e-05,
+      "loss": 0.048,
+      "step": 12914
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3762387288953416e-05,
+      "loss": 0.0525,
+      "step": 12915
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.374411629262742e-05,
+      "loss": 0.0382,
+      "step": 12916
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.372584961486649e-05,
+      "loss": 0.056,
+      "step": 12917
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.370758725634933e-05,
+      "loss": 0.0355,
+      "step": 12918
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.3689329217754236e-05,
+      "loss": 0.0771,
+      "step": 12919
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.367107549975952e-05,
+      "loss": 0.0359,
+      "step": 12920
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.365282610304322e-05,
+      "loss": 0.0267,
+      "step": 12921
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.363458102828322e-05,
+      "loss": 0.0193,
+      "step": 12922
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.361634027615732e-05,
+      "loss": 0.0296,
+      "step": 12923
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.359810384734306e-05,
+      "loss": 0.0326,
+      "step": 12924
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.357987174251791e-05,
+      "loss": 0.0458,
+      "step": 12925
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.356164396235911e-05,
+      "loss": 0.0462,
+      "step": 12926
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.354342050754385e-05,
+      "loss": 0.0525,
+      "step": 12927
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3525201378748944e-05,
+      "loss": 0.0231,
+      "step": 12928
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.350698657665129e-05,
+      "loss": 0.053,
+      "step": 12929
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3488776101927416e-05,
+      "loss": 0.0539,
+      "step": 12930
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3470569955253833e-05,
+      "loss": 0.0483,
+      "step": 12931
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.345236813730696e-05,
+      "loss": 0.0171,
+      "step": 12932
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.343417064876276e-05,
+      "loss": 0.0456,
+      "step": 12933
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.341597749029732e-05,
+      "loss": 0.0439,
+      "step": 12934
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.33977886625864e-05,
+      "loss": 0.0586,
+      "step": 12935
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.337960416630574e-05,
+      "loss": 0.0224,
+      "step": 12936
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.33614240021308e-05,
+      "loss": 0.0202,
+      "step": 12937
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.334324817073687e-05,
+      "loss": 0.0507,
+      "step": 12938
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.332507667279923e-05,
+      "loss": 0.0482,
+      "step": 12939
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.33069095089928e-05,
+      "loss": 0.0286,
+      "step": 12940
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.328874667999252e-05,
+      "loss": 0.0306,
+      "step": 12941
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3270588186473057e-05,
+      "loss": 0.0218,
+      "step": 12942
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.325243402910893e-05,
+      "loss": 0.0428,
+      "step": 12943
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.323428420857447e-05,
+      "loss": 0.0468,
+      "step": 12944
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3216138725543993e-05,
+      "loss": 0.0441,
+      "step": 12945
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3197997580691455e-05,
+      "loss": 0.0592,
+      "step": 12946
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3179860774690826e-05,
+      "loss": 0.0536,
+      "step": 12947
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.316172830821578e-05,
+      "loss": 0.0335,
+      "step": 12948
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.314360018193988e-05,
+      "loss": 0.0289,
+      "step": 12949
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.31254763965366e-05,
+      "loss": 0.0279,
+      "step": 12950
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.310735695267909e-05,
+      "loss": 0.0454,
+      "step": 12951
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.308924185104057e-05,
+      "loss": 0.0622,
+      "step": 12952
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3071131092293795e-05,
+      "loss": 0.0545,
+      "step": 12953
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3053024677111595e-05,
+      "loss": 0.0286,
+      "step": 12954
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3034922606166645e-05,
+      "loss": 0.0285,
+      "step": 12955
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.3016824880131266e-05,
+      "loss": 0.0247,
+      "step": 12956
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.299873149967789e-05,
+      "loss": 0.0384,
+      "step": 12957
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.298064246547845e-05,
+      "loss": 0.0098,
+      "step": 12958
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.296255777820502e-05,
+      "loss": 0.0115,
+      "step": 12959
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.294447743852932e-05,
+      "loss": 0.0331,
+      "step": 12960
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.292640144712305e-05,
+      "loss": 0.0387,
+      "step": 12961
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2908329804657675e-05,
+      "loss": 0.0392,
+      "step": 12962
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2890262511804414e-05,
+      "loss": 0.0196,
+      "step": 12963
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2872199569234544e-05,
+      "loss": 0.0437,
+      "step": 12964
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.285414097761893e-05,
+      "loss": 0.0447,
+      "step": 12965
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2836086737628495e-05,
+      "loss": 0.0518,
+      "step": 12966
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2818036849933854e-05,
+      "loss": 0.038,
+      "step": 12967
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2799991315205517e-05,
+      "loss": 0.0441,
+      "step": 12968
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.278195013411378e-05,
+      "loss": 0.0302,
+      "step": 12969
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2763913307328885e-05,
+      "loss": 0.0103,
+      "step": 12970
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.274588083552078e-05,
+      "loss": 0.0106,
+      "step": 12971
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.272785271935941e-05,
+      "loss": 0.091,
+      "step": 12972
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.27098289595144e-05,
+      "loss": 0.0671,
+      "step": 12973
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.269180955665526e-05,
+      "loss": 0.053,
+      "step": 12974
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.267379451145143e-05,
+      "loss": 0.0155,
+      "step": 12975
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.265578382457203e-05,
+      "loss": 0.0188,
+      "step": 12976
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2637777496686214e-05,
+      "loss": 0.0492,
+      "step": 12977
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.261977552846278e-05,
+      "loss": 0.0498,
+      "step": 12978
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.260177792057044e-05,
+      "loss": 0.0428,
+      "step": 12979
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.258378467367782e-05,
+      "loss": 0.0502,
+      "step": 12980
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.256579578845324e-05,
+      "loss": 0.0406,
+      "step": 12981
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2547811265565046e-05,
+      "loss": 0.079,
+      "step": 12982
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.252983110568116e-05,
+      "loss": 0.0773,
+      "step": 12983
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.251185530946962e-05,
+      "loss": 0.0096,
+      "step": 12984
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.249388387759808e-05,
+      "loss": 0.0328,
+      "step": 12985
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.247591681073421e-05,
+      "loss": 0.0236,
+      "step": 12986
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2457954109545356e-05,
+      "loss": 0.0462,
+      "step": 12987
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.243999577469886e-05,
+      "loss": 0.0293,
+      "step": 12988
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.2422041806861774e-05,
+      "loss": 0.0444,
+      "step": 12989
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.240409220670099e-05,
+      "loss": 0.0375,
+      "step": 12990
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.23861469748834e-05,
+      "loss": 0.0424,
+      "step": 12991
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.236820611207549e-05,
+      "loss": 0.03,
+      "step": 12992
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.235026961894385e-05,
+      "loss": 0.0462,
+      "step": 12993
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.233233749615462e-05,
+      "loss": 0.0449,
+      "step": 12994
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.231440974437398e-05,
+      "loss": 0.0315,
+      "step": 12995
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2296486364267955e-05,
+      "loss": 0.089,
+      "step": 12996
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.227856735650225e-05,
+      "loss": 0.0303,
+      "step": 12997
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.226065272174266e-05,
+      "loss": 0.0681,
+      "step": 12998
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.224274246065445e-05,
+      "loss": 0.0323,
+      "step": 12999
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.222483657390309e-05,
+      "loss": 0.091,
+      "step": 13000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2206935062153625e-05,
+      "loss": 0.0423,
+      "step": 13001
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2189037926071144e-05,
+      "loss": 0.0477,
+      "step": 13002
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.217114516632042e-05,
+      "loss": 0.0227,
+      "step": 13003
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.21532567835661e-05,
+      "loss": 0.0512,
+      "step": 13004
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2135372778472726e-05,
+      "loss": 0.0135,
+      "step": 13005
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.211749315170459e-05,
+      "loss": 0.0154,
+      "step": 13006
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.209961790392595e-05,
+      "loss": 0.057,
+      "step": 13007
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.208174703580074e-05,
+      "loss": 0.0131,
+      "step": 13008
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.206388054799284e-05,
+      "loss": 0.0534,
+      "step": 13009
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.2046018441165896e-05,
+      "loss": 0.0302,
+      "step": 13010
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.202816071598351e-05,
+      "loss": 0.0881,
+      "step": 13011
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.201030737310897e-05,
+      "loss": 0.0541,
+      "step": 13012
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.199245841320555e-05,
+      "loss": 0.0217,
+      "step": 13013
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.197461383693623e-05,
+      "loss": 0.0603,
+      "step": 13014
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.195677364496387e-05,
+      "loss": 0.0292,
+      "step": 13015
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1938937837951244e-05,
+      "loss": 0.0259,
+      "step": 13016
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1921106416560834e-05,
+      "loss": 0.0481,
+      "step": 13017
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.190327938145511e-05,
+      "loss": 0.0535,
+      "step": 13018
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1885456733296214e-05,
+      "loss": 0.0471,
+      "step": 13019
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.18676384727462e-05,
+      "loss": 0.0249,
+      "step": 13020
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.184982460046704e-05,
+      "loss": 0.0246,
+      "step": 13021
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.183201511712039e-05,
+      "loss": 0.0403,
+      "step": 13022
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.181421002336793e-05,
+      "loss": 0.0526,
+      "step": 13023
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.179640931987091e-05,
+      "loss": 0.0324,
+      "step": 13024
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.1778613007290694e-05,
+      "loss": 0.0477,
+      "step": 13025
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1760821086288276e-05,
+      "loss": 0.0425,
+      "step": 13026
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1743033557524676e-05,
+      "loss": 0.0835,
+      "step": 13027
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.172525042166057e-05,
+      "loss": 0.0763,
+      "step": 13028
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1707471679356556e-05,
+      "loss": 0.0182,
+      "step": 13029
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.168969733127311e-05,
+      "loss": 0.0233,
+      "step": 13030
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.167192737807043e-05,
+      "loss": 0.0599,
+      "step": 13031
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.165416182040869e-05,
+      "loss": 0.0318,
+      "step": 13032
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1636400658947805e-05,
+      "loss": 0.0345,
+      "step": 13033
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.161864389434752e-05,
+      "loss": 0.0433,
+      "step": 13034
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1600891527267425e-05,
+      "loss": 0.0551,
+      "step": 13035
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1583143558367016e-05,
+      "loss": 0.0297,
+      "step": 13036
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.156539998830559e-05,
+      "loss": 0.0739,
+      "step": 13037
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.154766081774228e-05,
+      "loss": 0.0436,
+      "step": 13038
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1529926047336e-05,
+      "loss": 0.0513,
+      "step": 13039
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.15121956777455e-05,
+      "loss": 0.0378,
+      "step": 13040
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.149446970962953e-05,
+      "loss": 0.0325,
+      "step": 13041
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1476748143646435e-05,
+      "loss": 0.0233,
+      "step": 13042
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.145903098045462e-05,
+      "loss": 0.0267,
+      "step": 13043
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1441318220712204e-05,
+      "loss": 0.0717,
+      "step": 13044
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.142360986507709e-05,
+      "loss": 0.0505,
+      "step": 13045
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1405905914207204e-05,
+      "loss": 0.0473,
+      "step": 13046
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1388206368760085e-05,
+      "loss": 0.0223,
+      "step": 13047
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.137051122939336e-05,
+      "loss": 0.0436,
+      "step": 13048
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.135282049676419e-05,
+      "loss": 0.0693,
+      "step": 13049
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.133513417152983e-05,
+      "loss": 0.0737,
+      "step": 13050
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.131745225434723e-05,
+      "loss": 0.0274,
+      "step": 13051
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.1299774745873276e-05,
+      "loss": 0.0602,
+      "step": 13052
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.128210164676457e-05,
+      "loss": 0.0316,
+      "step": 13053
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.126443295767769e-05,
+      "loss": 0.017,
+      "step": 13054
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.124676867926893e-05,
+      "loss": 0.0004,
+      "step": 13055
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.122910881219444e-05,
+      "loss": 0.0177,
+      "step": 13056
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.12114533571103e-05,
+      "loss": 0.0335,
+      "step": 13057
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.11938023146723e-05,
+      "loss": 0.0239,
+      "step": 13058
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.117615568553617e-05,
+      "loss": 0.0527,
+      "step": 13059
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1158513470357415e-05,
+      "loss": 0.0102,
+      "step": 13060
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.114087566979134e-05,
+      "loss": 0.0196,
+      "step": 13061
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.112324228449322e-05,
+      "loss": 0.0592,
+      "step": 13062
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.110561331511801e-05,
+      "loss": 0.0243,
+      "step": 13063
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1087988762320686e-05,
+      "loss": 0.0512,
+      "step": 13064
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.107036862675581e-05,
+      "loss": 0.0267,
+      "step": 13065
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1052752909078006e-05,
+      "loss": 0.0278,
+      "step": 13066
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.1035141609941584e-05,
+      "loss": 0.0631,
+      "step": 13067
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.101753473000082e-05,
+      "loss": 0.0243,
+      "step": 13068
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.099993226990974e-05,
+      "loss": 0.0402,
+      "step": 13069
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0982334230322166e-05,
+      "loss": 0.0258,
+      "step": 13070
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0964740611891906e-05,
+      "loss": 0.041,
+      "step": 13071
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.094715141527242e-05,
+      "loss": 0.0493,
+      "step": 13072
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0929566641117174e-05,
+      "loss": 0.0803,
+      "step": 13073
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.091198629007937e-05,
+      "loss": 0.056,
+      "step": 13074
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.089441036281205e-05,
+      "loss": 0.0212,
+      "step": 13075
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0876838859968055e-05,
+      "loss": 0.0287,
+      "step": 13076
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.085927178220018e-05,
+      "loss": 0.0231,
+      "step": 13077
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.084170913016103e-05,
+      "loss": 0.0345,
+      "step": 13078
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.082415090450294e-05,
+      "loss": 0.047,
+      "step": 13079
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0806597105878186e-05,
+      "loss": 0.0203,
+      "step": 13080
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.078904773493875e-05,
+      "loss": 0.0347,
+      "step": 13081
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.077150279233666e-05,
+      "loss": 0.0707,
+      "step": 13082
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0753962278723564e-05,
+      "loss": 0.0278,
+      "step": 13083
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.073642619475114e-05,
+      "loss": 0.0525,
+      "step": 13084
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0718894541070734e-05,
+      "loss": 0.0783,
+      "step": 13085
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0701367318333556e-05,
+      "loss": 0.0401,
+      "step": 13086
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.06838445271908e-05,
+      "loss": 0.0362,
+      "step": 13087
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.066632616829328e-05,
+      "loss": 0.0525,
+      "step": 13088
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.064881224229188e-05,
+      "loss": 0.064,
+      "step": 13089
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0631302749837026e-05,
+      "loss": 0.0212,
+      "step": 13090
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.061379769157927e-05,
+      "loss": 0.0321,
+      "step": 13091
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0596297068168794e-05,
+      "loss": 0.0488,
+      "step": 13092
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0578800880255774e-05,
+      "loss": 0.0362,
+      "step": 13093
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.056130912849011e-05,
+      "loss": 0.0307,
+      "step": 13094
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.05438218135215e-05,
+      "loss": 0.0423,
+      "step": 13095
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0526338935999674e-05,
+      "loss": 0.0634,
+      "step": 13096
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.050886049657395e-05,
+      "loss": 0.0372,
+      "step": 13097
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0491386495893683e-05,
+      "loss": 0.0313,
+      "step": 13098
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.047391693460797e-05,
+      "loss": 0.0441,
+      "step": 13099
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0456451813365683e-05,
+      "loss": 0.0321,
+      "step": 13100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0438991132815698e-05,
+      "loss": 0.0557,
+      "step": 13101
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0421534893606542e-05,
+      "loss": 0.0436,
+      "step": 13102
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0404083096386745e-05,
+      "loss": 0.0557,
+      "step": 13103
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0386635741804537e-05,
+      "loss": 0.0162,
+      "step": 13104
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0369192830508072e-05,
+      "loss": 0.1158,
+      "step": 13105
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0351754363145224e-05,
+      "loss": 0.0309,
+      "step": 13106
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0334320340363866e-05,
+      "loss": 0.05,
+      "step": 13107
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0316890762811574e-05,
+      "loss": 0.0568,
+      "step": 13108
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.029946563113585e-05,
+      "loss": 0.019,
+      "step": 13109
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.028204494598396e-05,
+      "loss": 0.0334,
+      "step": 13110
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.026462870800298e-05,
+      "loss": 0.055,
+      "step": 13111
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0247216917839966e-05,
+      "loss": 0.0149,
+      "step": 13112
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.022980957614165e-05,
+      "loss": 0.054,
+      "step": 13113
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0212406683554696e-05,
+      "loss": 0.0214,
+      "step": 13114
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0195008240725588e-05,
+      "loss": 0.05,
+      "step": 13115
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0177614248300575e-05,
+      "loss": 0.072,
+      "step": 13116
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.016022470692579e-05,
+      "loss": 0.0424,
+      "step": 13117
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0142839617247232e-05,
+      "loss": 0.0526,
+      "step": 13118
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.012545897991074e-05,
+      "loss": 0.0472,
+      "step": 13119
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.010808279556192e-05,
+      "loss": 0.0472,
+      "step": 13120
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0090711064846233e-05,
+      "loss": 0.0246,
+      "step": 13121
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0073343788408977e-05,
+      "loss": 0.0473,
+      "step": 13122
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.005598096689535e-05,
+      "loss": 0.0416,
+      "step": 13123
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0038622600950253e-05,
+      "loss": 0.0251,
+      "step": 13124
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.002126869121857e-05,
+      "loss": 0.0112,
+      "step": 13125
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0003919238344935e-05,
+      "loss": 0.0322,
+      "step": 13126
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9986574242973787e-05,
+      "loss": 0.0453,
+      "step": 13127
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.996923370574949e-05,
+      "loss": 0.0146,
+      "step": 13128
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9951897627316147e-05,
+      "loss": 0.0584,
+      "step": 13129
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.993456600831784e-05,
+      "loss": 0.0121,
+      "step": 13130
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.991723884939824e-05,
+      "loss": 0.032,
+      "step": 13131
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9899916151201116e-05,
+      "loss": 0.0446,
+      "step": 13132
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9882597914369867e-05,
+      "loss": 0.026,
+      "step": 13133
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9865284139547912e-05,
+      "loss": 0.0317,
+      "step": 13134
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9847974827378356e-05,
+      "loss": 0.0441,
+      "step": 13135
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9830669978504167e-05,
+      "loss": 0.0345,
+      "step": 13136
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9813369593568214e-05,
+      "loss": 0.0429,
+      "step": 13137
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9796073673213117e-05,
+      "loss": 0.0262,
+      "step": 13138
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.977878221808141e-05,
+      "loss": 0.0599,
+      "step": 13139
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9761495228815413e-05,
+      "loss": 0.0213,
+      "step": 13140
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.974421270605723e-05,
+      "loss": 0.0437,
+      "step": 13141
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.972693465044893e-05,
+      "loss": 0.042,
+      "step": 13142
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9709661062632267e-05,
+      "loss": 0.0238,
+      "step": 13143
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9692391943248994e-05,
+      "loss": 0.057,
+      "step": 13144
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9675127292940548e-05,
+      "loss": 0.0389,
+      "step": 13145
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9657867112348298e-05,
+      "loss": 0.0765,
+      "step": 13146
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.964061140211332e-05,
+      "loss": 0.0528,
+      "step": 13147
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9623360162876723e-05,
+      "loss": 0.0664,
+      "step": 13148
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9606113395279275e-05,
+      "loss": 0.0189,
+      "step": 13149
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9588871099961687e-05,
+      "loss": 0.035,
+      "step": 13150
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9571633277564433e-05,
+      "loss": 0.0618,
+      "step": 13151
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.95543999287278e-05,
+      "loss": 0.0439,
+      "step": 13152
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9537171054092062e-05,
+      "loss": 0.0554,
+      "step": 13153
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9519946654297123e-05,
+      "loss": 0.0346,
+      "step": 13154
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9502726729982934e-05,
+      "loss": 0.0471,
+      "step": 13155
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.948551128178902e-05,
+      "loss": 0.0447,
+      "step": 13156
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9468300310354986e-05,
+      "loss": 0.0283,
+      "step": 13157
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9451093816320103e-05,
+      "loss": 0.0331,
+      "step": 13158
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9433891800323595e-05,
+      "loss": 0.0218,
+      "step": 13159
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.94166942630045e-05,
+      "loss": 0.0432,
+      "step": 13160
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.939950120500155e-05,
+      "loss": 0.0435,
+      "step": 13161
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9382312626953497e-05,
+      "loss": 0.0475,
+      "step": 13162
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.936512852949879e-05,
+      "loss": 0.0643,
+      "step": 13163
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9347948913275832e-05,
+      "loss": 0.0321,
+      "step": 13164
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9330773778922767e-05,
+      "loss": 0.0506,
+      "step": 13165
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.931360312707755e-05,
+      "loss": 0.0572,
+      "step": 13166
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9296436958378123e-05,
+      "loss": 0.0591,
+      "step": 13167
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9279275273462054e-05,
+      "loss": 0.0473,
+      "step": 13168
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.926211807296694e-05,
+      "loss": 0.0403,
+      "step": 13169
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9244965357530066e-05,
+      "loss": 0.0219,
+      "step": 13170
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9227817127788623e-05,
+      "loss": 0.0456,
+      "step": 13171
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9210673384379585e-05,
+      "loss": 0.0261,
+      "step": 13172
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9193534127939845e-05,
+      "loss": 0.0479,
+      "step": 13173
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9176399359106023e-05,
+      "loss": 0.0529,
+      "step": 13174
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.915926907851468e-05,
+      "loss": 0.0449,
+      "step": 13175
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9142143286802138e-05,
+      "loss": 0.0239,
+      "step": 13176
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.912502198460451e-05,
+      "loss": 0.0228,
+      "step": 13177
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9107905172557894e-05,
+      "loss": 0.0277,
+      "step": 13178
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9090792851298057e-05,
+      "loss": 0.0351,
+      "step": 13179
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9073685021460737e-05,
+      "loss": 0.0103,
+      "step": 13180
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9056581683681407e-05,
+      "loss": 0.0371,
+      "step": 13181
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9039482838595364e-05,
+      "loss": 0.035,
+      "step": 13182
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9022388486837844e-05,
+      "loss": 0.0459,
+      "step": 13183
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.900529862904381e-05,
+      "loss": 0.0485,
+      "step": 13184
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8988213265848155e-05,
+      "loss": 0.038,
+      "step": 13185
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.89711323978855e-05,
+      "loss": 0.0214,
+      "step": 13186
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8954056025790372e-05,
+      "loss": 0.0628,
+      "step": 13187
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.893698415019705e-05,
+      "loss": 0.032,
+      "step": 13188
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.8919916771739793e-05,
+      "loss": 0.0329,
+      "step": 13189
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.890285389105254e-05,
+      "loss": 0.0227,
+      "step": 13190
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.888579550876917e-05,
+      "loss": 0.0228,
+      "step": 13191
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8868741625523344e-05,
+      "loss": 0.0381,
+      "step": 13192
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8851692241948522e-05,
+      "loss": 0.0329,
+      "step": 13193
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8834647358678092e-05,
+      "loss": 0.0118,
+      "step": 13194
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8817606976345154e-05,
+      "loss": 0.0244,
+      "step": 13195
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8800571095582847e-05,
+      "loss": 0.0429,
+      "step": 13196
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.878353971702382e-05,
+      "loss": 0.0467,
+      "step": 13197
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8766512841300875e-05,
+      "loss": 0.0239,
+      "step": 13198
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8749490469046415e-05,
+      "loss": 0.0333,
+      "step": 13199
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.873247260089283e-05,
+      "loss": 0.0717,
+      "step": 13200
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.871545923747235e-05,
+      "loss": 0.0174,
+      "step": 13201
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8698450379416814e-05,
+      "loss": 0.0093,
+      "step": 13202
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8681446027358174e-05,
+      "loss": 0.0173,
+      "step": 13203
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8664446181928002e-05,
+      "loss": 0.037,
+      "step": 13204
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8647450843757897e-05,
+      "loss": 0.0025,
+      "step": 13205
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8630460013479117e-05,
+      "loss": 0.0422,
+      "step": 13206
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8613473691722794e-05,
+      "loss": 0.0393,
+      "step": 13207
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8596491879120004e-05,
+      "loss": 0.0182,
+      "step": 13208
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8579514576301493e-05,
+      "loss": 0.015,
+      "step": 13209
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.856254178389799e-05,
+      "loss": 0.0397,
+      "step": 13210
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.854557350253994e-05,
+      "loss": 0.0282,
+      "step": 13211
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8528609732857678e-05,
+      "loss": 0.0259,
+      "step": 13212
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8511650475481325e-05,
+      "loss": 0.0234,
+      "step": 13213
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8494695731040913e-05,
+      "loss": 0.0528,
+      "step": 13214
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8477745500166206e-05,
+      "loss": 0.0175,
+      "step": 13215
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8460799783486947e-05,
+      "loss": 0.0249,
+      "step": 13216
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8443858581632557e-05,
+      "loss": 0.0342,
+      "step": 13217
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8426921895232325e-05,
+      "loss": 0.0246,
+      "step": 13218
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8409989724915466e-05,
+      "loss": 0.0116,
+      "step": 13219
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8393062071310894e-05,
+      "loss": 0.0185,
+      "step": 13220
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8376138935047526e-05,
+      "loss": 0.0338,
+      "step": 13221
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.8359220316753872e-05,
+      "loss": 0.0232,
+      "step": 13222
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.834230621705848e-05,
+      "loss": 0.0238,
+      "step": 13223
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8325396636589682e-05,
+      "loss": 0.0169,
+      "step": 13224
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8308491575975557e-05,
+      "loss": 0.0206,
+      "step": 13225
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8291591035844185e-05,
+      "loss": 0.0465,
+      "step": 13226
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8274695016823218e-05,
+      "loss": 0.0231,
+      "step": 13227
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.825780351954042e-05,
+      "loss": 0.0322,
+      "step": 13228
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.824091654462318e-05,
+      "loss": 0.0204,
+      "step": 13229
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.822403409269886e-05,
+      "loss": 0.025,
+      "step": 13230
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8207156164394583e-05,
+      "loss": 0.0281,
+      "step": 13231
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8190282760337252e-05,
+      "loss": 0.008,
+      "step": 13232
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.817341388115376e-05,
+      "loss": 0.0294,
+      "step": 13233
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8156549527470647e-05,
+      "loss": 0.0196,
+      "step": 13234
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8139689699914452e-05,
+      "loss": 0.0435,
+      "step": 13235
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8122834399111433e-05,
+      "loss": 0.0051,
+      "step": 13236
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8105983625687715e-05,
+      "loss": 0.0257,
+      "step": 13237
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8089137380269204e-05,
+      "loss": 0.0134,
+      "step": 13238
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8072295663481797e-05,
+      "loss": 0.0566,
+      "step": 13239
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.805545847595101e-05,
+      "loss": 0.035,
+      "step": 13240
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.803862581830234e-05,
+      "loss": 0.0187,
+      "step": 13241
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.802179769116114e-05,
+      "loss": 0.0212,
+      "step": 13242
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.8004974095152394e-05,
+      "loss": 0.0179,
+      "step": 13243
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.798815503090115e-05,
+      "loss": 0.0264,
+      "step": 13244
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7971340499032107e-05,
+      "loss": 0.025,
+      "step": 13245
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7954530500169954e-05,
+      "loss": 0.0209,
+      "step": 13246
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.79377250349391e-05,
+      "loss": 0.0342,
+      "step": 13247
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.792092410396379e-05,
+      "loss": 0.0378,
+      "step": 13248
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7904127707868178e-05,
+      "loss": 0.0236,
+      "step": 13249
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7887335847276143e-05,
+      "loss": 0.0335,
+      "step": 13250
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7870548522811532e-05,
+      "loss": 0.0225,
+      "step": 13251
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7853765735097894e-05,
+      "loss": 0.0332,
+      "step": 13252
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.7836987484758683e-05,
+      "loss": 0.0243,
+      "step": 13253
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.78202137724171e-05,
+      "loss": 0.0405,
+      "step": 13254
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.780344459869631e-05,
+      "loss": 0.0089,
+      "step": 13255
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.778667996421919e-05,
+      "loss": 0.0341,
+      "step": 13256
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7769919869608566e-05,
+      "loss": 0.0159,
+      "step": 13257
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7753164315486974e-05,
+      "loss": 0.0299,
+      "step": 13258
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7736413302476807e-05,
+      "loss": 0.0199,
+      "step": 13259
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7719666831200392e-05,
+      "loss": 0.0255,
+      "step": 13260
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.770292490227972e-05,
+      "loss": 0.0079,
+      "step": 13261
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7686187516336845e-05,
+      "loss": 0.0195,
+      "step": 13262
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.766945467399333e-05,
+      "loss": 0.023,
+      "step": 13263
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7652726375870865e-05,
+      "loss": 0.0118,
+      "step": 13264
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7636002622590853e-05,
+      "loss": 0.0353,
+      "step": 13265
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.761928341477448e-05,
+      "loss": 0.0238,
+      "step": 13266
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.760256875304293e-05,
+      "loss": 0.0171,
+      "step": 13267
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7585858638016944e-05,
+      "loss": 0.0275,
+      "step": 13268
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7569153070317374e-05,
+      "loss": 0.0208,
+      "step": 13269
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7552452050564705e-05,
+      "loss": 0.0229,
+      "step": 13270
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7535755579379394e-05,
+      "loss": 0.0169,
+      "step": 13271
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7519063657381658e-05,
+      "loss": 0.0359,
+      "step": 13272
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7502376285191484e-05,
+      "loss": 0.0155,
+      "step": 13273
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.748569346342885e-05,
+      "loss": 0.0259,
+      "step": 13274
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.74690151927134e-05,
+      "loss": 0.0101,
+      "step": 13275
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7452341473664748e-05,
+      "loss": 0.0342,
+      "step": 13276
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.743567230690224e-05,
+      "loss": 0.0192,
+      "step": 13277
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7419007693045104e-05,
+      "loss": 0.0243,
+      "step": 13278
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7402347632712323e-05,
+      "loss": 0.0285,
+      "step": 13279
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.738569212652284e-05,
+      "loss": 0.0128,
+      "step": 13280
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7369041175095307e-05,
+      "loss": 0.0407,
+      "step": 13281
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7352394779048314e-05,
+      "loss": 0.0434,
+      "step": 13282
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7335752939000182e-05,
+      "loss": 0.0084,
+      "step": 13283
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.73191156555691e-05,
+      "loss": 0.0398,
+      "step": 13284
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7302482929373144e-05,
+      "loss": 0.0332,
+      "step": 13285
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7285854761030097e-05,
+      "loss": 0.0396,
+      "step": 13286
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.7269231151157733e-05,
+      "loss": 0.0168,
+      "step": 13287
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.725261210037351e-05,
+      "loss": 0.0314,
+      "step": 13288
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.723599760929477e-05,
+      "loss": 0.0053,
+      "step": 13289
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.721938767853876e-05,
+      "loss": 0.0272,
+      "step": 13290
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.72027823087224e-05,
+      "loss": 0.0207,
+      "step": 13291
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.718618150046265e-05,
+      "loss": 0.0422,
+      "step": 13292
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7169585254376043e-05,
+      "loss": 0.0149,
+      "step": 13293
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7152993571079178e-05,
+      "loss": 0.0275,
+      "step": 13294
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7136406451188318e-05,
+      "loss": 0.0222,
+      "step": 13295
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7119823895319704e-05,
+      "loss": 0.0122,
+      "step": 13296
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7103245904089288e-05,
+      "loss": 0.0401,
+      "step": 13297
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7086672478112863e-05,
+      "loss": 0.011,
+      "step": 13298
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7070103618006157e-05,
+      "loss": 0.0281,
+      "step": 13299
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7053539324384553e-05,
+      "loss": 0.0191,
+      "step": 13300
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7036979597863482e-05,
+      "loss": 0.0304,
+      "step": 13301
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7020424439057985e-05,
+      "loss": 0.0258,
+      "step": 13302
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.7003873848583167e-05,
+      "loss": 0.0255,
+      "step": 13303
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.698732782705367e-05,
+      "loss": 0.03,
+      "step": 13304
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6970786375084225e-05,
+      "loss": 0.0222,
+      "step": 13305
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6954249493289316e-05,
+      "loss": 0.0183,
+      "step": 13306
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.693771718228317e-05,
+      "loss": 0.0312,
+      "step": 13307
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.692118944268003e-05,
+      "loss": 0.0138,
+      "step": 13308
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6904666275093706e-05,
+      "loss": 0.01,
+      "step": 13309
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6888147680138088e-05,
+      "loss": 0.0047,
+      "step": 13310
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6871633658426722e-05,
+      "loss": 0.0105,
+      "step": 13311
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.685512421057315e-05,
+      "loss": 0.0204,
+      "step": 13312
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6838619337190575e-05,
+      "loss": 0.0486,
+      "step": 13313
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.682211903889208e-05,
+      "loss": 0.0335,
+      "step": 13314
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6805623316290708e-05,
+      "loss": 0.0275,
+      "step": 13315
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6789132169999118e-05,
+      "loss": 0.0422,
+      "step": 13316
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6772645600629984e-05,
+      "loss": 0.0176,
+      "step": 13317
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6756163608795717e-05,
+      "loss": 0.0276,
+      "step": 13318
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6739686195108546e-05,
+      "loss": 0.0374,
+      "step": 13319
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6723213360180545e-05,
+      "loss": 0.0162,
+      "step": 13320
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.670674510462371e-05,
+      "loss": 0.0261,
+      "step": 13321
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.6690281429049688e-05,
+      "loss": 0.017,
+      "step": 13322
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.667382233407015e-05,
+      "loss": 0.0036,
+      "step": 13323
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.665736782029646e-05,
+      "loss": 0.031,
+      "step": 13324
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.664091788833983e-05,
+      "loss": 0.0311,
+      "step": 13325
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6624472538811385e-05,
+      "loss": 0.0196,
+      "step": 13326
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6608031772321942e-05,
+      "loss": 0.0153,
+      "step": 13327
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.659159558948232e-05,
+      "loss": 0.019,
+      "step": 13328
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6575163990903026e-05,
+      "loss": 0.0233,
+      "step": 13329
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6558736977194412e-05,
+      "loss": 0.0424,
+      "step": 13330
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6542314548966776e-05,
+      "loss": 0.0264,
+      "step": 13331
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.652589670683008e-05,
+      "loss": 0.0493,
+      "step": 13332
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6509483451394304e-05,
+      "loss": 0.0356,
+      "step": 13333
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.649307478326901e-05,
+      "loss": 0.0108,
+      "step": 13334
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.647667070306384e-05,
+      "loss": 0.0115,
+      "step": 13335
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6460271211388096e-05,
+      "loss": 0.0091,
+      "step": 13336
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6443876308851035e-05,
+      "loss": 0.0183,
+      "step": 13337
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.642748599606164e-05,
+      "loss": 0.02,
+      "step": 13338
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6411100273628744e-05,
+      "loss": 0.0324,
+      "step": 13339
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.639471914216107e-05,
+      "loss": 0.0336,
+      "step": 13340
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6378342602267083e-05,
+      "loss": 0.0351,
+      "step": 13341
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.636197065455518e-05,
+      "loss": 0.0312,
+      "step": 13342
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.634560329963351e-05,
+      "loss": 0.0146,
+      "step": 13343
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6329240538110068e-05,
+      "loss": 0.037,
+      "step": 13344
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6312882370592636e-05,
+      "loss": 0.0121,
+      "step": 13345
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.629652879768891e-05,
+      "loss": 0.0252,
+      "step": 13346
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.628017982000642e-05,
+      "loss": 0.0265,
+      "step": 13347
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6263835438152447e-05,
+      "loss": 0.0104,
+      "step": 13348
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6247495652734125e-05,
+      "loss": 0.0335,
+      "step": 13349
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6231160464358415e-05,
+      "loss": 0.0248,
+      "step": 13350
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6214829873632182e-05,
+      "loss": 0.0566,
+      "step": 13351
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6198503881161976e-05,
+      "loss": 0.0156,
+      "step": 13352
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.6182182487554342e-05,
+      "loss": 0.0371,
+      "step": 13353
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.616586569341555e-05,
+      "loss": 0.0171,
+      "step": 13354
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.614955349935166e-05,
+      "loss": 0.0461,
+      "step": 13355
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6133245905968692e-05,
+      "loss": 0.021,
+      "step": 13356
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.611694291387239e-05,
+      "loss": 0.0181,
+      "step": 13357
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.610064452366844e-05,
+      "loss": 0.0425,
+      "step": 13358
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.608435073596214e-05,
+      "loss": 0.0307,
+      "step": 13359
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6068061551358853e-05,
+      "loss": 0.0281,
+      "step": 13360
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6051776970463633e-05,
+      "loss": 0.0145,
+      "step": 13361
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.603549699388146e-05,
+      "loss": 0.0213,
+      "step": 13362
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6019221622217e-05,
+      "loss": 0.0163,
+      "step": 13363
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.6002950856074937e-05,
+      "loss": 0.0065,
+      "step": 13364
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5986684696059624e-05,
+      "loss": 0.0275,
+      "step": 13365
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.597042314277527e-05,
+      "loss": 0.0295,
+      "step": 13366
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5954166196826032e-05,
+      "loss": 0.0237,
+      "step": 13367
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5937913858815708e-05,
+      "loss": 0.0386,
+      "step": 13368
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.592166612934811e-05,
+      "loss": 0.0156,
+      "step": 13369
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5905423009026783e-05,
+      "loss": 0.0229,
+      "step": 13370
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.588918449845502e-05,
+      "loss": 0.023,
+      "step": 13371
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5872950598236153e-05,
+      "loss": 0.0334,
+      "step": 13372
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.585672130897314e-05,
+      "loss": 0.0074,
+      "step": 13373
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5840496631268943e-05,
+      "loss": 0.0121,
+      "step": 13374
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5824276565726133e-05,
+      "loss": 0.0258,
+      "step": 13375
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5808061112947342e-05,
+      "loss": 0.0162,
+      "step": 13376
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.579185027353486e-05,
+      "loss": 0.0233,
+      "step": 13377
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.577564404809093e-05,
+      "loss": 0.0499,
+      "step": 13378
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5759442437217548e-05,
+      "loss": 0.0516,
+      "step": 13379
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5743245441516507e-05,
+      "loss": 0.02,
+      "step": 13380
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.572705306158955e-05,
+      "loss": 0.0186,
+      "step": 13381
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.571086529803812e-05,
+      "loss": 0.0209,
+      "step": 13382
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5694682151463593e-05,
+      "loss": 0.0266,
+      "step": 13383
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.56785036224671e-05,
+      "loss": 0.0511,
+      "step": 13384
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5662329711649648e-05,
+      "loss": 0.0299,
+      "step": 13385
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5646160419611973e-05,
+      "loss": 0.0249,
+      "step": 13386
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5629995746954822e-05,
+      "loss": 0.0171,
+      "step": 13387
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.5613835694278573e-05,
+      "loss": 0.0219,
+      "step": 13388
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5597680262183605e-05,
+      "loss": 0.0336,
+      "step": 13389
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.558152945127001e-05,
+      "loss": 0.0253,
+      "step": 13390
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5565383262137708e-05,
+      "loss": 0.0121,
+      "step": 13391
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5549241695386546e-05,
+      "loss": 0.0264,
+      "step": 13392
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5533104751616084e-05,
+      "loss": 0.0224,
+      "step": 13393
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.551697243142581e-05,
+      "loss": 0.0397,
+      "step": 13394
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5500844735414972e-05,
+      "loss": 0.0252,
+      "step": 13395
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.548472166418261e-05,
+      "loss": 0.0044,
+      "step": 13396
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.546860321832775e-05,
+      "loss": 0.0383,
+      "step": 13397
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5452489398449056e-05,
+      "loss": 0.0388,
+      "step": 13398
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5436380205145235e-05,
+      "loss": 0.0147,
+      "step": 13399
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5420275639014526e-05,
+      "loss": 0.0324,
+      "step": 13400
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.540417570065528e-05,
+      "loss": 0.0347,
+      "step": 13401
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5388080390665494e-05,
+      "loss": 0.0249,
+      "step": 13402
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5371989709643146e-05,
+      "loss": 0.0184,
+      "step": 13403
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.53559036581859e-05,
+      "loss": 0.0391,
+      "step": 13404
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5339822236891288e-05,
+      "loss": 0.0084,
+      "step": 13405
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.532374544635677e-05,
+      "loss": 0.027,
+      "step": 13406
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.530767328717943e-05,
+      "loss": 0.0226,
+      "step": 13407
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.529160575995643e-05,
+      "loss": 0.0278,
+      "step": 13408
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.527554286528457e-05,
+      "loss": 0.0179,
+      "step": 13409
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5259484603760494e-05,
+      "loss": 0.0333,
+      "step": 13410
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5243430975980806e-05,
+      "loss": 0.0507,
+      "step": 13411
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.522738198254178e-05,
+      "loss": 0.026,
+      "step": 13412
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.521133762403965e-05,
+      "loss": 0.0214,
+      "step": 13413
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5195297901070398e-05,
+      "loss": 0.0285,
+      "step": 13414
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5179262814229845e-05,
+      "loss": 0.05,
+      "step": 13415
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5163232364113593e-05,
+      "loss": 0.0368,
+      "step": 13416
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5147206551317223e-05,
+      "loss": 0.0308,
+      "step": 13417
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.513118537643597e-05,
+      "loss": 0.014,
+      "step": 13418
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5115168840065026e-05,
+      "loss": 0.0083,
+      "step": 13419
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5099156942799348e-05,
+      "loss": 0.0265,
+      "step": 13420
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.5083149685233677e-05,
+      "loss": 0.0298,
+      "step": 13421
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.506714706796274e-05,
+      "loss": 0.0242,
+      "step": 13422
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.505114909158086e-05,
+      "loss": 0.0346,
+      "step": 13423
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5035155756682474e-05,
+      "loss": 0.0189,
+      "step": 13424
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5019167063861513e-05,
+      "loss": 0.045,
+      "step": 13425
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.5003183013712032e-05,
+      "loss": 0.0396,
+      "step": 13426
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.498720360682772e-05,
+      "loss": 0.0522,
+      "step": 13427
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4971228843802227e-05,
+      "loss": 0.0261,
+      "step": 13428
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4955258725228905e-05,
+      "loss": 0.0367,
+      "step": 13429
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4939293251701066e-05,
+      "loss": 0.0206,
+      "step": 13430
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.492333242381175e-05,
+      "loss": 0.0034,
+      "step": 13431
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.49073762421538e-05,
+      "loss": 0.0276,
+      "step": 13432
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.489142470732004e-05,
+      "loss": 0.0332,
+      "step": 13433
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4875477819902932e-05,
+      "loss": 0.016,
+      "step": 13434
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.485953558049494e-05,
+      "loss": 0.0318,
+      "step": 13435
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.484359798968823e-05,
+      "loss": 0.0202,
+      "step": 13436
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4827665048074802e-05,
+      "loss": 0.0315,
+      "step": 13437
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.481173675624658e-05,
+      "loss": 0.0109,
+      "step": 13438
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4795813114795193e-05,
+      "loss": 0.026,
+      "step": 13439
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4779894124312266e-05,
+      "loss": 0.0112,
+      "step": 13440
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4763979785388994e-05,
+      "loss": 0.0294,
+      "step": 13441
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.474807009861665e-05,
+      "loss": 0.0301,
+      "step": 13442
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4732165064586185e-05,
+      "loss": 0.0164,
+      "step": 13443
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.471626468388847e-05,
+      "loss": 0.0383,
+      "step": 13444
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.470036895711412e-05,
+      "loss": 0.0124,
+      "step": 13445
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.468447788485361e-05,
+      "loss": 0.0376,
+      "step": 13446
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4668591467697284e-05,
+      "loss": 0.0244,
+      "step": 13447
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4652709706235224e-05,
+      "loss": 0.0221,
+      "step": 13448
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4636832601057437e-05,
+      "loss": 0.0479,
+      "step": 13449
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4620960152753723e-05,
+      "loss": 0.0155,
+      "step": 13450
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.460509236191362e-05,
+      "loss": 0.0169,
+      "step": 13451
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4589229229126645e-05,
+      "loss": 0.019,
+      "step": 13452
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.4573370754982006e-05,
+      "loss": 0.0431,
+      "step": 13453
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.455751694006889e-05,
+      "loss": 0.0311,
+      "step": 13454
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4541667784976148e-05,
+      "loss": 0.0327,
+      "step": 13455
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4525823290292534e-05,
+      "loss": 0.0098,
+      "step": 13456
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4509983456606606e-05,
+      "loss": 0.0105,
+      "step": 13457
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.449414828450685e-05,
+      "loss": 0.02,
+      "step": 13458
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4478317774581387e-05,
+      "loss": 0.0157,
+      "step": 13459
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4462491927418367e-05,
+      "loss": 0.056,
+      "step": 13460
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4446670743605636e-05,
+      "loss": 0.0245,
+      "step": 13461
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4430854223730872e-05,
+      "loss": 0.0061,
+      "step": 13462
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4415042368381677e-05,
+      "loss": 0.0258,
+      "step": 13463
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.439923517814536e-05,
+      "loss": 0.0148,
+      "step": 13464
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.43834326536092e-05,
+      "loss": 0.0245,
+      "step": 13465
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4367634795360068e-05,
+      "loss": 0.0245,
+      "step": 13466
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.435184160398494e-05,
+      "loss": 0.0352,
+      "step": 13467
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4336053080070394e-05,
+      "loss": 0.019,
+      "step": 13468
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4320269224202997e-05,
+      "loss": 0.0477,
+      "step": 13469
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4304490036969065e-05,
+      "loss": 0.0108,
+      "step": 13470
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4288715518954675e-05,
+      "loss": 0.0159,
+      "step": 13471
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4272945670745898e-05,
+      "loss": 0.0172,
+      "step": 13472
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4257180492928473e-05,
+      "loss": 0.0066,
+      "step": 13473
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4241419986088073e-05,
+      "loss": 0.0199,
+      "step": 13474
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.422566415081016e-05,
+      "loss": 0.0251,
+      "step": 13475
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4209912987679932e-05,
+      "loss": 0.013,
+      "step": 13476
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4194166497282618e-05,
+      "loss": 0.0455,
+      "step": 13477
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.417842468020304e-05,
+      "loss": 0.0148,
+      "step": 13478
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.416268753702606e-05,
+      "loss": 0.044,
+      "step": 13479
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4146955068336227e-05,
+      "loss": 0.0159,
+      "step": 13480
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4131227274717956e-05,
+      "loss": 0.0176,
+      "step": 13481
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.411550415675545e-05,
+      "loss": 0.0315,
+      "step": 13482
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.409978571503284e-05,
+      "loss": 0.0211,
+      "step": 13483
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4084071950133953e-05,
+      "loss": 0.0177,
+      "step": 13484
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4068362862642583e-05,
+      "loss": 0.0225,
+      "step": 13485
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.4052658453142236e-05,
+      "loss": 0.0165,
+      "step": 13486
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.403695872221627e-05,
+      "loss": 0.0499,
+      "step": 13487
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.4021263670447932e-05,
+      "loss": 0.0295,
+      "step": 13488
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.400557329842019e-05,
+      "loss": 0.0235,
+      "step": 13489
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3989887606715958e-05,
+      "loss": 0.0175,
+      "step": 13490
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.397420659591787e-05,
+      "loss": 0.0339,
+      "step": 13491
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.395853026660845e-05,
+      "loss": 0.0187,
+      "step": 13492
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3942858619369975e-05,
+      "loss": 0.0456,
+      "step": 13493
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3927191654784633e-05,
+      "loss": 0.0319,
+      "step": 13494
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3911529373434474e-05,
+      "loss": 0.0138,
+      "step": 13495
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.389587177590123e-05,
+      "loss": 0.0185,
+      "step": 13496
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.388021886276656e-05,
+      "loss": 0.0267,
+      "step": 13497
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3864570634611873e-05,
+      "loss": 0.0112,
+      "step": 13498
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3848927092018544e-05,
+      "loss": 0.0106,
+      "step": 13499
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3833288235567606e-05,
+      "loss": 0.0269,
+      "step": 13500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3817654065840042e-05,
+      "loss": 0.0404,
+      "step": 13501
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3802024583416625e-05,
+      "loss": 0.0279,
+      "step": 13502
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3786399788877885e-05,
+      "loss": 0.0145,
+      "step": 13503
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3770779682804296e-05,
+      "loss": 0.0284,
+      "step": 13504
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.375516426577606e-05,
+      "loss": 0.0132,
+      "step": 13505
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.373955353837331e-05,
+      "loss": 0.0215,
+      "step": 13506
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.372394750117584e-05,
+      "loss": 0.0242,
+      "step": 13507
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.370834615476344e-05,
+      "loss": 0.0188,
+      "step": 13508
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3692749499715607e-05,
+      "loss": 0.0228,
+      "step": 13509
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3677157536611774e-05,
+      "loss": 0.0428,
+      "step": 13510
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3661570266031092e-05,
+      "loss": 0.0228,
+      "step": 13511
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.364598768855257e-05,
+      "loss": 0.0088,
+      "step": 13512
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.363040980475509e-05,
+      "loss": 0.0308,
+      "step": 13513
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.361483661521729e-05,
+      "loss": 0.0552,
+      "step": 13514
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.359926812051771e-05,
+      "loss": 0.021,
+      "step": 13515
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.358370432123467e-05,
+      "loss": 0.0245,
+      "step": 13516
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3568145217946255e-05,
+      "loss": 0.025,
+      "step": 13517
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.355259081123052e-05,
+      "loss": 0.0277,
+      "step": 13518
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3537041101665205e-05,
+      "loss": 0.0206,
+      "step": 13519
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3521496089827997e-05,
+      "loss": 0.0202,
+      "step": 13520
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3505955776296316e-05,
+      "loss": 0.0122,
+      "step": 13521
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.349042016164743e-05,
+      "loss": 0.0326,
+      "step": 13522
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3474889246458406e-05,
+      "loss": 0.0216,
+      "step": 13523
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3459363031306273e-05,
+      "loss": 0.0123,
+      "step": 13524
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.344384151676767e-05,
+      "loss": 0.0196,
+      "step": 13525
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.342832470341927e-05,
+      "loss": 0.0301,
+      "step": 13526
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3412812591837448e-05,
+      "loss": 0.0327,
+      "step": 13527
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3397305182598393e-05,
+      "loss": 0.041,
+      "step": 13528
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3381802476278223e-05,
+      "loss": 0.0207,
+      "step": 13529
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3366304473452747e-05,
+      "loss": 0.0273,
+      "step": 13530
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.335081117469777e-05,
+      "loss": 0.0317,
+      "step": 13531
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.33353225805887e-05,
+      "loss": 0.0222,
+      "step": 13532
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3319838691700993e-05,
+      "loss": 0.0269,
+      "step": 13533
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3304359508609754e-05,
+      "loss": 0.0342,
+      "step": 13534
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3288885031890027e-05,
+      "loss": 0.029,
+      "step": 13535
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3273415262116728e-05,
+      "loss": 0.0413,
+      "step": 13536
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.325795019986434e-05,
+      "loss": 0.0142,
+      "step": 13537
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.324248984570748e-05,
+      "loss": 0.0142,
+      "step": 13538
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3227034200220374e-05,
+      "loss": 0.0114,
+      "step": 13539
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.321158326397721e-05,
+      "loss": 0.0099,
+      "step": 13540
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3196137037551936e-05,
+      "loss": 0.0157,
+      "step": 13541
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.318069552151829e-05,
+      "loss": 0.0196,
+      "step": 13542
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3165258716449942e-05,
+      "loss": 0.0418,
+      "step": 13543
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3149826622920255e-05,
+      "loss": 0.0105,
+      "step": 13544
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3134399241502565e-05,
+      "loss": 0.0603,
+      "step": 13545
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3118976572769904e-05,
+      "loss": 0.0184,
+      "step": 13546
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.310355861729519e-05,
+      "loss": 0.0244,
+      "step": 13547
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3088145375651123e-05,
+      "loss": 0.0278,
+      "step": 13548
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3072736848410323e-05,
+      "loss": 0.0173,
+      "step": 13549
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.30573330361451e-05,
+      "loss": 0.0253,
+      "step": 13550
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3041933939427725e-05,
+      "loss": 0.0127,
+      "step": 13551
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.3026539558830215e-05,
+      "loss": 0.0382,
+      "step": 13552
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.3011149894924374e-05,
+      "loss": 0.0193,
+      "step": 13553
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2995764948281962e-05,
+      "loss": 0.024,
+      "step": 13554
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.298038471947441e-05,
+      "loss": 0.0094,
+      "step": 13555
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.296500920907312e-05,
+      "loss": 0.0119,
+      "step": 13556
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2949638417649207e-05,
+      "loss": 0.0209,
+      "step": 13557
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2934272345773613e-05,
+      "loss": 0.0036,
+      "step": 13558
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2918910994017232e-05,
+      "loss": 0.0174,
+      "step": 13559
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2903554362950605e-05,
+      "loss": 0.018,
+      "step": 13560
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.288820245314426e-05,
+      "loss": 0.0284,
+      "step": 13561
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2872855265168444e-05,
+      "loss": 0.0221,
+      "step": 13562
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2857512799593263e-05,
+      "loss": 0.0171,
+      "step": 13563
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2842175056988598e-05,
+      "loss": 0.0301,
+      "step": 13564
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2826842037924275e-05,
+      "loss": 0.0199,
+      "step": 13565
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2811513742969805e-05,
+      "loss": 0.0312,
+      "step": 13566
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.279619017269465e-05,
+      "loss": 0.0266,
+      "step": 13567
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2780871327668022e-05,
+      "loss": 0.0078,
+      "step": 13568
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.276555720845891e-05,
+      "loss": 0.0253,
+      "step": 13569
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2750247815636285e-05,
+      "loss": 0.0175,
+      "step": 13570
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2734943149768753e-05,
+      "loss": 0.0301,
+      "step": 13571
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2719643211424942e-05,
+      "loss": 0.0073,
+      "step": 13572
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2704348001173083e-05,
+      "loss": 0.02,
+      "step": 13573
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.268905751958144e-05,
+      "loss": 0.015,
+      "step": 13574
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2673771767217934e-05,
+      "loss": 0.0116,
+      "step": 13575
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.265849074465042e-05,
+      "loss": 0.0339,
+      "step": 13576
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.264321445244662e-05,
+      "loss": 0.0488,
+      "step": 13577
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2627942891173873e-05,
+      "loss": 0.0106,
+      "step": 13578
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2612676061399564e-05,
+      "loss": 0.0186,
+      "step": 13579
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2597413963690747e-05,
+      "loss": 0.0256,
+      "step": 13580
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2582156598614415e-05,
+      "loss": 0.0328,
+      "step": 13581
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.256690396673733e-05,
+      "loss": 0.0458,
+      "step": 13582
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.255165606862603e-05,
+      "loss": 0.0174,
+      "step": 13583
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2536412904847e-05,
+      "loss": 0.0144,
+      "step": 13584
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2521174475966413e-05,
+      "loss": 0.0143,
+      "step": 13585
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2505940782550396e-05,
+      "loss": 0.0313,
+      "step": 13586
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2490711825164814e-05,
+      "loss": 0.0215,
+      "step": 13587
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2475487604375382e-05,
+      "loss": 0.029,
+      "step": 13588
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.246026812074756e-05,
+      "loss": 0.0269,
+      "step": 13589
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2445053374846815e-05,
+      "loss": 0.0171,
+      "step": 13590
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2429843367238264e-05,
+      "loss": 0.021,
+      "step": 13591
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2414638098486955e-05,
+      "loss": 0.0115,
+      "step": 13592
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2399437569157713e-05,
+      "loss": 0.0265,
+      "step": 13593
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2384241779815127e-05,
+      "loss": 0.0478,
+      "step": 13594
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2369050731023774e-05,
+      "loss": 0.0376,
+      "step": 13595
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.235386442334787e-05,
+      "loss": 0.0082,
+      "step": 13596
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.233868285735167e-05,
+      "loss": 0.0368,
+      "step": 13597
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2323506033598965e-05,
+      "loss": 0.0472,
+      "step": 13598
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.230833395265359e-05,
+      "loss": 0.0397,
+      "step": 13599
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2293166615079194e-05,
+      "loss": 0.0535,
+      "step": 13600
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.227800402143913e-05,
+      "loss": 0.0347,
+      "step": 13601
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.226284617229674e-05,
+      "loss": 0.0061,
+      "step": 13602
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2247693068214963e-05,
+      "loss": 0.0186,
+      "step": 13603
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2232544709756796e-05,
+      "loss": 0.0117,
+      "step": 13604
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.221740109748487e-05,
+      "loss": 0.026,
+      "step": 13605
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2202262231961824e-05,
+      "loss": 0.0216,
+      "step": 13606
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2187128113749958e-05,
+      "loss": 0.0216,
+      "step": 13607
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.217199874341145e-05,
+      "loss": 0.0135,
+      "step": 13608
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2156874121508374e-05,
+      "loss": 0.02,
+      "step": 13609
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.214175424860249e-05,
+      "loss": 0.0297,
+      "step": 13610
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2126639125255518e-05,
+      "loss": 0.0322,
+      "step": 13611
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.211152875202893e-05,
+      "loss": 0.0199,
+      "step": 13612
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.209642312948402e-05,
+      "loss": 0.0402,
+      "step": 13613
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2081322258181878e-05,
+      "loss": 0.0322,
+      "step": 13614
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.206622613868352e-05,
+      "loss": 0.0046,
+      "step": 13615
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2051134771549677e-05,
+      "loss": 0.0333,
+      "step": 13616
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2036048157340973e-05,
+      "loss": 0.0475,
+      "step": 13617
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2020966296617892e-05,
+      "loss": 0.0466,
+      "step": 13618
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.2005889189940534e-05,
+      "loss": 0.0154,
+      "step": 13619
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.19908168378691e-05,
+      "loss": 0.048,
+      "step": 13620
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1975749240963416e-05,
+      "loss": 0.0253,
+      "step": 13621
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.196068639978324e-05,
+      "loss": 0.0205,
+      "step": 13622
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.194562831488808e-05,
+      "loss": 0.0331,
+      "step": 13623
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1930574986837306e-05,
+      "loss": 0.0255,
+      "step": 13624
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.191552641619012e-05,
+      "loss": 0.0415,
+      "step": 13625
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.190048260350551e-05,
+      "loss": 0.0369,
+      "step": 13626
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.188544354934234e-05,
+      "loss": 0.0301,
+      "step": 13627
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.187040925425925e-05,
+      "loss": 0.028,
+      "step": 13628
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.185537971881473e-05,
+      "loss": 0.0187,
+      "step": 13629
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1840354943567028e-05,
+      "loss": 0.0216,
+      "step": 13630
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1825334929074352e-05,
+      "loss": 0.0173,
+      "step": 13631
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1810319675894584e-05,
+      "loss": 0.0226,
+      "step": 13632
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1795309184585565e-05,
+      "loss": 0.021,
+      "step": 13633
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1780303455704844e-05,
+      "loss": 0.0106,
+      "step": 13634
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1765302489809818e-05,
+      "loss": 0.0333,
+      "step": 13635
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1750306287457782e-05,
+      "loss": 0.0332,
+      "step": 13636
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.173531484920576e-05,
+      "loss": 0.0341,
+      "step": 13637
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1720328175610722e-05,
+      "loss": 0.0244,
+      "step": 13638
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1705346267229245e-05,
+      "loss": 0.0092,
+      "step": 13639
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.169036912461793e-05,
+      "loss": 0.0216,
+      "step": 13640
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.167539674833318e-05,
+      "loss": 0.0144,
+      "step": 13641
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1660429138931085e-05,
+      "loss": 0.0204,
+      "step": 13642
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1645466296967768e-05,
+      "loss": 0.0151,
+      "step": 13643
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1630508222998917e-05,
+      "loss": 0.0254,
+      "step": 13644
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1615554917580286e-05,
+      "loss": 0.0407,
+      "step": 13645
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1600606381267267e-05,
+      "loss": 0.0216,
+      "step": 13646
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.158566261461523e-05,
+      "loss": 0.0256,
+      "step": 13647
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1570723618179247e-05,
+      "loss": 0.0432,
+      "step": 13648
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.155578939251423e-05,
+      "loss": 0.0153,
+      "step": 13649
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.154085993817501e-05,
+      "loss": 0.0163,
+      "step": 13650
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.1525935255716115e-05,
+      "loss": 0.025,
+      "step": 13651
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1511015345692e-05,
+      "loss": 0.0135,
+      "step": 13652
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1496100208656892e-05,
+      "loss": 0.0188,
+      "step": 13653
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.148118984516481e-05,
+      "loss": 0.0298,
+      "step": 13654
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.146628425576963e-05,
+      "loss": 0.0518,
+      "step": 13655
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1451383441025088e-05,
+      "loss": 0.018,
+      "step": 13656
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1436487401484647e-05,
+      "loss": 0.0136,
+      "step": 13657
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1421596137701747e-05,
+      "loss": 0.0059,
+      "step": 13658
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1406709650229474e-05,
+      "loss": 0.041,
+      "step": 13659
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.139182793962081e-05,
+      "loss": 0.0442,
+      "step": 13660
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.137695100642864e-05,
+      "loss": 0.0226,
+      "step": 13661
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1362078851205523e-05,
+      "loss": 0.0373,
+      "step": 13662
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1347211474503982e-05,
+      "loss": 0.0186,
+      "step": 13663
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1332348876876266e-05,
+      "loss": 0.0169,
+      "step": 13664
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1317491058874427e-05,
+      "loss": 0.0215,
+      "step": 13665
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.130263802105049e-05,
+      "loss": 0.027,
+      "step": 13666
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1287789763956098e-05,
+      "loss": 0.026,
+      "step": 13667
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1272946288142943e-05,
+      "loss": 0.0443,
+      "step": 13668
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.125810759416227e-05,
+      "loss": 0.039,
+      "step": 13669
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.124327368256542e-05,
+      "loss": 0.0437,
+      "step": 13670
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1228444553903334e-05,
+      "loss": 0.0098,
+      "step": 13671
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.121362020872694e-05,
+      "loss": 0.0217,
+      "step": 13672
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1198800647586894e-05,
+      "loss": 0.0265,
+      "step": 13673
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1183985871033675e-05,
+      "loss": 0.0292,
+      "step": 13674
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.116917587961766e-05,
+      "loss": 0.0225,
+      "step": 13675
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1154370673888926e-05,
+      "loss": 0.0387,
+      "step": 13676
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.113957025439753e-05,
+      "loss": 0.0269,
+      "step": 13677
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1124774621693168e-05,
+      "loss": 0.0335,
+      "step": 13678
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1109983776325584e-05,
+      "loss": 0.0119,
+      "step": 13679
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.109519771884407e-05,
+      "loss": 0.0252,
+      "step": 13680
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1080416449797954e-05,
+      "loss": 0.0197,
+      "step": 13681
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1065639969736346e-05,
+      "loss": 0.0157,
+      "step": 13682
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.1050868279208078e-05,
+      "loss": 0.0212,
+      "step": 13683
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.103610137876199e-05,
+      "loss": 0.0485,
+      "step": 13684
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.1021339268946474e-05,
+      "loss": 0.0129,
+      "step": 13685
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.1006581950310015e-05,
+      "loss": 0.0352,
+      "step": 13686
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.099182942340072e-05,
+      "loss": 0.0264,
+      "step": 13687
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0977081688766702e-05,
+      "loss": 0.0207,
+      "step": 13688
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.096233874695574e-05,
+      "loss": 0.0115,
+      "step": 13689
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.094760059851544e-05,
+      "loss": 0.0155,
+      "step": 13690
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.093286724399339e-05,
+      "loss": 0.0485,
+      "step": 13691
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0918138683936797e-05,
+      "loss": 0.0326,
+      "step": 13692
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0903414918892843e-05,
+      "loss": 0.028,
+      "step": 13693
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0888695949408468e-05,
+      "loss": 0.0069,
+      "step": 13694
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0873981776030422e-05,
+      "loss": 0.0371,
+      "step": 13695
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0859272399305255e-05,
+      "loss": 0.0336,
+      "step": 13696
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.084456781977944e-05,
+      "loss": 0.0597,
+      "step": 13697
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0829868037999176e-05,
+      "loss": 0.0452,
+      "step": 13698
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.081517305451054e-05,
+      "loss": 0.0191,
+      "step": 13699
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0800482869859404e-05,
+      "loss": 0.0316,
+      "step": 13700
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.078579748459142e-05,
+      "loss": 0.0386,
+      "step": 13701
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.077111689925216e-05,
+      "loss": 0.0296,
+      "step": 13702
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0756441114386936e-05,
+      "loss": 0.0315,
+      "step": 13703
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.074177013054094e-05,
+      "loss": 0.0272,
+      "step": 13704
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0727103948259143e-05,
+      "loss": 0.0488,
+      "step": 13705
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.071244256808631e-05,
+      "loss": 0.0355,
+      "step": 13706
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0697785990567145e-05,
+      "loss": 0.0389,
+      "step": 13707
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0683134216246005e-05,
+      "loss": 0.0093,
+      "step": 13708
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0668487245667288e-05,
+      "loss": 0.0216,
+      "step": 13709
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0653845079374938e-05,
+      "loss": 0.017,
+      "step": 13710
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.063920771791297e-05,
+      "loss": 0.0094,
+      "step": 13711
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0624575161825052e-05,
+      "loss": 0.035,
+      "step": 13712
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0609947411654813e-05,
+      "loss": 0.032,
+      "step": 13713
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0595324467945606e-05,
+      "loss": 0.0185,
+      "step": 13714
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0580706331240576e-05,
+      "loss": 0.0547,
+      "step": 13715
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0566093002082824e-05,
+      "loss": 0.0383,
+      "step": 13716
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.0551484481015114e-05,
+      "loss": 0.038,
+      "step": 13717
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0536880768580195e-05,
+      "loss": 0.0572,
+      "step": 13718
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.05222818653205e-05,
+      "loss": 0.0264,
+      "step": 13719
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0507687771778347e-05,
+      "loss": 0.0366,
+      "step": 13720
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0493098488495817e-05,
+      "loss": 0.049,
+      "step": 13721
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0478514016014912e-05,
+      "loss": 0.0669,
+      "step": 13722
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0463934354877415e-05,
+      "loss": 0.0254,
+      "step": 13723
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0449359505624906e-05,
+      "loss": 0.029,
+      "step": 13724
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.043478946879879e-05,
+      "loss": 0.0099,
+      "step": 13725
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0420224244940265e-05,
+      "loss": 0.0249,
+      "step": 13726
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0405663834590446e-05,
+      "loss": 0.0289,
+      "step": 13727
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0391108238290155e-05,
+      "loss": 0.0063,
+      "step": 13728
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.037655745658014e-05,
+      "loss": 0.0061,
+      "step": 13729
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.036201149000092e-05,
+      "loss": 0.0164,
+      "step": 13730
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.034747033909276e-05,
+      "loss": 0.0219,
+      "step": 13731
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.033293400439591e-05,
+      "loss": 0.0477,
+      "step": 13732
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0318402486450297e-05,
+      "loss": 0.0392,
+      "step": 13733
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0303875785795798e-05,
+      "loss": 0.0301,
+      "step": 13734
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0289353902971915e-05,
+      "loss": 0.0108,
+      "step": 13735
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0274836838518206e-05,
+      "loss": 0.0366,
+      "step": 13736
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0260324592973847e-05,
+      "loss": 0.0206,
+      "step": 13737
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0245817166877998e-05,
+      "loss": 0.0055,
+      "step": 13738
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0231314560769525e-05,
+      "loss": 0.0129,
+      "step": 13739
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0216816775187196e-05,
+      "loss": 0.0185,
+      "step": 13740
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0202323810669546e-05,
+      "loss": 0.0164,
+      "step": 13741
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.01878356677549e-05,
+      "loss": 0.0192,
+      "step": 13742
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.017335234698152e-05,
+      "loss": 0.0242,
+      "step": 13743
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0158873848887347e-05,
+      "loss": 0.0433,
+      "step": 13744
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.01444001740103e-05,
+      "loss": 0.0331,
+      "step": 13745
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0129931322887984e-05,
+      "loss": 0.058,
+      "step": 13746
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.011546729605783e-05,
+      "loss": 0.0132,
+      "step": 13747
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.010100809405723e-05,
+      "loss": 0.0548,
+      "step": 13748
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.008655371742322e-05,
+      "loss": 0.0352,
+      "step": 13749
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.0072104166692836e-05,
+      "loss": 0.026,
+      "step": 13750
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.0057659442402717e-05,
+      "loss": 0.0237,
+      "step": 13751
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.0043219545089517e-05,
+      "loss": 0.0087,
+      "step": 13752
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.0028784475289587e-05,
+      "loss": 0.0274,
+      "step": 13753
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.0014354233539205e-05,
+      "loss": 0.0124,
+      "step": 13754
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9999928820374382e-05,
+      "loss": 0.0253,
+      "step": 13755
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9985508236330966e-05,
+      "loss": 0.0107,
+      "step": 13756
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.997109248194467e-05,
+      "loss": 0.0208,
+      "step": 13757
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9956681557750958e-05,
+      "loss": 0.0183,
+      "step": 13758
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9942275464285195e-05,
+      "loss": 0.0372,
+      "step": 13759
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9927874202082524e-05,
+      "loss": 0.0431,
+      "step": 13760
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9913477771677876e-05,
+      "loss": 0.0192,
+      "step": 13761
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9899086173606032e-05,
+      "loss": 0.0144,
+      "step": 13762
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9884699408401628e-05,
+      "loss": 0.0292,
+      "step": 13763
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9870317476599102e-05,
+      "loss": 0.0152,
+      "step": 13764
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9855940378732676e-05,
+      "loss": 0.0833,
+      "step": 13765
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9841568115336427e-05,
+      "loss": 0.0106,
+      "step": 13766
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9827200686944207e-05,
+      "loss": 0.0261,
+      "step": 13767
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.981283809408979e-05,
+      "loss": 0.0177,
+      "step": 13768
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9798480337306614e-05,
+      "loss": 0.0197,
+      "step": 13769
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9784127417128127e-05,
+      "loss": 0.019,
+      "step": 13770
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.976977933408746e-05,
+      "loss": 0.0341,
+      "step": 13771
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9755436088717542e-05,
+      "loss": 0.0362,
+      "step": 13772
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9741097681551276e-05,
+      "loss": 0.034,
+      "step": 13773
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9726764113121223e-05,
+      "loss": 0.0152,
+      "step": 13774
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9712435383959923e-05,
+      "loss": 0.0285,
+      "step": 13775
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.96981114945995e-05,
+      "loss": 0.053,
+      "step": 13776
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9683792445572176e-05,
+      "loss": 0.0577,
+      "step": 13777
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.966947823740978e-05,
+      "loss": 0.0285,
+      "step": 13778
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9655168870644116e-05,
+      "loss": 0.0295,
+      "step": 13779
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9640864345806695e-05,
+      "loss": 0.0181,
+      "step": 13780
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.962656466342884e-05,
+      "loss": 0.0291,
+      "step": 13781
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9612269824041844e-05,
+      "loss": 0.008,
+      "step": 13782
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.9597979828176612e-05,
+      "loss": 0.0324,
+      "step": 13783
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9583694676364076e-05,
+      "loss": 0.0248,
+      "step": 13784
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.956941436913484e-05,
+      "loss": 0.0121,
+      "step": 13785
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.955513890701933e-05,
+      "loss": 0.0328,
+      "step": 13786
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9540868290547924e-05,
+      "loss": 0.0154,
+      "step": 13787
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.952660252025066e-05,
+      "loss": 0.0219,
+      "step": 13788
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9512341596657538e-05,
+      "loss": 0.0322,
+      "step": 13789
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9498085520298263e-05,
+      "loss": 0.0148,
+      "step": 13790
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9483834291702427e-05,
+      "loss": 0.0296,
+      "step": 13791
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9469587911399376e-05,
+      "loss": 0.0335,
+      "step": 13792
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9455346379918386e-05,
+      "loss": 0.0301,
+      "step": 13793
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9441109697788433e-05,
+      "loss": 0.0258,
+      "step": 13794
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9426877865538427e-05,
+      "loss": 0.038,
+      "step": 13795
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.941265088369701e-05,
+      "loss": 0.0415,
+      "step": 13796
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9398428752792618e-05,
+      "loss": 0.0367,
+      "step": 13797
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9384211473353666e-05,
+      "loss": 0.0147,
+      "step": 13798
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9369999045908176e-05,
+      "loss": 0.0243,
+      "step": 13799
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9355791470984235e-05,
+      "loss": 0.0228,
+      "step": 13800
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9341588749109455e-05,
+      "loss": 0.0232,
+      "step": 13801
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9327390880811534e-05,
+      "loss": 0.0532,
+      "step": 13802
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9313197866617815e-05,
+      "loss": 0.0184,
+      "step": 13803
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9299009707055573e-05,
+      "loss": 0.0157,
+      "step": 13804
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.928482640265186e-05,
+      "loss": 0.0269,
+      "step": 13805
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9270647953933533e-05,
+      "loss": 0.0214,
+      "step": 13806
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.925647436142727e-05,
+      "loss": 0.0361,
+      "step": 13807
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9242305625659556e-05,
+      "loss": 0.0512,
+      "step": 13808
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.922814174715676e-05,
+      "loss": 0.0243,
+      "step": 13809
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9213982726445004e-05,
+      "loss": 0.0201,
+      "step": 13810
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.919982856405028e-05,
+      "loss": 0.0173,
+      "step": 13811
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.918567926049835e-05,
+      "loss": 0.0305,
+      "step": 13812
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.917153481631481e-05,
+      "loss": 0.0223,
+      "step": 13813
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9157395232025123e-05,
+      "loss": 0.0329,
+      "step": 13814
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.914326050815449e-05,
+      "loss": 0.0244,
+      "step": 13815
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.9129130645228054e-05,
+      "loss": 0.0019,
+      "step": 13816
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.911500564377056e-05,
+      "loss": 0.0413,
+      "step": 13817
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.910088550430684e-05,
+      "loss": 0.0216,
+      "step": 13818
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9086770227361325e-05,
+      "loss": 0.0233,
+      "step": 13819
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.907265981345843e-05,
+      "loss": 0.0397,
+      "step": 13820
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9058554263122282e-05,
+      "loss": 0.007,
+      "step": 13821
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.904445357687682e-05,
+      "loss": 0.0232,
+      "step": 13822
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.903035775524593e-05,
+      "loss": 0.0415,
+      "step": 13823
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.9016266798753133e-05,
+      "loss": 0.0244,
+      "step": 13824
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.900218070792196e-05,
+      "loss": 0.0292,
+      "step": 13825
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8988099483275627e-05,
+      "loss": 0.0486,
+      "step": 13826
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8974023125337177e-05,
+      "loss": 0.0449,
+      "step": 13827
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8959951634629584e-05,
+      "loss": 0.0235,
+      "step": 13828
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8945885011675455e-05,
+      "loss": 0.0348,
+      "step": 13829
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.893182325699744e-05,
+      "loss": 0.0318,
+      "step": 13830
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8917766371117827e-05,
+      "loss": 0.0259,
+      "step": 13831
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8903714354558808e-05,
+      "loss": 0.0381,
+      "step": 13832
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8889667207842318e-05,
+      "loss": 0.01,
+      "step": 13833
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.887562493149024e-05,
+      "loss": 0.0231,
+      "step": 13834
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8861587526024157e-05,
+      "loss": 0.0139,
+      "step": 13835
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.884755499196557e-05,
+      "loss": 0.017,
+      "step": 13836
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8833527329835706e-05,
+      "loss": 0.0338,
+      "step": 13837
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8819504540155637e-05,
+      "loss": 0.0111,
+      "step": 13838
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8805486623446315e-05,
+      "loss": 0.0146,
+      "step": 13839
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8791473580228415e-05,
+      "loss": 0.0317,
+      "step": 13840
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8777465411022563e-05,
+      "loss": 0.0456,
+      "step": 13841
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8763462116349005e-05,
+      "loss": 0.0187,
+      "step": 13842
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8749463696728024e-05,
+      "loss": 0.0304,
+      "step": 13843
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.873547015267952e-05,
+      "loss": 0.0265,
+      "step": 13844
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8721481484723384e-05,
+      "loss": 0.0196,
+      "step": 13845
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.870749769337931e-05,
+      "loss": 0.0219,
+      "step": 13846
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8693518779166617e-05,
+      "loss": 0.0308,
+      "step": 13847
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8679544742604674e-05,
+      "loss": 0.0131,
+      "step": 13848
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.8665575584212516e-05,
+      "loss": 0.0313,
+      "step": 13849
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.865161130450913e-05,
+      "loss": 0.0102,
+      "step": 13850
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8637651904013195e-05,
+      "loss": 0.005,
+      "step": 13851
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8623697383243253e-05,
+      "loss": 0.0422,
+      "step": 13852
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8609747742717734e-05,
+      "loss": 0.0368,
+      "step": 13853
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8595802982954743e-05,
+      "loss": 0.0172,
+      "step": 13854
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8581863104472384e-05,
+      "loss": 0.0365,
+      "step": 13855
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8567928107788404e-05,
+      "loss": 0.0302,
+      "step": 13856
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8553997993420495e-05,
+      "loss": 0.0289,
+      "step": 13857
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8540072761886054e-05,
+      "loss": 0.0327,
+      "step": 13858
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.852615241370244e-05,
+      "loss": 0.0456,
+      "step": 13859
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8512236949386685e-05,
+      "loss": 0.034,
+      "step": 13860
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8498326369455775e-05,
+      "loss": 0.0369,
+      "step": 13861
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8484420674426413e-05,
+      "loss": 0.0383,
+      "step": 13862
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.847051986481514e-05,
+      "loss": 0.0309,
+      "step": 13863
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.845662394113836e-05,
+      "loss": 0.0219,
+      "step": 13864
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.844273290391224e-05,
+      "loss": 0.0327,
+      "step": 13865
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.842884675365283e-05,
+      "loss": 0.0071,
+      "step": 13866
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.841496549087593e-05,
+      "loss": 0.0135,
+      "step": 13867
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.840108911609716e-05,
+      "loss": 0.0188,
+      "step": 13868
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.838721762983204e-05,
+      "loss": 0.004,
+      "step": 13869
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.837335103259582e-05,
+      "loss": 0.0245,
+      "step": 13870
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.835948932490363e-05,
+      "loss": 0.0099,
+      "step": 13871
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8345632507270404e-05,
+      "loss": 0.0145,
+      "step": 13872
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8331780580210835e-05,
+      "loss": 0.0357,
+      "step": 13873
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.831793354423949e-05,
+      "loss": 0.0274,
+      "step": 13874
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8304091399870778e-05,
+      "loss": 0.0149,
+      "step": 13875
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.829025414761885e-05,
+      "loss": 0.0352,
+      "step": 13876
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.827642178799777e-05,
+      "loss": 0.0201,
+      "step": 13877
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8262594321521344e-05,
+      "loss": 0.0225,
+      "step": 13878
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.82487717487032e-05,
+      "loss": 0.029,
+      "step": 13879
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8234954070056857e-05,
+      "loss": 0.0188,
+      "step": 13880
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8221141286095535e-05,
+      "loss": 0.0239,
+      "step": 13881
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.8207333397332447e-05,
+      "loss": 0.0186,
+      "step": 13882
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8193530404280376e-05,
+      "loss": 0.0291,
+      "step": 13883
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8179732307452166e-05,
+      "loss": 0.0302,
+      "step": 13884
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8165939107360306e-05,
+      "loss": 0.0214,
+      "step": 13885
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.81521508045172e-05,
+      "loss": 0.0153,
+      "step": 13886
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.813836739943513e-05,
+      "loss": 0.016,
+      "step": 13887
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.812458889262596e-05,
+      "loss": 0.0283,
+      "step": 13888
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.811081528460163e-05,
+      "loss": 0.0259,
+      "step": 13889
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.809704657587371e-05,
+      "loss": 0.0426,
+      "step": 13890
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8083282766953745e-05,
+      "loss": 0.0152,
+      "step": 13891
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.806952385835298e-05,
+      "loss": 0.0028,
+      "step": 13892
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8055769850582475e-05,
+      "loss": 0.0436,
+      "step": 13893
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8042020744153253e-05,
+      "loss": 0.0245,
+      "step": 13894
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8028276539575955e-05,
+      "loss": 0.0293,
+      "step": 13895
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8014537237361203e-05,
+      "loss": 0.025,
+      "step": 13896
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.8000802838019362e-05,
+      "loss": 0.0313,
+      "step": 13897
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7987073342060604e-05,
+      "loss": 0.0132,
+      "step": 13898
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7973348749994902e-05,
+      "loss": 0.0467,
+      "step": 13899
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7959629062332188e-05,
+      "loss": 0.0411,
+      "step": 13900
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7945914279581997e-05,
+      "loss": 0.0308,
+      "step": 13901
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7932204402253892e-05,
+      "loss": 0.0338,
+      "step": 13902
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7918499430857097e-05,
+      "loss": 0.0222,
+      "step": 13903
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7904799365900692e-05,
+      "loss": 0.0227,
+      "step": 13904
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.789110420789365e-05,
+      "loss": 0.0292,
+      "step": 13905
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7877413957344662e-05,
+      "loss": 0.005,
+      "step": 13906
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7863728614762357e-05,
+      "loss": 0.0216,
+      "step": 13907
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.785004818065498e-05,
+      "loss": 0.0201,
+      "step": 13908
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.783637265553079e-05,
+      "loss": 0.0169,
+      "step": 13909
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.782270203989783e-05,
+      "loss": 0.0076,
+      "step": 13910
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7809036334263843e-05,
+      "loss": 0.0509,
+      "step": 13911
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.779537553913656e-05,
+      "loss": 0.0273,
+      "step": 13912
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7781719655023336e-05,
+      "loss": 0.0289,
+      "step": 13913
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7768068682431534e-05,
+      "loss": 0.0155,
+      "step": 13914
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.7754422621868185e-05,
+      "loss": 0.0306,
+      "step": 13915
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.774078147384025e-05,
+      "loss": 0.0291,
+      "step": 13916
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.772714523885445e-05,
+      "loss": 0.0279,
+      "step": 13917
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.771351391741727e-05,
+      "loss": 0.0441,
+      "step": 13918
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.769988751003517e-05,
+      "loss": 0.0467,
+      "step": 13919
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7686266017214245e-05,
+      "loss": 0.0183,
+      "step": 13920
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7672649439460556e-05,
+      "loss": 0.0212,
+      "step": 13921
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.76590377772799e-05,
+      "loss": 0.0353,
+      "step": 13922
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7645431031177887e-05,
+      "loss": 0.0336,
+      "step": 13923
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.763182920165998e-05,
+      "loss": 0.0141,
+      "step": 13924
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.761823228923146e-05,
+      "loss": 0.0145,
+      "step": 13925
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7604640294397387e-05,
+      "loss": 0.0088,
+      "step": 13926
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.759105321766271e-05,
+      "loss": 0.0241,
+      "step": 13927
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.757747105953212e-05,
+      "loss": 0.0068,
+      "step": 13928
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7563893820510124e-05,
+      "loss": 0.0125,
+      "step": 13929
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.755032150110112e-05,
+      "loss": 0.0184,
+      "step": 13930
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7536754101809253e-05,
+      "loss": 0.0212,
+      "step": 13931
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7523191623138562e-05,
+      "loss": 0.0271,
+      "step": 13932
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.750963406559281e-05,
+      "loss": 0.0257,
+      "step": 13933
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7496081429675597e-05,
+      "loss": 0.0184,
+      "step": 13934
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.748253371589042e-05,
+      "loss": 0.0231,
+      "step": 13935
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.746899092474049e-05,
+      "loss": 0.0087,
+      "step": 13936
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7455453056728924e-05,
+      "loss": 0.0049,
+      "step": 13937
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7441920112358603e-05,
+      "loss": 0.0215,
+      "step": 13938
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.742839209213221e-05,
+      "loss": 0.0273,
+      "step": 13939
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.741486899655227e-05,
+      "loss": 0.019,
+      "step": 13940
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7401350826121173e-05,
+      "loss": 0.045,
+      "step": 13941
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.738783758134103e-05,
+      "loss": 0.0205,
+      "step": 13942
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7374329262713855e-05,
+      "loss": 0.0158,
+      "step": 13943
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.736082587074143e-05,
+      "loss": 0.0399,
+      "step": 13944
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.7347327405925337e-05,
+      "loss": 0.0137,
+      "step": 13945
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.733383386876706e-05,
+      "loss": 0.0308,
+      "step": 13946
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.732034525976778e-05,
+      "loss": 0.0203,
+      "step": 13947
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7306861579428648e-05,
+      "loss": 0.0272,
+      "step": 13948
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7293382828250446e-05,
+      "loss": 0.0241,
+      "step": 13949
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.727990900673389e-05,
+      "loss": 0.0579,
+      "step": 13950
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.726644011537956e-05,
+      "loss": 0.0134,
+      "step": 13951
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7252976154687708e-05,
+      "loss": 0.0344,
+      "step": 13952
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7239517125158587e-05,
+      "loss": 0.0303,
+      "step": 13953
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7226063027292008e-05,
+      "loss": 0.0328,
+      "step": 13954
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7212613861587878e-05,
+      "loss": 0.0202,
+      "step": 13955
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7199169628545706e-05,
+      "loss": 0.0314,
+      "step": 13956
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.718573032866498e-05,
+      "loss": 0.0531,
+      "step": 13957
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7172295962444904e-05,
+      "loss": 0.0359,
+      "step": 13958
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.715886653038449e-05,
+      "loss": 0.0115,
+      "step": 13959
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.714544203298266e-05,
+      "loss": 0.0259,
+      "step": 13960
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7132022470738027e-05,
+      "loss": 0.0669,
+      "step": 13961
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7118607844149164e-05,
+      "loss": 0.0154,
+      "step": 13962
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7105198153714333e-05,
+      "loss": 0.0234,
+      "step": 13963
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.709179339993169e-05,
+      "loss": 0.0331,
+      "step": 13964
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7078393583299156e-05,
+      "loss": 0.0119,
+      "step": 13965
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7064998704314524e-05,
+      "loss": 0.0221,
+      "step": 13966
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.705160876347534e-05,
+      "loss": 0.0391,
+      "step": 13967
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7038223761279046e-05,
+      "loss": 0.0185,
+      "step": 13968
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7024843698222836e-05,
+      "loss": 0.0429,
+      "step": 13969
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.7011468574803704e-05,
+      "loss": 0.0172,
+      "step": 13970
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.699809839151856e-05,
+      "loss": 0.0157,
+      "step": 13971
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6984733148864005e-05,
+      "loss": 0.0089,
+      "step": 13972
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.697137284733659e-05,
+      "loss": 0.0246,
+      "step": 13973
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6958017487432552e-05,
+      "loss": 0.0199,
+      "step": 13974
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.694466706964801e-05,
+      "loss": 0.0179,
+      "step": 13975
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.693132159447892e-05,
+      "loss": 0.0188,
+      "step": 13976
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6917981062420983e-05,
+      "loss": 0.0288,
+      "step": 13977
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6904645473969853e-05,
+      "loss": 0.0703,
+      "step": 13978
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.68913148296208e-05,
+      "loss": 0.0234,
+      "step": 13979
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.687798912986908e-05,
+      "loss": 0.0238,
+      "step": 13980
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6864668375209643e-05,
+      "loss": 0.0326,
+      "step": 13981
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6851352566137416e-05,
+      "loss": 0.0148,
+      "step": 13982
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6838041703146965e-05,
+      "loss": 0.0405,
+      "step": 13983
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6824735786732747e-05,
+      "loss": 0.0319,
+      "step": 13984
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6811434817389097e-05,
+      "loss": 0.033,
+      "step": 13985
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.679813879561004e-05,
+      "loss": 0.0304,
+      "step": 13986
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6784847721889543e-05,
+      "loss": 0.0296,
+      "step": 13987
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.67715615967213e-05,
+      "loss": 0.0504,
+      "step": 13988
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6758280420598862e-05,
+      "loss": 0.034,
+      "step": 13989
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6745004194015537e-05,
+      "loss": 0.0275,
+      "step": 13990
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6731732917464545e-05,
+      "loss": 0.0146,
+      "step": 13991
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6718466591438895e-05,
+      "loss": 0.0218,
+      "step": 13992
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6705205216431356e-05,
+      "loss": 0.0583,
+      "step": 13993
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.66919487929346e-05,
+      "loss": 0.0164,
+      "step": 13994
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.667869732144097e-05,
+      "loss": 0.017,
+      "step": 13995
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6665450802442803e-05,
+      "loss": 0.0486,
+      "step": 13996
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6652209236432108e-05,
+      "loss": 0.0155,
+      "step": 13997
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.663897262390082e-05,
+      "loss": 0.0254,
+      "step": 13998
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6625740965340633e-05,
+      "loss": 0.0235,
+      "step": 13999
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6612514261243022e-05,
+      "loss": 0.022,
+      "step": 14000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6599292512099373e-05,
+      "loss": 0.0426,
+      "step": 14001
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6586075718400795e-05,
+      "loss": 0.0415,
+      "step": 14002
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6572863880638292e-05,
+      "loss": 0.0304,
+      "step": 14003
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.655965699930263e-05,
+      "loss": 0.0104,
+      "step": 14004
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.654645507488439e-05,
+      "loss": 0.0224,
+      "step": 14005
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.653325810787397e-05,
+      "loss": 0.0113,
+      "step": 14006
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6520066098761658e-05,
+      "loss": 0.0287,
+      "step": 14007
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.650687904803743e-05,
+      "loss": 0.0117,
+      "step": 14008
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.649369695619121e-05,
+      "loss": 0.0321,
+      "step": 14009
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.648051982371264e-05,
+      "loss": 0.0302,
+      "step": 14010
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6467347651091197e-05,
+      "loss": 0.0099,
+      "step": 14011
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6454180438816234e-05,
+      "loss": 0.0316,
+      "step": 14012
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6441018187376813e-05,
+      "loss": 0.0293,
+      "step": 14013
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.642786089726193e-05,
+      "loss": 0.0193,
+      "step": 14014
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.641470856896032e-05,
+      "loss": 0.0442,
+      "step": 14015
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6401561202960535e-05,
+      "loss": 0.0381,
+      "step": 14016
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.638841879975101e-05,
+      "loss": 0.0217,
+      "step": 14017
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.637528135981987e-05,
+      "loss": 0.0402,
+      "step": 14018
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6362148883655228e-05,
+      "loss": 0.0334,
+      "step": 14019
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.634902137174483e-05,
+      "loss": 0.0136,
+      "step": 14020
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6335898824576364e-05,
+      "loss": 0.0488,
+      "step": 14021
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6322781242637277e-05,
+      "loss": 0.02,
+      "step": 14022
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6309668626414895e-05,
+      "loss": 0.0056,
+      "step": 14023
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6296560976396277e-05,
+      "loss": 0.0465,
+      "step": 14024
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6283458293068303e-05,
+      "loss": 0.0107,
+      "step": 14025
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6270360576917762e-05,
+      "loss": 0.013,
+      "step": 14026
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6257267828431135e-05,
+      "loss": 0.0359,
+      "step": 14027
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6244180048094834e-05,
+      "loss": 0.0224,
+      "step": 14028
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.623109723639502e-05,
+      "loss": 0.0241,
+      "step": 14029
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6218019393817667e-05,
+      "loss": 0.0372,
+      "step": 14030
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6204946520848556e-05,
+      "loss": 0.0249,
+      "step": 14031
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.619187861797333e-05,
+      "loss": 0.0099,
+      "step": 14032
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.617881568567744e-05,
+      "loss": 0.0536,
+      "step": 14033
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6165757724446137e-05,
+      "loss": 0.0334,
+      "step": 14034
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.615270473476446e-05,
+      "loss": 0.0309,
+      "step": 14035
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6139656717117267e-05,
+      "loss": 0.0247,
+      "step": 14036
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6126613671989318e-05,
+      "loss": 0.0843,
+      "step": 14037
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6113575599865063e-05,
+      "loss": 0.0304,
+      "step": 14038
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.610054250122888e-05,
+      "loss": 0.024,
+      "step": 14039
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.60875143765649e-05,
+      "loss": 0.0151,
+      "step": 14040
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.607449122635701e-05,
+      "loss": 0.0348,
+      "step": 14041
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6061473051089085e-05,
+      "loss": 0.0108,
+      "step": 14042
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6048459851244634e-05,
+      "loss": 0.0369,
+      "step": 14043
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.603545162730715e-05,
+      "loss": 0.0401,
+      "step": 14044
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6022448379759733e-05,
+      "loss": 0.0286,
+      "step": 14045
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6009450109085497e-05,
+      "loss": 0.011,
+      "step": 14046
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5996456815767238e-05,
+      "loss": 0.0238,
+      "step": 14047
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.598346850028767e-05,
+      "loss": 0.0061,
+      "step": 14048
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.597048516312926e-05,
+      "loss": 0.0193,
+      "step": 14049
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.595750680477425e-05,
+      "loss": 0.0337,
+      "step": 14050
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.594453342570481e-05,
+      "loss": 0.0251,
+      "step": 14051
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5931565026402816e-05,
+      "loss": 0.023,
+      "step": 14052
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5918601607350067e-05,
+      "loss": 0.0499,
+      "step": 14053
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.590564316902803e-05,
+      "loss": 0.024,
+      "step": 14054
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5892689711918156e-05,
+      "loss": 0.0263,
+      "step": 14055
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5879741236501603e-05,
+      "loss": 0.0237,
+      "step": 14056
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5866797743259323e-05,
+      "loss": 0.0214,
+      "step": 14057
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5853859232672184e-05,
+      "loss": 0.0226,
+      "step": 14058
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.584092570522078e-05,
+      "loss": 0.0426,
+      "step": 14059
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5827997161385614e-05,
+      "loss": 0.0283,
+      "step": 14060
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.581507360164685e-05,
+      "loss": 0.0144,
+      "step": 14061
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5802155026484636e-05,
+      "loss": 0.0427,
+      "step": 14062
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.578924143637879e-05,
+      "loss": 0.0409,
+      "step": 14063
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5776332831809096e-05,
+      "loss": 0.0164,
+      "step": 14064
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.576342921325502e-05,
+      "loss": 0.0179,
+      "step": 14065
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5750530581195892e-05,
+      "loss": 0.0311,
+      "step": 14066
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.573763693611088e-05,
+      "loss": 0.0147,
+      "step": 14067
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5724748278478904e-05,
+      "loss": 0.0248,
+      "step": 14068
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.571186460877881e-05,
+      "loss": 0.0435,
+      "step": 14069
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5698985927489153e-05,
+      "loss": 0.0358,
+      "step": 14070
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.568611223508833e-05,
+      "loss": 0.0166,
+      "step": 14071
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.567324353205453e-05,
+      "loss": 0.0388,
+      "step": 14072
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5660379818865865e-05,
+      "loss": 0.0203,
+      "step": 14073
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5647521096000105e-05,
+      "loss": 0.0207,
+      "step": 14074
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5634667363934968e-05,
+      "loss": 0.0188,
+      "step": 14075
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5621818623147936e-05,
+      "loss": 0.0152,
+      "step": 14076
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.560897487411623e-05,
+      "loss": 0.0147,
+      "step": 14077
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5596136117317055e-05,
+      "loss": 0.0113,
+      "step": 14078
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5583302353227256e-05,
+      "loss": 0.035,
+      "step": 14079
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5570473582323624e-05,
+      "loss": 0.0299,
+      "step": 14080
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.555764980508269e-05,
+      "loss": 0.0141,
+      "step": 14081
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5544831021980792e-05,
+      "loss": 0.0102,
+      "step": 14082
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5532017233494166e-05,
+      "loss": 0.0206,
+      "step": 14083
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.551920844009873e-05,
+      "loss": 0.0424,
+      "step": 14084
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.550640464227042e-05,
+      "loss": 0.0283,
+      "step": 14085
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.549360584048471e-05,
+      "loss": 0.023,
+      "step": 14086
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5480812035217128e-05,
+      "loss": 0.0067,
+      "step": 14087
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5468023226942882e-05,
+      "loss": 0.0213,
+      "step": 14088
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5455239416137093e-05,
+      "loss": 0.0297,
+      "step": 14089
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.54424606032746e-05,
+      "loss": 0.0228,
+      "step": 14090
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5429686788830082e-05,
+      "loss": 0.0246,
+      "step": 14091
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5416917973278114e-05,
+      "loss": 0.0181,
+      "step": 14092
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.540415415709295e-05,
+      "loss": 0.0262,
+      "step": 14093
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5391395340748798e-05,
+      "loss": 0.0181,
+      "step": 14094
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5378641524719554e-05,
+      "loss": 0.0073,
+      "step": 14095
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5365892709478977e-05,
+      "loss": 0.0328,
+      "step": 14096
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5353148895500723e-05,
+      "loss": 0.0103,
+      "step": 14097
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.534041008325809e-05,
+      "loss": 0.0233,
+      "step": 14098
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5327676273224375e-05,
+      "loss": 0.039,
+      "step": 14099
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5314947465872546e-05,
+      "loss": 0.0346,
+      "step": 14100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.530222366167547e-05,
+      "loss": 0.0208,
+      "step": 14101
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.528950486110576e-05,
+      "loss": 0.011,
+      "step": 14102
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.527679106463595e-05,
+      "loss": 0.0224,
+      "step": 14103
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5264082272738232e-05,
+      "loss": 0.0314,
+      "step": 14104
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5251378485884791e-05,
+      "loss": 0.015,
+      "step": 14105
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5238679704547485e-05,
+      "loss": 0.0355,
+      "step": 14106
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5225985929198003e-05,
+      "loss": 0.0318,
+      "step": 14107
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5213297160307969e-05,
+      "loss": 0.0283,
+      "step": 14108
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5200613398348655e-05,
+      "loss": 0.0155,
+      "step": 14109
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.51879346437913e-05,
+      "loss": 0.0285,
+      "step": 14110
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5175260897106795e-05,
+      "loss": 0.048,
+      "step": 14111
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5162592158766013e-05,
+      "loss": 0.0259,
+      "step": 14112
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5149928429239477e-05,
+      "loss": 0.0261,
+      "step": 14113
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5137269708997695e-05,
+      "loss": 0.0193,
+      "step": 14114
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5124615998510841e-05,
+      "loss": 0.038,
+      "step": 14115
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5111967298249007e-05,
+      "loss": 0.0295,
+      "step": 14116
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5099323608682029e-05,
+      "loss": 0.0382,
+      "step": 14117
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5086684930279552e-05,
+      "loss": 0.0145,
+      "step": 14118
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5074051263511133e-05,
+      "loss": 0.0155,
+      "step": 14119
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5061422608846013e-05,
+      "loss": 0.0171,
+      "step": 14120
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5048798966753367e-05,
+      "loss": 0.01,
+      "step": 14121
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5036180337702102e-05,
+      "loss": 0.019,
+      "step": 14122
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.502356672216093e-05,
+      "loss": 0.026,
+      "step": 14123
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5010958120598472e-05,
+      "loss": 0.0092,
+      "step": 14124
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4998354533483037e-05,
+      "loss": 0.0212,
+      "step": 14125
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4985755961282886e-05,
+      "loss": 0.0143,
+      "step": 14126
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4973162404465927e-05,
+      "loss": 0.0068,
+      "step": 14127
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4960573863500052e-05,
+      "loss": 0.0277,
+      "step": 14128
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.494799033885284e-05,
+      "loss": 0.015,
+      "step": 14129
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4935411830991762e-05,
+      "loss": 0.0568,
+      "step": 14130
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4922838340384064e-05,
+      "loss": 0.0555,
+      "step": 14131
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4910269867496788e-05,
+      "loss": 0.0185,
+      "step": 14132
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4897706412796861e-05,
+      "loss": 0.0371,
+      "step": 14133
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4885147976750943e-05,
+      "loss": 0.0126,
+      "step": 14134
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4872594559825563e-05,
+      "loss": 0.0518,
+      "step": 14135
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.486004616248706e-05,
+      "loss": 0.0304,
+      "step": 14136
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4847502785201499e-05,
+      "loss": 0.0561,
+      "step": 14137
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.483496442843492e-05,
+      "loss": 0.0111,
+      "step": 14138
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4822431092653003e-05,
+      "loss": 0.0205,
+      "step": 14139
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4809902778321408e-05,
+      "loss": 0.0302,
+      "step": 14140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.479737948590548e-05,
+      "loss": 0.0272,
+      "step": 14141
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.478486121587043e-05,
+      "loss": 0.0464,
+      "step": 14142
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4772347968681237e-05,
+      "loss": 0.0303,
+      "step": 14143
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4759839744802776e-05,
+      "loss": 0.0104,
+      "step": 14144
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.4747336544699679e-05,
+      "loss": 0.03,
+      "step": 14145
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4734838368836422e-05,
+      "loss": 0.0403,
+      "step": 14146
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.472234521767725e-05,
+      "loss": 0.0404,
+      "step": 14147
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4709857091686228e-05,
+      "loss": 0.0354,
+      "step": 14148
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4697373991327316e-05,
+      "loss": 0.0292,
+      "step": 14149
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.468489591706416e-05,
+      "loss": 0.0133,
+      "step": 14150
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4672422869360374e-05,
+      "loss": 0.029,
+      "step": 14151
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4659954848679167e-05,
+      "loss": 0.0308,
+      "step": 14152
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.464749185548379e-05,
+      "loss": 0.0373,
+      "step": 14153
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4635033890237135e-05,
+      "loss": 0.0253,
+      "step": 14154
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4622580953402052e-05,
+      "loss": 0.0311,
+      "step": 14155
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4610133045441103e-05,
+      "loss": 0.0176,
+      "step": 14156
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4597690166816651e-05,
+      "loss": 0.0195,
+      "step": 14157
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4585252317990959e-05,
+      "loss": 0.0362,
+      "step": 14158
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4572819499426026e-05,
+      "loss": 0.0204,
+      "step": 14159
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4560391711583747e-05,
+      "loss": 0.0068,
+      "step": 14160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4547968954925737e-05,
+      "loss": 0.0153,
+      "step": 14161
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4535551229913429e-05,
+      "loss": 0.0163,
+      "step": 14162
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.45231385370082e-05,
+      "loss": 0.0142,
+      "step": 14163
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.451073087667105e-05,
+      "loss": 0.0252,
+      "step": 14164
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4498328249362945e-05,
+      "loss": 0.0246,
+      "step": 14165
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4485930655544614e-05,
+      "loss": 0.0224,
+      "step": 14166
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4473538095676556e-05,
+      "loss": 0.0301,
+      "step": 14167
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4461150570219088e-05,
+      "loss": 0.0278,
+      "step": 14168
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4448768079632456e-05,
+      "loss": 0.0288,
+      "step": 14169
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.443639062437656e-05,
+      "loss": 0.0331,
+      "step": 14170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4424018204911248e-05,
+      "loss": 0.0263,
+      "step": 14171
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4411650821696069e-05,
+      "loss": 0.0215,
+      "step": 14172
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.439928847519044e-05,
+      "loss": 0.0123,
+      "step": 14173
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4386931165853628e-05,
+      "loss": 0.0332,
+      "step": 14174
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.437457889414461e-05,
+      "loss": 0.0126,
+      "step": 14175
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.436223166052231e-05,
+      "loss": 0.0246,
+      "step": 14176
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4349889465445307e-05,
+      "loss": 0.0289,
+      "step": 14177
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.4337552309372152e-05,
+      "loss": 0.0218,
+      "step": 14178
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4325220192761061e-05,
+      "loss": 0.0295,
+      "step": 14179
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4312893116070184e-05,
+      "loss": 0.0138,
+      "step": 14180
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4300571079757456e-05,
+      "loss": 0.0214,
+      "step": 14181
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4288254084280575e-05,
+      "loss": 0.022,
+      "step": 14182
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4275942130097096e-05,
+      "loss": 0.0309,
+      "step": 14183
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4263635217664316e-05,
+      "loss": 0.0268,
+      "step": 14184
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.425133334743947e-05,
+      "loss": 0.0359,
+      "step": 14185
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4239036519879492e-05,
+      "loss": 0.0378,
+      "step": 14186
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.422674473544122e-05,
+      "loss": 0.0184,
+      "step": 14187
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.421445799458122e-05,
+      "loss": 0.023,
+      "step": 14188
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4202176297755891e-05,
+      "loss": 0.0246,
+      "step": 14189
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4189899645421521e-05,
+      "loss": 0.0494,
+      "step": 14190
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4177628038034095e-05,
+      "loss": 0.0198,
+      "step": 14191
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4165361476049547e-05,
+      "loss": 0.0144,
+      "step": 14192
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.415309995992343e-05,
+      "loss": 0.0404,
+      "step": 14193
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4140843490111315e-05,
+      "loss": 0.0195,
+      "step": 14194
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4128592067068416e-05,
+      "loss": 0.0263,
+      "step": 14195
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4116345691249924e-05,
+      "loss": 0.0069,
+      "step": 14196
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4104104363110707e-05,
+      "loss": 0.0069,
+      "step": 14197
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4091868083105484e-05,
+      "loss": 0.0302,
+      "step": 14198
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4079636851688842e-05,
+      "loss": 0.0198,
+      "step": 14199
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4067410669315066e-05,
+      "loss": 0.034,
+      "step": 14200
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4055189536438394e-05,
+      "loss": 0.015,
+      "step": 14201
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4042973453512796e-05,
+      "loss": 0.0264,
+      "step": 14202
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4030762420991993e-05,
+      "loss": 0.021,
+      "step": 14203
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.4018556439329687e-05,
+      "loss": 0.0272,
+      "step": 14204
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.40063555089792e-05,
+      "loss": 0.0351,
+      "step": 14205
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3994159630393853e-05,
+      "loss": 0.0189,
+      "step": 14206
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3981968804026633e-05,
+      "loss": 0.0322,
+      "step": 14207
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3969783030330395e-05,
+      "loss": 0.0381,
+      "step": 14208
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3957602309757793e-05,
+      "loss": 0.0232,
+      "step": 14209
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3945426642761348e-05,
+      "loss": 0.006,
+      "step": 14210
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3933256029793299e-05,
+      "loss": 0.0187,
+      "step": 14211
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3921090471305802e-05,
+      "loss": 0.0419,
+      "step": 14212
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3908929967750759e-05,
+      "loss": 0.0211,
+      "step": 14213
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3896774519579845e-05,
+      "loss": 0.02,
+      "step": 14214
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3884624127244664e-05,
+      "loss": 0.0233,
+      "step": 14215
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3872478791196523e-05,
+      "loss": 0.0095,
+      "step": 14216
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.386033851188666e-05,
+      "loss": 0.0115,
+      "step": 14217
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.384820328976593e-05,
+      "loss": 0.0352,
+      "step": 14218
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3836073125285225e-05,
+      "loss": 0.0224,
+      "step": 14219
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3823948018895065e-05,
+      "loss": 0.0159,
+      "step": 14220
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3811827971045907e-05,
+      "loss": 0.0144,
+      "step": 14221
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3799712982188027e-05,
+      "loss": 0.0279,
+      "step": 14222
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3787603052771362e-05,
+      "loss": 0.0525,
+      "step": 14223
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.377549818324582e-05,
+      "loss": 0.0205,
+      "step": 14224
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3763398374061024e-05,
+      "loss": 0.0081,
+      "step": 14225
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3751303625666482e-05,
+      "loss": 0.0115,
+      "step": 14226
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3739213938511483e-05,
+      "loss": 0.0292,
+      "step": 14227
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3727129313045072e-05,
+      "loss": 0.0316,
+      "step": 14228
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3715049749716201e-05,
+      "loss": 0.0244,
+      "step": 14229
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3702975248973563e-05,
+      "loss": 0.0378,
+      "step": 14230
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3690905811265734e-05,
+      "loss": 0.0313,
+      "step": 14231
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3678841437041038e-05,
+      "loss": 0.0288,
+      "step": 14232
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3666782126747617e-05,
+      "loss": 0.0025,
+      "step": 14233
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3654727880833426e-05,
+      "loss": 0.0059,
+      "step": 14234
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3642678699746295e-05,
+      "loss": 0.0319,
+      "step": 14235
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3630634583933747e-05,
+      "loss": 0.0349,
+      "step": 14236
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3618595533843257e-05,
+      "loss": 0.0335,
+      "step": 14237
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3606561549922002e-05,
+      "loss": 0.0093,
+      "step": 14238
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.359453263261699e-05,
+      "loss": 0.0272,
+      "step": 14239
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3582508782375112e-05,
+      "loss": 0.0114,
+      "step": 14240
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3570489999642965e-05,
+      "loss": 0.0254,
+      "step": 14241
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.355847628486707e-05,
+      "loss": 0.0254,
+      "step": 14242
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3546467638493674e-05,
+      "loss": 0.0126,
+      "step": 14243
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3534464060968819e-05,
+      "loss": 0.0233,
+      "step": 14244
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3522465552738482e-05,
+      "loss": 0.0353,
+      "step": 14245
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3510472114248289e-05,
+      "loss": 0.0304,
+      "step": 14246
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.349848374594385e-05,
+      "loss": 0.0238,
+      "step": 14247
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3486500448270443e-05,
+      "loss": 0.0451,
+      "step": 14248
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3474522221673211e-05,
+      "loss": 0.0074,
+      "step": 14249
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3462549066597118e-05,
+      "loss": 0.0314,
+      "step": 14250
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3450580983486936e-05,
+      "loss": 0.0356,
+      "step": 14251
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.343861797278723e-05,
+      "loss": 0.0497,
+      "step": 14252
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3426660034942427e-05,
+      "loss": 0.0312,
+      "step": 14253
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3414707170396721e-05,
+      "loss": 0.0523,
+      "step": 14254
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3402759379594057e-05,
+      "loss": 0.042,
+      "step": 14255
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3390816662978366e-05,
+      "loss": 0.0196,
+      "step": 14256
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3378879020993189e-05,
+      "loss": 0.0364,
+      "step": 14257
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3366946454082072e-05,
+      "loss": 0.0321,
+      "step": 14258
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.335501896268818e-05,
+      "loss": 0.0057,
+      "step": 14259
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.334309654725464e-05,
+      "loss": 0.0222,
+      "step": 14260
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3331179208224296e-05,
+      "loss": 0.0201,
+      "step": 14261
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3319266946039864e-05,
+      "loss": 0.0127,
+      "step": 14262
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.330735976114392e-05,
+      "loss": 0.0199,
+      "step": 14263
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3295457653978647e-05,
+      "loss": 0.0354,
+      "step": 14264
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3283560624986272e-05,
+      "loss": 0.0506,
+      "step": 14265
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.327166867460866e-05,
+      "loss": 0.0045,
+      "step": 14266
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3259781803287656e-05,
+      "loss": 0.044,
+      "step": 14267
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3247900011464758e-05,
+      "loss": 0.0191,
+      "step": 14268
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3236023299581311e-05,
+      "loss": 0.0115,
+      "step": 14269
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3224151668078581e-05,
+      "loss": 0.0195,
+      "step": 14270
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3212285117397514e-05,
+      "loss": 0.032,
+      "step": 14271
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3200423647978925e-05,
+      "loss": 0.0368,
+      "step": 14272
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.318856726026346e-05,
+      "loss": 0.0196,
+      "step": 14273
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3176715954691519e-05,
+      "loss": 0.0412,
+      "step": 14274
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3164869731703315e-05,
+      "loss": 0.0083,
+      "step": 14275
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3153028591738963e-05,
+      "loss": 0.012,
+      "step": 14276
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3141192535238276e-05,
+      "loss": 0.0172,
+      "step": 14277
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3129361562640988e-05,
+      "loss": 0.0268,
+      "step": 14278
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3117535674386531e-05,
+      "loss": 0.0236,
+      "step": 14279
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3105714870914203e-05,
+      "loss": 0.0124,
+      "step": 14280
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3093899152663151e-05,
+      "loss": 0.0067,
+      "step": 14281
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3082088520072243e-05,
+      "loss": 0.0339,
+      "step": 14282
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3070282973580293e-05,
+      "loss": 0.0196,
+      "step": 14283
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3058482513625734e-05,
+      "loss": 0.0548,
+      "step": 14284
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3046687140646966e-05,
+      "loss": 0.0113,
+      "step": 14285
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3034896855082188e-05,
+      "loss": 0.0313,
+      "step": 14286
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3023111657369334e-05,
+      "loss": 0.04,
+      "step": 14287
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3011331547946235e-05,
+      "loss": 0.0337,
+      "step": 14288
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.299955652725041e-05,
+      "loss": 0.0257,
+      "step": 14289
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2987786595719323e-05,
+      "loss": 0.0174,
+      "step": 14290
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2976021753790161e-05,
+      "loss": 0.0276,
+      "step": 14291
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2964262001900005e-05,
+      "loss": 0.0215,
+      "step": 14292
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2952507340485657e-05,
+      "loss": 0.026,
+      "step": 14293
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2940757769983751e-05,
+      "loss": 0.0254,
+      "step": 14294
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2929013290830787e-05,
+      "loss": 0.0129,
+      "step": 14295
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2917273903462999e-05,
+      "loss": 0.0368,
+      "step": 14296
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2905539608316524e-05,
+      "loss": 0.0364,
+      "step": 14297
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.289381040582721e-05,
+      "loss": 0.0115,
+      "step": 14298
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2882086296430794e-05,
+      "loss": 0.0163,
+      "step": 14299
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.287036728056276e-05,
+      "loss": 0.0348,
+      "step": 14300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.285865335865846e-05,
+      "loss": 0.0107,
+      "step": 14301
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2846944531153014e-05,
+      "loss": 0.0221,
+      "step": 14302
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2835240798481373e-05,
+      "loss": 0.0377,
+      "step": 14303
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2823542161078353e-05,
+      "loss": 0.0267,
+      "step": 14304
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2811848619378412e-05,
+      "loss": 0.0085,
+      "step": 14305
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2800160173816032e-05,
+      "loss": 0.012,
+      "step": 14306
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.278847682482535e-05,
+      "loss": 0.011,
+      "step": 14307
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2776798572840402e-05,
+      "loss": 0.029,
+      "step": 14308
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2765125418294975e-05,
+      "loss": 0.0064,
+      "step": 14309
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2753457361622688e-05,
+      "loss": 0.0248,
+      "step": 14310
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.274179440325701e-05,
+      "loss": 0.032,
+      "step": 14311
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2730136543631131e-05,
+      "loss": 0.0409,
+      "step": 14312
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2718483783178168e-05,
+      "loss": 0.0273,
+      "step": 14313
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2706836122330977e-05,
+      "loss": 0.0298,
+      "step": 14314
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2695193561522194e-05,
+      "loss": 0.0165,
+      "step": 14315
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2683556101184306e-05,
+      "loss": 0.0154,
+      "step": 14316
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2671923741749651e-05,
+      "loss": 0.041,
+      "step": 14317
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2660296483650301e-05,
+      "loss": 0.0415,
+      "step": 14318
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2648674327318209e-05,
+      "loss": 0.0122,
+      "step": 14319
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2637057273185097e-05,
+      "loss": 0.0301,
+      "step": 14320
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2625445321682453e-05,
+      "loss": 0.021,
+      "step": 14321
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2613838473241716e-05,
+      "loss": 0.0247,
+      "step": 14322
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2602236728293957e-05,
+      "loss": 0.029,
+      "step": 14323
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2590640087270232e-05,
+      "loss": 0.0284,
+      "step": 14324
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2579048550601245e-05,
+      "loss": 0.0131,
+      "step": 14325
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2567462118717602e-05,
+      "loss": 0.0099,
+      "step": 14326
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.255588079204976e-05,
+      "loss": 0.0371,
+      "step": 14327
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2544304571027875e-05,
+      "loss": 0.0155,
+      "step": 14328
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2532733456082034e-05,
+      "loss": 0.0235,
+      "step": 14329
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2521167447641977e-05,
+      "loss": 0.0481,
+      "step": 14330
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2509606546137413e-05,
+      "loss": 0.0038,
+      "step": 14331
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2498050751997763e-05,
+      "loss": 0.0216,
+      "step": 14332
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2486500065652333e-05,
+      "loss": 0.0109,
+      "step": 14333
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2474954487530164e-05,
+      "loss": 0.0193,
+      "step": 14334
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.246341401806013e-05,
+      "loss": 0.0118,
+      "step": 14335
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2451878657670954e-05,
+      "loss": 0.0166,
+      "step": 14336
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2440348406791112e-05,
+      "loss": 0.0279,
+      "step": 14337
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2428823265848958e-05,
+      "loss": 0.0303,
+      "step": 14338
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2417303235272602e-05,
+      "loss": 0.018,
+      "step": 14339
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2405788315489984e-05,
+      "loss": 0.0318,
+      "step": 14340
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2394278506928796e-05,
+      "loss": 0.0302,
+      "step": 14341
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2382773810016661e-05,
+      "loss": 0.0258,
+      "step": 14342
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.2371274225180889e-05,
+      "loss": 0.0441,
+      "step": 14343
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.235977975284872e-05,
+      "loss": 0.0222,
+      "step": 14344
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2348290393447113e-05,
+      "loss": 0.0131,
+      "step": 14345
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2336806147402828e-05,
+      "loss": 0.0253,
+      "step": 14346
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2325327015142522e-05,
+      "loss": 0.0459,
+      "step": 14347
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2313852997092572e-05,
+      "loss": 0.0051,
+      "step": 14348
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2302384093679236e-05,
+      "loss": 0.025,
+      "step": 14349
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2290920305328556e-05,
+      "loss": 0.03,
+      "step": 14350
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2279461632466308e-05,
+      "loss": 0.0281,
+      "step": 14351
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2268008075518237e-05,
+      "loss": 0.0292,
+      "step": 14352
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2256559634909736e-05,
+      "loss": 0.0229,
+      "step": 14353
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2245116311066178e-05,
+      "loss": 0.0264,
+      "step": 14354
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2233678104412526e-05,
+      "loss": 0.0238,
+      "step": 14355
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2222245015373771e-05,
+      "loss": 0.0071,
+      "step": 14356
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.221081704437456e-05,
+      "loss": 0.0124,
+      "step": 14357
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2199394191839451e-05,
+      "loss": 0.0282,
+      "step": 14358
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2187976458192739e-05,
+      "loss": 0.0517,
+      "step": 14359
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2176563843858566e-05,
+      "loss": 0.0085,
+      "step": 14360
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2165156349260896e-05,
+      "loss": 0.0305,
+      "step": 14361
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2153753974823455e-05,
+      "loss": 0.0444,
+      "step": 14362
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2142356720969837e-05,
+      "loss": 0.0284,
+      "step": 14363
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2130964588123405e-05,
+      "loss": 0.0232,
+      "step": 14364
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2119577576707334e-05,
+      "loss": 0.0256,
+      "step": 14365
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2108195687144623e-05,
+      "loss": 0.0159,
+      "step": 14366
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2096818919858065e-05,
+      "loss": 0.0167,
+      "step": 14367
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2085447275270305e-05,
+      "loss": 0.0106,
+      "step": 14368
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2074080753803723e-05,
+      "loss": 0.0269,
+      "step": 14369
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.206271935588063e-05,
+      "loss": 0.0225,
+      "step": 14370
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2051363081922972e-05,
+      "loss": 0.0193,
+      "step": 14371
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2040011932352662e-05,
+      "loss": 0.0406,
+      "step": 14372
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2028665907591329e-05,
+      "loss": 0.0336,
+      "step": 14373
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2017325008060468e-05,
+      "loss": 0.0078,
+      "step": 14374
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2005989234181362e-05,
+      "loss": 0.0166,
+      "step": 14375
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.1994658586375072e-05,
+      "loss": 0.0432,
+      "step": 14376
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1983333065062528e-05,
+      "loss": 0.0141,
+      "step": 14377
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1972012670664427e-05,
+      "loss": 0.0217,
+      "step": 14378
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.19606974036013e-05,
+      "loss": 0.0234,
+      "step": 14379
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1949387264293476e-05,
+      "loss": 0.0184,
+      "step": 14380
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.193808225316107e-05,
+      "loss": 0.0226,
+      "step": 14381
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1926782370624043e-05,
+      "loss": 0.0282,
+      "step": 14382
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1915487617102165e-05,
+      "loss": 0.0251,
+      "step": 14383
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1904197993014963e-05,
+      "loss": 0.0141,
+      "step": 14384
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1892913498781887e-05,
+      "loss": 0.0198,
+      "step": 14385
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1881634134822066e-05,
+      "loss": 0.0218,
+      "step": 14386
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1870359901554483e-05,
+      "loss": 0.0063,
+      "step": 14387
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1859090799397985e-05,
+      "loss": 0.0192,
+      "step": 14388
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1847826828771157e-05,
+      "loss": 0.0341,
+      "step": 14389
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1836567990092461e-05,
+      "loss": 0.0302,
+      "step": 14390
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1825314283780097e-05,
+      "loss": 0.0217,
+      "step": 14391
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.181406571025208e-05,
+      "loss": 0.0385,
+      "step": 14392
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1802822269926327e-05,
+      "loss": 0.0064,
+      "step": 14393
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1791583963220452e-05,
+      "loss": 0.0122,
+      "step": 14394
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1780350790551974e-05,
+      "loss": 0.0198,
+      "step": 14395
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1769122752338107e-05,
+      "loss": 0.0266,
+      "step": 14396
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1757899848995984e-05,
+      "loss": 0.0161,
+      "step": 14397
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1746682080942455e-05,
+      "loss": 0.0292,
+      "step": 14398
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1735469448594304e-05,
+      "loss": 0.0447,
+      "step": 14399
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1724261952367997e-05,
+      "loss": 0.0355,
+      "step": 14400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1713059592679852e-05,
+      "loss": 0.03,
+      "step": 14401
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1701862369946036e-05,
+      "loss": 0.0155,
+      "step": 14402
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1690670284582448e-05,
+      "loss": 0.0351,
+      "step": 14403
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1679483337004908e-05,
+      "loss": 0.0345,
+      "step": 14404
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1668301527628931e-05,
+      "loss": 0.0215,
+      "step": 14405
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1657124856869905e-05,
+      "loss": 0.021,
+      "step": 14406
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1645953325142977e-05,
+      "loss": 0.0347,
+      "step": 14407
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1634786932863166e-05,
+      "loss": 0.0224,
+      "step": 14408
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.1623625680445292e-05,
+      "loss": 0.0079,
+      "step": 14409
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1612469568303923e-05,
+      "loss": 0.002,
+      "step": 14410
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1601318596853509e-05,
+      "loss": 0.0185,
+      "step": 14411
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1590172766508222e-05,
+      "loss": 0.0112,
+      "step": 14412
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1579032077682177e-05,
+      "loss": 0.0086,
+      "step": 14413
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1567896530789128e-05,
+      "loss": 0.0152,
+      "step": 14414
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1556766126242812e-05,
+      "loss": 0.0292,
+      "step": 14415
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1545640864456645e-05,
+      "loss": 0.0312,
+      "step": 14416
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1534520745843883e-05,
+      "loss": 0.0044,
+      "step": 14417
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1523405770817662e-05,
+      "loss": 0.0214,
+      "step": 14418
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.15122959397908e-05,
+      "loss": 0.0336,
+      "step": 14419
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1501191253176084e-05,
+      "loss": 0.0133,
+      "step": 14420
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1490091711385919e-05,
+      "loss": 0.0306,
+      "step": 14421
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1478997314832705e-05,
+      "loss": 0.0133,
+      "step": 14422
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.14679080639285e-05,
+      "loss": 0.0266,
+      "step": 14423
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1456823959085287e-05,
+      "loss": 0.0302,
+      "step": 14424
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.144574500071479e-05,
+      "loss": 0.0291,
+      "step": 14425
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1434671189228544e-05,
+      "loss": 0.0313,
+      "step": 14426
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1423602525037939e-05,
+      "loss": 0.0229,
+      "step": 14427
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.141253900855411e-05,
+      "loss": 0.0629,
+      "step": 14428
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1401480640188082e-05,
+      "loss": 0.0181,
+      "step": 14429
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1390427420350574e-05,
+      "loss": 0.0132,
+      "step": 14430
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1379379349452256e-05,
+      "loss": 0.0123,
+      "step": 14431
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1368336427903485e-05,
+      "loss": 0.0403,
+      "step": 14432
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1357298656114467e-05,
+      "loss": 0.0272,
+      "step": 14433
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1346266034495255e-05,
+      "loss": 0.0242,
+      "step": 14434
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1335238563455624e-05,
+      "loss": 0.0182,
+      "step": 14435
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.132421624340531e-05,
+      "loss": 0.033,
+      "step": 14436
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1313199074753654e-05,
+      "loss": 0.0303,
+      "step": 14437
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1302187057909978e-05,
+      "loss": 0.0077,
+      "step": 14438
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1291180193283306e-05,
+      "loss": 0.0313,
+      "step": 14439
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1280178481282542e-05,
+      "loss": 0.0033,
+      "step": 14440
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.1269181922316362e-05,
+      "loss": 0.0383,
+      "step": 14441
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.125819051679322e-05,
+      "loss": 0.0548,
+      "step": 14442
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1247204265121474e-05,
+      "loss": 0.0199,
+      "step": 14443
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1236223167709163e-05,
+      "loss": 0.0346,
+      "step": 14444
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1225247224964279e-05,
+      "loss": 0.02,
+      "step": 14445
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1214276437294511e-05,
+      "loss": 0.0046,
+      "step": 14446
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1203310805107368e-05,
+      "loss": 0.045,
+      "step": 14447
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1192350328810207e-05,
+      "loss": 0.0285,
+      "step": 14448
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1181395008810168e-05,
+      "loss": 0.0159,
+      "step": 14449
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1170444845514242e-05,
+      "loss": 0.043,
+      "step": 14450
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1159499839329189e-05,
+      "loss": 0.0332,
+      "step": 14451
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1148559990661566e-05,
+      "loss": 0.0214,
+      "step": 14452
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1137625299917729e-05,
+      "loss": 0.029,
+      "step": 14453
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.112669576750394e-05,
+      "loss": 0.0482,
+      "step": 14454
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1115771393826122e-05,
+      "loss": 0.0414,
+      "step": 14455
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1104852179290152e-05,
+      "loss": 0.0109,
+      "step": 14456
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.109393812430162e-05,
+      "loss": 0.0084,
+      "step": 14457
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1083029229265921e-05,
+      "loss": 0.0264,
+      "step": 14458
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1072125494588347e-05,
+      "loss": 0.0117,
+      "step": 14459
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1061226920673871e-05,
+      "loss": 0.0294,
+      "step": 14460
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1050333507927439e-05,
+      "loss": 0.0102,
+      "step": 14461
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1039445256753594e-05,
+      "loss": 0.0494,
+      "step": 14462
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1028562167556892e-05,
+      "loss": 0.052,
+      "step": 14463
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1017684240741549e-05,
+      "loss": 0.01,
+      "step": 14464
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.1006811476711702e-05,
+      "loss": 0.0178,
+      "step": 14465
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0995943875871199e-05,
+      "loss": 0.0354,
+      "step": 14466
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0985081438623749e-05,
+      "loss": 0.0266,
+      "step": 14467
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.097422416537288e-05,
+      "loss": 0.0363,
+      "step": 14468
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0963372056521869e-05,
+      "loss": 0.0304,
+      "step": 14469
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0952525112473892e-05,
+      "loss": 0.0133,
+      "step": 14470
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0941683333631845e-05,
+      "loss": 0.0615,
+      "step": 14471
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0930846720398474e-05,
+      "loss": 0.0214,
+      "step": 14472
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0920015273176336e-05,
+      "loss": 0.0457,
+      "step": 14473
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.0909188992367763e-05,
+      "loss": 0.012,
+      "step": 14474
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0898367878374964e-05,
+      "loss": 0.0156,
+      "step": 14475
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0887551931599886e-05,
+      "loss": 0.0278,
+      "step": 14476
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0876741152444307e-05,
+      "loss": 0.0082,
+      "step": 14477
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.086593554130979e-05,
+      "loss": 0.0237,
+      "step": 14478
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0855135098597794e-05,
+      "loss": 0.0462,
+      "step": 14479
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0844339824709452e-05,
+      "loss": 0.0063,
+      "step": 14480
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0833549720045859e-05,
+      "loss": 0.0163,
+      "step": 14481
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0822764785007776e-05,
+      "loss": 0.0218,
+      "step": 14482
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0811985019995834e-05,
+      "loss": 0.0284,
+      "step": 14483
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0801210425410495e-05,
+      "loss": 0.0234,
+      "step": 14484
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.079044100165199e-05,
+      "loss": 0.0299,
+      "step": 14485
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0779676749120398e-05,
+      "loss": 0.0088,
+      "step": 14486
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0768917668215516e-05,
+      "loss": 0.0208,
+      "step": 14487
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0758163759337073e-05,
+      "loss": 0.0188,
+      "step": 14488
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0747415022884503e-05,
+      "loss": 0.0122,
+      "step": 14489
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.07366714592571e-05,
+      "loss": 0.008,
+      "step": 14490
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0725933068853997e-05,
+      "loss": 0.0109,
+      "step": 14491
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0715199852074058e-05,
+      "loss": 0.0172,
+      "step": 14492
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0704471809316011e-05,
+      "loss": 0.0215,
+      "step": 14493
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0693748940978309e-05,
+      "loss": 0.016,
+      "step": 14494
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0683031247459362e-05,
+      "loss": 0.0319,
+      "step": 14495
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0672318729157237e-05,
+      "loss": 0.0179,
+      "step": 14496
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0661611386469915e-05,
+      "loss": 0.0222,
+      "step": 14497
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0650909219795112e-05,
+      "loss": 0.0352,
+      "step": 14498
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0640212229530392e-05,
+      "loss": 0.015,
+      "step": 14499
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0629520416073123e-05,
+      "loss": 0.0446,
+      "step": 14500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0618833779820452e-05,
+      "loss": 0.0318,
+      "step": 14501
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0608152321169428e-05,
+      "loss": 0.0237,
+      "step": 14502
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0597476040516734e-05,
+      "loss": 0.0027,
+      "step": 14503
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0586804938259035e-05,
+      "loss": 0.032,
+      "step": 14504
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0576139014792667e-05,
+      "loss": 0.0204,
+      "step": 14505
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0565478270513928e-05,
+      "loss": 0.0065,
+      "step": 14506
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.0554822705818766e-05,
+      "loss": 0.0102,
+      "step": 14507
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0544172321103e-05,
+      "loss": 0.0227,
+      "step": 14508
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0533527116762296e-05,
+      "loss": 0.0208,
+      "step": 14509
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0522887093192073e-05,
+      "loss": 0.0178,
+      "step": 14510
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0512252250787595e-05,
+      "loss": 0.035,
+      "step": 14511
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0501622589943897e-05,
+      "loss": 0.0256,
+      "step": 14512
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0490998111055816e-05,
+      "loss": 0.0276,
+      "step": 14513
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0480378814518082e-05,
+      "loss": 0.0169,
+      "step": 14514
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0469764700725115e-05,
+      "loss": 0.016,
+      "step": 14515
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0459155770071247e-05,
+      "loss": 0.0268,
+      "step": 14516
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0448552022950534e-05,
+      "loss": 0.0315,
+      "step": 14517
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0437953459756875e-05,
+      "loss": 0.0104,
+      "step": 14518
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0427360080883972e-05,
+      "loss": 0.022,
+      "step": 14519
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.041677188672536e-05,
+      "loss": 0.0285,
+      "step": 14520
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0406188877674343e-05,
+      "loss": 0.0091,
+      "step": 14521
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.039561105412407e-05,
+      "loss": 0.0309,
+      "step": 14522
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0385038416467445e-05,
+      "loss": 0.0119,
+      "step": 14523
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.037447096509722e-05,
+      "loss": 0.0288,
+      "step": 14524
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0363908700405965e-05,
+      "loss": 0.0237,
+      "step": 14525
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0353351622786e-05,
+      "loss": 0.017,
+      "step": 14526
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0342799732629558e-05,
+      "loss": 0.0209,
+      "step": 14527
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.033225303032853e-05,
+      "loss": 0.0616,
+      "step": 14528
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0321711516274733e-05,
+      "loss": 0.022,
+      "step": 14529
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0311175190859739e-05,
+      "loss": 0.0575,
+      "step": 14530
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0300644054474948e-05,
+      "loss": 0.0171,
+      "step": 14531
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.029011810751162e-05,
+      "loss": 0.015,
+      "step": 14532
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0279597350360653e-05,
+      "loss": 0.0368,
+      "step": 14533
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0269081783412952e-05,
+      "loss": 0.0196,
+      "step": 14534
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0258571407059073e-05,
+      "loss": 0.0427,
+      "step": 14535
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0248066221689488e-05,
+      "loss": 0.0577,
+      "step": 14536
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0237566227694449e-05,
+      "loss": 0.0186,
+      "step": 14537
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0227071425463929e-05,
+      "loss": 0.0178,
+      "step": 14538
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0216581815387864e-05,
+      "loss": 0.0217,
+      "step": 14539
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.0206097397855845e-05,
+      "loss": 0.0585,
+      "step": 14540
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0195618173257391e-05,
+      "loss": 0.024,
+      "step": 14541
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0185144141981743e-05,
+      "loss": 0.0404,
+      "step": 14542
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0174675304417989e-05,
+      "loss": 0.0185,
+      "step": 14543
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0164211660955001e-05,
+      "loss": 0.015,
+      "step": 14544
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0153753211981485e-05,
+      "loss": 0.0262,
+      "step": 14545
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0143299957885947e-05,
+      "loss": 0.0353,
+      "step": 14546
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0132851899056693e-05,
+      "loss": 0.0266,
+      "step": 14547
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.012240903588183e-05,
+      "loss": 0.0141,
+      "step": 14548
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0111971368749278e-05,
+      "loss": 0.0357,
+      "step": 14549
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0101538898046779e-05,
+      "loss": 0.0375,
+      "step": 14550
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.009111162416184e-05,
+      "loss": 0.0111,
+      "step": 14551
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0080689547481885e-05,
+      "loss": 0.05,
+      "step": 14552
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0070272668393937e-05,
+      "loss": 0.0061,
+      "step": 14553
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0059860987285034e-05,
+      "loss": 0.0267,
+      "step": 14554
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0049454504541938e-05,
+      "loss": 0.0258,
+      "step": 14555
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0039053220551186e-05,
+      "loss": 0.0178,
+      "step": 14556
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0028657135699203e-05,
+      "loss": 0.021,
+      "step": 14557
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0018266250372132e-05,
+      "loss": 0.035,
+      "step": 14558
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.0007880564955995e-05,
+      "loss": 0.0226,
+      "step": 14559
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.997500079836535e-06,
+      "loss": 0.0171,
+      "step": 14560
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.98712479539941e-06,
+      "loss": 0.0212,
+      "step": 14561
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.97675471203001e-06,
+      "loss": 0.0227,
+      "step": 14562
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.966389830113564e-06,
+      "loss": 0.0445,
+      "step": 14563
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.95603015003511e-06,
+      "loss": 0.0081,
+      "step": 14564
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.945675672179426e-06,
+      "loss": 0.027,
+      "step": 14565
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.935326396931203e-06,
+      "loss": 0.0328,
+      "step": 14566
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.924982324674852e-06,
+      "loss": 0.0282,
+      "step": 14567
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.91464345579468e-06,
+      "loss": 0.0243,
+      "step": 14568
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.904309790674664e-06,
+      "loss": 0.0196,
+      "step": 14569
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.893981329698747e-06,
+      "loss": 0.0199,
+      "step": 14570
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.883658073250522e-06,
+      "loss": 0.0266,
+      "step": 14571
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.873340021713515e-06,
+      "loss": 0.0556,
+      "step": 14572
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.863027175471072e-06,
+      "loss": 0.016,
+      "step": 14573
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.852719534906166e-06,
+      "loss": 0.013,
+      "step": 14574
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.842417100401779e-06,
+      "loss": 0.0287,
+      "step": 14575
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.832119872340549e-06,
+      "loss": 0.0136,
+      "step": 14576
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.821827851105057e-06,
+      "loss": 0.0682,
+      "step": 14577
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.811541037077596e-06,
+      "loss": 0.0286,
+      "step": 14578
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.801259430640263e-06,
+      "loss": 0.0284,
+      "step": 14579
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.790983032175031e-06,
+      "loss": 0.0321,
+      "step": 14580
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.7807118420636e-06,
+      "loss": 0.0228,
+      "step": 14581
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.770445860687559e-06,
+      "loss": 0.0365,
+      "step": 14582
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.760185088428225e-06,
+      "loss": 0.0094,
+      "step": 14583
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.749929525666772e-06,
+      "loss": 0.0218,
+      "step": 14584
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.739679172784149e-06,
+      "loss": 0.0282,
+      "step": 14585
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.729434030161132e-06,
+      "loss": 0.0219,
+      "step": 14586
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.719194098178302e-06,
+      "loss": 0.025,
+      "step": 14587
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.708959377216036e-06,
+      "loss": 0.0481,
+      "step": 14588
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.698729867654547e-06,
+      "loss": 0.0187,
+      "step": 14589
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.6885055698738e-06,
+      "loss": 0.0349,
+      "step": 14590
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.678286484253622e-06,
+      "loss": 0.0113,
+      "step": 14591
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.668072611173577e-06,
+      "loss": 0.0122,
+      "step": 14592
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.657863951013179e-06,
+      "loss": 0.0283,
+      "step": 14593
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.647660504151521e-06,
+      "loss": 0.0222,
+      "step": 14594
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.637462270967705e-06,
+      "loss": 0.0114,
+      "step": 14595
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.627269251840592e-06,
+      "loss": 0.0322,
+      "step": 14596
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.617081447148745e-06,
+      "loss": 0.0402,
+      "step": 14597
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.606898857270712e-06,
+      "loss": 0.0363,
+      "step": 14598
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.596721482584624e-06,
+      "loss": 0.0317,
+      "step": 14599
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.586549323468645e-06,
+      "loss": 0.0305,
+      "step": 14600
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.576382380300574e-06,
+      "loss": 0.02,
+      "step": 14601
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.566220653458124e-06,
+      "loss": 0.0428,
+      "step": 14602
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.556064143318759e-06,
+      "loss": 0.0248,
+      "step": 14603
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.545912850259745e-06,
+      "loss": 0.0327,
+      "step": 14604
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.535766774658198e-06,
+      "loss": 0.0323,
+      "step": 14605
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.525625916891e-06,
+      "loss": 0.0195,
+      "step": 14606
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.51549027733488e-06,
+      "loss": 0.0173,
+      "step": 14607
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.50535985636634e-06,
+      "loss": 0.021,
+      "step": 14608
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.495234654361678e-06,
+      "loss": 0.015,
+      "step": 14609
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.485114671696992e-06,
+      "loss": 0.0164,
+      "step": 14610
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.474999908748265e-06,
+      "loss": 0.0152,
+      "step": 14611
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.464890365891181e-06,
+      "loss": 0.0329,
+      "step": 14612
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.454786043501321e-06,
+      "loss": 0.0285,
+      "step": 14613
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.444686941954022e-06,
+      "loss": 0.0203,
+      "step": 14614
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.434593061624412e-06,
+      "loss": 0.0236,
+      "step": 14615
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.424504402887478e-06,
+      "loss": 0.0348,
+      "step": 14616
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.414420966117952e-06,
+      "loss": 0.0183,
+      "step": 14617
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.404342751690437e-06,
+      "loss": 0.016,
+      "step": 14618
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.394269759979312e-06,
+      "loss": 0.0063,
+      "step": 14619
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.384201991358713e-06,
+      "loss": 0.0077,
+      "step": 14620
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.374139446202678e-06,
+      "loss": 0.0786,
+      "step": 14621
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.364082124884954e-06,
+      "loss": 0.0137,
+      "step": 14622
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.354030027779192e-06,
+      "loss": 0.0123,
+      "step": 14623
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.343983155258777e-06,
+      "loss": 0.0086,
+      "step": 14624
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.333941507696913e-06,
+      "loss": 0.047,
+      "step": 14625
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.323905085466615e-06,
+      "loss": 0.0058,
+      "step": 14626
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.31387388894072e-06,
+      "loss": 0.0165,
+      "step": 14627
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.30384791849183e-06,
+      "loss": 0.0286,
+      "step": 14628
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.293827174492429e-06,
+      "loss": 0.0264,
+      "step": 14629
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.283811657314737e-06,
+      "loss": 0.0233,
+      "step": 14630
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.273801367330758e-06,
+      "loss": 0.054,
+      "step": 14631
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.263796304912425e-06,
+      "loss": 0.0297,
+      "step": 14632
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.253796470431308e-06,
+      "loss": 0.0125,
+      "step": 14633
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.243801864258976e-06,
+      "loss": 0.046,
+      "step": 14634
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.233812486766584e-06,
+      "loss": 0.0071,
+      "step": 14635
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.223828338325284e-06,
+      "loss": 0.0274,
+      "step": 14636
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.213849419305946e-06,
+      "loss": 0.0272,
+      "step": 14637
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.203875730079224e-06,
+      "loss": 0.0196,
+      "step": 14638
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.193907271015704e-06,
+      "loss": 0.007,
+      "step": 14639
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.183944042485559e-06,
+      "loss": 0.0177,
+      "step": 14640
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.173986044858972e-06,
+      "loss": 0.0237,
+      "step": 14641
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.164033278505816e-06,
+      "loss": 0.0187,
+      "step": 14642
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.154085743795847e-06,
+      "loss": 0.0284,
+      "step": 14643
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.144143441098567e-06,
+      "loss": 0.0206,
+      "step": 14644
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.134206370783282e-06,
+      "loss": 0.0085,
+      "step": 14645
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.124274533219162e-06,
+      "loss": 0.0337,
+      "step": 14646
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.114347928775112e-06,
+      "loss": 0.026,
+      "step": 14647
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.104426557819905e-06,
+      "loss": 0.0537,
+      "step": 14648
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.094510420722095e-06,
+      "loss": 0.024,
+      "step": 14649
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.084599517850006e-06,
+      "loss": 0.0216,
+      "step": 14650
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.07469384957179e-06,
+      "loss": 0.0269,
+      "step": 14651
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.064793416255473e-06,
+      "loss": 0.0279,
+      "step": 14652
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.054898218268757e-06,
+      "loss": 0.0323,
+      "step": 14653
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.045008255979285e-06,
+      "loss": 0.0067,
+      "step": 14654
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.035123529754413e-06,
+      "loss": 0.0448,
+      "step": 14655
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.025244039961293e-06,
+      "loss": 0.0259,
+      "step": 14656
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.015369786966987e-06,
+      "loss": 0.0349,
+      "step": 14657
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.00550077113823e-06,
+      "loss": 0.0062,
+      "step": 14658
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.995636992841666e-06,
+      "loss": 0.0344,
+      "step": 14659
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.985778452443715e-06,
+      "loss": 0.0472,
+      "step": 14660
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.975925150310537e-06,
+      "loss": 0.018,
+      "step": 14661
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.966077086808222e-06,
+      "loss": 0.0102,
+      "step": 14662
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.956234262302541e-06,
+      "loss": 0.0148,
+      "step": 14663
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.94639667715919e-06,
+      "loss": 0.0158,
+      "step": 14664
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.936564331743523e-06,
+      "loss": 0.0276,
+      "step": 14665
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.92673722642085e-06,
+      "loss": 0.0343,
+      "step": 14666
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.916915361556176e-06,
+      "loss": 0.0102,
+      "step": 14667
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.907098737514378e-06,
+      "loss": 0.0486,
+      "step": 14668
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.897287354660132e-06,
+      "loss": 0.0199,
+      "step": 14669
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.887481213357844e-06,
+      "loss": 0.0093,
+      "step": 14670
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.877680313971857e-06,
+      "loss": 0.0381,
+      "step": 14671
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 8.867884656866181e-06,
+      "loss": 0.0167,
+      "step": 14672
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.85809424240474e-06,
+      "loss": 0.0191,
+      "step": 14673
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.84830907095121e-06,
+      "loss": 0.0188,
+      "step": 14674
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.838529142869067e-06,
+      "loss": 0.0087,
+      "step": 14675
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.828754458521586e-06,
+      "loss": 0.0302,
+      "step": 14676
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.818985018271912e-06,
+      "loss": 0.0244,
+      "step": 14677
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.809220822482937e-06,
+      "loss": 0.0324,
+      "step": 14678
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.79946187151737e-06,
+      "loss": 0.0193,
+      "step": 14679
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.789708165737758e-06,
+      "loss": 0.0274,
+      "step": 14680
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.779959705506356e-06,
+      "loss": 0.0264,
+      "step": 14681
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.770216491185328e-06,
+      "loss": 0.0232,
+      "step": 14682
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.7604785231366e-06,
+      "loss": 0.0618,
+      "step": 14683
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.750745801721932e-06,
+      "loss": 0.0327,
+      "step": 14684
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.74101832730285e-06,
+      "loss": 0.0237,
+      "step": 14685
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.731296100240687e-06,
+      "loss": 0.0034,
+      "step": 14686
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.721579120896615e-06,
+      "loss": 0.0171,
+      "step": 14687
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.71186738963155e-06,
+      "loss": 0.0145,
+      "step": 14688
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.702160906806333e-06,
+      "loss": 0.0126,
+      "step": 14689
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.692459672781476e-06,
+      "loss": 0.0195,
+      "step": 14690
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.68276368791736e-06,
+      "loss": 0.0229,
+      "step": 14691
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.673072952574145e-06,
+      "loss": 0.0191,
+      "step": 14692
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.663387467111843e-06,
+      "loss": 0.0362,
+      "step": 14693
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.653707231890217e-06,
+      "loss": 0.0122,
+      "step": 14694
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.644032247268895e-06,
+      "loss": 0.0302,
+      "step": 14695
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.634362513607256e-06,
+      "loss": 0.0137,
+      "step": 14696
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.624698031264466e-06,
+      "loss": 0.0161,
+      "step": 14697
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.6150388005996e-06,
+      "loss": 0.0116,
+      "step": 14698
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.605384821971406e-06,
+      "loss": 0.0161,
+      "step": 14699
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.595736095738548e-06,
+      "loss": 0.0294,
+      "step": 14700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.586092622259421e-06,
+      "loss": 0.0185,
+      "step": 14701
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.576454401892257e-06,
+      "loss": 0.0523,
+      "step": 14702
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.566821434995103e-06,
+      "loss": 0.0213,
+      "step": 14703
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.55719372192577e-06,
+      "loss": 0.0235,
+      "step": 14704
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 8.547571263041941e-06,
+      "loss": 0.0313,
+      "step": 14705
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.537954058700997e-06,
+      "loss": 0.0441,
+      "step": 14706
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.52834210926025e-06,
+      "loss": 0.0177,
+      "step": 14707
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.518735415076716e-06,
+      "loss": 0.029,
+      "step": 14708
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.509133976507293e-06,
+      "loss": 0.0284,
+      "step": 14709
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.499537793908611e-06,
+      "loss": 0.0227,
+      "step": 14710
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.489946867637138e-06,
+      "loss": 0.0174,
+      "step": 14711
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.480361198049184e-06,
+      "loss": 0.0237,
+      "step": 14712
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.470780785500787e-06,
+      "loss": 0.0227,
+      "step": 14713
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.461205630347872e-06,
+      "loss": 0.0345,
+      "step": 14714
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.45163573294611e-06,
+      "loss": 0.0251,
+      "step": 14715
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.442071093650999e-06,
+      "loss": 0.0417,
+      "step": 14716
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.432511712817818e-06,
+      "loss": 0.0194,
+      "step": 14717
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.422957590801687e-06,
+      "loss": 0.0191,
+      "step": 14718
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.413408727957504e-06,
+      "loss": 0.0387,
+      "step": 14719
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.403865124640002e-06,
+      "loss": 0.0162,
+      "step": 14720
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.39432678120368e-06,
+      "loss": 0.0359,
+      "step": 14721
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.384793698002839e-06,
+      "loss": 0.0274,
+      "step": 14722
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.375265875391663e-06,
+      "loss": 0.0118,
+      "step": 14723
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.365743313724017e-06,
+      "loss": 0.0525,
+      "step": 14724
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.356226013353684e-06,
+      "loss": 0.0593,
+      "step": 14725
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.3467139746342e-06,
+      "loss": 0.0291,
+      "step": 14726
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.337207197918867e-06,
+      "loss": 0.0348,
+      "step": 14727
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.327705683560898e-06,
+      "loss": 0.0205,
+      "step": 14728
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.318209431913181e-06,
+      "loss": 0.0163,
+      "step": 14729
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.308718443328549e-06,
+      "loss": 0.0258,
+      "step": 14730
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.29923271815947e-06,
+      "loss": 0.0099,
+      "step": 14731
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.289752256758397e-06,
+      "loss": 0.02,
+      "step": 14732
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.28027705947743e-06,
+      "loss": 0.0239,
+      "step": 14733
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.270807126668622e-06,
+      "loss": 0.0238,
+      "step": 14734
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.261342458683689e-06,
+      "loss": 0.0223,
+      "step": 14735
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.251883055874237e-06,
+      "loss": 0.0178,
+      "step": 14736
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.242428918591665e-06,
+      "loss": 0.0359,
+      "step": 14737
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 8.232980047187143e-06,
+      "loss": 0.0098,
+      "step": 14738
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.223536442011708e-06,
+      "loss": 0.0328,
+      "step": 14739
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.214098103416127e-06,
+      "loss": 0.0277,
+      "step": 14740
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.204665031751007e-06,
+      "loss": 0.0261,
+      "step": 14741
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.19523722736678e-06,
+      "loss": 0.018,
+      "step": 14742
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.185814690613635e-06,
+      "loss": 0.0185,
+      "step": 14743
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.17639742184164e-06,
+      "loss": 0.0103,
+      "step": 14744
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.166985421400552e-06,
+      "loss": 0.0135,
+      "step": 14745
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.157578689640088e-06,
+      "loss": 0.0265,
+      "step": 14746
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.148177226909569e-06,
+      "loss": 0.015,
+      "step": 14747
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.138781033558318e-06,
+      "loss": 0.0172,
+      "step": 14748
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.129390109935318e-06,
+      "loss": 0.0309,
+      "step": 14749
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.120004456389478e-06,
+      "loss": 0.0158,
+      "step": 14750
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.1106240732694e-06,
+      "loss": 0.0079,
+      "step": 14751
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.10124896092354e-06,
+      "loss": 0.0098,
+      "step": 14752
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.091879119700184e-06,
+      "loss": 0.0366,
+      "step": 14753
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.082514549947355e-06,
+      "loss": 0.0488,
+      "step": 14754
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.073155252012959e-06,
+      "loss": 0.023,
+      "step": 14755
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.063801226244648e-06,
+      "loss": 0.0163,
+      "step": 14756
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.054452472989897e-06,
+      "loss": 0.0331,
+      "step": 14757
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.045108992595978e-06,
+      "loss": 0.0177,
+      "step": 14758
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.035770785409979e-06,
+      "loss": 0.0373,
+      "step": 14759
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.02643785177879e-06,
+      "loss": 0.0406,
+      "step": 14760
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.017110192049097e-06,
+      "loss": 0.0297,
+      "step": 14761
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 8.007787806567428e-06,
+      "loss": 0.0297,
+      "step": 14762
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.998470695680004e-06,
+      "loss": 0.0213,
+      "step": 14763
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.989158859733013e-06,
+      "loss": 0.0238,
+      "step": 14764
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.979852299072314e-06,
+      "loss": 0.0491,
+      "step": 14765
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.97055101404363e-06,
+      "loss": 0.0525,
+      "step": 14766
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.9612550049925e-06,
+      "loss": 0.015,
+      "step": 14767
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.951964272264183e-06,
+      "loss": 0.0402,
+      "step": 14768
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.942678816203885e-06,
+      "loss": 0.0168,
+      "step": 14769
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.933398637156445e-06,
+      "loss": 0.024,
+      "step": 14770
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 7.924123735466693e-06,
+      "loss": 0.0326,
+      "step": 14771
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.914854111479064e-06,
+      "loss": 0.0177,
+      "step": 14772
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.905589765537967e-06,
+      "loss": 0.0266,
+      "step": 14773
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.896330697987496e-06,
+      "loss": 0.0041,
+      "step": 14774
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.88707690917164e-06,
+      "loss": 0.0276,
+      "step": 14775
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.877828399434144e-06,
+      "loss": 0.0155,
+      "step": 14776
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.868585169118529e-06,
+      "loss": 0.0164,
+      "step": 14777
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.859347218568207e-06,
+      "loss": 0.0414,
+      "step": 14778
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.850114548126285e-06,
+      "loss": 0.0371,
+      "step": 14779
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.840887158135772e-06,
+      "loss": 0.0319,
+      "step": 14780
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.831665048939428e-06,
+      "loss": 0.0509,
+      "step": 14781
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.822448220879795e-06,
+      "loss": 0.0201,
+      "step": 14782
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.813236674299318e-06,
+      "loss": 0.0373,
+      "step": 14783
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.804030409540102e-06,
+      "loss": 0.043,
+      "step": 14784
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.794829426944194e-06,
+      "loss": 0.0341,
+      "step": 14785
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.785633726853352e-06,
+      "loss": 0.0194,
+      "step": 14786
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.776443309609187e-06,
+      "loss": 0.0062,
+      "step": 14787
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.767258175553077e-06,
+      "loss": 0.0415,
+      "step": 14788
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.75807832502623e-06,
+      "loss": 0.0204,
+      "step": 14789
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.748903758369641e-06,
+      "loss": 0.0307,
+      "step": 14790
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.739734475924153e-06,
+      "loss": 0.0141,
+      "step": 14791
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.730570478030345e-06,
+      "loss": 0.0151,
+      "step": 14792
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.721411765028629e-06,
+      "loss": 0.0219,
+      "step": 14793
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.712258337259247e-06,
+      "loss": 0.0158,
+      "step": 14794
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.703110195062213e-06,
+      "loss": 0.0132,
+      "step": 14795
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.693967338777384e-06,
+      "loss": 0.0074,
+      "step": 14796
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.68482976874431e-06,
+      "loss": 0.0265,
+      "step": 14797
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.675697485302501e-06,
+      "loss": 0.0337,
+      "step": 14798
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.666570488791152e-06,
+      "loss": 0.0607,
+      "step": 14799
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.657448779549324e-06,
+      "loss": 0.0064,
+      "step": 14800
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.648332357915881e-06,
+      "loss": 0.0116,
+      "step": 14801
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.63922122422942e-06,
+      "loss": 0.0367,
+      "step": 14802
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.630115378828433e-06,
+      "loss": 0.0292,
+      "step": 14803
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.6210148220511525e-06,
+      "loss": 0.0253,
+      "step": 14804
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.611919554235657e-06,
+      "loss": 0.0336,
+      "step": 14805
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.602829575719793e-06,
+      "loss": 0.0166,
+      "step": 14806
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.593744886841241e-06,
+      "loss": 0.0302,
+      "step": 14807
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.58466548793748e-06,
+      "loss": 0.0529,
+      "step": 14808
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.575591379345741e-06,
+      "loss": 0.0261,
+      "step": 14809
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.566522561403138e-06,
+      "loss": 0.0085,
+      "step": 14810
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.557459034446517e-06,
+      "loss": 0.0255,
+      "step": 14811
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.548400798812643e-06,
+      "loss": 0.0475,
+      "step": 14812
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.539347854837879e-06,
+      "loss": 0.0185,
+      "step": 14813
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.530300202858625e-06,
+      "loss": 0.0077,
+      "step": 14814
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.5212578432108925e-06,
+      "loss": 0.0233,
+      "step": 14815
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.512220776230632e-06,
+      "loss": 0.0284,
+      "step": 14816
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.50318900225354e-06,
+      "loss": 0.0239,
+      "step": 14817
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.494162521615082e-06,
+      "loss": 0.038,
+      "step": 14818
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.485141334650608e-06,
+      "loss": 0.0134,
+      "step": 14819
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.476125441695197e-06,
+      "loss": 0.0106,
+      "step": 14820
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.467114843083799e-06,
+      "loss": 0.0172,
+      "step": 14821
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.458109539151114e-06,
+      "loss": 0.038,
+      "step": 14822
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.449109530231656e-06,
+      "loss": 0.0324,
+      "step": 14823
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.440114816659726e-06,
+      "loss": 0.048,
+      "step": 14824
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.431125398769489e-06,
+      "loss": 0.0469,
+      "step": 14825
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.4221412768948775e-06,
+      "loss": 0.0187,
+      "step": 14826
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.413162451369609e-06,
+      "loss": 0.0323,
+      "step": 14827
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.404188922527216e-06,
+      "loss": 0.0238,
+      "step": 14828
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.395220690701048e-06,
+      "loss": 0.0234,
+      "step": 14829
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.38625775622424e-06,
+      "loss": 0.0125,
+      "step": 14830
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.377300119429741e-06,
+      "loss": 0.0289,
+      "step": 14831
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.368347780650319e-06,
+      "loss": 0.0693,
+      "step": 14832
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.359400740218524e-06,
+      "loss": 0.0669,
+      "step": 14833
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.350458998466674e-06,
+      "loss": 0.0136,
+      "step": 14834
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.34152255572697e-06,
+      "loss": 0.0337,
+      "step": 14835
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.332591412331345e-06,
+      "loss": 0.0125,
+      "step": 14836
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 7.32366556861162e-06,
+      "loss": 0.0253,
+      "step": 14837
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.3147450248992765e-06,
+      "loss": 0.0356,
+      "step": 14838
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.3058297815257685e-06,
+      "loss": 0.0225,
+      "step": 14839
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.296919838822213e-06,
+      "loss": 0.0236,
+      "step": 14840
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.2880151971196285e-06,
+      "loss": 0.014,
+      "step": 14841
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.279115856748785e-06,
+      "loss": 0.0366,
+      "step": 14842
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.270221818040234e-06,
+      "loss": 0.0218,
+      "step": 14843
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.2613330813244275e-06,
+      "loss": 0.0027,
+      "step": 14844
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.252449646931486e-06,
+      "loss": 0.0187,
+      "step": 14845
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.243571515191443e-06,
+      "loss": 0.0053,
+      "step": 14846
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.234698686434104e-06,
+      "loss": 0.0123,
+      "step": 14847
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.225831160989037e-06,
+      "loss": 0.0207,
+      "step": 14848
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.2169689391856796e-06,
+      "loss": 0.0117,
+      "step": 14849
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.208112021353185e-06,
+      "loss": 0.0106,
+      "step": 14850
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.199260407820639e-06,
+      "loss": 0.0144,
+      "step": 14851
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.190414098916797e-06,
+      "loss": 0.0214,
+      "step": 14852
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.181573094970278e-06,
+      "loss": 0.0072,
+      "step": 14853
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.172737396309486e-06,
+      "loss": 0.0346,
+      "step": 14854
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.163907003262692e-06,
+      "loss": 0.0267,
+      "step": 14855
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.155081916157868e-06,
+      "loss": 0.0104,
+      "step": 14856
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.146262135322883e-06,
+      "loss": 0.0295,
+      "step": 14857
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.137447661085344e-06,
+      "loss": 0.0162,
+      "step": 14858
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.128638493772654e-06,
+      "loss": 0.0135,
+      "step": 14859
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.119834633712101e-06,
+      "loss": 0.0385,
+      "step": 14860
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.111036081230675e-06,
+      "loss": 0.0069,
+      "step": 14861
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.102242836655297e-06,
+      "loss": 0.035,
+      "step": 14862
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.093454900312506e-06,
+      "loss": 0.0293,
+      "step": 14863
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.084672272528807e-06,
+      "loss": 0.0132,
+      "step": 14864
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.075894953630423e-06,
+      "loss": 0.0237,
+      "step": 14865
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.0671229439434265e-06,
+      "loss": 0.0117,
+      "step": 14866
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.058356243793672e-06,
+      "loss": 0.0337,
+      "step": 14867
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.049594853506818e-06,
+      "loss": 0.0134,
+      "step": 14868
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.040838773408319e-06,
+      "loss": 0.026,
+      "step": 14869
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 7.032088003823417e-06,
+      "loss": 0.0157,
+      "step": 14870
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.023342545077198e-06,
+      "loss": 0.0338,
+      "step": 14871
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.014602397494523e-06,
+      "loss": 0.0373,
+      "step": 14872
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 7.0058675614000804e-06,
+      "loss": 0.0261,
+      "step": 14873
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.997138037118344e-06,
+      "loss": 0.024,
+      "step": 14874
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.988413824973554e-06,
+      "loss": 0.0244,
+      "step": 14875
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.9796949252898195e-06,
+      "loss": 0.0218,
+      "step": 14876
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.970981338391013e-06,
+      "loss": 0.0268,
+      "step": 14877
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.962273064600843e-06,
+      "loss": 0.0354,
+      "step": 14878
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.953570104242734e-06,
+      "loss": 0.0178,
+      "step": 14879
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.944872457640044e-06,
+      "loss": 0.005,
+      "step": 14880
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.936180125115831e-06,
+      "loss": 0.0361,
+      "step": 14881
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.927493106992987e-06,
+      "loss": 0.0237,
+      "step": 14882
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.918811403594237e-06,
+      "loss": 0.0382,
+      "step": 14883
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.910135015242024e-06,
+      "loss": 0.0527,
+      "step": 14884
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.901463942258723e-06,
+      "loss": 0.0351,
+      "step": 14885
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.892798184966375e-06,
+      "loss": 0.0232,
+      "step": 14886
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.8841377436869404e-06,
+      "loss": 0.0547,
+      "step": 14887
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.875482618742112e-06,
+      "loss": 0.0306,
+      "step": 14888
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.866832810453365e-06,
+      "loss": 0.0014,
+      "step": 14889
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.8581883191420605e-06,
+      "loss": 0.0438,
+      "step": 14890
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.84954914512929e-06,
+      "loss": 0.0292,
+      "step": 14891
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.840915288735998e-06,
+      "loss": 0.0213,
+      "step": 14892
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.832286750282895e-06,
+      "loss": 0.0325,
+      "step": 14893
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.823663530090506e-06,
+      "loss": 0.0246,
+      "step": 14894
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.815045628479127e-06,
+      "loss": 0.0165,
+      "step": 14895
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.80643304576895e-06,
+      "loss": 0.0387,
+      "step": 14896
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.797825782279853e-06,
+      "loss": 0.0251,
+      "step": 14897
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.789223838331598e-06,
+      "loss": 0.0171,
+      "step": 14898
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.7806272142437276e-06,
+      "loss": 0.0252,
+      "step": 14899
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.772035910335538e-06,
+      "loss": 0.0376,
+      "step": 14900
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.763449926926223e-06,
+      "loss": 0.0329,
+      "step": 14901
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.754869264334695e-06,
+      "loss": 0.0365,
+      "step": 14902
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 6.746293922879748e-06,
+      "loss": 0.0118,
+      "step": 14903
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.737723902879843e-06,
+      "loss": 0.0361,
+      "step": 14904
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.729159204653412e-06,
+      "loss": 0.0113,
+      "step": 14905
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.720599828518547e-06,
+      "loss": 0.038,
+      "step": 14906
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.712045774793245e-06,
+      "loss": 0.0255,
+      "step": 14907
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.703497043795303e-06,
+      "loss": 0.0206,
+      "step": 14908
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.694953635842182e-06,
+      "loss": 0.0246,
+      "step": 14909
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.6864155512513115e-06,
+      "loss": 0.034,
+      "step": 14910
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.6778827903398225e-06,
+      "loss": 0.014,
+      "step": 14911
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.669355353424727e-06,
+      "loss": 0.0194,
+      "step": 14912
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.6608332408227546e-06,
+      "loss": 0.0274,
+      "step": 14913
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.65231645285047e-06,
+      "loss": 0.0492,
+      "step": 14914
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.643804989824286e-06,
+      "loss": 0.006,
+      "step": 14915
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.63529885206035e-06,
+      "loss": 0.0231,
+      "step": 14916
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.626798039874659e-06,
+      "loss": 0.0103,
+      "step": 14917
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.618302553582977e-06,
+      "loss": 0.0448,
+      "step": 14918
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.609812393500902e-06,
+      "loss": 0.0199,
+      "step": 14919
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.6013275599437806e-06,
+      "loss": 0.0469,
+      "step": 14920
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.592848053226846e-06,
+      "loss": 0.034,
+      "step": 14921
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.584373873665061e-06,
+      "loss": 0.0498,
+      "step": 14922
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.575905021573225e-06,
+      "loss": 0.0189,
+      "step": 14923
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.5674414972659515e-06,
+      "loss": 0.0305,
+      "step": 14924
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.558983301057574e-06,
+      "loss": 0.0141,
+      "step": 14925
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.550530433262357e-06,
+      "loss": 0.021,
+      "step": 14926
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.5420828941942674e-06,
+      "loss": 0.0111,
+      "step": 14927
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.533640684167102e-06,
+      "loss": 0.035,
+      "step": 14928
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.525203803494495e-06,
+      "loss": 0.0262,
+      "step": 14929
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.51677225248981e-06,
+      "loss": 0.0224,
+      "step": 14930
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.508346031466283e-06,
+      "loss": 0.0479,
+      "step": 14931
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.499925140736911e-06,
+      "loss": 0.0113,
+      "step": 14932
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.491509580614512e-06,
+      "loss": 0.0061,
+      "step": 14933
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.483099351411703e-06,
+      "loss": 0.0314,
+      "step": 14934
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.4746944534408986e-06,
+      "loss": 0.0052,
+      "step": 14935
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 6.4662948870142835e-06,
+      "loss": 0.0437,
+      "step": 14936
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.457900652443926e-06,
+      "loss": 0.0087,
+      "step": 14937
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.449511750041608e-06,
+      "loss": 0.0121,
+      "step": 14938
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.441128180118998e-06,
+      "loss": 0.0078,
+      "step": 14939
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.432749942987481e-06,
+      "loss": 0.0152,
+      "step": 14940
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.424377038958273e-06,
+      "loss": 0.0326,
+      "step": 14941
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.416009468342459e-06,
+      "loss": 0.0197,
+      "step": 14942
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.407647231450807e-06,
+      "loss": 0.0299,
+      "step": 14943
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.39929032859402e-06,
+      "loss": 0.0252,
+      "step": 14944
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.390938760082465e-06,
+      "loss": 0.0637,
+      "step": 14945
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.38259252622641e-06,
+      "loss": 0.0302,
+      "step": 14946
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.374251627335875e-06,
+      "loss": 0.0202,
+      "step": 14947
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.3659160637207105e-06,
+      "loss": 0.0196,
+      "step": 14948
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.357585835690604e-06,
+      "loss": 0.0468,
+      "step": 14949
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.349260943554907e-06,
+      "loss": 0.0206,
+      "step": 14950
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.3409413876229386e-06,
+      "loss": 0.041,
+      "step": 14951
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.3326271682037026e-06,
+      "loss": 0.0211,
+      "step": 14952
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.324318285606083e-06,
+      "loss": 0.0266,
+      "step": 14953
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.316014740138719e-06,
+      "loss": 0.0153,
+      "step": 14954
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.3077165321100465e-06,
+      "loss": 0.0127,
+      "step": 14955
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.299423661828334e-06,
+      "loss": 0.0495,
+      "step": 14956
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.29113612960162e-06,
+      "loss": 0.0489,
+      "step": 14957
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.282853935737792e-06,
+      "loss": 0.0271,
+      "step": 14958
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.2745770805445015e-06,
+      "loss": 0.0297,
+      "step": 14959
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.266305564329188e-06,
+      "loss": 0.0072,
+      "step": 14960
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.258039387399122e-06,
+      "loss": 0.026,
+      "step": 14961
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.249778550061374e-06,
+      "loss": 0.0296,
+      "step": 14962
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.241523052622815e-06,
+      "loss": 0.0267,
+      "step": 14963
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.233272895390102e-06,
+      "loss": 0.0361,
+      "step": 14964
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.225028078669719e-06,
+      "loss": 0.0108,
+      "step": 14965
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.2167886027679085e-06,
+      "loss": 0.0412,
+      "step": 14966
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.208554467990773e-06,
+      "loss": 0.0159,
+      "step": 14967
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 6.200325674644169e-06,
+      "loss": 0.0145,
+      "step": 14968
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.192102223033801e-06,
+      "loss": 0.0349,
+      "step": 14969
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.183884113465093e-06,
+      "loss": 0.0097,
+      "step": 14970
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.175671346243349e-06,
+      "loss": 0.0329,
+      "step": 14971
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.167463921673676e-06,
+      "loss": 0.0234,
+      "step": 14972
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.159261840060914e-06,
+      "loss": 0.028,
+      "step": 14973
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.1510651017098015e-06,
+      "loss": 0.0336,
+      "step": 14974
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.142873706924745e-06,
+      "loss": 0.0297,
+      "step": 14975
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.134687656010085e-06,
+      "loss": 0.0176,
+      "step": 14976
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.126506949269894e-06,
+      "loss": 0.0263,
+      "step": 14977
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.11833158700808e-06,
+      "loss": 0.04,
+      "step": 14978
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.110161569528299e-06,
+      "loss": 0.0444,
+      "step": 14979
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.101996897134059e-06,
+      "loss": 0.0395,
+      "step": 14980
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.093837570128684e-06,
+      "loss": 0.0282,
+      "step": 14981
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.085683588815215e-06,
+      "loss": 0.0232,
+      "step": 14982
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.077534953496593e-06,
+      "loss": 0.0217,
+      "step": 14983
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.069391664475492e-06,
+      "loss": 0.0089,
+      "step": 14984
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.0612537220544375e-06,
+      "loss": 0.0257,
+      "step": 14985
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.053121126535671e-06,
+      "loss": 0.0277,
+      "step": 14986
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.044993878221366e-06,
+      "loss": 0.0222,
+      "step": 14987
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.036871977413366e-06,
+      "loss": 0.0131,
+      "step": 14988
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.028755424413428e-06,
+      "loss": 0.0237,
+      "step": 14989
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.02064421952303e-06,
+      "loss": 0.0088,
+      "step": 14990
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.012538363043462e-06,
+      "loss": 0.0319,
+      "step": 14991
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 6.004437855275884e-06,
+      "loss": 0.0217,
+      "step": 14992
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.996342696521139e-06,
+      "loss": 0.0241,
+      "step": 14993
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.98825288708002e-06,
+      "loss": 0.0319,
+      "step": 14994
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.980168427252968e-06,
+      "loss": 0.029,
+      "step": 14995
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.972089317340329e-06,
+      "loss": 0.0202,
+      "step": 14996
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.96401555764221e-06,
+      "loss": 0.0516,
+      "step": 14997
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.95594714845854e-06,
+      "loss": 0.0169,
+      "step": 14998
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.947884090089028e-06,
+      "loss": 0.0193,
+      "step": 14999
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.939826382833185e-06,
+      "loss": 0.0195,
+      "step": 15000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 5.9317740269903544e-06,
+      "loss": 0.0344,
+      "step": 15001
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.9237270228596145e-06,
+      "loss": 0.0269,
+      "step": 15002
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.915685370739942e-06,
+      "loss": 0.0322,
+      "step": 15003
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.907649070930015e-06,
+      "loss": 0.0321,
+      "step": 15004
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.899618123728395e-06,
+      "loss": 0.0216,
+      "step": 15005
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.891592529433392e-06,
+      "loss": 0.0108,
+      "step": 15006
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.883572288343103e-06,
+      "loss": 0.008,
+      "step": 15007
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.875557400755504e-06,
+      "loss": 0.0271,
+      "step": 15008
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.8675478669683085e-06,
+      "loss": 0.0193,
+      "step": 15009
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.859543687279061e-06,
+      "loss": 0.0159,
+      "step": 15010
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.8515448619850415e-06,
+      "loss": 0.0393,
+      "step": 15011
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.843551391383428e-06,
+      "loss": 0.0141,
+      "step": 15012
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.8355632757711654e-06,
+      "loss": 0.0106,
+      "step": 15013
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.827580515444952e-06,
+      "loss": 0.0189,
+      "step": 15014
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.819603110701365e-06,
+      "loss": 0.0086,
+      "step": 15015
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.811631061836686e-06,
+      "loss": 0.0133,
+      "step": 15016
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.803664369147093e-06,
+      "loss": 0.0148,
+      "step": 15017
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.795703032928517e-06,
+      "loss": 0.0355,
+      "step": 15018
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.787747053476704e-06,
+      "loss": 0.0296,
+      "step": 15019
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.7797964310871855e-06,
+      "loss": 0.0382,
+      "step": 15020
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.771851166055292e-06,
+      "loss": 0.0164,
+      "step": 15021
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.763911258676202e-06,
+      "loss": 0.0206,
+      "step": 15022
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.75597670924483e-06,
+      "loss": 0.0185,
+      "step": 15023
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.748047518055943e-06,
+      "loss": 0.0251,
+      "step": 15024
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.740123685404069e-06,
+      "loss": 0.0148,
+      "step": 15025
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.732205211583574e-06,
+      "loss": 0.0135,
+      "step": 15026
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.724292096888555e-06,
+      "loss": 0.0161,
+      "step": 15027
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.716384341613045e-06,
+      "loss": 0.0296,
+      "step": 15028
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.7084819460507085e-06,
+      "loss": 0.0087,
+      "step": 15029
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.700584910495159e-06,
+      "loss": 0.0139,
+      "step": 15030
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.692693235239731e-06,
+      "loss": 0.0132,
+      "step": 15031
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.684806920577556e-06,
+      "loss": 0.0177,
+      "step": 15032
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.676925966801615e-06,
+      "loss": 0.0137,
+      "step": 15033
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 5.669050374204626e-06,
+      "loss": 0.0273,
+      "step": 15034
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.661180143079202e-06,
+      "loss": 0.0143,
+      "step": 15035
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.653315273717663e-06,
+      "loss": 0.0241,
+      "step": 15036
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.645455766412138e-06,
+      "loss": 0.0094,
+      "step": 15037
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.637601621454646e-06,
+      "loss": 0.0187,
+      "step": 15038
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.629752839136886e-06,
+      "loss": 0.0158,
+      "step": 15039
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.621909419750492e-06,
+      "loss": 0.0186,
+      "step": 15040
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.614071363586731e-06,
+      "loss": 0.0171,
+      "step": 15041
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.606238670936836e-06,
+      "loss": 0.0308,
+      "step": 15042
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.598411342091725e-06,
+      "loss": 0.0127,
+      "step": 15043
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.590589377342197e-06,
+      "loss": 0.0184,
+      "step": 15044
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.5827727769787886e-06,
+      "loss": 0.0141,
+      "step": 15045
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.5749615412918665e-06,
+      "loss": 0.0124,
+      "step": 15046
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.567155670571616e-06,
+      "loss": 0.019,
+      "step": 15047
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.559355165107954e-06,
+      "loss": 0.0228,
+      "step": 15048
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.551560025190716e-06,
+      "loss": 0.0219,
+      "step": 15049
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.54377025110942e-06,
+      "loss": 0.0156,
+      "step": 15050
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.535985843153451e-06,
+      "loss": 0.0094,
+      "step": 15051
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.528206801611945e-06,
+      "loss": 0.0232,
+      "step": 15052
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.520433126773904e-06,
+      "loss": 0.0177,
+      "step": 15053
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.512664818928097e-06,
+      "loss": 0.0261,
+      "step": 15054
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.504901878363077e-06,
+      "loss": 0.0157,
+      "step": 15055
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.497144305367263e-06,
+      "loss": 0.0472,
+      "step": 15056
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.48939210022874e-06,
+      "loss": 0.0245,
+      "step": 15057
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.481645263235546e-06,
+      "loss": 0.017,
+      "step": 15058
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.473903794675416e-06,
+      "loss": 0.027,
+      "step": 15059
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.466167694835971e-06,
+      "loss": 0.0367,
+      "step": 15060
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.458436964004548e-06,
+      "loss": 0.0377,
+      "step": 15061
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.4507116024683e-06,
+      "loss": 0.0086,
+      "step": 15062
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.4429916105142634e-06,
+      "loss": 0.0028,
+      "step": 15063
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.4352769884291436e-06,
+      "loss": 0.0143,
+      "step": 15064
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.427567736499578e-06,
+      "loss": 0.0181,
+      "step": 15065
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.419863855011919e-06,
+      "loss": 0.0254,
+      "step": 15066
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 5.41216534425234e-06,
+      "loss": 0.0347,
+      "step": 15067
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.4044722045067934e-06,
+      "loss": 0.0427,
+      "step": 15068
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.396784436061102e-06,
+      "loss": 0.0284,
+      "step": 15069
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.3891020392008045e-06,
+      "loss": 0.0532,
+      "step": 15070
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.381425014211321e-06,
+      "loss": 0.0332,
+      "step": 15071
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.373753361377792e-06,
+      "loss": 0.0179,
+      "step": 15072
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.366087080985221e-06,
+      "loss": 0.0297,
+      "step": 15073
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.358426173318381e-06,
+      "loss": 0.0195,
+      "step": 15074
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.350770638661828e-06,
+      "loss": 0.0389,
+      "step": 15075
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.343120477299984e-06,
+      "loss": 0.0229,
+      "step": 15076
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.3354756895170215e-06,
+      "loss": 0.0426,
+      "step": 15077
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.327836275596897e-06,
+      "loss": 0.0192,
+      "step": 15078
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.320202235823418e-06,
+      "loss": 0.0235,
+      "step": 15079
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.312573570480138e-06,
+      "loss": 0.0272,
+      "step": 15080
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.3049502798504835e-06,
+      "loss": 0.0149,
+      "step": 15081
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.297332364217594e-06,
+      "loss": 0.0441,
+      "step": 15082
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.289719823864474e-06,
+      "loss": 0.0219,
+      "step": 15083
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.2821126590739e-06,
+      "loss": 0.061,
+      "step": 15084
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.274510870128462e-06,
+      "loss": 0.0028,
+      "step": 15085
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.266914457310551e-06,
+      "loss": 0.0475,
+      "step": 15086
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.259323420902323e-06,
+      "loss": 0.0246,
+      "step": 15087
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.251737761185787e-06,
+      "loss": 0.0324,
+      "step": 15088
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.244157478442718e-06,
+      "loss": 0.0278,
+      "step": 15089
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.236582572954723e-06,
+      "loss": 0.0178,
+      "step": 15090
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.229013045003161e-06,
+      "loss": 0.0273,
+      "step": 15091
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.22144889486924e-06,
+      "loss": 0.0236,
+      "step": 15092
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.213890122833903e-06,
+      "loss": 0.0163,
+      "step": 15093
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.206336729177974e-06,
+      "loss": 0.0274,
+      "step": 15094
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.198788714182045e-06,
+      "loss": 0.0277,
+      "step": 15095
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.191246078126493e-06,
+      "loss": 0.0287,
+      "step": 15096
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.183708821291493e-06,
+      "loss": 0.0363,
+      "step": 15097
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.176176943957039e-06,
+      "loss": 0.0232,
+      "step": 15098
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.168650446402922e-06,
+      "loss": 0.0243,
+      "step": 15099
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.16112932890872e-06,
+      "loss": 0.0124,
+      "step": 15100
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.1536135917538245e-06,
+      "loss": 0.0256,
+      "step": 15101
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.146103235217447e-06,
+      "loss": 0.0241,
+      "step": 15102
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.13859825957853e-06,
+      "loss": 0.0171,
+      "step": 15103
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.131098665115902e-06,
+      "loss": 0.0182,
+      "step": 15104
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.123604452108104e-06,
+      "loss": 0.0574,
+      "step": 15105
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.116115620833599e-06,
+      "loss": 0.0117,
+      "step": 15106
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.108632171570498e-06,
+      "loss": 0.0371,
+      "step": 15107
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.101154104596844e-06,
+      "loss": 0.0277,
+      "step": 15108
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.093681420190382e-06,
+      "loss": 0.0506,
+      "step": 15109
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.086214118628723e-06,
+      "loss": 0.0293,
+      "step": 15110
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.0787522001892625e-06,
+      "loss": 0.0338,
+      "step": 15111
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.071295665149178e-06,
+      "loss": 0.0358,
+      "step": 15112
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.063844513785464e-06,
+      "loss": 0.0083,
+      "step": 15113
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.056398746374885e-06,
+      "loss": 0.015,
+      "step": 15114
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.048958363194066e-06,
+      "loss": 0.0038,
+      "step": 15115
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.041523364519373e-06,
+      "loss": 0.0108,
+      "step": 15116
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.034093750626999e-06,
+      "loss": 0.0113,
+      "step": 15117
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.026669521792942e-06,
+      "loss": 0.0272,
+      "step": 15118
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.019250678292963e-06,
+      "loss": 0.0323,
+      "step": 15119
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.011837220402692e-06,
+      "loss": 0.0437,
+      "step": 15120
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 5.004429148397459e-06,
+      "loss": 0.0277,
+      "step": 15121
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.997026462552544e-06,
+      "loss": 0.0114,
+      "step": 15122
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.989629163142827e-06,
+      "loss": 0.0174,
+      "step": 15123
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.982237250443172e-06,
+      "loss": 0.0229,
+      "step": 15124
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.974850724728125e-06,
+      "loss": 0.0246,
+      "step": 15125
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.967469586272116e-06,
+      "loss": 0.0252,
+      "step": 15126
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.9600938353492934e-06,
+      "loss": 0.0308,
+      "step": 15127
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.952723472233655e-06,
+      "loss": 0.0274,
+      "step": 15128
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.9453584971989975e-06,
+      "loss": 0.0156,
+      "step": 15129
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.937998910518887e-06,
+      "loss": 0.0406,
+      "step": 15130
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.930644712466753e-06,
+      "loss": 0.003,
+      "step": 15131
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.923295903315743e-06,
+      "loss": 0.0169,
+      "step": 15132
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.915952483338859e-06,
+      "loss": 0.0125,
+      "step": 15133
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.9086144528088786e-06,
+      "loss": 0.0324,
+      "step": 15134
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.901281811998387e-06,
+      "loss": 0.0794,
+      "step": 15135
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.89395456117978e-06,
+      "loss": 0.0169,
+      "step": 15136
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.886632700625259e-06,
+      "loss": 0.0156,
+      "step": 15137
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.879316230606789e-06,
+      "loss": 0.025,
+      "step": 15138
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.872005151396136e-06,
+      "loss": 0.0225,
+      "step": 15139
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.864699463264932e-06,
+      "loss": 0.0412,
+      "step": 15140
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.8573991664845105e-06,
+      "loss": 0.0423,
+      "step": 15141
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.850104261326104e-06,
+      "loss": 0.029,
+      "step": 15142
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.842814748060681e-06,
+      "loss": 0.031,
+      "step": 15143
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.835530626959005e-06,
+      "loss": 0.0315,
+      "step": 15144
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.82825189829168e-06,
+      "loss": 0.0372,
+      "step": 15145
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.820978562329087e-06,
+      "loss": 0.0228,
+      "step": 15146
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.813710619341427e-06,
+      "loss": 0.0348,
+      "step": 15147
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.806448069598618e-06,
+      "loss": 0.0207,
+      "step": 15148
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.79919091337051e-06,
+      "loss": 0.0166,
+      "step": 15149
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.791939150926655e-06,
+      "loss": 0.0158,
+      "step": 15150
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.7846927825364535e-06,
+      "loss": 0.0177,
+      "step": 15151
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.777451808469074e-06,
+      "loss": 0.032,
+      "step": 15152
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.770216228993485e-06,
+      "loss": 0.0154,
+      "step": 15153
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.762986044378486e-06,
+      "loss": 0.0246,
+      "step": 15154
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.755761254892648e-06,
+      "loss": 0.0504,
+      "step": 15155
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.748541860804355e-06,
+      "loss": 0.0258,
+      "step": 15156
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.7413278623817945e-06,
+      "loss": 0.0247,
+      "step": 15157
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.734119259892933e-06,
+      "loss": 0.0228,
+      "step": 15158
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.726916053605556e-06,
+      "loss": 0.014,
+      "step": 15159
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.719718243787218e-06,
+      "loss": 0.0039,
+      "step": 15160
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.712525830705338e-06,
+      "loss": 0.0388,
+      "step": 15161
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.705338814627069e-06,
+      "loss": 0.0382,
+      "step": 15162
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.69815719581938e-06,
+      "loss": 0.0349,
+      "step": 15163
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.690980974549058e-06,
+      "loss": 0.0302,
+      "step": 15164
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.68381015108269e-06,
+      "loss": 0.0346,
+      "step": 15165
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.676644725686612e-06,
+      "loss": 0.0392,
+      "step": 15166
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.669484698627046e-06,
+      "loss": 0.0213,
+      "step": 15167
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.662330070169962e-06,
+      "loss": 0.0641,
+      "step": 15168
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.65518084058108e-06,
+      "loss": 0.0239,
+      "step": 15169
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.648037010126038e-06,
+      "loss": 0.0363,
+      "step": 15170
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.640898579070157e-06,
+      "loss": 0.0472,
+      "step": 15171
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.633765547678658e-06,
+      "loss": 0.0102,
+      "step": 15172
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.626637916216463e-06,
+      "loss": 0.0313,
+      "step": 15173
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.619515684948377e-06,
+      "loss": 0.0106,
+      "step": 15174
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.6123988541389544e-06,
+      "loss": 0.0195,
+      "step": 15175
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.605287424052551e-06,
+      "loss": 0.0267,
+      "step": 15176
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.598181394953404e-06,
+      "loss": 0.0413,
+      "step": 15177
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.591080767105387e-06,
+      "loss": 0.0294,
+      "step": 15178
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.583985540772322e-06,
+      "loss": 0.0508,
+      "step": 15179
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.576895716217766e-06,
+      "loss": 0.014,
+      "step": 15180
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.569811293705089e-06,
+      "loss": 0.0027,
+      "step": 15181
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.562732273497433e-06,
+      "loss": 0.0208,
+      "step": 15182
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.555658655857802e-06,
+      "loss": 0.0355,
+      "step": 15183
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.548590441048955e-06,
+      "loss": 0.0212,
+      "step": 15184
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.541527629333397e-06,
+      "loss": 0.0221,
+      "step": 15185
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.534470220973568e-06,
+      "loss": 0.0078,
+      "step": 15186
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.527418216231576e-06,
+      "loss": 0.0032,
+      "step": 15187
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.520371615369428e-06,
+      "loss": 0.0433,
+      "step": 15188
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.513330418648847e-06,
+      "loss": 0.004,
+      "step": 15189
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.506294626331408e-06,
+      "loss": 0.024,
+      "step": 15190
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.499264238678435e-06,
+      "loss": 0.0202,
+      "step": 15191
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.4922392559511525e-06,
+      "loss": 0.0185,
+      "step": 15192
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.485219678410484e-06,
+      "loss": 0.0204,
+      "step": 15193
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.478205506317173e-06,
+      "loss": 0.0141,
+      "step": 15194
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.471196739931793e-06,
+      "loss": 0.014,
+      "step": 15195
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.464193379514685e-06,
+      "loss": 0.0189,
+      "step": 15196
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.4571954253260265e-06,
+      "loss": 0.013,
+      "step": 15197
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.450202877625758e-06,
+      "loss": 0.0297,
+      "step": 15198
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.443215736673605e-06,
+      "loss": 0.0131,
+      "step": 15199
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.436234002729178e-06,
+      "loss": 0.0128,
+      "step": 15200
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.4292576760517676e-06,
+      "loss": 0.0355,
+      "step": 15201
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.422286756900584e-06,
+      "loss": 0.0119,
+      "step": 15202
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.41532124553452e-06,
+      "loss": 0.0214,
+      "step": 15203
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.408361142212369e-06,
+      "loss": 0.0284,
+      "step": 15204
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.40140644719264e-06,
+      "loss": 0.0456,
+      "step": 15205
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.3944571607336945e-06,
+      "loss": 0.0682,
+      "step": 15206
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.387513283093674e-06,
+      "loss": 0.0334,
+      "step": 15207
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.380574814530557e-06,
+      "loss": 0.0254,
+      "step": 15208
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.373641755302054e-06,
+      "loss": 0.0235,
+      "step": 15209
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.3667141056656915e-06,
+      "loss": 0.0234,
+      "step": 15210
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.359791865878864e-06,
+      "loss": 0.0268,
+      "step": 15211
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.3528750361986495e-06,
+      "loss": 0.0169,
+      "step": 15212
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.345963616882059e-06,
+      "loss": 0.02,
+      "step": 15213
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.33905760818577e-06,
+      "loss": 0.0214,
+      "step": 15214
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.332157010366361e-06,
+      "loss": 0.007,
+      "step": 15215
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.325261823680126e-06,
+      "loss": 0.0158,
+      "step": 15216
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.3183720483832285e-06,
+      "loss": 0.0425,
+      "step": 15217
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.31148768473163e-06,
+      "loss": 0.027,
+      "step": 15218
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.3046087329810075e-06,
+      "loss": 0.0217,
+      "step": 15219
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.297735193386925e-06,
+      "loss": 0.0246,
+      "step": 15220
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.290867066204711e-06,
+      "loss": 0.0278,
+      "step": 15221
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.284004351689513e-06,
+      "loss": 0.018,
+      "step": 15222
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.277147050096241e-06,
+      "loss": 0.052,
+      "step": 15223
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.27029516167961e-06,
+      "loss": 0.0277,
+      "step": 15224
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.263448686694199e-06,
+      "loss": 0.0083,
+      "step": 15225
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.256607625394254e-06,
+      "loss": 0.0218,
+      "step": 15226
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.249771978033989e-06,
+      "loss": 0.0164,
+      "step": 15227
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.242941744867267e-06,
+      "loss": 0.036,
+      "step": 15228
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.236116926147837e-06,
+      "loss": 0.0331,
+      "step": 15229
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.22929752212921e-06,
+      "loss": 0.0357,
+      "step": 15230
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.222483533064702e-06,
+      "loss": 0.042,
+      "step": 15231
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.215674959207443e-06,
+      "loss": 0.0463,
+      "step": 15232
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.208871800810365e-06,
+      "loss": 0.0191,
+      "step": 15233
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.202074058126164e-06,
+      "loss": 0.0257,
+      "step": 15234
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.195281731407357e-06,
+      "loss": 0.0371,
+      "step": 15235
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.1884948209062745e-06,
+      "loss": 0.0121,
+      "step": 15236
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.181713326875013e-06,
+      "loss": 0.026,
+      "step": 15237
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.1749372495655075e-06,
+      "loss": 0.0249,
+      "step": 15238
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.168166589229438e-06,
+      "loss": 0.0399,
+      "step": 15239
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.1614013461183204e-06,
+      "loss": 0.0242,
+      "step": 15240
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.154641520483504e-06,
+      "loss": 0.0612,
+      "step": 15241
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.147887112576037e-06,
+      "loss": 0.0219,
+      "step": 15242
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.141138122646887e-06,
+      "loss": 0.0319,
+      "step": 15243
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.134394550946718e-06,
+      "loss": 0.0428,
+      "step": 15244
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.127656397726048e-06,
+      "loss": 0.0284,
+      "step": 15245
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.12092366323516e-06,
+      "loss": 0.0338,
+      "step": 15246
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.114196347724186e-06,
+      "loss": 0.0307,
+      "step": 15247
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.107474451442994e-06,
+      "loss": 0.0234,
+      "step": 15248
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.100757974641317e-06,
+      "loss": 0.0265,
+      "step": 15249
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.094046917568622e-06,
+      "loss": 0.0198,
+      "step": 15250
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.087341280474227e-06,
+      "loss": 0.0096,
+      "step": 15251
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.080641063607215e-06,
+      "loss": 0.0267,
+      "step": 15252
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.0739462672164714e-06,
+      "loss": 0.0198,
+      "step": 15253
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.067256891550729e-06,
+      "loss": 0.029,
+      "step": 15254
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.060572936858425e-06,
+      "loss": 0.0243,
+      "step": 15255
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.0538944033878915e-06,
+      "loss": 0.0266,
+      "step": 15256
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.047221291387165e-06,
+      "loss": 0.0177,
+      "step": 15257
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.04055360110418e-06,
+      "loss": 0.0096,
+      "step": 15258
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.033891332786621e-06,
+      "loss": 0.0258,
+      "step": 15259
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.02723448668194e-06,
+      "loss": 0.0207,
+      "step": 15260
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.020583063037458e-06,
+      "loss": 0.0261,
+      "step": 15261
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.013937062100209e-06,
+      "loss": 0.0605,
+      "step": 15262
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.007296484117112e-06,
+      "loss": 0.0334,
+      "step": 15263
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.00066132933482e-06,
+      "loss": 0.0393,
+      "step": 15264
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.9940315979998194e-06,
+      "loss": 0.0168,
+      "step": 15265
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.987407290358397e-06,
+      "loss": 0.0185,
+      "step": 15266
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.9807884066566044e-06,
+      "loss": 0.0382,
+      "step": 15267
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.9741749471403295e-06,
+      "loss": 0.0158,
+      "step": 15268
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.967566912055259e-06,
+      "loss": 0.0272,
+      "step": 15269
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.960964301646813e-06,
+      "loss": 0.0222,
+      "step": 15270
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.954367116160295e-06,
+      "loss": 0.0152,
+      "step": 15271
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.947775355840776e-06,
+      "loss": 0.0177,
+      "step": 15272
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.941189020933078e-06,
+      "loss": 0.0214,
+      "step": 15273
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.93460811168192e-06,
+      "loss": 0.0066,
+      "step": 15274
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.92803262833174e-06,
+      "loss": 0.0402,
+      "step": 15275
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.921462571126777e-06,
+      "loss": 0.0475,
+      "step": 15276
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.914897940311118e-06,
+      "loss": 0.0161,
+      "step": 15277
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.908338736128602e-06,
+      "loss": 0.0169,
+      "step": 15278
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.901784958822917e-06,
+      "loss": 0.0125,
+      "step": 15279
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8952366086374535e-06,
+      "loss": 0.0192,
+      "step": 15280
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8886936858155136e-06,
+      "loss": 0.0232,
+      "step": 15281
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.8821561906001385e-06,
+      "loss": 0.0186,
+      "step": 15282
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.875624123234167e-06,
+      "loss": 0.0406,
+      "step": 15283
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.869097483960287e-06,
+      "loss": 0.0329,
+      "step": 15284
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.862576273020873e-06,
+      "loss": 0.0331,
+      "step": 15285
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.85606049065823e-06,
+      "loss": 0.032,
+      "step": 15286
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.849550137114349e-06,
+      "loss": 0.0487,
+      "step": 15287
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.843045212631119e-06,
+      "loss": 0.0237,
+      "step": 15288
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.836545717450163e-06,
+      "loss": 0.0155,
+      "step": 15289
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.830051651812887e-06,
+      "loss": 0.0262,
+      "step": 15290
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.823563015960585e-06,
+      "loss": 0.0141,
+      "step": 15291
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.817079810134227e-06,
+      "loss": 0.0576,
+      "step": 15292
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.810602034574689e-06,
+      "loss": 0.0164,
+      "step": 15293
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.804129689522595e-06,
+      "loss": 0.0539,
+      "step": 15294
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.7976627752183687e-06,
+      "loss": 0.0093,
+      "step": 15295
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.7912012919022183e-06,
+      "loss": 0.0174,
+      "step": 15296
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.7847452398142018e-06,
+      "loss": 0.0142,
+      "step": 15297
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.778294619194111e-06,
+      "loss": 0.0251,
+      "step": 15298
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7718494302816037e-06,
+      "loss": 0.039,
+      "step": 15299
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7654096733160723e-06,
+      "loss": 0.0542,
+      "step": 15300
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7589753485367413e-06,
+      "loss": 0.0079,
+      "step": 15301
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7525464561826203e-06,
+      "loss": 0.0378,
+      "step": 15302
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7461229964925344e-06,
+      "loss": 0.0089,
+      "step": 15303
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7397049697050928e-06,
+      "loss": 0.0415,
+      "step": 15304
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.733292376058722e-06,
+      "loss": 0.044,
+      "step": 15305
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7268852157915973e-06,
+      "loss": 0.0259,
+      "step": 15306
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.720483489141746e-06,
+      "loss": 0.0223,
+      "step": 15307
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7140871963469763e-06,
+      "loss": 0.0266,
+      "step": 15308
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7076963376449e-06,
+      "loss": 0.0201,
+      "step": 15309
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.7013109132728925e-06,
+      "loss": 0.022,
+      "step": 15310
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6949309234681814e-06,
+      "loss": 0.0263,
+      "step": 15311
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6885563684677268e-06,
+      "loss": 0.0237,
+      "step": 15312
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6821872485083557e-06,
+      "loss": 0.0246,
+      "step": 15313
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6758235638266466e-06,
+      "loss": 0.0348,
+      "step": 15314
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.669465314659026e-06,
+      "loss": 0.0287,
+      "step": 15315
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.663112501241655e-06,
+      "loss": 0.0212,
+      "step": 15316
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6567651238104954e-06,
+      "loss": 0.0233,
+      "step": 15317
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.650423182601392e-06,
+      "loss": 0.0339,
+      "step": 15318
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.644086677849889e-06,
+      "loss": 0.0403,
+      "step": 15319
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6377556097913985e-06,
+      "loss": 0.0343,
+      "step": 15320
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.631429978661049e-06,
+      "loss": 0.0219,
+      "step": 15321
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6251097846938692e-06,
+      "loss": 0.0181,
+      "step": 15322
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.618795028124638e-06,
+      "loss": 0.012,
+      "step": 15323
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.6124857091878845e-06,
+      "loss": 0.0332,
+      "step": 15324
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.606181828118038e-06,
+      "loss": 0.0202,
+      "step": 15325
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5998833851492106e-06,
+      "loss": 0.0169,
+      "step": 15326
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.593590380515432e-06,
+      "loss": 0.0122,
+      "step": 15327
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.587302814450399e-06,
+      "loss": 0.0242,
+      "step": 15328
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.5810206871877413e-06,
+      "loss": 0.0478,
+      "step": 15329
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.574743998960805e-06,
+      "loss": 0.0322,
+      "step": 15330
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.568472750002721e-06,
+      "loss": 0.0333,
+      "step": 15331
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5622069405464694e-06,
+      "loss": 0.0138,
+      "step": 15332
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.555946570824797e-06,
+      "loss": 0.0163,
+      "step": 15333
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.549691641070285e-06,
+      "loss": 0.0252,
+      "step": 15334
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.543442151515263e-06,
+      "loss": 0.0214,
+      "step": 15335
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5371981023918805e-06,
+      "loss": 0.0176,
+      "step": 15336
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.530959493932084e-06,
+      "loss": 0.0169,
+      "step": 15337
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5247263263676384e-06,
+      "loss": 0.023,
+      "step": 15338
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5184985999300417e-06,
+      "loss": 0.0232,
+      "step": 15339
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5122763148507096e-06,
+      "loss": 0.0258,
+      "step": 15340
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.5060594713607228e-06,
+      "loss": 0.0253,
+      "step": 15341
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4998480696910304e-06,
+      "loss": 0.0392,
+      "step": 15342
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4936421100723976e-06,
+      "loss": 0.0413,
+      "step": 15343
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.487441592735307e-06,
+      "loss": 0.0374,
+      "step": 15344
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4812465179101245e-06,
+      "loss": 0.0052,
+      "step": 15345
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.475056885826982e-06,
+      "loss": 0.0096,
+      "step": 15346
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4688726967157797e-06,
+      "loss": 0.0224,
+      "step": 15347
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4626939508062667e-06,
+      "loss": 0.0323,
+      "step": 15348
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4565206483279595e-06,
+      "loss": 0.0285,
+      "step": 15349
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4503527895101914e-06,
+      "loss": 0.0246,
+      "step": 15350
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4441903745820285e-06,
+      "loss": 0.0195,
+      "step": 15351
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4380334037724387e-06,
+      "loss": 0.0092,
+      "step": 15352
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4318818773101055e-06,
+      "loss": 0.0258,
+      "step": 15353
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4257357954235787e-06,
+      "loss": 0.0161,
+      "step": 15354
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4195951583411263e-06,
+      "loss": 0.0205,
+      "step": 15355
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.413459966290866e-06,
+      "loss": 0.0408,
+      "step": 15356
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4073302195007146e-06,
+      "loss": 0.02,
+      "step": 15357
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4012059181983576e-06,
+      "loss": 0.0222,
+      "step": 15358
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3950870626113123e-06,
+      "loss": 0.0072,
+      "step": 15359
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.388973652966864e-06,
+      "loss": 0.0328,
+      "step": 15360
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3828656894921145e-06,
+      "loss": 0.0216,
+      "step": 15361
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.3767631724139486e-06,
+      "loss": 0.0274,
+      "step": 15362
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.370666101959052e-06,
+      "loss": 0.0223,
+      "step": 15363
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.364574478353943e-06,
+      "loss": 0.0298,
+      "step": 15364
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.358488301824891e-06,
+      "loss": 0.0312,
+      "step": 15365
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.352407572597965e-06,
+      "loss": 0.021,
+      "step": 15366
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.346332290899051e-06,
+      "loss": 0.0266,
+      "step": 15367
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.340262456953835e-06,
+      "loss": 0.0318,
+      "step": 15368
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3341980709877867e-06,
+      "loss": 0.0204,
+      "step": 15369
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.32813913322621e-06,
+      "loss": 0.0272,
+      "step": 15370
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.3220856438941403e-06,
+      "loss": 0.0297,
+      "step": 15371
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.316037603216448e-06,
+      "loss": 0.0165,
+      "step": 15372
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.309995011417821e-06,
+      "loss": 0.02,
+      "step": 15373
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.303957868722712e-06,
+      "loss": 0.0173,
+      "step": 15374
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.297926175355392e-06,
+      "loss": 0.019,
+      "step": 15375
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2918999315399154e-06,
+      "loss": 0.0425,
+      "step": 15376
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.285879137500136e-06,
+      "loss": 0.0214,
+      "step": 15377
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2798637934597085e-06,
+      "loss": 0.0231,
+      "step": 15378
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2738538996420872e-06,
+      "loss": 0.0386,
+      "step": 15379
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2678494562705103e-06,
+      "loss": 0.0166,
+      "step": 15380
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2618504635680665e-06,
+      "loss": 0.0262,
+      "step": 15381
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2558569217575604e-06,
+      "loss": 0.037,
+      "step": 15382
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2498688310616307e-06,
+      "loss": 0.0214,
+      "step": 15383
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2438861917027493e-06,
+      "loss": 0.0236,
+      "step": 15384
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.237909003903122e-06,
+      "loss": 0.0194,
+      "step": 15385
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.231937267884821e-06,
+      "loss": 0.0366,
+      "step": 15386
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2259709838696356e-06,
+      "loss": 0.0195,
+      "step": 15387
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2200101520792222e-06,
+      "loss": 0.0366,
+      "step": 15388
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2140547727350197e-06,
+      "loss": 0.0227,
+      "step": 15389
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.208104846058218e-06,
+      "loss": 0.033,
+      "step": 15390
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.2021603722698743e-06,
+      "loss": 0.0276,
+      "step": 15391
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.1962213515907776e-06,
+      "loss": 0.0181,
+      "step": 15392
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.190287784241552e-06,
+      "loss": 0.0167,
+      "step": 15393
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.1843596704426213e-06,
+      "loss": 0.0178,
+      "step": 15394
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.1784370104142086e-06,
+      "loss": 0.037,
+      "step": 15395
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.1725198043763223e-06,
+      "loss": 0.0171,
+      "step": 15396
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.1666080525487193e-06,
+      "loss": 0.0129,
+      "step": 15397
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.160701755151074e-06,
+      "loss": 0.0143,
+      "step": 15398
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1548009124027275e-06,
+      "loss": 0.012,
+      "step": 15399
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1489055245229376e-06,
+      "loss": 0.0354,
+      "step": 15400
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.143015591730663e-06,
+      "loss": 0.0197,
+      "step": 15401
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.137131114244695e-06,
+      "loss": 0.0314,
+      "step": 15402
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.131252092283643e-06,
+      "loss": 0.0215,
+      "step": 15403
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1253785260658805e-06,
+      "loss": 0.03,
+      "step": 15404
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.119510415809584e-06,
+      "loss": 0.0422,
+      "step": 15405
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1136477617327793e-06,
+      "loss": 0.0309,
+      "step": 15406
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.1077905640532087e-06,
+      "loss": 0.022,
+      "step": 15407
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.101938822988448e-06,
+      "loss": 0.0219,
+      "step": 15408
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0960925387559067e-06,
+      "loss": 0.0262,
+      "step": 15409
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.090251711572728e-06,
+      "loss": 0.0357,
+      "step": 15410
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0844163416558876e-06,
+      "loss": 0.0222,
+      "step": 15411
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.078586429222163e-06,
+      "loss": 0.0241,
+      "step": 15412
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0727619744880807e-06,
+      "loss": 0.0242,
+      "step": 15413
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.066942977670067e-06,
+      "loss": 0.0503,
+      "step": 15414
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0611294389842168e-06,
+      "loss": 0.0081,
+      "step": 15415
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.05532135864654e-06,
+      "loss": 0.021,
+      "step": 15416
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.04951873687273e-06,
+      "loss": 0.0154,
+      "step": 15417
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0437215738783994e-06,
+      "loss": 0.0395,
+      "step": 15418
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.037929869878841e-06,
+      "loss": 0.0239,
+      "step": 15419
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.032143625089234e-06,
+      "loss": 0.0091,
+      "step": 15420
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0263628397245223e-06,
+      "loss": 0.0336,
+      "step": 15421
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0205875139994174e-06,
+      "loss": 0.0247,
+      "step": 15422
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.014817648128498e-06,
+      "loss": 0.0248,
+      "step": 15423
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0090532423260426e-06,
+      "loss": 0.0306,
+      "step": 15424
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.0032942968062303e-06,
+      "loss": 0.01,
+      "step": 15425
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.9975408117829723e-06,
+      "loss": 0.0547,
+      "step": 15426
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.9917927874699822e-06,
+      "loss": 0.0347,
+      "step": 15427
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.9860502240808048e-06,
+      "loss": 0.0235,
+      "step": 15428
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.9803131218287202e-06,
+      "loss": 0.0302,
+      "step": 15429
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.9745814809269073e-06,
+      "loss": 0.0239,
+      "step": 15430
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.968855301588213e-06,
+      "loss": 0.0425,
+      "step": 15431
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.963134584025417e-06,
+      "loss": 0.0365,
+      "step": 15432
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9574193284509494e-06,
+      "loss": 0.013,
+      "step": 15433
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9517095350771736e-06,
+      "loss": 0.0242,
+      "step": 15434
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.94600520411617e-06,
+      "loss": 0.0087,
+      "step": 15435
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.940306335779835e-06,
+      "loss": 0.0351,
+      "step": 15436
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9346129302798838e-06,
+      "loss": 0.0026,
+      "step": 15437
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.92892498782778e-06,
+      "loss": 0.0267,
+      "step": 15438
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9232425086348375e-06,
+      "loss": 0.0158,
+      "step": 15439
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9175654929121384e-06,
+      "loss": 0.0228,
+      "step": 15440
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.911893940870563e-06,
+      "loss": 0.0328,
+      "step": 15441
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9062278527208107e-06,
+      "loss": 0.0071,
+      "step": 15442
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9005672286733284e-06,
+      "loss": 0.0242,
+      "step": 15443
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8949120689384153e-06,
+      "loss": 0.0128,
+      "step": 15444
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8892623737261366e-06,
+      "loss": 0.013,
+      "step": 15445
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8836181432463414e-06,
+      "loss": 0.0344,
+      "step": 15446
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8779793777087445e-06,
+      "loss": 0.0179,
+      "step": 15447
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.872346077322779e-06,
+      "loss": 0.0317,
+      "step": 15448
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8667182422976943e-06,
+      "loss": 0.0279,
+      "step": 15449
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8610958728425728e-06,
+      "loss": 0.0205,
+      "step": 15450
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.855478969166247e-06,
+      "loss": 0.0187,
+      "step": 15451
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8498675314773845e-06,
+      "loss": 0.0104,
+      "step": 15452
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8442615599844344e-06,
+      "loss": 0.0228,
+      "step": 15453
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.83866105489563e-06,
+      "loss": 0.0395,
+      "step": 15454
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.833066016419039e-06,
+      "loss": 0.0236,
+      "step": 15455
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.827476444762461e-06,
+      "loss": 0.0368,
+      "step": 15456
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8218923401335637e-06,
+      "loss": 0.0075,
+      "step": 15457
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8163137027397643e-06,
+      "loss": 0.0358,
+      "step": 15458
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.810740532788297e-06,
+      "loss": 0.0117,
+      "step": 15459
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.8051728304861967e-06,
+      "loss": 0.0364,
+      "step": 15460
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.7996105960402804e-06,
+      "loss": 0.05,
+      "step": 15461
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.794053829657167e-06,
+      "loss": 0.0347,
+      "step": 15462
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7885025315432743e-06,
+      "loss": 0.0315,
+      "step": 15463
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7829567019048203e-06,
+      "loss": 0.0314,
+      "step": 15464
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.777416340947791e-06,
+      "loss": 0.0456,
+      "step": 15465
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.771881448878055e-06,
+      "loss": 0.0216,
+      "step": 15466
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.766352025901164e-06,
+      "loss": 0.0213,
+      "step": 15467
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7608280722225372e-06,
+      "loss": 0.0145,
+      "step": 15468
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7553095880473773e-06,
+      "loss": 0.0188,
+      "step": 15469
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.749796573580654e-06,
+      "loss": 0.0128,
+      "step": 15470
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7442890290272025e-06,
+      "loss": 0.0342,
+      "step": 15471
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.738786954591593e-06,
+      "loss": 0.0188,
+      "step": 15472
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7332903504781955e-06,
+      "loss": 0.0308,
+      "step": 15473
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7277992168911965e-06,
+      "loss": 0.0149,
+      "step": 15474
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7223135540345996e-06,
+      "loss": 0.0181,
+      "step": 15475
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7168333621121587e-06,
+      "loss": 0.0559,
+      "step": 15476
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.71135864132746e-06,
+      "loss": 0.0839,
+      "step": 15477
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7058893918838754e-06,
+      "loss": 0.0372,
+      "step": 15478
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.7004256139845416e-06,
+      "loss": 0.0094,
+      "step": 15479
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.694967307832463e-06,
+      "loss": 0.0252,
+      "step": 15480
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6895144736303608e-06,
+      "loss": 0.0181,
+      "step": 15481
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.68406711158084e-06,
+      "loss": 0.0206,
+      "step": 15482
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.678625221886205e-06,
+      "loss": 0.0214,
+      "step": 15483
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.673188804748644e-06,
+      "loss": 0.0171,
+      "step": 15484
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.667757860370062e-06,
+      "loss": 0.0219,
+      "step": 15485
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.662332388952232e-06,
+      "loss": 0.0197,
+      "step": 15486
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.656912390696708e-06,
+      "loss": 0.0204,
+      "step": 15487
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.651497865804797e-06,
+      "loss": 0.0274,
+      "step": 15488
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.646088814477637e-06,
+      "loss": 0.0347,
+      "step": 15489
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.640685236916168e-06,
+      "loss": 0.0219,
+      "step": 15490
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.635287133321129e-06,
+      "loss": 0.016,
+      "step": 15491
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6298945038930265e-06,
+      "loss": 0.0213,
+      "step": 15492
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6245073488321667e-06,
+      "loss": 0.0367,
+      "step": 15493
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6191256683387064e-06,
+      "loss": 0.0209,
+      "step": 15494
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.613749462612519e-06,
+      "loss": 0.0097,
+      "step": 15495
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6083787318533445e-06,
+      "loss": 0.03,
+      "step": 15496
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.603013476260657e-06,
+      "loss": 0.0527,
+      "step": 15497
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5976536960338134e-06,
+      "loss": 0.0245,
+      "step": 15498
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.592299391371855e-06,
+      "loss": 0.0467,
+      "step": 15499
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.586950562473722e-06,
+      "loss": 0.0244,
+      "step": 15500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5816072095380724e-06,
+      "loss": 0.0502,
+      "step": 15501
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.576269332763431e-06,
+      "loss": 0.0099,
+      "step": 15502
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5709369323480555e-06,
+      "loss": 0.0134,
+      "step": 15503
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5656100084900377e-06,
+      "loss": 0.038,
+      "step": 15504
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.560288561387269e-06,
+      "loss": 0.0052,
+      "step": 15505
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5549725912374084e-06,
+      "loss": 0.0388,
+      "step": 15506
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5496620982379644e-06,
+      "loss": 0.0215,
+      "step": 15507
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5443570825861626e-06,
+      "loss": 0.0274,
+      "step": 15508
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5390575444790786e-06,
+      "loss": 0.0425,
+      "step": 15509
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5337634841135713e-06,
+      "loss": 0.0073,
+      "step": 15510
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5284749016863173e-06,
+      "loss": 0.0199,
+      "step": 15511
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5231917973937755e-06,
+      "loss": 0.0348,
+      "step": 15512
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.517914171432189e-06,
+      "loss": 0.0193,
+      "step": 15513
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.512642023997602e-06,
+      "loss": 0.0236,
+      "step": 15514
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.50737535528584e-06,
+      "loss": 0.0215,
+      "step": 15515
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.5021141654925803e-06,
+      "loss": 0.0152,
+      "step": 15516
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.496858454813233e-06,
+      "loss": 0.0,
+      "step": 15517
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.491608223443059e-06,
+      "loss": 0.0176,
+      "step": 15518
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4863634715770854e-06,
+      "loss": 0.0191,
+      "step": 15519
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.481124199410106e-06,
+      "loss": 0.0113,
+      "step": 15520
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4758904071367657e-06,
+      "loss": 0.0195,
+      "step": 15521
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4706620949514923e-06,
+      "loss": 0.0549,
+      "step": 15522
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4654392630485133e-06,
+      "loss": 0.036,
+      "step": 15523
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.460221911621807e-06,
+      "loss": 0.0447,
+      "step": 15524
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.455010040865202e-06,
+      "loss": 0.0672,
+      "step": 15525
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4498036509722927e-06,
+      "loss": 0.0248,
+      "step": 15526
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.444602742136509e-06,
+      "loss": 0.0201,
+      "step": 15527
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.4394073145510283e-06,
+      "loss": 0.0472,
+      "step": 15528
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4342173684088304e-06,
+      "loss": 0.0322,
+      "step": 15529
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4290329039027444e-06,
+      "loss": 0.0043,
+      "step": 15530
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.423853921225316e-06,
+      "loss": 0.026,
+      "step": 15531
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.418680420568975e-06,
+      "loss": 0.0175,
+      "step": 15532
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4135124021258667e-06,
+      "loss": 0.0342,
+      "step": 15533
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.4083498660879717e-06,
+      "loss": 0.0389,
+      "step": 15534
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.403192812647087e-06,
+      "loss": 0.045,
+      "step": 15535
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3980412419947413e-06,
+      "loss": 0.0224,
+      "step": 15536
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3928951543223494e-06,
+      "loss": 0.0337,
+      "step": 15537
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3877545498210413e-06,
+      "loss": 0.0314,
+      "step": 15538
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.382619428681781e-06,
+      "loss": 0.0269,
+      "step": 15539
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3774897910953162e-06,
+      "loss": 0.0202,
+      "step": 15540
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3723656372522105e-06,
+      "loss": 0.0043,
+      "step": 15541
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3672469673427953e-06,
+      "loss": 0.0211,
+      "step": 15542
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.362133781557252e-06,
+      "loss": 0.0214,
+      "step": 15543
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3570260800854777e-06,
+      "loss": 0.0222,
+      "step": 15544
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3519238631172223e-06,
+      "loss": 0.0205,
+      "step": 15545
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.346827130842033e-06,
+      "loss": 0.0434,
+      "step": 15546
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3417358834492085e-06,
+      "loss": 0.0205,
+      "step": 15547
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.336650121127931e-06,
+      "loss": 0.0344,
+      "step": 15548
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3315698440670496e-06,
+      "loss": 0.0193,
+      "step": 15549
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3264950524553127e-06,
+      "loss": 0.0321,
+      "step": 15550
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.321425746481237e-06,
+      "loss": 0.0302,
+      "step": 15551
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3163619263331378e-06,
+      "loss": 0.0281,
+      "step": 15552
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.311303592199132e-06,
+      "loss": 0.0257,
+      "step": 15553
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3062507442670685e-06,
+      "loss": 0.0337,
+      "step": 15554
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3012033827247145e-06,
+      "loss": 0.0155,
+      "step": 15555
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.296161507759503e-06,
+      "loss": 0.0289,
+      "step": 15556
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2911251195587674e-06,
+      "loss": 0.0095,
+      "step": 15557
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2860942183095753e-06,
+      "loss": 0.0343,
+      "step": 15558
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2810688041988266e-06,
+      "loss": 0.0281,
+      "step": 15559
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.276048877413189e-06,
+      "loss": 0.0145,
+      "step": 15560
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2710344381391134e-06,
+      "loss": 0.0341,
+      "step": 15561
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2660254865629333e-06,
+      "loss": 0.017,
+      "step": 15562
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.261022022870651e-06,
+      "loss": 0.0272,
+      "step": 15563
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.256024047248184e-06,
+      "loss": 0.0548,
+      "step": 15564
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.25103155988115e-06,
+      "loss": 0.0292,
+      "step": 15565
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2460445609550348e-06,
+      "loss": 0.013,
+      "step": 15566
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2410630506550733e-06,
+      "loss": 0.02,
+      "step": 15567
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.236087029166317e-06,
+      "loss": 0.0253,
+      "step": 15568
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2311164966736184e-06,
+      "loss": 0.0335,
+      "step": 15569
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.226151453361613e-06,
+      "loss": 0.0236,
+      "step": 15570
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.221191899414737e-06,
+      "loss": 0.0077,
+      "step": 15571
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2162378350172094e-06,
+      "loss": 0.0447,
+      "step": 15572
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2112892603530996e-06,
+      "loss": 0.0124,
+      "step": 15573
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.206346175606194e-06,
+      "loss": 0.0234,
+      "step": 15574
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2014085809601288e-06,
+      "loss": 0.0205,
+      "step": 15575
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1964764765983244e-06,
+      "loss": 0.0219,
+      "step": 15576
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1915498627039673e-06,
+      "loss": 0.0306,
+      "step": 15577
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1866287394601114e-06,
+      "loss": 0.053,
+      "step": 15578
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.181713107049543e-06,
+      "loss": 0.0211,
+      "step": 15579
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.17680296565485e-06,
+      "loss": 0.0272,
+      "step": 15580
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.17189831545842e-06,
+      "loss": 0.0248,
+      "step": 15581
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1669991566424907e-06,
+      "loss": 0.0293,
+      "step": 15582
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.162105489388999e-06,
+      "loss": 0.0416,
+      "step": 15583
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1572173138797834e-06,
+      "loss": 0.0052,
+      "step": 15584
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1523346302963986e-06,
+      "loss": 0.0264,
+      "step": 15585
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.147457438820216e-06,
+      "loss": 0.0133,
+      "step": 15586
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1425857396324076e-06,
+      "loss": 0.0143,
+      "step": 15587
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.137719532913945e-06,
+      "loss": 0.0134,
+      "step": 15588
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1328588188456343e-06,
+      "loss": 0.0113,
+      "step": 15589
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.12800359760798e-06,
+      "loss": 0.0183,
+      "step": 15590
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1231538693813555e-06,
+      "loss": 0.016,
+      "step": 15591
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.118309634345916e-06,
+      "loss": 0.0186,
+      "step": 15592
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.113470892681618e-06,
+      "loss": 0.0308,
+      "step": 15593
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1086376445682174e-06,
+      "loss": 0.014,
+      "step": 15594
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.103809890185221e-06,
+      "loss": 0.0342,
+      "step": 15595
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.098987629712001e-06,
+      "loss": 0.0196,
+      "step": 15596
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0941708633276655e-06,
+      "loss": 0.0349,
+      "step": 15597
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0893595912111538e-06,
+      "loss": 0.0217,
+      "step": 15598
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.08455381354119e-06,
+      "loss": 0.028,
+      "step": 15599
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0797535304962975e-06,
+      "loss": 0.025,
+      "step": 15600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.074958742254784e-06,
+      "loss": 0.0247,
+      "step": 15601
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0701694489947574e-06,
+      "loss": 0.0263,
+      "step": 15602
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.065385650894158e-06,
+      "loss": 0.0145,
+      "step": 15603
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.060607348130644e-06,
+      "loss": 0.0434,
+      "step": 15604
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.055834540881757e-06,
+      "loss": 0.0268,
+      "step": 15605
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0510672293247543e-06,
+      "loss": 0.0308,
+      "step": 15606
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.046305413636762e-06,
+      "loss": 0.028,
+      "step": 15607
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0415490939946542e-06,
+      "loss": 0.0257,
+      "step": 15608
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0367982705751074e-06,
+      "loss": 0.0238,
+      "step": 15609
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.032052943554613e-06,
+      "loss": 0.0254,
+      "step": 15610
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.02731311310943e-06,
+      "loss": 0.0166,
+      "step": 15611
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0225787794156345e-06,
+      "loss": 0.0244,
+      "step": 15612
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.017849942649102e-06,
+      "loss": 0.0184,
+      "step": 15613
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0131266029855097e-06,
+      "loss": 0.0346,
+      "step": 15614
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0084087606002665e-06,
+      "loss": 0.0171,
+      "step": 15615
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0036964156686654e-06,
+      "loss": 0.0171,
+      "step": 15616
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9989895683657498e-06,
+      "loss": 0.0276,
+      "step": 15617
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9942882188663635e-06,
+      "loss": 0.0203,
+      "step": 15618
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9895923673451498e-06,
+      "loss": 0.0179,
+      "step": 15619
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.984902013976536e-06,
+      "loss": 0.0257,
+      "step": 15620
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.980217158934766e-06,
+      "loss": 0.018,
+      "step": 15621
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9755378023938674e-06,
+      "loss": 0.0158,
+      "step": 15622
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.970863944527651e-06,
+      "loss": 0.0143,
+      "step": 15623
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9661955855097443e-06,
+      "loss": 0.0256,
+      "step": 15624
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9615327255135927e-06,
+      "loss": 0.0073,
+      "step": 15625
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.956875364712357e-06,
+      "loss": 0.0396,
+      "step": 15626
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.9522235032790655e-06,
+      "loss": 0.033,
+      "step": 15627
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.947577141386547e-06,
+      "loss": 0.0258,
+      "step": 15628
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9429362792073465e-06,
+      "loss": 0.013,
+      "step": 15629
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.938300916913926e-06,
+      "loss": 0.0169,
+      "step": 15630
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.933671054678415e-06,
+      "loss": 0.0157,
+      "step": 15631
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9290466926728254e-06,
+      "loss": 0.0102,
+      "step": 15632
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9244278310689366e-06,
+      "loss": 0.0135,
+      "step": 15633
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.919814470038328e-06,
+      "loss": 0.0265,
+      "step": 15634
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9152066097523788e-06,
+      "loss": 0.023,
+      "step": 15635
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9106042503822363e-06,
+      "loss": 0.0009,
+      "step": 15636
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9060073920988961e-06,
+      "loss": 0.024,
+      "step": 15637
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9014160350730722e-06,
+      "loss": 0.0172,
+      "step": 15638
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8968301794753615e-06,
+      "loss": 0.0382,
+      "step": 15639
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.892249825476111e-06,
+      "loss": 0.0186,
+      "step": 15640
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8876749732454345e-06,
+      "loss": 0.0246,
+      "step": 15641
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8831056229533126e-06,
+      "loss": 0.0268,
+      "step": 15642
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8785417747694598e-06,
+      "loss": 0.0254,
+      "step": 15643
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8739834288634237e-06,
+      "loss": 0.0044,
+      "step": 15644
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8694305854045355e-06,
+      "loss": 0.0134,
+      "step": 15645
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8648832445618934e-06,
+      "loss": 0.0315,
+      "step": 15646
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8603414065044452e-06,
+      "loss": 0.011,
+      "step": 15647
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8558050714009066e-06,
+      "loss": 0.0041,
+      "step": 15648
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8512742394197588e-06,
+      "loss": 0.0176,
+      "step": 15649
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8467489107293509e-06,
+      "loss": 0.0457,
+      "step": 15650
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8422290854977484e-06,
+      "loss": 0.0271,
+      "step": 15651
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8377147638928668e-06,
+      "loss": 0.0361,
+      "step": 15652
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8332059460824223e-06,
+      "loss": 0.0315,
+      "step": 15653
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8287026322338471e-06,
+      "loss": 0.0286,
+      "step": 15654
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8242048225144912e-06,
+      "loss": 0.0276,
+      "step": 15655
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8197125170913873e-06,
+      "loss": 0.0426,
+      "step": 15656
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8152257161314355e-06,
+      "loss": 0.0145,
+      "step": 15657
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8107444198013022e-06,
+      "loss": 0.0337,
+      "step": 15658
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8062686282674376e-06,
+      "loss": 0.0283,
+      "step": 15659
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.8017983416961424e-06,
+      "loss": 0.0312,
+      "step": 15660
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7973335602534334e-06,
+      "loss": 0.0221,
+      "step": 15661
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7928742841051946e-06,
+      "loss": 0.0274,
+      "step": 15662
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7884205134170437e-06,
+      "loss": 0.0317,
+      "step": 15663
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7839722483544483e-06,
+      "loss": 0.0039,
+      "step": 15664
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7795294890826594e-06,
+      "loss": 0.0195,
+      "step": 15665
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7750922357666786e-06,
+      "loss": 0.0208,
+      "step": 15666
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7706604885713738e-06,
+      "loss": 0.0342,
+      "step": 15667
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.76623424766133e-06,
+      "loss": 0.0132,
+      "step": 15668
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7618135132010158e-06,
+      "loss": 0.0424,
+      "step": 15669
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7573982853546164e-06,
+      "loss": 0.0513,
+      "step": 15670
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7529885642861508e-06,
+      "loss": 0.0458,
+      "step": 15671
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.748584350159421e-06,
+      "loss": 0.0398,
+      "step": 15672
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.744185643138063e-06,
+      "loss": 0.0404,
+      "step": 15673
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7397924433854293e-06,
+      "loss": 0.0291,
+      "step": 15674
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7354047510647395e-06,
+      "loss": 0.0047,
+      "step": 15675
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7310225663389965e-06,
+      "loss": 0.0357,
+      "step": 15676
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7266458893709535e-06,
+      "loss": 0.0125,
+      "step": 15677
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7222747203232134e-06,
+      "loss": 0.0129,
+      "step": 15678
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7179090593581302e-06,
+      "loss": 0.0251,
+      "step": 15679
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7135489066379237e-06,
+      "loss": 0.0169,
+      "step": 15680
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7091942623245147e-06,
+      "loss": 0.0151,
+      "step": 15681
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7048451265796737e-06,
+      "loss": 0.0304,
+      "step": 15682
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7005014995649713e-06,
+      "loss": 0.0297,
+      "step": 15683
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6961633814417453e-06,
+      "loss": 0.0179,
+      "step": 15684
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.69183077237115e-06,
+      "loss": 0.0328,
+      "step": 15685
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.68750367251414e-06,
+      "loss": 0.0312,
+      "step": 15686
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6831820820314533e-06,
+      "loss": 0.0382,
+      "step": 15687
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6788660010836118e-06,
+      "loss": 0.0361,
+      "step": 15688
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6745554298309372e-06,
+      "loss": 0.025,
+      "step": 15689
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6702503684335844e-06,
+      "loss": 0.0332,
+      "step": 15690
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6659508170514424e-06,
+      "loss": 0.0276,
+      "step": 15691
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6616567758442501e-06,
+      "loss": 0.0085,
+      "step": 15692
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.6573682449715131e-06,
+      "loss": 0.0118,
+      "step": 15693
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6530852245925375e-06,
+      "loss": 0.0284,
+      "step": 15694
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.648807714866396e-06,
+      "loss": 0.0513,
+      "step": 15695
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6445357159520444e-06,
+      "loss": 0.0217,
+      "step": 15696
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.640269228008123e-06,
+      "loss": 0.0233,
+      "step": 15697
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6360082511931382e-06,
+      "loss": 0.0224,
+      "step": 15698
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.63175278566538e-06,
+      "loss": 0.0343,
+      "step": 15699
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.627502831582922e-06,
+      "loss": 0.0129,
+      "step": 15700
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6232583891036378e-06,
+      "loss": 0.0107,
+      "step": 15701
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6190194583851846e-06,
+      "loss": 0.0152,
+      "step": 15702
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6147860395850364e-06,
+      "loss": 0.0185,
+      "step": 15703
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.610558132860451e-06,
+      "loss": 0.022,
+      "step": 15704
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.606335738368486e-06,
+      "loss": 0.022,
+      "step": 15705
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.6021188562659826e-06,
+      "loss": 0.0145,
+      "step": 15706
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.597907486709582e-06,
+      "loss": 0.0237,
+      "step": 15707
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5937016298557593e-06,
+      "loss": 0.0168,
+      "step": 15708
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5895012858606893e-06,
+      "loss": 0.0118,
+      "step": 15709
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5853064548804639e-06,
+      "loss": 0.0171,
+      "step": 15710
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5811171370708753e-06,
+      "loss": 0.0356,
+      "step": 15711
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5769333325875487e-06,
+      "loss": 0.0412,
+      "step": 15712
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5727550415858936e-06,
+      "loss": 0.0337,
+      "step": 15713
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5685822642211355e-06,
+      "loss": 0.0122,
+      "step": 15714
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.564415000648267e-06,
+      "loss": 0.0093,
+      "step": 15715
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5602532510221144e-06,
+      "loss": 0.018,
+      "step": 15716
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5560970154972542e-06,
+      "loss": 0.0295,
+      "step": 15717
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5519462942280625e-06,
+      "loss": 0.0097,
+      "step": 15718
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5478010873687662e-06,
+      "loss": 0.0198,
+      "step": 15719
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.543661395073309e-06,
+      "loss": 0.0297,
+      "step": 15720
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.539527217495501e-06,
+      "loss": 0.0331,
+      "step": 15721
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5353985547889026e-06,
+      "loss": 0.0223,
+      "step": 15722
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5312754071068745e-06,
+      "loss": 0.0194,
+      "step": 15723
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5271577746025776e-06,
+      "loss": 0.0222,
+      "step": 15724
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5230456574289728e-06,
+      "loss": 0.0135,
+      "step": 15725
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5189390557388381e-06,
+      "loss": 0.0397,
+      "step": 15726
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5148379696846845e-06,
+      "loss": 0.0232,
+      "step": 15727
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5107423994188738e-06,
+      "loss": 0.0419,
+      "step": 15728
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5066523450935342e-06,
+      "loss": 0.0126,
+      "step": 15729
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5025678068606273e-06,
+      "loss": 0.0428,
+      "step": 15730
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4984887848718486e-06,
+      "loss": 0.0436,
+      "step": 15731
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4944152792787267e-06,
+      "loss": 0.0332,
+      "step": 15732
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4903472902326075e-06,
+      "loss": 0.0241,
+      "step": 15733
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4862848178845698e-06,
+      "loss": 0.0289,
+      "step": 15734
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4822278623855432e-06,
+      "loss": 0.019,
+      "step": 15735
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4781764238862237e-06,
+      "loss": 0.0173,
+      "step": 15736
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4741305025371242e-06,
+      "loss": 0.0212,
+      "step": 15737
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4700900984885078e-06,
+      "loss": 0.0541,
+      "step": 15738
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4660552118904877e-06,
+      "loss": 0.0178,
+      "step": 15739
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.462025842892961e-06,
+      "loss": 0.0107,
+      "step": 15740
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.458001991645591e-06,
+      "loss": 0.0275,
+      "step": 15741
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4539836582978415e-06,
+      "loss": 0.0265,
+      "step": 15742
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4499708429989765e-06,
+      "loss": 0.0307,
+      "step": 15743
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4459635458980934e-06,
+      "loss": 0.02,
+      "step": 15744
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4419617671440397e-06,
+      "loss": 0.0259,
+      "step": 15745
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4379655068854468e-06,
+      "loss": 0.0348,
+      "step": 15746
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4339747652707956e-06,
+      "loss": 0.0292,
+      "step": 15747
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.429989542448301e-06,
+      "loss": 0.0312,
+      "step": 15748
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.426009838566028e-06,
+      "loss": 0.0324,
+      "step": 15749
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4220356537717914e-06,
+      "loss": 0.0369,
+      "step": 15750
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.41806698821324e-06,
+      "loss": 0.0378,
+      "step": 15751
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4141038420377892e-06,
+      "loss": 0.0263,
+      "step": 15752
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.410146215392638e-06,
+      "loss": 0.0118,
+      "step": 15753
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.406194108424835e-06,
+      "loss": 0.0215,
+      "step": 15754
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.4022475212811635e-06,
+      "loss": 0.0322,
+      "step": 15755
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.3983064541082389e-06,
+      "loss": 0.0221,
+      "step": 15756
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.3943709070524612e-06,
+      "loss": 0.0361,
+      "step": 15757
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.3904408802600297e-06,
+      "loss": 0.0138,
+      "step": 15758
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.3865163738769114e-06,
+      "loss": 0.0134,
+      "step": 15759
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3825973880489062e-06,
+      "loss": 0.0153,
+      "step": 15760
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3786839229215974e-06,
+      "loss": 0.033,
+      "step": 15761
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3747759786403523e-06,
+      "loss": 0.0265,
+      "step": 15762
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.370873555350338e-06,
+      "loss": 0.0155,
+      "step": 15763
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3669766531965054e-06,
+      "loss": 0.0152,
+      "step": 15764
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3630852723236552e-06,
+      "loss": 0.0194,
+      "step": 15765
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3591994128762884e-06,
+      "loss": 0.0425,
+      "step": 15766
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3553190749988063e-06,
+      "loss": 0.042,
+      "step": 15767
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3514442588352936e-06,
+      "loss": 0.0222,
+      "step": 15768
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.347574964529735e-06,
+      "loss": 0.0242,
+      "step": 15769
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3437111922258326e-06,
+      "loss": 0.0226,
+      "step": 15770
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3398529420671378e-06,
+      "loss": 0.0527,
+      "step": 15771
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3360002141969694e-06,
+      "loss": 0.019,
+      "step": 15772
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3321530087584298e-06,
+      "loss": 0.0254,
+      "step": 15773
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3283113258944545e-06,
+      "loss": 0.0535,
+      "step": 15774
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3244751657477293e-06,
+      "loss": 0.0296,
+      "step": 15775
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.320644528460757e-06,
+      "loss": 0.021,
+      "step": 15776
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3168194141758571e-06,
+      "loss": 0.0276,
+      "step": 15777
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.312999823035099e-06,
+      "loss": 0.0176,
+      "step": 15778
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3091857551803697e-06,
+      "loss": 0.0124,
+      "step": 15779
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3053772107533555e-06,
+      "loss": 0.0253,
+      "step": 15780
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.3015741898955434e-06,
+      "loss": 0.0284,
+      "step": 15781
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2977766927482037e-06,
+      "loss": 0.0231,
+      "step": 15782
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.29398471945239e-06,
+      "loss": 0.0398,
+      "step": 15783
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2901982701489567e-06,
+      "loss": 0.0335,
+      "step": 15784
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.286417344978574e-06,
+      "loss": 0.023,
+      "step": 15785
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2826419440816804e-06,
+      "loss": 0.0259,
+      "step": 15786
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2788720675985465e-06,
+      "loss": 0.0257,
+      "step": 15787
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.275107715669177e-06,
+      "loss": 0.0202,
+      "step": 15788
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2713488884334267e-06,
+      "loss": 0.0259,
+      "step": 15789
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2675955860309339e-06,
+      "loss": 0.0141,
+      "step": 15790
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.263847808601104e-06,
+      "loss": 0.0323,
+      "step": 15791
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2601055562831752e-06,
+      "loss": 0.0267,
+      "step": 15792
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.25636882921612e-06,
+      "loss": 0.0594,
+      "step": 15793
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.252637627538794e-06,
+      "loss": 0.009,
+      "step": 15794
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.248911951389786e-06,
+      "loss": 0.0303,
+      "step": 15795
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2451918009074858e-06,
+      "loss": 0.0192,
+      "step": 15796
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2414771762300824e-06,
+      "loss": 0.0309,
+      "step": 15797
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2377680774955822e-06,
+      "loss": 0.0207,
+      "step": 15798
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2340645048417586e-06,
+      "loss": 0.0209,
+      "step": 15799
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.230366458406168e-06,
+      "loss": 0.0304,
+      "step": 15800
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.226673938326217e-06,
+      "loss": 0.0296,
+      "step": 15801
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.222986944739046e-06,
+      "loss": 0.0418,
+      "step": 15802
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.219305477781629e-06,
+      "loss": 0.0408,
+      "step": 15803
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2156295375907233e-06,
+      "loss": 0.0296,
+      "step": 15804
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2119591243028526e-06,
+      "loss": 0.0171,
+      "step": 15805
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.208294238054408e-06,
+      "loss": 0.0308,
+      "step": 15806
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2046348789814808e-06,
+      "loss": 0.0263,
+      "step": 15807
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2009810472200621e-06,
+      "loss": 0.0207,
+      "step": 15808
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1973327429058265e-06,
+      "loss": 0.0485,
+      "step": 15809
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.193689966174316e-06,
+      "loss": 0.0443,
+      "step": 15810
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1900527171608553e-06,
+      "loss": 0.0228,
+      "step": 15811
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1864209960005532e-06,
+      "loss": 0.038,
+      "step": 15812
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1827948028283352e-06,
+      "loss": 0.0135,
+      "step": 15813
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1791741377788599e-06,
+      "loss": 0.0093,
+      "step": 15814
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.17555900098667e-06,
+      "loss": 0.0088,
+      "step": 15815
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1719493925860413e-06,
+      "loss": 0.0171,
+      "step": 15816
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1683453127110498e-06,
+      "loss": 0.0435,
+      "step": 15817
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1647467614955885e-06,
+      "loss": 0.0275,
+      "step": 15818
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1611537390733338e-06,
+      "loss": 0.0164,
+      "step": 15819
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1575662455777457e-06,
+      "loss": 0.024,
+      "step": 15820
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.153984281142084e-06,
+      "loss": 0.0257,
+      "step": 15821
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1504078458994425e-06,
+      "loss": 0.032,
+      "step": 15822
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1468369399826315e-06,
+      "loss": 0.027,
+      "step": 15823
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1432715635243284e-06,
+      "loss": 0.0253,
+      "step": 15824
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.139711716656977e-06,
+      "loss": 0.035,
+      "step": 15825
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1361573995127882e-06,
+      "loss": 0.0219,
+      "step": 15826
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1326086122238232e-06,
+      "loss": 0.0061,
+      "step": 15827
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.129065354921893e-06,
+      "loss": 0.0219,
+      "step": 15828
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.125527627738626e-06,
+      "loss": 0.0541,
+      "step": 15829
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1219954308054336e-06,
+      "loss": 0.0369,
+      "step": 15830
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.118468764253544e-06,
+      "loss": 0.0517,
+      "step": 15831
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1149476282139358e-06,
+      "loss": 0.0377,
+      "step": 15832
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1114320228174378e-06,
+      "loss": 0.0196,
+      "step": 15833
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1079219481945957e-06,
+      "loss": 0.0229,
+      "step": 15834
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1044174044758546e-06,
+      "loss": 0.0132,
+      "step": 15835
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1009183917913778e-06,
+      "loss": 0.0263,
+      "step": 15836
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0974249102711274e-06,
+      "loss": 0.015,
+      "step": 15837
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0939369600449e-06,
+      "loss": 0.0144,
+      "step": 15838
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0904545412422249e-06,
+      "loss": 0.0105,
+      "step": 15839
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0869776539925156e-06,
+      "loss": 0.0175,
+      "step": 15840
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.083506298424902e-06,
+      "loss": 0.0201,
+      "step": 15841
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0800404746683311e-06,
+      "loss": 0.0241,
+      "step": 15842
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0765801828515496e-06,
+      "loss": 0.0334,
+      "step": 15843
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0731254231030883e-06,
+      "loss": 0.0261,
+      "step": 15844
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0696761955513111e-06,
+      "loss": 0.005,
+      "step": 15845
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0662325003243156e-06,
+      "loss": 0.0043,
+      "step": 15846
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0627943375500491e-06,
+      "loss": 0.0067,
+      "step": 15847
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0593617073562266e-06,
+      "loss": 0.0108,
+      "step": 15848
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.055934609870329e-06,
+      "loss": 0.0249,
+      "step": 15849
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.052513045219705e-06,
+      "loss": 0.0331,
+      "step": 15850
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0490970135314193e-06,
+      "loss": 0.0049,
+      "step": 15851
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0456865149324035e-06,
+      "loss": 0.0096,
+      "step": 15852
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0422815495493064e-06,
+      "loss": 0.0157,
+      "step": 15853
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0388821175086604e-06,
+      "loss": 0.015,
+      "step": 15854
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0354882189366974e-06,
+      "loss": 0.0151,
+      "step": 15855
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.032099853959517e-06,
+      "loss": 0.0066,
+      "step": 15856
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0287170227029852e-06,
+      "loss": 0.0374,
+      "step": 15857
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.0253397252927676e-06,
+      "loss": 0.0134,
+      "step": 15858
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.021967961854314e-06,
+      "loss": 0.0224,
+      "step": 15859
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0186017325128749e-06,
+      "loss": 0.0271,
+      "step": 15860
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0152410373934827e-06,
+      "loss": 0.0,
+      "step": 15861
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0118858766210048e-06,
+      "loss": 0.0024,
+      "step": 15862
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0085362503200744e-06,
+      "loss": 0.0212,
+      "step": 15863
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0051921586150758e-06,
+      "loss": 0.0182,
+      "step": 15864
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.0018536016302925e-06,
+      "loss": 0.0303,
+      "step": 15865
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.985205794896923e-07,
+      "loss": 0.0308,
+      "step": 15866
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.95193092317126e-07,
+      "loss": 0.012,
+      "step": 15867
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.918711402361946e-07,
+      "loss": 0.0151,
+      "step": 15868
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.885547233702662e-07,
+      "loss": 0.0513,
+      "step": 15869
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.85243841842559e-07,
+      "loss": 0.031,
+      "step": 15870
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.819384957760578e-07,
+      "loss": 0.0171,
+      "step": 15871
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.786386852935645e-07,
+      "loss": 0.0352,
+      "step": 15872
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.75344410517631e-07,
+      "loss": 0.0309,
+      "step": 15873
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.720556715706597e-07,
+      "loss": 0.0336,
+      "step": 15874
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.687724685747856e-07,
+      "loss": 0.018,
+      "step": 15875
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.654948016519782e-07,
+      "loss": 0.042,
+      "step": 15876
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.622226709240066e-07,
+      "loss": 0.0543,
+      "step": 15877
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.58956076512407e-07,
+      "loss": 0.0235,
+      "step": 15878
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.556950185385492e-07,
+      "loss": 0.0108,
+      "step": 15879
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.524394971235194e-07,
+      "loss": 0.0149,
+      "step": 15880
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.491895123883042e-07,
+      "loss": 0.0279,
+      "step": 15881
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.459450644536071e-07,
+      "loss": 0.0322,
+      "step": 15882
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.427061534399649e-07,
+      "loss": 0.0158,
+      "step": 15883
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.394727794676815e-07,
+      "loss": 0.0229,
+      "step": 15884
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.362449426568441e-07,
+      "loss": 0.0295,
+      "step": 15885
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.330226431274069e-07,
+      "loss": 0.0398,
+      "step": 15886
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.298058809990238e-07,
+      "loss": 0.0368,
+      "step": 15887
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.265946563912163e-07,
+      "loss": 0.0448,
+      "step": 15888
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.233889694232721e-07,
+      "loss": 0.0553,
+      "step": 15889
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.20188820214246e-07,
+      "loss": 0.0292,
+      "step": 15890
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.169942088830429e-07,
+      "loss": 0.022,
+      "step": 15891
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.138051355483178e-07,
+      "loss": 0.0288,
+      "step": 15892
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.106216003285427e-07,
+      "loss": 0.0324,
+      "step": 15893
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.07443603341973e-07,
+      "loss": 0.0273,
+      "step": 15894
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.042711447066809e-07,
+      "loss": 0.0083,
+      "step": 15895
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.01104224540472e-07,
+      "loss": 0.0378,
+      "step": 15896
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.979428429610191e-07,
+      "loss": 0.0172,
+      "step": 15897
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.947870000857449e-07,
+      "loss": 0.0281,
+      "step": 15898
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.916366960319055e-07,
+      "loss": 0.0135,
+      "step": 15899
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.884919309164906e-07,
+      "loss": 0.0226,
+      "step": 15900
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.853527048563402e-07,
+      "loss": 0.0113,
+      "step": 15901
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.822190179680611e-07,
+      "loss": 0.0366,
+      "step": 15902
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.7909087036806e-07,
+      "loss": 0.0485,
+      "step": 15903
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.759682621725439e-07,
+      "loss": 0.0332,
+      "step": 15904
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.728511934975036e-07,
+      "loss": 0.0012,
+      "step": 15905
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.697396644587295e-07,
+      "loss": 0.0229,
+      "step": 15906
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.666336751717961e-07,
+      "loss": 0.0265,
+      "step": 15907
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.635332257520944e-07,
+      "loss": 0.0082,
+      "step": 15908
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.604383163148154e-07,
+      "loss": 0.0224,
+      "step": 15909
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.57348946974884e-07,
+      "loss": 0.0232,
+      "step": 15910
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.54265117847075e-07,
+      "loss": 0.0408,
+      "step": 15911
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.511868290459634e-07,
+      "loss": 0.0146,
+      "step": 15912
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.481140806858911e-07,
+      "loss": 0.0126,
+      "step": 15913
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.450468728809833e-07,
+      "loss": 0.0404,
+      "step": 15914
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.419852057451992e-07,
+      "loss": 0.015,
+      "step": 15915
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.389290793922643e-07,
+      "loss": 0.0276,
+      "step": 15916
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.358784939357044e-07,
+      "loss": 0.0351,
+      "step": 15917
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.32833449488829e-07,
+      "loss": 0.0253,
+      "step": 15918
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.29793946164764e-07,
+      "loss": 0.029,
+      "step": 15919
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.267599840764194e-07,
+      "loss": 0.0161,
+      "step": 15920
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.237315633364883e-07,
+      "loss": 0.0482,
+      "step": 15921
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.207086840574806e-07,
+      "loss": 0.0322,
+      "step": 15922
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.176913463516732e-07,
+      "loss": 0.0126,
+      "step": 15923
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 8.146795503311765e-07,
+      "loss": 0.0437,
+      "step": 15924
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.116732961078343e-07,
+      "loss": 0.0291,
+      "step": 15925
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.086725837933405e-07,
+      "loss": 0.0348,
+      "step": 15926
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.056774134991562e-07,
+      "loss": 0.0286,
+      "step": 15927
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 8.026877853365587e-07,
+      "loss": 0.0113,
+      "step": 15928
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.997036994165928e-07,
+      "loss": 0.0235,
+      "step": 15929
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.967251558501031e-07,
+      "loss": 0.0173,
+      "step": 15930
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.937521547477343e-07,
+      "loss": 0.0222,
+      "step": 15931
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.907846962199316e-07,
+      "loss": 0.0334,
+      "step": 15932
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.878227803769399e-07,
+      "loss": 0.0467,
+      "step": 15933
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.848664073287548e-07,
+      "loss": 0.0349,
+      "step": 15934
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.819155771852215e-07,
+      "loss": 0.0204,
+      "step": 15935
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.789702900559358e-07,
+      "loss": 0.025,
+      "step": 15936
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.760305460503269e-07,
+      "loss": 0.038,
+      "step": 15937
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.73096345277574e-07,
+      "loss": 0.0104,
+      "step": 15938
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.7016768784669e-07,
+      "loss": 0.0156,
+      "step": 15939
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.672445738664879e-07,
+      "loss": 0.0272,
+      "step": 15940
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.643270034454973e-07,
+      "loss": 0.0158,
+      "step": 15941
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.614149766921485e-07,
+      "loss": 0.0161,
+      "step": 15942
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.58508493714588e-07,
+      "loss": 0.0347,
+      "step": 15943
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.556075546207796e-07,
+      "loss": 0.0263,
+      "step": 15944
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.527121595185204e-07,
+      "loss": 0.0237,
+      "step": 15945
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.498223085153243e-07,
+      "loss": 0.0191,
+      "step": 15946
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.469380017185555e-07,
+      "loss": 0.0001,
+      "step": 15947
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.440592392353617e-07,
+      "loss": 0.0215,
+      "step": 15948
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.411860211726739e-07,
+      "loss": 0.036,
+      "step": 15949
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.383183476372401e-07,
+      "loss": 0.0189,
+      "step": 15950
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.354562187355584e-07,
+      "loss": 0.0162,
+      "step": 15951
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.325996345739771e-07,
+      "loss": 0.0156,
+      "step": 15952
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.29748595258578e-07,
+      "loss": 0.0185,
+      "step": 15953
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.269031008953097e-07,
+      "loss": 0.0347,
+      "step": 15954
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.240631515898377e-07,
+      "loss": 0.0008,
+      "step": 15955
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.212287474476774e-07,
+      "loss": 0.0281,
+      "step": 15956
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 7.183998885740949e-07,
+      "loss": 0.0099,
+      "step": 15957
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.155765750742227e-07,
+      "loss": 0.0313,
+      "step": 15958
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.127588070528767e-07,
+      "loss": 0.0558,
+      "step": 15959
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.099465846147734e-07,
+      "loss": 0.0271,
+      "step": 15960
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.071399078643791e-07,
+      "loss": 0.0051,
+      "step": 15961
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.043387769059106e-07,
+      "loss": 0.0124,
+      "step": 15962
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 7.015431918434844e-07,
+      "loss": 0.0306,
+      "step": 15963
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.987531527808843e-07,
+      "loss": 0.017,
+      "step": 15964
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.959686598218106e-07,
+      "loss": 0.0345,
+      "step": 15965
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.931897130696473e-07,
+      "loss": 0.0365,
+      "step": 15966
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.904163126276451e-07,
+      "loss": 0.0096,
+      "step": 15967
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.876484585988384e-07,
+      "loss": 0.0276,
+      "step": 15968
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.848861510860448e-07,
+      "loss": 0.0082,
+      "step": 15969
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.821293901918657e-07,
+      "loss": 0.0482,
+      "step": 15970
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.793781760187022e-07,
+      "loss": 0.0222,
+      "step": 15971
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.766325086687563e-07,
+      "loss": 0.0284,
+      "step": 15972
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.738923882440295e-07,
+      "loss": 0.0457,
+      "step": 15973
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.711578148463237e-07,
+      "loss": 0.0315,
+      "step": 15974
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.684287885771745e-07,
+      "loss": 0.0296,
+      "step": 15975
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.657053095380005e-07,
+      "loss": 0.0238,
+      "step": 15976
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.629873778299543e-07,
+      "loss": 0.0094,
+      "step": 15977
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.602749935539886e-07,
+      "loss": 0.0179,
+      "step": 15978
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.575681568108893e-07,
+      "loss": 0.032,
+      "step": 15979
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.548668677011759e-07,
+      "loss": 0.0267,
+      "step": 15980
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.521711263252183e-07,
+      "loss": 0.0269,
+      "step": 15981
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.494809327831196e-07,
+      "loss": 0.005,
+      "step": 15982
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.467962871748666e-07,
+      "loss": 0.0299,
+      "step": 15983
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.441171896001462e-07,
+      "loss": 0.0092,
+      "step": 15984
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.414436401584788e-07,
+      "loss": 0.0277,
+      "step": 15985
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.387756389492016e-07,
+      "loss": 0.0153,
+      "step": 15986
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.361131860714185e-07,
+      "loss": 0.0429,
+      "step": 15987
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.334562816240174e-07,
+      "loss": 0.0492,
+      "step": 15988
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.308049257056857e-07,
+      "loss": 0.0559,
+      "step": 15989
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.281591184149615e-07,
+      "loss": 0.0572,
+      "step": 15990
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.25518859850066e-07,
+      "loss": 0.0227,
+      "step": 15991
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.228841501091208e-07,
+      "loss": 0.0135,
+      "step": 15992
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.202549892899977e-07,
+      "loss": 0.0118,
+      "step": 15993
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.176313774903186e-07,
+      "loss": 0.0139,
+      "step": 15994
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.150133148076053e-07,
+      "loss": 0.0305,
+      "step": 15995
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.124008013390636e-07,
+      "loss": 0.0362,
+      "step": 15996
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.097938371817657e-07,
+      "loss": 0.0293,
+      "step": 15997
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.071924224325342e-07,
+      "loss": 0.0519,
+      "step": 15998
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.045965571880085e-07,
+      "loss": 0.0269,
+      "step": 15999
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 6.020062415446281e-07,
+      "loss": 0.0264,
+      "step": 16000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.99421475598616e-07,
+      "loss": 0.0086,
+      "step": 16001
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.968422594459788e-07,
+      "loss": 0.0337,
+      "step": 16002
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.942685931825231e-07,
+      "loss": 0.035,
+      "step": 16003
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.917004769038725e-07,
+      "loss": 0.0386,
+      "step": 16004
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.891379107054007e-07,
+      "loss": 0.0071,
+      "step": 16005
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.865808946823314e-07,
+      "loss": 0.0323,
+      "step": 16006
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.840294289296221e-07,
+      "loss": 0.0124,
+      "step": 16007
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.814835135420471e-07,
+      "loss": 0.0408,
+      "step": 16008
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.789431486141971e-07,
+      "loss": 0.0316,
+      "step": 16009
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.764083342404469e-07,
+      "loss": 0.0071,
+      "step": 16010
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.738790705149543e-07,
+      "loss": 0.0407,
+      "step": 16011
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.713553575316442e-07,
+      "loss": 0.0235,
+      "step": 16012
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.688371953842919e-07,
+      "loss": 0.0379,
+      "step": 16013
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.663245841664388e-07,
+      "loss": 0.0412,
+      "step": 16014
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.638175239714105e-07,
+      "loss": 0.0305,
+      "step": 16015
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.61316014892349e-07,
+      "loss": 0.0542,
+      "step": 16016
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.588200570221634e-07,
+      "loss": 0.0249,
+      "step": 16017
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.563296504535964e-07,
+      "loss": 0.0173,
+      "step": 16018
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.538447952791236e-07,
+      "loss": 0.0256,
+      "step": 16019
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.513654915910881e-07,
+      "loss": 0.0306,
+      "step": 16020
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.488917394815661e-07,
+      "loss": 0.0327,
+      "step": 16021
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.46423539042451e-07,
+      "loss": 0.0165,
+      "step": 16022
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.439608903654357e-07,
+      "loss": 0.0315,
+      "step": 16023
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.415037935419975e-07,
+      "loss": 0.0146,
+      "step": 16024
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.390522486633964e-07,
+      "loss": 0.0315,
+      "step": 16025
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.366062558207262e-07,
+      "loss": 0.0111,
+      "step": 16026
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.341658151048311e-07,
+      "loss": 0.024,
+      "step": 16027
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.317309266063718e-07,
+      "loss": 0.0119,
+      "step": 16028
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.293015904157927e-07,
+      "loss": 0.0126,
+      "step": 16029
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.268778066233382e-07,
+      "loss": 0.0012,
+      "step": 16030
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.244595753190695e-07,
+      "loss": 0.015,
+      "step": 16031
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.220468965927815e-07,
+      "loss": 0.0362,
+      "step": 16032
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.196397705341027e-07,
+      "loss": 0.0122,
+      "step": 16033
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.172381972324613e-07,
+      "loss": 0.0213,
+      "step": 16034
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.148421767770694e-07,
+      "loss": 0.0306,
+      "step": 16035
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.124517092569225e-07,
+      "loss": 0.0207,
+      "step": 16036
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.100667947608328e-07,
+      "loss": 0.0144,
+      "step": 16037
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.076874333773961e-07,
+      "loss": 0.0354,
+      "step": 16038
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.053136251949752e-07,
+      "loss": 0.0315,
+      "step": 16039
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.029453703017661e-07,
+      "loss": 0.0286,
+      "step": 16040
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.005826687857317e-07,
+      "loss": 0.0443,
+      "step": 16041
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.98225520734652e-07,
+      "loss": 0.0245,
+      "step": 16042
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.958739262360901e-07,
+      "loss": 0.013,
+      "step": 16043
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.935278853773927e-07,
+      "loss": 0.0233,
+      "step": 16044
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.911873982457071e-07,
+      "loss": 0.0287,
+      "step": 16045
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.888524649279968e-07,
+      "loss": 0.0319,
+      "step": 16046
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.865230855109592e-07,
+      "loss": 0.0297,
+      "step": 16047
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.841992600811418e-07,
+      "loss": 0.0308,
+      "step": 16048
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.818809887248754e-07,
+      "loss": 0.0269,
+      "step": 16049
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.795682715282911e-07,
+      "loss": 0.0062,
+      "step": 16050
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.772611085772537e-07,
+      "loss": 0.0276,
+      "step": 16051
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.7495949995751124e-07,
+      "loss": 0.013,
+      "step": 16052
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.7266344575452864e-07,
+      "loss": 0.014,
+      "step": 16053
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.703729460536376e-07,
+      "loss": 0.0281,
+      "step": 16054
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 4.680880009398702e-07,
+      "loss": 0.0449,
+      "step": 16055
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6580861049815845e-07,
+      "loss": 0.0303,
+      "step": 16056
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.635347748131513e-07,
+      "loss": 0.0143,
+      "step": 16057
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6126649396929785e-07,
+      "loss": 0.01,
+      "step": 16058
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.5900376805088066e-07,
+      "loss": 0.0305,
+      "step": 16059
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.567465971419493e-07,
+      "loss": 0.042,
+      "step": 16060
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.5449498132636985e-07,
+      "loss": 0.0278,
+      "step": 16061
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.522489206877422e-07,
+      "loss": 0.0114,
+      "step": 16062
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.500084153095329e-07,
+      "loss": 0.0677,
+      "step": 16063
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.477734652749421e-07,
+      "loss": 0.0495,
+      "step": 16064
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.455440706670199e-07,
+      "loss": 0.0349,
+      "step": 16065
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.4332023156856686e-07,
+      "loss": 0.0087,
+      "step": 16066
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.4110194806220025e-07,
+      "loss": 0.0368,
+      "step": 16067
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.388892202303207e-07,
+      "loss": 0.0377,
+      "step": 16068
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.366820481551125e-07,
+      "loss": 0.0236,
+      "step": 16069
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.344804319185935e-07,
+      "loss": 0.0166,
+      "step": 16070
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.3228437160251484e-07,
+      "loss": 0.0221,
+      "step": 16071
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.300938672884613e-07,
+      "loss": 0.0304,
+      "step": 16072
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.2790891905781775e-07,
+      "loss": 0.0055,
+      "step": 16073
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.2572952699175267e-07,
+      "loss": 0.0446,
+      "step": 16074
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.2355569117120124e-07,
+      "loss": 0.0297,
+      "step": 16075
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.2138741167693224e-07,
+      "loss": 0.0219,
+      "step": 16076
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1922468858948123e-07,
+      "loss": 0.0346,
+      "step": 16077
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.170675219891839e-07,
+      "loss": 0.0175,
+      "step": 16078
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.149159119561929e-07,
+      "loss": 0.0387,
+      "step": 16079
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.127698585704109e-07,
+      "loss": 0.0297,
+      "step": 16080
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1062936191157413e-07,
+      "loss": 0.0214,
+      "step": 16081
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.0849442205920234e-07,
+      "loss": 0.0071,
+      "step": 16082
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.0636503909256546e-07,
+      "loss": 0.0172,
+      "step": 16083
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.0424121309080016e-07,
+      "loss": 0.0313,
+      "step": 16084
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.021229441327767e-07,
+      "loss": 0.0437,
+      "step": 16085
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.0001023229719876e-07,
+      "loss": 0.0284,
+      "step": 16086
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.979030776625536e-07,
+      "loss": 0.027,
+      "step": 16087
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.9580148030709524e-07,
+      "loss": 0.0187,
+      "step": 16088
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.9370544030891125e-07,
+      "loss": 0.0325,
+      "step": 16089
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.916149577458394e-07,
+      "loss": 0.0112,
+      "step": 16090
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.895300326955508e-07,
+      "loss": 0.0128,
+      "step": 16091
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.874506652355003e-07,
+      "loss": 0.0343,
+      "step": 16092
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.853768554429093e-07,
+      "loss": 0.0186,
+      "step": 16093
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.8330860339483294e-07,
+      "loss": 0.0138,
+      "step": 16094
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.812459091681097e-07,
+      "loss": 0.0114,
+      "step": 16095
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.791887728393284e-07,
+      "loss": 0.0491,
+      "step": 16096
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.771371944849277e-07,
+      "loss": 0.0593,
+      "step": 16097
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.750911741811302e-07,
+      "loss": 0.0066,
+      "step": 16098
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.730507120039083e-07,
+      "loss": 0.0407,
+      "step": 16099
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.710158080290848e-07,
+      "loss": 0.043,
+      "step": 16100
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.6898646233223253e-07,
+      "loss": 0.0258,
+      "step": 16101
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.669626749887411e-07,
+      "loss": 0.0284,
+      "step": 16102
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.6494444607380046e-07,
+      "loss": 0.031,
+      "step": 16103
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.629317756623673e-07,
+      "loss": 0.0419,
+      "step": 16104
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.6092466382921514e-07,
+      "loss": 0.0257,
+      "step": 16105
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.5892311064888435e-07,
+      "loss": 0.0364,
+      "step": 16106
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.5692711619576544e-07,
+      "loss": 0.0321,
+      "step": 16107
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.549366805439657e-07,
+      "loss": 0.0247,
+      "step": 16108
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.529518037674428e-07,
+      "loss": 0.0207,
+      "step": 16109
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.509724859399043e-07,
+      "loss": 0.0259,
+      "step": 16110
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.4899872713492483e-07,
+      "loss": 0.0267,
+      "step": 16111
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.4703052742577896e-07,
+      "loss": 0.0181,
+      "step": 16112
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.450678868855916e-07,
+      "loss": 0.0189,
+      "step": 16113
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.4311080558728777e-07,
+      "loss": 0.0534,
+      "step": 16114
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.411592836035426e-07,
+      "loss": 0.0457,
+      "step": 16115
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3921332100686484e-07,
+      "loss": 0.0068,
+      "step": 16116
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.372729178695299e-07,
+      "loss": 0.0497,
+      "step": 16117
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3533807426363026e-07,
+      "loss": 0.023,
+      "step": 16118
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.33408790261025e-07,
+      "loss": 0.0225,
+      "step": 16119
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.314850659333901e-07,
+      "loss": 0.0172,
+      "step": 16120
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.295669013521851e-07,
+      "loss": 0.0292,
+      "step": 16121
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.276542965886697e-07,
+      "loss": 0.0304,
+      "step": 16122
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.2574725171388704e-07,
+      "loss": 0.0135,
+      "step": 16123
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.2384576679868047e-07,
+      "loss": 0.0291,
+      "step": 16124
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.219498419136768e-07,
+      "loss": 0.0443,
+      "step": 16125
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.20059477129303e-07,
+      "loss": 0.0292,
+      "step": 16126
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.181746725157863e-07,
+      "loss": 0.0402,
+      "step": 16127
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.16295428143154e-07,
+      "loss": 0.0355,
+      "step": 16128
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.1442174408120024e-07,
+      "loss": 0.0127,
+      "step": 16129
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.12553620399536e-07,
+      "loss": 0.027,
+      "step": 16130
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.1069105716753915e-07,
+      "loss": 0.0343,
+      "step": 16131
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.0883405445440436e-07,
+      "loss": 0.0218,
+      "step": 16132
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.0698261232912636e-07,
+      "loss": 0.0407,
+      "step": 16133
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.051367308604835e-07,
+      "loss": 0.0053,
+      "step": 16134
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.03296410117021e-07,
+      "loss": 0.024,
+      "step": 16135
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.0146165016711743e-07,
+      "loss": 0.0104,
+      "step": 16136
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.996324510789183e-07,
+      "loss": 0.044,
+      "step": 16137
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.978088129203859e-07,
+      "loss": 0.0176,
+      "step": 16138
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.959907357592661e-07,
+      "loss": 0.0086,
+      "step": 16139
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.941782196630882e-07,
+      "loss": 0.025,
+      "step": 16140
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.923712646991649e-07,
+      "loss": 0.0327,
+      "step": 16141
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.9056987093464267e-07,
+      "loss": 0.027,
+      "step": 16142
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.887740384364345e-07,
+      "loss": 0.0297,
+      "step": 16143
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.8698376727125383e-07,
+      "loss": 0.0423,
+      "step": 16144
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.851990575055807e-07,
+      "loss": 0.0126,
+      "step": 16145
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.8341990920572876e-07,
+      "loss": 0.0279,
+      "step": 16146
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.8164632243779516e-07,
+      "loss": 0.0197,
+      "step": 16147
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.798782972676605e-07,
+      "loss": 0.0316,
+      "step": 16148
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.78115833760989e-07,
+      "loss": 0.0149,
+      "step": 16149
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.7635893198326153e-07,
+      "loss": 0.0604,
+      "step": 16150
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.7460759199972593e-07,
+      "loss": 0.0647,
+      "step": 16151
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.728618138754635e-07,
+      "loss": 0.0153,
+      "step": 16152
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.711215976753056e-07,
+      "loss": 0.0503,
+      "step": 16153
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.6938694346391733e-07,
+      "loss": 0.0501,
+      "step": 16154
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.676578513057137e-07,
+      "loss": 0.0146,
+      "step": 16155
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6593432126492676e-07,
+      "loss": 0.0112,
+      "step": 16156
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.642163534056052e-07,
+      "loss": 0.0243,
+      "step": 16157
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.6250394779154805e-07,
+      "loss": 0.0399,
+      "step": 16158
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.607971044863544e-07,
+      "loss": 0.0234,
+      "step": 16159
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.590958235534568e-07,
+      "loss": 0.0374,
+      "step": 16160
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.574001050560215e-07,
+      "loss": 0.0316,
+      "step": 16161
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5570994905706464e-07,
+      "loss": 0.0349,
+      "step": 16162
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5402535561936943e-07,
+      "loss": 0.0191,
+      "step": 16163
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.523463248054858e-07,
+      "loss": 0.0251,
+      "step": 16164
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.506728566778138e-07,
+      "loss": 0.0426,
+      "step": 16165
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.490049512985204e-07,
+      "loss": 0.0071,
+      "step": 16166
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.473426087295394e-07,
+      "loss": 0.0233,
+      "step": 16167
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.45685829032638e-07,
+      "loss": 0.0326,
+      "step": 16168
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4403461226936706e-07,
+      "loss": 0.0427,
+      "step": 16169
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.423889585010441e-07,
+      "loss": 0.0333,
+      "step": 16170
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.4074886778880367e-07,
+      "loss": 0.0083,
+      "step": 16171
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.391143401935969e-07,
+      "loss": 0.0263,
+      "step": 16172
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.374853757761086e-07,
+      "loss": 0.0088,
+      "step": 16173
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3586197459685706e-07,
+      "loss": 0.0162,
+      "step": 16174
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3424413671616073e-07,
+      "loss": 0.0198,
+      "step": 16175
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3263186219412144e-07,
+      "loss": 0.036,
+      "step": 16176
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.3102515109060805e-07,
+      "loss": 0.0362,
+      "step": 16177
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2942400346532275e-07,
+      "loss": 0.0192,
+      "step": 16178
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2782841937771802e-07,
+      "loss": 0.0488,
+      "step": 16179
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2623839888711304e-07,
+      "loss": 0.0259,
+      "step": 16180
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.246539420525273e-07,
+      "loss": 0.0333,
+      "step": 16181
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2307504893283034e-07,
+      "loss": 0.0165,
+      "step": 16182
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.2150171958667528e-07,
+      "loss": 0.0131,
+      "step": 16183
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.19933954072532e-07,
+      "loss": 0.0514,
+      "step": 16184
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.1837175244860394e-07,
+      "loss": 0.0156,
+      "step": 16185
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.16815114772928e-07,
+      "loss": 0.0152,
+      "step": 16186
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.1526404110334128e-07,
+      "loss": 0.0283,
+      "step": 16187
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.1371853149744766e-07,
+      "loss": 0.0283,
+      "step": 16188
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.1217858601268455e-07,
+      "loss": 0.0385,
+      "step": 16189
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.106442047062229e-07,
+      "loss": 0.0407,
+      "step": 16190
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0911538763508374e-07,
+      "loss": 0.0266,
+      "step": 16191
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0759213485603832e-07,
+      "loss": 0.0241,
+      "step": 16192
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0607444642569136e-07,
+      "loss": 0.0211,
+      "step": 16193
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0456232240041448e-07,
+      "loss": 0.0096,
+      "step": 16194
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.030557628363627e-07,
+      "loss": 0.0164,
+      "step": 16195
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.015547677895246e-07,
+      "loss": 0.0269,
+      "step": 16196
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.0005933731563895e-07,
+      "loss": 0.0373,
+      "step": 16197
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9856947147026125e-07,
+      "loss": 0.0205,
+      "step": 16198
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9708517030874727e-07,
+      "loss": 0.0128,
+      "step": 16199
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.956064338862362e-07,
+      "loss": 0.029,
+      "step": 16200
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9413326225763416e-07,
+      "loss": 0.01,
+      "step": 16201
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9266565547768064e-07,
+      "loss": 0.0127,
+      "step": 16202
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.912036136008821e-07,
+      "loss": 0.033,
+      "step": 16203
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8974713668157838e-07,
+      "loss": 0.0415,
+      "step": 16204
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8829622477384287e-07,
+      "loss": 0.0355,
+      "step": 16205
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.868508779315825e-07,
+      "loss": 0.0255,
+      "step": 16206
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.854110962084876e-07,
+      "loss": 0.0216,
+      "step": 16207
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8397687965804874e-07,
+      "loss": 0.0341,
+      "step": 16208
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8254822833353998e-07,
+      "loss": 0.0177,
+      "step": 16209
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.8112514228801888e-07,
+      "loss": 0.0365,
+      "step": 16210
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7970762157435982e-07,
+      "loss": 0.0152,
+      "step": 16211
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7829566624522062e-07,
+      "loss": 0.0439,
+      "step": 16212
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7688927635305938e-07,
+      "loss": 0.0542,
+      "step": 16213
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7548845195010096e-07,
+      "loss": 0.0094,
+      "step": 16214
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.740931930883871e-07,
+      "loss": 0.0224,
+      "step": 16215
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.7270349981974297e-07,
+      "loss": 0.0242,
+      "step": 16216
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.713193721958106e-07,
+      "loss": 0.026,
+      "step": 16217
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.699408102679989e-07,
+      "loss": 0.0436,
+      "step": 16218
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.685678140875002e-07,
+      "loss": 0.0139,
+      "step": 16219
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6720038370532373e-07,
+      "loss": 0.0319,
+      "step": 16220
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.658385191722955e-07,
+      "loss": 0.0075,
+      "step": 16221
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6448222053895844e-07,
+      "loss": 0.0098,
+      "step": 16222
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6313148785573883e-07,
+      "loss": 0.0195,
+      "step": 16223
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.617863211727799e-07,
+      "loss": 0.0299,
+      "step": 16224
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6044672054005836e-07,
+      "loss": 0.044,
+      "step": 16225
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5911268600735106e-07,
+      "loss": 0.0353,
+      "step": 16226
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5778421762420169e-07,
+      "loss": 0.0208,
+      "step": 16227
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.564613154399541e-07,
+      "loss": 0.0059,
+      "step": 16228
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5514397950375234e-07,
+      "loss": 0.0152,
+      "step": 16229
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5383220986454058e-07,
+      "loss": 0.0357,
+      "step": 16230
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5252600657104653e-07,
+      "loss": 0.0142,
+      "step": 16231
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5122536967178135e-07,
+      "loss": 0.0476,
+      "step": 16232
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.499302992150564e-07,
+      "loss": 0.0096,
+      "step": 16233
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.486407952490165e-07,
+      "loss": 0.0272,
+      "step": 16234
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4735685782150673e-07,
+      "loss": 0.0227,
+      "step": 16235
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4607848698027224e-07,
+      "loss": 0.0422,
+      "step": 16236
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.448056827727584e-07,
+      "loss": 0.0271,
+      "step": 16237
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4353844524627734e-07,
+      "loss": 0.0221,
+      "step": 16238
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4227677444789143e-07,
+      "loss": 0.0399,
+      "step": 16239
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.4102067042447984e-07,
+      "loss": 0.039,
+      "step": 16240
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.397701332226886e-07,
+      "loss": 0.0196,
+      "step": 16241
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3852516288896387e-07,
+      "loss": 0.0148,
+      "step": 16242
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3728575946956865e-07,
+      "loss": 0.0359,
+      "step": 16243
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3605192301054947e-07,
+      "loss": 0.0449,
+      "step": 16244
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3482365355770296e-07,
+      "loss": 0.0194,
+      "step": 16245
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3360095115670931e-07,
+      "loss": 0.0257,
+      "step": 16246
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3238381585294888e-07,
+      "loss": 0.0417,
+      "step": 16247
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.3117224769165214e-07,
+      "loss": 0.024,
+      "step": 16248
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.299662467178164e-07,
+      "loss": 0.0332,
+      "step": 16249
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2876581297623923e-07,
+      "loss": 0.0443,
+      "step": 16250
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2757094651151823e-07,
+      "loss": 0.0194,
+      "step": 16251
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2638164736805124e-07,
+      "loss": 0.0321,
+      "step": 16252
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2519791558998626e-07,
+      "loss": 0.0118,
+      "step": 16253
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2401975122133812e-07,
+      "loss": 0.0113,
+      "step": 16254
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2284715430583847e-07,
+      "loss": 0.0373,
+      "step": 16255
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.216801248870525e-07,
+      "loss": 0.0067,
+      "step": 16256
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2051866300832879e-07,
+      "loss": 0.0469,
+      "step": 16257
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.193627687128329e-07,
+      "loss": 0.0142,
+      "step": 16258
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1821244204348047e-07,
+      "loss": 0.0177,
+      "step": 16259
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.170676830430206e-07,
+      "loss": 0.0104,
+      "step": 16260
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.159284917539527e-07,
+      "loss": 0.0272,
+      "step": 16261
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1479486821862615e-07,
+      "loss": 0.0326,
+      "step": 16262
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1366681247912402e-07,
+      "loss": 0.0656,
+      "step": 16263
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1254432457736273e-07,
+      "loss": 0.0201,
+      "step": 16264
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1142740455504229e-07,
+      "loss": 0.0325,
+      "step": 16265
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.1031605245364616e-07,
+      "loss": 0.0157,
+      "step": 16266
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.09210268314458e-07,
+      "loss": 0.0117,
+      "step": 16267
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0811005217856161e-07,
+      "loss": 0.0085,
+      "step": 16268
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.070154040868243e-07,
+      "loss": 0.0399,
+      "step": 16269
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0592632407991352e-07,
+      "loss": 0.0412,
+      "step": 16270
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0484281219826363e-07,
+      "loss": 0.0399,
+      "step": 16271
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0376486848215904e-07,
+      "loss": 0.0437,
+      "step": 16272
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.026924929716011e-07,
+      "loss": 0.0372,
+      "step": 16273
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.016256857064579e-07,
+      "loss": 0.0366,
+      "step": 16274
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0056444672634778e-07,
+      "loss": 0.0371,
+      "step": 16275
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.950877607068919e-08,
+      "loss": 0.0573,
+      "step": 16276
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.845867377870076e-08,
+      "loss": 0.0304,
+      "step": 16277
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.741413988938462e-08,
+      "loss": 0.0123,
+      "step": 16278
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.637517444155973e-08,
+      "loss": 0.0244,
+      "step": 16279
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.534177747379523e-08,
+      "loss": 0.0462,
+      "step": 16280
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.431394902451039e-08,
+      "loss": 0.0361,
+      "step": 16281
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.329168913184138e-08,
+      "loss": 0.0342,
+      "step": 16282
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.227499783380776e-08,
+      "loss": 0.0156,
+      "step": 16283
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.126387516814604e-08,
+      "loss": 0.027,
+      "step": 16284
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.02583211724428e-08,
+      "loss": 0.0382,
+      "step": 16285
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.92583358840182e-08,
+      "loss": 0.0285,
+      "step": 16286
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.826391934004251e-08,
+      "loss": 0.0261,
+      "step": 16287
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.727507157745284e-08,
+      "loss": 0.0117,
+      "step": 16288
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.629179263298647e-08,
+      "loss": 0.0323,
+      "step": 16289
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.531408254314753e-08,
+      "loss": 0.0283,
+      "step": 16290
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.434194134429029e-08,
+      "loss": 0.045,
+      "step": 16291
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.337536907248588e-08,
+      "loss": 0.0264,
+      "step": 16292
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.241436576365556e-08,
+      "loss": 0.0449,
+      "step": 16293
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.145893145352078e-08,
+      "loss": 0.0116,
+      "step": 16294
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 8.05090661775365e-08,
+      "loss": 0.0151,
+      "step": 16295
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.956476997102446e-08,
+      "loss": 0.0275,
+      "step": 16296
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.86260428690233e-08,
+      "loss": 0.0229,
+      "step": 16297
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.769288490643843e-08,
+      "loss": 0.0395,
+      "step": 16298
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.676529611790883e-08,
+      "loss": 0.0243,
+      "step": 16299
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.584327653790689e-08,
+      "loss": 0.0219,
+      "step": 16300
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.492682620067192e-08,
+      "loss": 0.014,
+      "step": 16301
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.401594514025999e-08,
+      "loss": 0.0364,
+      "step": 16302
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.311063339051071e-08,
+      "loss": 0.0136,
+      "step": 16303
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.221089098503052e-08,
+      "loss": 0.0198,
+      "step": 16304
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.131671795725935e-08,
+      "loss": 0.0157,
+      "step": 16305
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 7.042811434042061e-08,
+      "loss": 0.0308,
+      "step": 16306
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.95450801675046e-08,
+      "loss": 0.0232,
+      "step": 16307
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.866761547131838e-08,
+      "loss": 0.0115,
+      "step": 16308
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.779572028445257e-08,
+      "loss": 0.0334,
+      "step": 16309
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.692939463931457e-08,
+      "loss": 0.0529,
+      "step": 16310
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.606863856807865e-08,
+      "loss": 0.0051,
+      "step": 16311
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.521345210270257e-08,
+      "loss": 0.0468,
+      "step": 16312
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.436383527497758e-08,
+      "loss": 0.0279,
+      "step": 16313
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.351978811644509e-08,
+      "loss": 0.0347,
+      "step": 16314
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.268131065848003e-08,
+      "loss": 0.0661,
+      "step": 16315
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.18484029322075e-08,
+      "loss": 0.03,
+      "step": 16316
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.10210649685694e-08,
+      "loss": 0.0149,
+      "step": 16317
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 6.019929679832447e-08,
+      "loss": 0.0199,
+      "step": 16318
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.938309845196498e-08,
+      "loss": 0.0107,
+      "step": 16319
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.8572469959833335e-08,
+      "loss": 0.0307,
+      "step": 16320
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.776741135203877e-08,
+      "loss": 0.0171,
+      "step": 16321
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.6967922658474055e-08,
+      "loss": 0.0242,
+      "step": 16322
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.617400390884874e-08,
+      "loss": 0.037,
+      "step": 16323
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.538565513265592e-08,
+      "loss": 0.0266,
+      "step": 16324
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.4602876359172155e-08,
+      "loss": 0.0206,
+      "step": 16325
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.3825667617490855e-08,
+      "loss": 0.017,
+      "step": 16326
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.305402893645561e-08,
+      "loss": 0.0156,
+      "step": 16327
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.228796034476013e-08,
+      "loss": 0.0091,
+      "step": 16328
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.152746187084833e-08,
+      "loss": 0.0203,
+      "step": 16329
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.0772533542964286e-08,
+      "loss": 0.05,
+      "step": 16330
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 5.002317538916889e-08,
+      "loss": 0.0528,
+      "step": 16331
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.9279387437273223e-08,
+      "loss": 0.025,
+      "step": 16332
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.8541169714938485e-08,
+      "loss": 0.0347,
+      "step": 16333
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.7808522249559446e-08,
+      "loss": 0.0272,
+      "step": 16334
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.708144506836431e-08,
+      "loss": 0.0304,
+      "step": 16335
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.6359938198348156e-08,
+      "loss": 0.029,
+      "step": 16336
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.564400166633952e-08,
+      "loss": 0.027,
+      "step": 16337
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.493363549891715e-08,
+      "loss": 0.0165,
+      "step": 16338
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.42288397224766e-08,
+      "loss": 0.0317,
+      "step": 16339
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.3529614363180255e-08,
+      "loss": 0.0119,
+      "step": 16340
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.2835959447024004e-08,
+      "loss": 0.02,
+      "step": 16341
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.214787499977057e-08,
+      "loss": 0.0106,
+      "step": 16342
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.1465361046966184e-08,
+      "loss": 0.0376,
+      "step": 16343
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.078841761397389e-08,
+      "loss": 0.0221,
+      "step": 16344
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 4.011704472594024e-08,
+      "loss": 0.0429,
+      "step": 16345
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.9451242407811945e-08,
+      "loss": 0.0195,
+      "step": 16346
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.879101068430257e-08,
+      "loss": 0.0228,
+      "step": 16347
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.8136349579959145e-08,
+      "loss": 0.0146,
+      "step": 16348
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.7487259119078906e-08,
+      "loss": 0.0244,
+      "step": 16349
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6843739325809193e-08,
+      "loss": 0.0388,
+      "step": 16350
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.620579022401427e-08,
+      "loss": 0.0355,
+      "step": 16351
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.5573411837408475e-08,
+      "loss": 0.0431,
+      "step": 16352
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.494660418948969e-08,
+      "loss": 0.0333,
+      "step": 16353
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.432536730352264e-08,
+      "loss": 0.0348,
+      "step": 16354
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.370970120260552e-08,
+      "loss": 0.0151,
+      "step": 16355
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3099605909603365e-08,
+      "loss": 0.0239,
+      "step": 16356
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.249508144718138e-08,
+      "loss": 0.0218,
+      "step": 16357
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.189612783778828e-08,
+      "loss": 0.0364,
+      "step": 16358
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.130274510367292e-08,
+      "loss": 0.0173,
+      "step": 16359
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.071493326688435e-08,
+      "loss": 0.0387,
+      "step": 16360
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.013269234923843e-08,
+      "loss": 0.0396,
+      "step": 16361
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.955602237240118e-08,
+      "loss": 0.0271,
+      "step": 16362
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.8984923357755486e-08,
+      "loss": 0.0328,
+      "step": 16363
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.841939532655102e-08,
+      "loss": 0.0347,
+      "step": 16364
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.7859438299770997e-08,
+      "loss": 0.039,
+      "step": 16365
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.7305052298215448e-08,
+      "loss": 0.021,
+      "step": 16366
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.6756237342501208e-08,
+      "loss": 0.0297,
+      "step": 16367
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.621299345297867e-08,
+      "loss": 0.0534,
+      "step": 16368
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5675320649864995e-08,
+      "loss": 0.0271,
+      "step": 16369
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.514321895311089e-08,
+      "loss": 0.0304,
+      "step": 16370
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4616688382500526e-08,
+      "loss": 0.025,
+      "step": 16371
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.4095728957568284e-08,
+      "loss": 0.0366,
+      "step": 16372
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3580340697682e-08,
+      "loss": 0.019,
+      "step": 16373
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.307052362199302e-08,
+      "loss": 0.0013,
+      "step": 16374
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2566277749419548e-08,
+      "loss": 0.0369,
+      "step": 16375
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.2067603098696595e-08,
+      "loss": 0.0253,
+      "step": 16376
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.157449968837599e-08,
+      "loss": 0.0202,
+      "step": 16377
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.1086967536743105e-08,
+      "loss": 0.0184,
+      "step": 16378
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.060500666191678e-08,
+      "loss": 0.0261,
+      "step": 16379
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.012861708179936e-08,
+      "loss": 0.0106,
+      "step": 16380
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.9657798814110006e-08,
+      "loss": 0.0282,
+      "step": 16381
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.919255187630142e-08,
+      "loss": 0.0112,
+      "step": 16382
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8732876285676436e-08,
+      "loss": 0.0038,
+      "step": 16383
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.8278772059321378e-08,
+      "loss": 0.0168,
+      "step": 16384
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.7830239214089435e-08,
+      "loss": 0.019,
+      "step": 16385
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7387277766633957e-08,
+      "loss": 0.0452,
+      "step": 16386
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6949887733425093e-08,
+      "loss": 0.0132,
+      "step": 16387
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6518069130716515e-08,
+      "loss": 0.0231,
+      "step": 16388
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.609182197452874e-08,
+      "loss": 0.0208,
+      "step": 16389
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5671146280699097e-08,
+      "loss": 0.0387,
+      "step": 16390
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5256042064881736e-08,
+      "loss": 0.0085,
+      "step": 16391
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.484650934246434e-08,
+      "loss": 0.02,
+      "step": 16392
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4442548128668074e-08,
+      "loss": 0.0337,
+      "step": 16393
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4044158438497599e-08,
+      "loss": 0.015,
+      "step": 16394
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.3651340286757739e-08,
+      "loss": 0.0185,
+      "step": 16395
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.326409368805348e-08,
+      "loss": 0.0352,
+      "step": 16396
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2882418656740001e-08,
+      "loss": 0.032,
+      "step": 16397
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.250631520702261e-08,
+      "loss": 0.0123,
+      "step": 16398
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2135783352856809e-08,
+      "loss": 0.0192,
+      "step": 16399
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1770823108014915e-08,
+      "loss": 0.0232,
+      "step": 16400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1411434486036098e-08,
+      "loss": 0.037,
+      "step": 16401
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.1057617500309646e-08,
+      "loss": 0.0178,
+      "step": 16402
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.070937216392509e-08,
+      "loss": 0.0426,
+      "step": 16403
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0366698489872038e-08,
+      "loss": 0.07,
+      "step": 16404
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0029596490840341e-08,
+      "loss": 0.0066,
+      "step": 16405
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.698066179369968e-09,
+      "loss": 0.0195,
+      "step": 16406
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.372107567767738e-09,
+      "loss": 0.0255,
+      "step": 16407
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.051720668157291e-09,
+      "loss": 0.0061,
+      "step": 16408
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.736905492429114e-09,
+      "loss": 0.0323,
+      "step": 16409
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.427662052273853e-09,
+      "loss": 0.009,
+      "step": 16410
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.123990359198973e-09,
+      "loss": 0.032,
+      "step": 16411
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.825890424462133e-09,
+      "loss": 0.0343,
+      "step": 16412
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.533362259137809e-09,
+      "loss": 0.0417,
+      "step": 16413
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.246405874100636e-09,
+      "loss": 0.0342,
+      "step": 16414
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.9650212800087536e-09,
+      "loss": 0.0264,
+      "step": 16415
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.689208487320463e-09,
+      "loss": 0.0325,
+      "step": 16416
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.418967506277573e-09,
+      "loss": 0.0315,
+      "step": 16417
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.1542983469220486e-09,
+      "loss": 0.0216,
+      "step": 16418
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.8952010190793654e-09,
+      "loss": 0.0148,
+      "step": 16419
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.6416755323585025e-09,
+      "loss": 0.0209,
+      "step": 16420
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.393721896218561e-09,
+      "loss": 0.0096,
+      "step": 16421
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.151340119835534e-09,
+      "loss": 0.0221,
+      "step": 16422
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.914530212218881e-09,
+      "loss": 0.0253,
+      "step": 16423
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.683292182178222e-09,
+      "loss": 0.0459,
+      "step": 16424
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.45762603829003e-09,
+      "loss": 0.0214,
+      "step": 16425
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.23753178894759e-09,
+      "loss": 0.0225,
+      "step": 16426
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.023009442311042e-09,
+      "loss": 0.0192,
+      "step": 16427
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.8140590063739925e-09,
+      "loss": 0.0476,
+      "step": 16428
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.6106804888802465e-09,
+      "loss": 0.0324,
+      "step": 16429
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.4128738973904222e-09,
+      "loss": 0.0044,
+      "step": 16430
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.220639239248646e-09,
+      "loss": 0.0141,
+      "step": 16431
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.0339765215992016e-09,
+      "loss": 0.0135,
+      "step": 16432
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.852885751369882e-09,
+      "loss": 0.007,
+      "step": 16433
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.6773669352886384e-09,
+      "loss": 0.0095,
+      "step": 16434
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5074200799002354e-09,
+      "loss": 0.0364,
+      "step": 16435
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.343045191482984e-09,
+      "loss": 0.0335,
+      "step": 16436
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.184242276148662e-09,
+      "loss": 0.0219,
+      "step": 16437
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.031011339825861e-09,
+      "loss": 0.0142,
+      "step": 16438
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.8833523881767175e-09,
+      "loss": 0.011,
+      "step": 16439
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.7412654266801828e-09,
+      "loss": 0.0068,
+      "step": 16440
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.6047504606486738e-09,
+      "loss": 0.0472,
+      "step": 16441
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4738074951281543e-09,
+      "loss": 0.0435,
+      "step": 16442
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3484365349980542e-09,
+      "loss": 0.0266,
+      "step": 16443
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2286375848880038e-09,
+      "loss": 0.0189,
+      "step": 16444
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1144106492777527e-09,
+      "loss": 0.0172,
+      "step": 16445
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.005755732397251e-09,
+      "loss": 0.0138,
+      "step": 16446
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.02672838276608e-10,
+      "loss": 0.0243,
+      "step": 16447
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.051619707627466e-10,
+      "loss": 0.0116,
+      "step": 16448
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.132231334694427e-10,
+      "loss": 0.024,
+      "step": 16449
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.268563298106321e-10,
+      "loss": 0.0286,
+      "step": 16450
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.460615629837572e-10,
+      "loss": 0.0314,
+      "step": 16451
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.708388360030735e-10,
+      "loss": 0.0269,
+      "step": 16452
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.011881516663429e-10,
+      "loss": 0.0308,
+      "step": 16453
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.37109512554834e-10,
+      "loss": 0.0575,
+      "step": 16454
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.786029210499752e-10,
+      "loss": 0.0372,
+      "step": 16455
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.2566837931670133e-10,
+      "loss": 0.0347,
+      "step": 16456
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.783058893367606e-10,
+      "loss": 0.0179,
+      "step": 16457
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.3651545287540754e-10,
+      "loss": 0.0193,
+      "step": 16458
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0029707144809662e-10,
+      "loss": 0.0301,
+      "step": 16459
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.965074643705549e-11,
+      "loss": 0.006,
+      "step": 16460
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.457647895805827e-11,
+      "loss": 0.0242,
+      "step": 16461
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5074269960345672e-11,
+      "loss": 0.0493,
+      "step": 16462
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.1144120160011538e-11,
+      "loss": 0.0081,
+      "step": 16463
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.78603005665623e-12,
+      "loss": 0.023,
+      "step": 16464
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.0655,
+      "step": 16465
+    },
+    {
+      "epoch": 5.0,
+      "step": 16465,
+      "total_flos": 5.860962792017756e+18,
+      "train_loss": 0.12529179257165163,
+      "train_runtime": 58502.878,
+      "train_samples_per_second": 2.252,
+      "train_steps_per_second": 0.281
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 16465,
+  "num_train_epochs": 5,
+  "save_steps": 0,
+  "total_flos": 5.860962792017756e+18,
+  "trial_name": null,
+  "trial_params": null
+}