diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,48738 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.0,
+  "eval_steps": 500,
+  "global_step": 8118,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.19672131147541e-07,
+      "loss": 0.4607,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.639344262295082e-06,
+      "loss": 0.4573,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.459016393442623e-06,
+      "loss": 0.4556,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.278688524590164e-06,
+      "loss": 0.4229,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.098360655737704e-06,
+      "loss": 0.4536,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.918032786885246e-06,
+      "loss": 0.4268,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.737704918032787e-06,
+      "loss": 0.4227,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.557377049180328e-06,
+      "loss": 0.4336,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3770491803278695e-06,
+      "loss": 0.4346,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.196721311475409e-06,
+      "loss": 0.4112,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.016393442622952e-06,
+      "loss": 0.3927,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.836065573770493e-06,
+      "loss": 0.3602,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0655737704918032e-05,
+      "loss": 0.3566,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1475409836065575e-05,
+      "loss": 0.3105,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2295081967213116e-05,
+      "loss": 0.3259,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3114754098360657e-05,
+      "loss": 0.2707,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3934426229508196e-05,
+      "loss": 0.269,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4754098360655739e-05,
+      "loss": 0.2458,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.557377049180328e-05,
+      "loss": 0.2265,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6393442622950818e-05,
+      "loss": 0.209,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.721311475409836e-05,
+      "loss": 0.1851,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8032786885245903e-05,
+      "loss": 0.1827,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8852459016393442e-05,
+      "loss": 0.1657,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9672131147540985e-05,
+      "loss": 0.1578,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0491803278688525e-05,
+      "loss": 0.1471,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1311475409836064e-05,
+      "loss": 0.1364,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2131147540983607e-05,
+      "loss": 0.1307,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.295081967213115e-05,
+      "loss": 0.1232,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.377049180327869e-05,
+      "loss": 0.1207,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.459016393442623e-05,
+      "loss": 0.113,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.540983606557377e-05,
+      "loss": 0.1065,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6229508196721314e-05,
+      "loss": 0.1084,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7049180327868856e-05,
+      "loss": 0.1084,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7868852459016392e-05,
+      "loss": 0.1059,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8688524590163935e-05,
+      "loss": 0.1004,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9508196721311478e-05,
+      "loss": 0.1012,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0327868852459017e-05,
+      "loss": 0.0987,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.114754098360656e-05,
+      "loss": 0.0958,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.19672131147541e-05,
+      "loss": 0.0947,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2786885245901635e-05,
+      "loss": 0.0946,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.360655737704918e-05,
+      "loss": 0.0896,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.442622950819672e-05,
+      "loss": 0.0949,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.524590163934427e-05,
+      "loss": 0.0947,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6065573770491806e-05,
+      "loss": 0.0892,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6885245901639346e-05,
+      "loss": 0.0872,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7704918032786885e-05,
+      "loss": 0.088,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8524590163934424e-05,
+      "loss": 0.0917,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.934426229508197e-05,
+      "loss": 0.0903,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.016393442622951e-05,
+      "loss": 0.0902,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.098360655737705e-05,
+      "loss": 0.0873,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1803278688524595e-05,
+      "loss": 0.0898,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.262295081967213e-05,
+      "loss": 0.0843,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3442622950819674e-05,
+      "loss": 0.0842,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.426229508196721e-05,
+      "loss": 0.0861,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.508196721311476e-05,
+      "loss": 0.0873,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.59016393442623e-05,
+      "loss": 0.0871,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.672131147540984e-05,
+      "loss": 0.0858,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.754098360655738e-05,
+      "loss": 0.0842,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.836065573770492e-05,
+      "loss": 0.0854,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.918032786885246e-05,
+      "loss": 0.0835,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5e-05,
+      "loss": 0.0812,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.081967213114754e-05,
+      "loss": 0.0831,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.163934426229509e-05,
+      "loss": 0.0796,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.245901639344263e-05,
+      "loss": 0.0796,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.327868852459017e-05,
+      "loss": 0.0843,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.409836065573771e-05,
+      "loss": 0.0836,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.491803278688525e-05,
+      "loss": 0.0821,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.5737704918032785e-05,
+      "loss": 0.0861,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.6557377049180324e-05,
+      "loss": 0.0777,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.737704918032787e-05,
+      "loss": 0.0806,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.819672131147541e-05,
+      "loss": 0.0809,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9016393442622956e-05,
+      "loss": 0.0793,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9836065573770495e-05,
+      "loss": 0.0795,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0655737704918034e-05,
+      "loss": 0.081,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.147540983606557e-05,
+      "loss": 0.0788,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.229508196721313e-05,
+      "loss": 0.0804,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.311475409836067e-05,
+      "loss": 0.0809,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.39344262295082e-05,
+      "loss": 0.0801,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.475409836065574e-05,
+      "loss": 0.0781,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.557377049180327e-05,
+      "loss": 0.0789,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.639344262295082e-05,
+      "loss": 0.08,
+      "step": 81
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.721311475409836e-05,
+      "loss": 0.0798,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.80327868852459e-05,
+      "loss": 0.0776,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.885245901639344e-05,
+      "loss": 0.076,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.967213114754098e-05,
+      "loss": 0.0828,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.049180327868853e-05,
+      "loss": 0.078,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.131147540983607e-05,
+      "loss": 0.0775,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.213114754098361e-05,
+      "loss": 0.0782,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.295081967213115e-05,
+      "loss": 0.0777,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.377049180327869e-05,
+      "loss": 0.0754,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.459016393442624e-05,
+      "loss": 0.0845,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.540983606557377e-05,
+      "loss": 0.076,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.622950819672131e-05,
+      "loss": 0.0767,
+      "step": 93
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.704918032786885e-05,
+      "loss": 0.0785,
+      "step": 94
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.78688524590164e-05,
+      "loss": 0.0796,
+      "step": 95
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.868852459016394e-05,
+      "loss": 0.0802,
+      "step": 96
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.950819672131148e-05,
+      "loss": 0.0787,
+      "step": 97
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.032786885245902e-05,
+      "loss": 0.0798,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.114754098360656e-05,
+      "loss": 0.0774,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.19672131147541e-05,
+      "loss": 0.0747,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.278688524590165e-05,
+      "loss": 0.0794,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.360655737704919e-05,
+      "loss": 0.076,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.442622950819673e-05,
+      "loss": 0.0772,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.524590163934426e-05,
+      "loss": 0.0779,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.606557377049181e-05,
+      "loss": 0.0804,
+      "step": 105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.688524590163935e-05,
+      "loss": 0.0793,
+      "step": 106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.770491803278689e-05,
+      "loss": 0.0794,
+      "step": 107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.852459016393443e-05,
+      "loss": 0.0771,
+      "step": 108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.934426229508197e-05,
+      "loss": 0.08,
+      "step": 109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.016393442622952e-05,
+      "loss": 0.0771,
+      "step": 110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.098360655737706e-05,
+      "loss": 0.0736,
+      "step": 111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.18032786885246e-05,
+      "loss": 0.078,
+      "step": 112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.262295081967214e-05,
+      "loss": 0.0769,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.344262295081968e-05,
+      "loss": 0.0746,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.426229508196722e-05,
+      "loss": 0.0754,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.508196721311476e-05,
+      "loss": 0.0774,
+      "step": 116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.59016393442623e-05,
+      "loss": 0.0773,
+      "step": 117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.672131147540983e-05,
+      "loss": 0.0774,
+      "step": 118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.754098360655737e-05,
+      "loss": 0.0763,
+      "step": 119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.836065573770493e-05,
+      "loss": 0.0776,
+      "step": 120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918032786885247e-05,
+      "loss": 0.0771,
+      "step": 121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001,
+      "loss": 0.0777,
+      "step": 122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00010081967213114753,
+      "loss": 0.0766,
+      "step": 123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00010163934426229508,
+      "loss": 0.078,
+      "step": 124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00010245901639344262,
+      "loss": 0.0744,
+      "step": 125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00010327868852459018,
+      "loss": 0.0785,
+      "step": 126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001040983606557377,
+      "loss": 0.0779,
+      "step": 127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00010491803278688525,
+      "loss": 0.0749,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001057377049180328,
+      "loss": 0.0753,
+      "step": 129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010655737704918035,
+      "loss": 0.0772,
+      "step": 130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010737704918032787,
+      "loss": 0.078,
+      "step": 131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010819672131147543,
+      "loss": 0.0757,
+      "step": 132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00010901639344262295,
+      "loss": 0.0787,
+      "step": 133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001098360655737705,
+      "loss": 0.0789,
+      "step": 134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011065573770491804,
+      "loss": 0.0771,
+      "step": 135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011147540983606557,
+      "loss": 0.0778,
+      "step": 136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011229508196721312,
+      "loss": 0.0777,
+      "step": 137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011311475409836065,
+      "loss": 0.079,
+      "step": 138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001139344262295082,
+      "loss": 0.078,
+      "step": 139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011475409836065574,
+      "loss": 0.0789,
+      "step": 140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001155737704918033,
+      "loss": 0.0769,
+      "step": 141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011639344262295082,
+      "loss": 0.0806,
+      "step": 142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011721311475409837,
+      "loss": 0.078,
+      "step": 143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011803278688524591,
+      "loss": 0.0782,
+      "step": 144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011885245901639346,
+      "loss": 0.0734,
+      "step": 145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011967213114754099,
+      "loss": 0.0772,
+      "step": 146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00012049180327868852,
+      "loss": 0.0773,
+      "step": 147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00012131147540983607,
+      "loss": 0.0767,
+      "step": 148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001221311475409836,
+      "loss": 0.0768,
+      "step": 149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00012295081967213115,
+      "loss": 0.0752,
+      "step": 150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001237704918032787,
+      "loss": 0.0737,
+      "step": 151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00012459016393442625,
+      "loss": 0.0798,
+      "step": 152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00012540983606557378,
+      "loss": 0.0768,
+      "step": 153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00012622950819672133,
+      "loss": 0.0772,
+      "step": 154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00012704918032786886,
+      "loss": 0.0767,
+      "step": 155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001278688524590164,
+      "loss": 0.077,
+      "step": 156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00012868852459016394,
+      "loss": 0.0756,
+      "step": 157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001295081967213115,
+      "loss": 0.077,
+      "step": 158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013032786885245902,
+      "loss": 0.0763,
+      "step": 159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013114754098360654,
+      "loss": 0.0759,
+      "step": 160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001319672131147541,
+      "loss": 0.0748,
+      "step": 161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013278688524590165,
+      "loss": 0.0767,
+      "step": 162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001336065573770492,
+      "loss": 0.0769,
+      "step": 163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013442622950819673,
+      "loss": 0.0773,
+      "step": 164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013524590163934428,
+      "loss": 0.0764,
+      "step": 165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001360655737704918,
+      "loss": 0.0756,
+      "step": 166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013688524590163936,
+      "loss": 0.077,
+      "step": 167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013770491803278688,
+      "loss": 0.0786,
+      "step": 168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00013852459016393444,
+      "loss": 0.0757,
+      "step": 169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00013934426229508196,
+      "loss": 0.0772,
+      "step": 170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014016393442622951,
+      "loss": 0.0821,
+      "step": 171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014098360655737707,
+      "loss": 0.0772,
+      "step": 172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001418032786885246,
+      "loss": 0.0778,
+      "step": 173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014262295081967215,
+      "loss": 0.0779,
+      "step": 174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014344262295081967,
+      "loss": 0.0785,
+      "step": 175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014426229508196722,
+      "loss": 0.0797,
+      "step": 176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014508196721311475,
+      "loss": 0.0771,
+      "step": 177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001459016393442623,
+      "loss": 0.0746,
+      "step": 178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014672131147540983,
+      "loss": 0.0817,
+      "step": 179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014754098360655738,
+      "loss": 0.0761,
+      "step": 180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014836065573770494,
+      "loss": 0.0767,
+      "step": 181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001491803278688525,
+      "loss": 0.0794,
+      "step": 182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.0765,
+      "step": 183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015081967213114754,
+      "loss": 0.0785,
+      "step": 184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001516393442622951,
+      "loss": 0.0784,
+      "step": 185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015245901639344262,
+      "loss": 0.0776,
+      "step": 186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015327868852459017,
+      "loss": 0.0791,
+      "step": 187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001540983606557377,
+      "loss": 0.0771,
+      "step": 188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015491803278688525,
+      "loss": 0.0758,
+      "step": 189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001557377049180328,
+      "loss": 0.0766,
+      "step": 190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015655737704918036,
+      "loss": 0.0796,
+      "step": 191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015737704918032788,
+      "loss": 0.0783,
+      "step": 192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015819672131147543,
+      "loss": 0.078,
+      "step": 193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00015901639344262296,
+      "loss": 0.0823,
+      "step": 194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001598360655737705,
+      "loss": 0.0768,
+      "step": 195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00016065573770491804,
+      "loss": 0.0792,
+      "step": 196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00016147540983606556,
+      "loss": 0.0756,
+      "step": 197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00016229508196721312,
+      "loss": 0.076,
+      "step": 198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00016311475409836064,
+      "loss": 0.0792,
+      "step": 199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001639344262295082,
+      "loss": 0.0738,
+      "step": 200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00016475409836065575,
+      "loss": 0.075,
+      "step": 201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001655737704918033,
+      "loss": 0.0787,
+      "step": 202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00016639344262295083,
+      "loss": 0.0762,
+      "step": 203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00016721311475409838,
+      "loss": 0.0784,
+      "step": 204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001680327868852459,
+      "loss": 0.0754,
+      "step": 205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00016885245901639346,
+      "loss": 0.0778,
+      "step": 206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00016967213114754099,
+      "loss": 0.0771,
+      "step": 207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001704918032786885,
+      "loss": 0.0754,
+      "step": 208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00017131147540983606,
+      "loss": 0.0765,
+      "step": 209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017213114754098362,
+      "loss": 0.0744,
+      "step": 210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017295081967213117,
+      "loss": 0.0778,
+      "step": 211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001737704918032787,
+      "loss": 0.0779,
+      "step": 212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017459016393442625,
+      "loss": 0.0791,
+      "step": 213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017540983606557377,
+      "loss": 0.0768,
+      "step": 214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017622950819672133,
+      "loss": 0.0752,
+      "step": 215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017704918032786885,
+      "loss": 0.0773,
+      "step": 216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001778688524590164,
+      "loss": 0.078,
+      "step": 217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017868852459016393,
+      "loss": 0.0809,
+      "step": 218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00017950819672131149,
+      "loss": 0.076,
+      "step": 219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018032786885245904,
+      "loss": 0.0767,
+      "step": 220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018114754098360656,
+      "loss": 0.0789,
+      "step": 221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018196721311475412,
+      "loss": 0.0773,
+      "step": 222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00018278688524590164,
+      "loss": 0.0768,
+      "step": 223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001836065573770492,
+      "loss": 0.0768,
+      "step": 224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018442622950819672,
+      "loss": 0.0769,
+      "step": 225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018524590163934427,
+      "loss": 0.0754,
+      "step": 226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001860655737704918,
+      "loss": 0.0734,
+      "step": 227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018688524590163935,
+      "loss": 0.0769,
+      "step": 228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018770491803278688,
+      "loss": 0.0781,
+      "step": 229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018852459016393443,
+      "loss": 0.0733,
+      "step": 230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018934426229508198,
+      "loss": 0.0773,
+      "step": 231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001901639344262295,
+      "loss": 0.077,
+      "step": 232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019098360655737706,
+      "loss": 0.077,
+      "step": 233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001918032786885246,
+      "loss": 0.0759,
+      "step": 234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019262295081967214,
+      "loss": 0.0774,
+      "step": 235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019344262295081967,
+      "loss": 0.0758,
+      "step": 236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019426229508196722,
+      "loss": 0.0764,
+      "step": 237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019508196721311475,
+      "loss": 0.0767,
+      "step": 238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001959016393442623,
+      "loss": 0.0765,
+      "step": 239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019672131147540985,
+      "loss": 0.0794,
+      "step": 240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001975409836065574,
+      "loss": 0.078,
+      "step": 241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019836065573770493,
+      "loss": 0.0744,
+      "step": 242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019918032786885248,
+      "loss": 0.073,
+      "step": 243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.0756,
+      "step": 244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999999920406258,
+      "loss": 0.0759,
+      "step": 245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019999996816250448,
+      "loss": 0.0772,
+      "step": 246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019999992836563978,
+      "loss": 0.0743,
+      "step": 247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019999987265003813,
+      "loss": 0.0778,
+      "step": 248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019999980101570835,
+      "loss": 0.0754,
+      "step": 249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001999997134626618,
+      "loss": 0.078,
+      "step": 250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019999960999091248,
+      "loss": 0.0749,
+      "step": 251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019999949060047688,
+      "loss": 0.0763,
+      "step": 252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019999935529137392,
+      "loss": 0.0762,
+      "step": 253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999992040636252,
+      "loss": 0.0783,
+      "step": 254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019999903691725476,
+      "loss": 0.0782,
+      "step": 255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019999885385228928,
+      "loss": 0.0791,
+      "step": 256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999986548687578,
+      "loss": 0.0761,
+      "step": 257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001999984399666921,
+      "loss": 0.0746,
+      "step": 258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019999820914612629,
+      "loss": 0.0762,
+      "step": 259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019999796240709717,
+      "loss": 0.071,
+      "step": 260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000199997699749644,
+      "loss": 0.0772,
+      "step": 261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019999742117380863,
+      "loss": 0.0779,
+      "step": 262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019999712667963536,
+      "loss": 0.0792,
+      "step": 263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999681626717104,
+      "loss": 0.0748,
+      "step": 264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999648993646518,
+      "loss": 0.0766,
+      "step": 265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999614768756967,
+      "loss": 0.0795,
+      "step": 266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999578952053896,
+      "loss": 0.0762,
+      "step": 267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999541543543016,
+      "loss": 0.0769,
+      "step": 268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999502543230273,
+      "loss": 0.0754,
+      "step": 269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999946195112188,
+      "loss": 0.075,
+      "step": 270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999419767224294,
+      "loss": 0.0759,
+      "step": 271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999375991544237,
+      "loss": 0.0777,
+      "step": 272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999330624088674,
+      "loss": 0.0765,
+      "step": 273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999283664864828,
+      "loss": 0.075,
+      "step": 274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001999923511388017,
+      "loss": 0.0753,
+      "step": 275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019999184971142433,
+      "loss": 0.0741,
+      "step": 276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000199991332366596,
+      "loss": 0.0748,
+      "step": 277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019999079910439904,
+      "loss": 0.0742,
+      "step": 278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019999024992491834,
+      "loss": 0.0764,
+      "step": 279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019998968482824131,
+      "loss": 0.0761,
+      "step": 280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019998910381445795,
+      "loss": 0.0768,
+      "step": 281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999885068836607,
+      "loss": 0.0748,
+      "step": 282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019998789403594464,
+      "loss": 0.0762,
+      "step": 283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999872652714073,
+      "loss": 0.0769,
+      "step": 284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019998662059014875,
+      "loss": 0.0753,
+      "step": 285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019998595999227162,
+      "loss": 0.0777,
+      "step": 286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019998528347788108,
+      "loss": 0.0769,
+      "step": 287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001999845910470848,
+      "loss": 0.0757,
+      "step": 288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019998388269999308,
+      "loss": 0.0752,
+      "step": 289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019998315843671862,
+      "loss": 0.0774,
+      "step": 290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019998241825737673,
+      "loss": 0.0722,
+      "step": 291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019998166216208524,
+      "loss": 0.0762,
+      "step": 292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019998089015096446,
+      "loss": 0.0762,
+      "step": 293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019998010222413735,
+      "loss": 0.0768,
+      "step": 294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019997929838172932,
+      "loss": 0.0743,
+      "step": 295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001999784786238683,
+      "loss": 0.0781,
+      "step": 296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019997764295068485,
+      "loss": 0.0746,
+      "step": 297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019997679136231193,
+      "loss": 0.0749,
+      "step": 298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019997592385888514,
+      "loss": 0.077,
+      "step": 299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001999750404405426,
+      "loss": 0.0787,
+      "step": 300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019997414110742488,
+      "loss": 0.0776,
+      "step": 301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019997322585967514,
+      "loss": 0.0769,
+      "step": 302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019997229469743914,
+      "loss": 0.0755,
+      "step": 303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019997134762086508,
+      "loss": 0.0765,
+      "step": 304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019997038463010373,
+      "loss": 0.0787,
+      "step": 305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001999694057253083,
+      "loss": 0.0753,
+      "step": 306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019996841090663477,
+      "loss": 0.0733,
+      "step": 307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001999674001742414,
+      "loss": 0.0774,
+      "step": 308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001999663735282891,
+      "loss": 0.0811,
+      "step": 309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019996533096894133,
+      "loss": 0.0757,
+      "step": 310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019996427249636403,
+      "loss": 0.1331,
+      "step": 311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019996319811072568,
+      "loss": 0.0859,
+      "step": 312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019996210781219738,
+      "loss": 0.1207,
+      "step": 313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001999610016009526,
+      "loss": 0.9687,
+      "step": 314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019995987947716746,
+      "loss": 0.1831,
+      "step": 315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001999587414410206,
+      "loss": 0.2362,
+      "step": 316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019995758749269326,
+      "loss": 0.1455,
+      "step": 317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019995641763236898,
+      "loss": 0.1452,
+      "step": 318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001999552318602341,
+      "loss": 0.1397,
+      "step": 319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001999540301764773,
+      "loss": 0.1435,
+      "step": 320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019995281258128995,
+      "loss": 0.1099,
+      "step": 321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019995157907486585,
+      "loss": 0.123,
+      "step": 322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019995032965740137,
+      "loss": 0.1075,
+      "step": 323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019994906432909537,
+      "loss": 0.1094,
+      "step": 324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019994778309014927,
+      "loss": 2.2499,
+      "step": 325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019994648594076706,
+      "loss": 1.8688,
+      "step": 326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019994517288115523,
+      "loss": 5.6017,
+      "step": 327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019994384391152276,
+      "loss": 1.6719,
+      "step": 328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019994249903208125,
+      "loss": 5.8566,
+      "step": 329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019994113824304478,
+      "loss": 5.1529,
+      "step": 330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019993976154462995,
+      "loss": 0.6591,
+      "step": 331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019993836893705594,
+      "loss": 0.1951,
+      "step": 332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019993696042054437,
+      "loss": 0.3168,
+      "step": 333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019993553599531957,
+      "loss": 0.2357,
+      "step": 334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019993409566160822,
+      "loss": 0.186,
+      "step": 335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001999326394196396,
+      "loss": 0.1816,
+      "step": 336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019993116726964554,
+      "loss": 0.133,
+      "step": 337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001999296792118604,
+      "loss": 0.1276,
+      "step": 338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000199928175246521,
+      "loss": 0.1303,
+      "step": 339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019992665537386687,
+      "loss": 0.1229,
+      "step": 340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019992511959413984,
+      "loss": 0.115,
+      "step": 341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019992356790758445,
+      "loss": 0.1157,
+      "step": 342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001999220003144477,
+      "loss": 0.1097,
+      "step": 343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001999204168149791,
+      "loss": 0.1054,
+      "step": 344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019991881740943072,
+      "loss": 0.1069,
+      "step": 345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019991720209805723,
+      "loss": 0.1003,
+      "step": 346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001999155708811157,
+      "loss": 0.1012,
+      "step": 347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019991392375886586,
+      "loss": 0.098,
+      "step": 348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019991226073156986,
+      "loss": 0.1052,
+      "step": 349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019991058179949246,
+      "loss": 0.0987,
+      "step": 350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019990888696290088,
+      "loss": 0.0945,
+      "step": 351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000199907176222065,
+      "loss": 0.0944,
+      "step": 352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019990544957725707,
+      "loss": 0.0983,
+      "step": 353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000199903707028752,
+      "loss": 0.0868,
+      "step": 354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019990194857682716,
+      "loss": 0.0906,
+      "step": 355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019990017422176246,
+      "loss": 0.0901,
+      "step": 356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001998983839638404,
+      "loss": 0.0921,
+      "step": 357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001998965778033459,
+      "loss": 0.0891,
+      "step": 358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019989475574056656,
+      "loss": 0.0973,
+      "step": 359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019989291777579237,
+      "loss": 0.089,
+      "step": 360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019989106390931594,
+      "loss": 0.0886,
+      "step": 361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019988919414143234,
+      "loss": 0.0893,
+      "step": 362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019988730847243926,
+      "loss": 0.0838,
+      "step": 363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001998854069026369,
+      "loss": 0.0863,
+      "step": 364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019988348943232787,
+      "loss": 0.088,
+      "step": 365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019988155606181747,
+      "loss": 0.0826,
+      "step": 366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019987960679141345,
+      "loss": 0.0859,
+      "step": 367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019987764162142613,
+      "loss": 0.0874,
+      "step": 368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019987566055216832,
+      "loss": 0.0838,
+      "step": 369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019987366358395543,
+      "loss": 0.0846,
+      "step": 370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019987165071710527,
+      "loss": 0.081,
+      "step": 371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019986962195193834,
+      "loss": 0.0807,
+      "step": 372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019986757728877757,
+      "loss": 0.0797,
+      "step": 373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001998655167279484,
+      "loss": 0.0825,
+      "step": 374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001998634402697789,
+      "loss": 0.0873,
+      "step": 375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001998613479145996,
+      "loss": 0.0778,
+      "step": 376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019985923966274355,
+      "loss": 0.0789,
+      "step": 377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019985711551454642,
+      "loss": 0.0793,
+      "step": 378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019985497547034628,
+      "loss": 0.0778,
+      "step": 379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019985281953048385,
+      "loss": 0.0807,
+      "step": 380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001998506476953023,
+      "loss": 0.0816,
+      "step": 381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019984845996514735,
+      "loss": 0.0795,
+      "step": 382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019984625634036728,
+      "loss": 0.0795,
+      "step": 383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019984403682131286,
+      "loss": 0.0822,
+      "step": 384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019984180140833745,
+      "loss": 0.0839,
+      "step": 385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019983955010179687,
+      "loss": 0.0804,
+      "step": 386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001998372829020495,
+      "loss": 0.0788,
+      "step": 387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019983499980945623,
+      "loss": 0.0861,
+      "step": 388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019983270082438054,
+      "loss": 0.0784,
+      "step": 389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001998303859471884,
+      "loss": 0.0767,
+      "step": 390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019982805517824827,
+      "loss": 0.0814,
+      "step": 391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019982570851793122,
+      "loss": 0.0806,
+      "step": 392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001998233459666108,
+      "loss": 0.081,
+      "step": 393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019982096752466306,
+      "loss": 0.0782,
+      "step": 394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019981857319246665,
+      "loss": 0.08,
+      "step": 395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019981616297040273,
+      "loss": 0.0822,
+      "step": 396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019981373685885493,
+      "loss": 0.0819,
+      "step": 397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019981129485820953,
+      "loss": 0.0795,
+      "step": 398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001998088369688552,
+      "loss": 0.0822,
+      "step": 399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019980636319118325,
+      "loss": 0.0787,
+      "step": 400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019980387352558742,
+      "loss": 0.078,
+      "step": 401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001998013679724641,
+      "loss": 0.0816,
+      "step": 402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019979884653221209,
+      "loss": 0.0807,
+      "step": 403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001997963092052328,
+      "loss": 0.0757,
+      "step": 404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019979375599193014,
+      "loss": 0.0758,
+      "step": 405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019979118689271053,
+      "loss": 0.0796,
+      "step": 406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019978860190798296,
+      "loss": 0.0775,
+      "step": 407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019978600103815893,
+      "loss": 0.0797,
+      "step": 408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001997833842836524,
+      "loss": 0.082,
+      "step": 409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019978075164488003,
+      "loss": 0.0783,
+      "step": 410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019977810312226085,
+      "loss": 0.0758,
+      "step": 411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019977543871621648,
+      "loss": 0.0759,
+      "step": 412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000199772758427171,
+      "loss": 0.079,
+      "step": 413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019977006225555118,
+      "loss": 0.083,
+      "step": 414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019976735020178616,
+      "loss": 0.0813,
+      "step": 415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019976462226630766,
+      "loss": 0.079,
+      "step": 416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019976187844954996,
+      "loss": 0.0843,
+      "step": 417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019975911875194983,
+      "loss": 0.0806,
+      "step": 418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019975634317394655,
+      "loss": 0.0791,
+      "step": 419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019975355171598203,
+      "loss": 0.0785,
+      "step": 420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019975074437850057,
+      "loss": 0.0763,
+      "step": 421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019974792116194906,
+      "loss": 0.0782,
+      "step": 422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000199745082066777,
+      "loss": 0.0782,
+      "step": 423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019974222709343625,
+      "loss": 0.077,
+      "step": 424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001997393562423813,
+      "loss": 0.0781,
+      "step": 425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001997364695140692,
+      "loss": 0.0818,
+      "step": 426
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019973356690895943,
+      "loss": 0.081,
+      "step": 427
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001997306484275141,
+      "loss": 0.0796,
+      "step": 428
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001997277140701977,
+      "loss": 0.0762,
+      "step": 429
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019972476383747748,
+      "loss": 0.0801,
+      "step": 430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000199721797729823,
+      "loss": 0.0775,
+      "step": 431
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001997188157477064,
+      "loss": 0.0775,
+      "step": 432
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019971581789160245,
+      "loss": 0.079,
+      "step": 433
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019971280416198833,
+      "loss": 0.0778,
+      "step": 434
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019970977455934376,
+      "loss": 0.0795,
+      "step": 435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019970672908415106,
+      "loss": 0.0807,
+      "step": 436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019970366773689504,
+      "loss": 0.0805,
+      "step": 437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019970059051806297,
+      "loss": 0.0792,
+      "step": 438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019969749742814474,
+      "loss": 0.077,
+      "step": 439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019969438846763278,
+      "loss": 0.0766,
+      "step": 440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019969126363702188,
+      "loss": 0.0776,
+      "step": 441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019968812293680958,
+      "loss": 0.0802,
+      "step": 442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019968496636749584,
+      "loss": 0.0753,
+      "step": 443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019968179392958305,
+      "loss": 0.078,
+      "step": 444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001996786056235763,
+      "loss": 0.0779,
+      "step": 445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001996754014499831,
+      "loss": 0.0786,
+      "step": 446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019967218140931356,
+      "loss": 0.0775,
+      "step": 447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001996689455020802,
+      "loss": 0.0784,
+      "step": 448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019966569372879817,
+      "loss": 0.0773,
+      "step": 449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019966242608998513,
+      "loss": 0.0759,
+      "step": 450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019965914258616123,
+      "loss": 0.0781,
+      "step": 451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019965584321784917,
+      "loss": 0.0842,
+      "step": 452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019965252798557413,
+      "loss": 0.0787,
+      "step": 453
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001996491968898639,
+      "loss": 0.0769,
+      "step": 454
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001996458499312487,
+      "loss": 0.0763,
+      "step": 455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001996424871102614,
+      "loss": 0.0785,
+      "step": 456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019963910842743724,
+      "loss": 0.0791,
+      "step": 457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019963571388331415,
+      "loss": 0.0785,
+      "step": 458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019963230347843242,
+      "loss": 0.0786,
+      "step": 459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019962887721333496,
+      "loss": 0.0771,
+      "step": 460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001996254350885672,
+      "loss": 0.0802,
+      "step": 461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001996219771046771,
+      "loss": 0.0757,
+      "step": 462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001996185032622151,
+      "loss": 0.0769,
+      "step": 463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019961501356173422,
+      "loss": 0.0785,
+      "step": 464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019961150800378995,
+      "loss": 0.0779,
+      "step": 465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019960798658894032,
+      "loss": 0.0763,
+      "step": 466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019960444931774594,
+      "loss": 0.0755,
+      "step": 467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019960089619076987,
+      "loss": 0.0779,
+      "step": 468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019959732720857773,
+      "loss": 0.0765,
+      "step": 469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019959374237173768,
+      "loss": 0.0773,
+      "step": 470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019959014168082032,
+      "loss": 0.0787,
+      "step": 471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001995865251363989,
+      "loss": 0.0755,
+      "step": 472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019958289273904907,
+      "loss": 0.0772,
+      "step": 473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019957924448934912,
+      "loss": 0.0763,
+      "step": 474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019957558038787978,
+      "loss": 0.0802,
+      "step": 475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001995719004352243,
+      "loss": 0.0804,
+      "step": 476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019956820463196855,
+      "loss": 0.0783,
+      "step": 477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019956449297870078,
+      "loss": 0.0807,
+      "step": 478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001995607654760119,
+      "loss": 0.083,
+      "step": 479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019955702212449524,
+      "loss": 0.0773,
+      "step": 480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001995532629247467,
+      "loss": 0.076,
+      "step": 481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019954948787736474,
+      "loss": 0.0751,
+      "step": 482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019954569698295025,
+      "loss": 0.0771,
+      "step": 483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019954189024210672,
+      "loss": 0.0767,
+      "step": 484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019953806765544011,
+      "loss": 0.0836,
+      "step": 485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019953422922355896,
+      "loss": 0.0788,
+      "step": 486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001995303749470743,
+      "loss": 0.0776,
+      "step": 487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019952650482659966,
+      "loss": 0.081,
+      "step": 488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001995226188627511,
+      "loss": 0.0781,
+      "step": 489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019951871705614727,
+      "loss": 0.0767,
+      "step": 490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019951479940740922,
+      "loss": 0.077,
+      "step": 491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001995108659171607,
+      "loss": 0.0777,
+      "step": 492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001995069165860277,
+      "loss": 0.0797,
+      "step": 493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019950295141463908,
+      "loss": 0.0825,
+      "step": 494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019949897040362594,
+      "loss": 0.081,
+      "step": 495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019949497355362206,
+      "loss": 0.078,
+      "step": 496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019949096086526366,
+      "loss": 0.079,
+      "step": 497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019948693233918952,
+      "loss": 0.0751,
+      "step": 498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019948288797604094,
+      "loss": 0.0773,
+      "step": 499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001994788277764617,
+      "loss": 0.0796,
+      "step": 500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019947475174109816,
+      "loss": 0.0751,
+      "step": 501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019947065987059915,
+      "loss": 0.0822,
+      "step": 502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019946655216561608,
+      "loss": 0.0785,
+      "step": 503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001994624286268028,
+      "loss": 0.0791,
+      "step": 504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001994582892548158,
+      "loss": 0.0803,
+      "step": 505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019945413405031393,
+      "loss": 0.0749,
+      "step": 506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001994499630139587,
+      "loss": 0.0764,
+      "step": 507
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019944577614641404,
+      "loss": 0.0774,
+      "step": 508
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019944157344834652,
+      "loss": 0.0782,
+      "step": 509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001994373549204251,
+      "loss": 0.0826,
+      "step": 510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019943312056332134,
+      "loss": 0.0766,
+      "step": 511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019942887037770928,
+      "loss": 0.0799,
+      "step": 512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019942460436426552,
+      "loss": 0.0772,
+      "step": 513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001994203225236691,
+      "loss": 0.0793,
+      "step": 514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019941602485660172,
+      "loss": 0.0776,
+      "step": 515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019941171136374745,
+      "loss": 0.0772,
+      "step": 516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000199407382045793,
+      "loss": 0.077,
+      "step": 517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019940303690342749,
+      "loss": 0.0768,
+      "step": 518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001993986759373426,
+      "loss": 0.0755,
+      "step": 519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001993942991482326,
+      "loss": 0.0797,
+      "step": 520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019938990653679416,
+      "loss": 0.0776,
+      "step": 521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001993854981037266,
+      "loss": 0.0801,
+      "step": 522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019938107384973166,
+      "loss": 0.079,
+      "step": 523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001993766337755136,
+      "loss": 0.0783,
+      "step": 524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001993721778817792,
+      "loss": 0.0783,
+      "step": 525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019936770616923785,
+      "loss": 0.077,
+      "step": 526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019936321863860137,
+      "loss": 0.0774,
+      "step": 527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019935871529058413,
+      "loss": 0.0765,
+      "step": 528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019935419612590293,
+      "loss": 0.077,
+      "step": 529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019934966114527726,
+      "loss": 0.0769,
+      "step": 530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000199345110349429,
+      "loss": 0.075,
+      "step": 531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019934054373908254,
+      "loss": 0.0796,
+      "step": 532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001993359613149649,
+      "loss": 0.0803,
+      "step": 533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019933136307780547,
+      "loss": 0.0782,
+      "step": 534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019932674902833629,
+      "loss": 0.0753,
+      "step": 535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019932211916729182,
+      "loss": 0.0818,
+      "step": 536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001993174734954091,
+      "loss": 0.0747,
+      "step": 537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019931281201342765,
+      "loss": 0.077,
+      "step": 538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019930813472208953,
+      "loss": 0.0738,
+      "step": 539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019930344162213932,
+      "loss": 0.0794,
+      "step": 540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019929873271432404,
+      "loss": 0.0781,
+      "step": 541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019929400799939336,
+      "loss": 0.0805,
+      "step": 542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019928926747809938,
+      "loss": 0.0757,
+      "step": 543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019928451115119668,
+      "loss": 0.0779,
+      "step": 544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019927973901944247,
+      "loss": 0.0767,
+      "step": 545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019927495108359642,
+      "loss": 0.0755,
+      "step": 546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019927014734442064,
+      "loss": 0.0761,
+      "step": 547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019926532780267987,
+      "loss": 0.0749,
+      "step": 548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019926049245914132,
+      "loss": 0.0773,
+      "step": 549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019925564131457476,
+      "loss": 0.0766,
+      "step": 550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001992507743697523,
+      "loss": 0.0778,
+      "step": 551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019924589162544884,
+      "loss": 0.0758,
+      "step": 552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019924099308244158,
+      "loss": 0.0776,
+      "step": 553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019923607874151032,
+      "loss": 0.078,
+      "step": 554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019923114860343735,
+      "loss": 0.0781,
+      "step": 555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001992262026690075,
+      "loss": 0.0777,
+      "step": 556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019922124093900808,
+      "loss": 0.0761,
+      "step": 557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019921626341422899,
+      "loss": 0.0781,
+      "step": 558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019921127009546254,
+      "loss": 0.0761,
+      "step": 559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001992062609835036,
+      "loss": 0.0784,
+      "step": 560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001992012360791496,
+      "loss": 0.0791,
+      "step": 561
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001991961953832004,
+      "loss": 0.0778,
+      "step": 562
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019919113889645845,
+      "loss": 0.0757,
+      "step": 563
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019918606661972863,
+      "loss": 0.0775,
+      "step": 564
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019918097855381843,
+      "loss": 0.0743,
+      "step": 565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001991758746995378,
+      "loss": 0.076,
+      "step": 566
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001991707550576992,
+      "loss": 0.0764,
+      "step": 567
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001991656196291176,
+      "loss": 0.0749,
+      "step": 568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019916046841461052,
+      "loss": 0.0743,
+      "step": 569
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019915530141499795,
+      "loss": 0.0754,
+      "step": 570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019915011863110243,
+      "loss": 0.0777,
+      "step": 571
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019914492006374901,
+      "loss": 0.085,
+      "step": 572
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019913970571376515,
+      "loss": 0.0769,
+      "step": 573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000199134475581981,
+      "loss": 0.0784,
+      "step": 574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019912922966922912,
+      "loss": 0.0759,
+      "step": 575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019912396797634458,
+      "loss": 0.0747,
+      "step": 576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019911869050416497,
+      "loss": 0.0776,
+      "step": 577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019911339725353037,
+      "loss": 0.0778,
+      "step": 578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019910808822528345,
+      "loss": 0.0759,
+      "step": 579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001991027634202693,
+      "loss": 0.0795,
+      "step": 580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001990974228393356,
+      "loss": 0.0778,
+      "step": 581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019909206648333245,
+      "loss": 0.0764,
+      "step": 582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001990866943531126,
+      "loss": 0.0782,
+      "step": 583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019908130644953116,
+      "loss": 0.0766,
+      "step": 584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019907590277344582,
+      "loss": 0.0781,
+      "step": 585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019907048332571677,
+      "loss": 0.077,
+      "step": 586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019906504810720675,
+      "loss": 0.0835,
+      "step": 587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000199059597118781,
+      "loss": 0.0798,
+      "step": 588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001990541303613072,
+      "loss": 0.0798,
+      "step": 589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001990486478356556,
+      "loss": 0.0753,
+      "step": 590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019904314954269898,
+      "loss": 0.0768,
+      "step": 591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019903763548331256,
+      "loss": 0.0781,
+      "step": 592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019903210565837411,
+      "loss": 0.0762,
+      "step": 593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019902656006876392,
+      "loss": 0.0761,
+      "step": 594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001990209987153648,
+      "loss": 0.074,
+      "step": 595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000199015421599062,
+      "loss": 0.0778,
+      "step": 596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001990098287207434,
+      "loss": 0.0749,
+      "step": 597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019900422008129923,
+      "loss": 0.0781,
+      "step": 598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001989985956816224,
+      "loss": 0.0777,
+      "step": 599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019899295552260816,
+      "loss": 0.0763,
+      "step": 600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001989872996051544,
+      "loss": 0.0771,
+      "step": 601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001989816279301615,
+      "loss": 0.0759,
+      "step": 602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019897594049853224,
+      "loss": 0.079,
+      "step": 603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019897023731117207,
+      "loss": 0.076,
+      "step": 604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001989645183689888,
+      "loss": 0.0772,
+      "step": 605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019895878367289286,
+      "loss": 0.0751,
+      "step": 606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019895303322379712,
+      "loss": 0.0751,
+      "step": 607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019894726702261698,
+      "loss": 0.075,
+      "step": 608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001989414850702704,
+      "loss": 0.0772,
+      "step": 609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001989356873676777,
+      "loss": 0.0742,
+      "step": 610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019892987391576187,
+      "loss": 0.0763,
+      "step": 611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019892404471544828,
+      "loss": 0.0799,
+      "step": 612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019891819976766492,
+      "loss": 0.0768,
+      "step": 613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019891233907334222,
+      "loss": 0.0774,
+      "step": 614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019890646263341313,
+      "loss": 0.0772,
+      "step": 615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019890057044881306,
+      "loss": 0.0749,
+      "step": 616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019889466252048007,
+      "loss": 0.079,
+      "step": 617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019888873884935454,
+      "loss": 0.0761,
+      "step": 618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019888279943637945,
+      "loss": 0.0798,
+      "step": 619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019887684428250037,
+      "loss": 0.078,
+      "step": 620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019887087338866518,
+      "loss": 0.0828,
+      "step": 621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001988648867558244,
+      "loss": 0.0773,
+      "step": 622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019885888438493106,
+      "loss": 0.0778,
+      "step": 623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019885286627694063,
+      "loss": 0.0762,
+      "step": 624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019884683243281116,
+      "loss": 0.0767,
+      "step": 625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001988407828535031,
+      "loss": 0.0764,
+      "step": 626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019883471753997947,
+      "loss": 0.0755,
+      "step": 627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001988286364932059,
+      "loss": 0.0753,
+      "step": 628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019882253971415027,
+      "loss": 0.0768,
+      "step": 629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019881642720378321,
+      "loss": 0.0778,
+      "step": 630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001988102989630777,
+      "loss": 0.0765,
+      "step": 631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019880415499300936,
+      "loss": 0.0764,
+      "step": 632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001987979952945561,
+      "loss": 0.074,
+      "step": 633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019879181986869857,
+      "loss": 0.0784,
+      "step": 634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019878562871641982,
+      "loss": 0.077,
+      "step": 635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019877942183870535,
+      "loss": 0.0753,
+      "step": 636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019877319923654324,
+      "loss": 0.0742,
+      "step": 637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001987669609109241,
+      "loss": 0.0741,
+      "step": 638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019876070686284088,
+      "loss": 0.0757,
+      "step": 639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019875443709328926,
+      "loss": 0.0776,
+      "step": 640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019874815160326724,
+      "loss": 0.0791,
+      "step": 641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019874185039377542,
+      "loss": 0.0783,
+      "step": 642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001987355334658169,
+      "loss": 0.0758,
+      "step": 643
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019872920082039718,
+      "loss": 0.0773,
+      "step": 644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001987228524585244,
+      "loss": 0.076,
+      "step": 645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001987164883812091,
+      "loss": 0.0776,
+      "step": 646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001987101085894644,
+      "loss": 0.0771,
+      "step": 647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019870371308430588,
+      "loss": 0.0764,
+      "step": 648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001986973018667516,
+      "loss": 0.0772,
+      "step": 649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019869087493782218,
+      "loss": 0.0805,
+      "step": 650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019868443229854066,
+      "loss": 0.0743,
+      "step": 651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001986779739499326,
+      "loss": 0.0773,
+      "step": 652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019867149989302623,
+      "loss": 0.0754,
+      "step": 653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000198665010128852,
+      "loss": 0.0755,
+      "step": 654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019865850465844304,
+      "loss": 0.0763,
+      "step": 655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019865198348283496,
+      "loss": 0.0792,
+      "step": 656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019864544660306584,
+      "loss": 0.0762,
+      "step": 657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019863889402017625,
+      "loss": 0.0813,
+      "step": 658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001986323257352093,
+      "loss": 0.0812,
+      "step": 659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019862574174921055,
+      "loss": 0.0766,
+      "step": 660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019861914206322816,
+      "loss": 0.0785,
+      "step": 661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019861252667831262,
+      "loss": 0.0758,
+      "step": 662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001986058955955171,
+      "loss": 0.0764,
+      "step": 663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019859924881589715,
+      "loss": 0.0749,
+      "step": 664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019859258634051082,
+      "loss": 0.0795,
+      "step": 665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019858590817041876,
+      "loss": 0.0766,
+      "step": 666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019857921430668402,
+      "loss": 0.0762,
+      "step": 667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019857250475037216,
+      "loss": 0.0777,
+      "step": 668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019856577950255129,
+      "loss": 0.0762,
+      "step": 669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019855903856429196,
+      "loss": 0.0789,
+      "step": 670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019855228193666724,
+      "loss": 0.0791,
+      "step": 671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019854550962075275,
+      "loss": 0.0778,
+      "step": 672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019853872161762647,
+      "loss": 0.0775,
+      "step": 673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019853191792836906,
+      "loss": 0.0735,
+      "step": 674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019852509855406352,
+      "loss": 0.0767,
+      "step": 675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019851826349579545,
+      "loss": 0.075,
+      "step": 676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019851141275465287,
+      "loss": 0.0774,
+      "step": 677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019850454633172631,
+      "loss": 0.0803,
+      "step": 678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001984976642281089,
+      "loss": 0.0789,
+      "step": 679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019849076644489613,
+      "loss": 0.0756,
+      "step": 680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019848385298318607,
+      "loss": 0.0754,
+      "step": 681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001984769238440792,
+      "loss": 0.0762,
+      "step": 682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001984699790286786,
+      "loss": 0.0755,
+      "step": 683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001984630185380898,
+      "loss": 0.0783,
+      "step": 684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019845604237342078,
+      "loss": 0.077,
+      "step": 685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019844905053578214,
+      "loss": 0.0765,
+      "step": 686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001984420430262868,
+      "loss": 0.0762,
+      "step": 687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019843501984605032,
+      "loss": 0.0764,
+      "step": 688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019842798099619068,
+      "loss": 0.0762,
+      "step": 689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019842092647782842,
+      "loss": 0.0792,
+      "step": 690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019841385629208646,
+      "loss": 0.0733,
+      "step": 691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019840677044009034,
+      "loss": 0.0763,
+      "step": 692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019839966892296802,
+      "loss": 0.0765,
+      "step": 693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019839255174185,
+      "loss": 0.0753,
+      "step": 694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001983854188978692,
+      "loss": 0.0761,
+      "step": 695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019837827039216108,
+      "loss": 0.0758,
+      "step": 696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019837110622586364,
+      "loss": 0.0783,
+      "step": 697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019836392640011727,
+      "loss": 0.0769,
+      "step": 698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019835673091606496,
+      "loss": 0.0774,
+      "step": 699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019834951977485213,
+      "loss": 0.0791,
+      "step": 700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019834229297762668,
+      "loss": 0.08,
+      "step": 701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019833505052553905,
+      "loss": 0.0756,
+      "step": 702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019832779241974212,
+      "loss": 0.0761,
+      "step": 703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019832051866139133,
+      "loss": 0.0762,
+      "step": 704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001983132292516445,
+      "loss": 0.0771,
+      "step": 705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001983059241916621,
+      "loss": 0.0774,
+      "step": 706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019829860348260695,
+      "loss": 0.0734,
+      "step": 707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001982912671256444,
+      "loss": 0.077,
+      "step": 708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001982839151219424,
+      "loss": 0.0746,
+      "step": 709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001982765474726712,
+      "loss": 0.0767,
+      "step": 710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019826916417900364,
+      "loss": 0.0774,
+      "step": 711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001982617652421151,
+      "loss": 0.0771,
+      "step": 712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001982543506631834,
+      "loss": 0.0777,
+      "step": 713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019824692044338878,
+      "loss": 0.0771,
+      "step": 714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001982394745839141,
+      "loss": 0.0778,
+      "step": 715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019823201308594464,
+      "loss": 0.0771,
+      "step": 716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019822453595066817,
+      "loss": 0.0813,
+      "step": 717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019821704317927492,
+      "loss": 0.0762,
+      "step": 718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001982095347729577,
+      "loss": 0.0779,
+      "step": 719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019820201073291174,
+      "loss": 0.0757,
+      "step": 720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019819447106033476,
+      "loss": 0.0756,
+      "step": 721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019818691575642698,
+      "loss": 0.0781,
+      "step": 722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019817934482239111,
+      "loss": 0.0782,
+      "step": 723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019817175825943237,
+      "loss": 0.0741,
+      "step": 724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019816415606875842,
+      "loss": 0.0757,
+      "step": 725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019815653825157946,
+      "loss": 0.0788,
+      "step": 726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019814890480910813,
+      "loss": 0.076,
+      "step": 727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019814125574255957,
+      "loss": 0.0774,
+      "step": 728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019813359105315143,
+      "loss": 0.0749,
+      "step": 729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019812591074210386,
+      "loss": 0.0748,
+      "step": 730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001981182148106394,
+      "loss": 0.0761,
+      "step": 731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001981105032599832,
+      "loss": 0.0739,
+      "step": 732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019810277609136288,
+      "loss": 0.0775,
+      "step": 733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019809503330600842,
+      "loss": 0.075,
+      "step": 734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019808727490515238,
+      "loss": 0.0745,
+      "step": 735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019807950089002988,
+      "loss": 0.0782,
+      "step": 736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019807171126187838,
+      "loss": 0.0768,
+      "step": 737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019806390602193793,
+      "loss": 0.0764,
+      "step": 738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019805608517145098,
+      "loss": 0.0765,
+      "step": 739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019804824871166255,
+      "loss": 0.0791,
+      "step": 740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019804039664382008,
+      "loss": 0.0773,
+      "step": 741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019803252896917356,
+      "loss": 0.0781,
+      "step": 742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001980246456889754,
+      "loss": 0.0777,
+      "step": 743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001980167468044805,
+      "loss": 0.0806,
+      "step": 744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001980088323169463,
+      "loss": 0.0773,
+      "step": 745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019800090222763265,
+      "loss": 0.0771,
+      "step": 746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019799295653780195,
+      "loss": 0.0793,
+      "step": 747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019798499524871908,
+      "loss": 0.0765,
+      "step": 748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001979770183616513,
+      "loss": 0.0769,
+      "step": 749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001979690258778685,
+      "loss": 0.074,
+      "step": 750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019796101779864296,
+      "loss": 0.0759,
+      "step": 751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019795299412524945,
+      "loss": 0.0769,
+      "step": 752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019794495485896528,
+      "loss": 0.0761,
+      "step": 753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019793690000107015,
+      "loss": 0.077,
+      "step": 754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001979288295528463,
+      "loss": 0.0783,
+      "step": 755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001979207435155785,
+      "loss": 0.0754,
+      "step": 756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019791264189055388,
+      "loss": 0.0761,
+      "step": 757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019790452467906215,
+      "loss": 0.0789,
+      "step": 758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019789639188239546,
+      "loss": 0.0763,
+      "step": 759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019788824350184845,
+      "loss": 0.0782,
+      "step": 760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019788007953871825,
+      "loss": 0.0771,
+      "step": 761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019787189999430444,
+      "loss": 0.0775,
+      "step": 762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001978637048699091,
+      "loss": 0.0782,
+      "step": 763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019785549416683685,
+      "loss": 0.0775,
+      "step": 764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019784726788639464,
+      "loss": 0.0799,
+      "step": 765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019783902602989208,
+      "loss": 0.0789,
+      "step": 766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019783076859864108,
+      "loss": 0.0774,
+      "step": 767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001978224955939562,
+      "loss": 0.0759,
+      "step": 768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019781420701715436,
+      "loss": 0.0764,
+      "step": 769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000197805902869555,
+      "loss": 0.0768,
+      "step": 770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019779758315248004,
+      "loss": 0.0757,
+      "step": 771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019778924786725388,
+      "loss": 0.0785,
+      "step": 772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001977808970152034,
+      "loss": 0.0757,
+      "step": 773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001977725305976579,
+      "loss": 0.0777,
+      "step": 774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019776414861594926,
+      "loss": 0.0752,
+      "step": 775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019775575107141178,
+      "loss": 0.0745,
+      "step": 776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019774733796538226,
+      "loss": 0.0766,
+      "step": 777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001977389092991999,
+      "loss": 0.0767,
+      "step": 778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001977304650742065,
+      "loss": 0.0782,
+      "step": 779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019772200529174624,
+      "loss": 0.0717,
+      "step": 780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019771352995316582,
+      "loss": 0.0762,
+      "step": 781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019770503905981444,
+      "loss": 0.0785,
+      "step": 782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001976965326130437,
+      "loss": 0.0745,
+      "step": 783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019768801061420774,
+      "loss": 0.076,
+      "step": 784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019767947306466316,
+      "loss": 0.0791,
+      "step": 785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000197670919965769,
+      "loss": 0.0753,
+      "step": 786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019766235131888683,
+      "loss": 0.0785,
+      "step": 787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001976537671253807,
+      "loss": 0.0796,
+      "step": 788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019764516738661705,
+      "loss": 0.073,
+      "step": 789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019763655210396487,
+      "loss": 0.0768,
+      "step": 790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019762792127879563,
+      "loss": 0.0757,
+      "step": 791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019761927491248322,
+      "loss": 0.0794,
+      "step": 792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019761061300640405,
+      "loss": 0.0763,
+      "step": 793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019760193556193697,
+      "loss": 0.0763,
+      "step": 794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019759324258046334,
+      "loss": 0.0747,
+      "step": 795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000197584534063367,
+      "loss": 0.0741,
+      "step": 796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019757581001203414,
+      "loss": 0.0771,
+      "step": 797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001975670704278536,
+      "loss": 0.073,
+      "step": 798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019755831531221663,
+      "loss": 0.0752,
+      "step": 799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019754954466651687,
+      "loss": 0.0772,
+      "step": 800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019754075849215055,
+      "loss": 0.0757,
+      "step": 801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019753195679051628,
+      "loss": 0.0751,
+      "step": 802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019752313956301518,
+      "loss": 0.0763,
+      "step": 803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001975143068110509,
+      "loss": 0.0742,
+      "step": 804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019750545853602943,
+      "loss": 0.0766,
+      "step": 805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019749659473935934,
+      "loss": 0.0766,
+      "step": 806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019748771542245166,
+      "loss": 0.0749,
+      "step": 807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001974788205867198,
+      "loss": 0.0785,
+      "step": 808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019746991023357976,
+      "loss": 0.0749,
+      "step": 809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019746098436444995,
+      "loss": 0.0805,
+      "step": 810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019745204298075123,
+      "loss": 0.0785,
+      "step": 811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000197443086083907,
+      "loss": 0.0749,
+      "step": 812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019743411367534306,
+      "loss": 0.075,
+      "step": 813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019742512575648768,
+      "loss": 0.0755,
+      "step": 814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001974161223287717,
+      "loss": 0.0784,
+      "step": 815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019740710339362824,
+      "loss": 0.0743,
+      "step": 816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001973980689524931,
+      "loss": 0.0752,
+      "step": 817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001973890190068044,
+      "loss": 0.0767,
+      "step": 818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019737995355800282,
+      "loss": 0.0724,
+      "step": 819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001973708726075314,
+      "loss": 0.0773,
+      "step": 820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019736177615683578,
+      "loss": 0.0762,
+      "step": 821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019735266420736397,
+      "loss": 0.0768,
+      "step": 822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001973435367605665,
+      "loss": 0.0779,
+      "step": 823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019733439381789628,
+      "loss": 0.0778,
+      "step": 824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019732523538080882,
+      "loss": 0.0769,
+      "step": 825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019731606145076202,
+      "loss": 0.0762,
+      "step": 826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019730687202921624,
+      "loss": 0.0774,
+      "step": 827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001972976671176343,
+      "loss": 0.0729,
+      "step": 828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019728844671748155,
+      "loss": 0.0777,
+      "step": 829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019727921083022575,
+      "loss": 0.0757,
+      "step": 830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019726995945733715,
+      "loss": 0.074,
+      "step": 831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001972606926002884,
+      "loss": 0.078,
+      "step": 832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001972514102605547,
+      "loss": 0.079,
+      "step": 833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019724211243961367,
+      "loss": 0.0756,
+      "step": 834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019723279913894546,
+      "loss": 0.0711,
+      "step": 835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019722347036003256,
+      "loss": 0.0763,
+      "step": 836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019721412610436003,
+      "loss": 0.0769,
+      "step": 837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001972047663734154,
+      "loss": 0.0771,
+      "step": 838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001971953911686885,
+      "loss": 0.0771,
+      "step": 839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019718600049167185,
+      "loss": 0.0751,
+      "step": 840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001971765943438603,
+      "loss": 0.0763,
+      "step": 841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019716717272675121,
+      "loss": 0.0759,
+      "step": 842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019715773564184437,
+      "loss": 0.0761,
+      "step": 843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000197148283090642,
+      "loss": 0.0774,
+      "step": 844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019713881507464888,
+      "loss": 0.0784,
+      "step": 845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019712933159537222,
+      "loss": 0.0734,
+      "step": 846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001971198326543216,
+      "loss": 0.0757,
+      "step": 847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001971103182530092,
+      "loss": 0.078,
+      "step": 848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019710078839294952,
+      "loss": 0.0772,
+      "step": 849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019709124307565969,
+      "loss": 0.077,
+      "step": 850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001970816823026591,
+      "loss": 0.0773,
+      "step": 851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001970721060754698,
+      "loss": 0.075,
+      "step": 852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019706251439561618,
+      "loss": 0.0784,
+      "step": 853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019705290726462508,
+      "loss": 0.0759,
+      "step": 854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019704328468402586,
+      "loss": 0.076,
+      "step": 855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019703364665535027,
+      "loss": 0.0741,
+      "step": 856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019702399318013265,
+      "loss": 0.078,
+      "step": 857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019701432425990963,
+      "loss": 0.0771,
+      "step": 858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019700463989622046,
+      "loss": 0.0791,
+      "step": 859
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001969949400906067,
+      "loss": 0.078,
+      "step": 860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019698522484461246,
+      "loss": 0.0773,
+      "step": 861
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001969754941597843,
+      "loss": 0.0775,
+      "step": 862
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019696574803767123,
+      "loss": 0.0771,
+      "step": 863
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019695598647982468,
+      "loss": 0.0773,
+      "step": 864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019694620948779856,
+      "loss": 0.0757,
+      "step": 865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019693641706314927,
+      "loss": 0.077,
+      "step": 866
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019692660920743566,
+      "loss": 0.075,
+      "step": 867
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000196916785922219,
+      "loss": 0.0777,
+      "step": 868
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000196906947209063,
+      "loss": 0.074,
+      "step": 869
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019689709306953393,
+      "loss": 0.0767,
+      "step": 870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019688722350520035,
+      "loss": 0.0738,
+      "step": 871
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019687733851763345,
+      "loss": 0.0755,
+      "step": 872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001968674381084068,
+      "loss": 0.0752,
+      "step": 873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019685752227909634,
+      "loss": 0.0766,
+      "step": 874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019684759103128066,
+      "loss": 0.0768,
+      "step": 875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001968376443665406,
+      "loss": 0.0772,
+      "step": 876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001968276822864596,
+      "loss": 0.0785,
+      "step": 877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019681770479262342,
+      "loss": 0.078,
+      "step": 878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019680771188662044,
+      "loss": 0.0789,
+      "step": 879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019679770357004136,
+      "loss": 0.0776,
+      "step": 880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019678767984447938,
+      "loss": 0.0742,
+      "step": 881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001967776407115302,
+      "loss": 0.0729,
+      "step": 882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019676758617279187,
+      "loss": 0.0759,
+      "step": 883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019675751622986494,
+      "loss": 0.0764,
+      "step": 884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019674743088435247,
+      "loss": 0.0756,
+      "step": 885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019673733013785986,
+      "loss": 0.0785,
+      "step": 886
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019672721399199508,
+      "loss": 0.0761,
+      "step": 887
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019671708244836845,
+      "loss": 0.0752,
+      "step": 888
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019670693550859282,
+      "loss": 0.0747,
+      "step": 889
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019669677317428344,
+      "loss": 0.077,
+      "step": 890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019668659544705801,
+      "loss": 0.0754,
+      "step": 891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001966764023285367,
+      "loss": 0.0796,
+      "step": 892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019666619382034217,
+      "loss": 0.0776,
+      "step": 893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019665596992409943,
+      "loss": 0.0779,
+      "step": 894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019664573064143604,
+      "loss": 0.0754,
+      "step": 895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019663547597398195,
+      "loss": 0.0784,
+      "step": 896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019662520592336958,
+      "loss": 0.0796,
+      "step": 897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019661492049123376,
+      "loss": 0.074,
+      "step": 898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019660461967921182,
+      "loss": 0.0743,
+      "step": 899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019659430348894355,
+      "loss": 0.074,
+      "step": 900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019658397192207113,
+      "loss": 0.0744,
+      "step": 901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019657362498023924,
+      "loss": 0.0786,
+      "step": 902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019656326266509494,
+      "loss": 0.0786,
+      "step": 903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019655288497828782,
+      "loss": 0.0719,
+      "step": 904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019654249192146988,
+      "loss": 0.0748,
+      "step": 905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019653208349629554,
+      "loss": 0.0757,
+      "step": 906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001965216597044217,
+      "loss": 0.0745,
+      "step": 907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019651122054750768,
+      "loss": 0.076,
+      "step": 908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001965007660272153,
+      "loss": 0.0759,
+      "step": 909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019649029614520876,
+      "loss": 0.0786,
+      "step": 910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019647981090315474,
+      "loss": 0.0792,
+      "step": 911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019646931030272238,
+      "loss": 0.0749,
+      "step": 912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001964587943455832,
+      "loss": 0.0776,
+      "step": 913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001964482630334113,
+      "loss": 0.0761,
+      "step": 914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000196437716367883,
+      "loss": 0.0795,
+      "step": 915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001964271543506773,
+      "loss": 0.075,
+      "step": 916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019641657698347552,
+      "loss": 0.0743,
+      "step": 917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001964059842679614,
+      "loss": 0.0756,
+      "step": 918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001963953762058212,
+      "loss": 0.0806,
+      "step": 919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001963847527987436,
+      "loss": 0.0784,
+      "step": 920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001963741140484197,
+      "loss": 0.0762,
+      "step": 921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019636345995654306,
+      "loss": 0.0755,
+      "step": 922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001963527905248097,
+      "loss": 0.0788,
+      "step": 923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000196342105754918,
+      "loss": 0.0769,
+      "step": 924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001963314056485689,
+      "loss": 0.0754,
+      "step": 925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019632069020746572,
+      "loss": 0.0767,
+      "step": 926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001963099594333142,
+      "loss": 0.0746,
+      "step": 927
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019629921332782254,
+      "loss": 0.0741,
+      "step": 928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019628845189270137,
+      "loss": 0.0776,
+      "step": 929
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019627767512966384,
+      "loss": 0.0747,
+      "step": 930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019626688304042543,
+      "loss": 0.0766,
+      "step": 931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001962560756267041,
+      "loss": 0.0773,
+      "step": 932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001962452528902203,
+      "loss": 0.0738,
+      "step": 933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019623441483269683,
+      "loss": 0.0785,
+      "step": 934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019622356145585897,
+      "loss": 0.0796,
+      "step": 935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019621269276143445,
+      "loss": 0.0778,
+      "step": 936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019620180875115346,
+      "loss": 0.0788,
+      "step": 937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001961909094267486,
+      "loss": 0.0789,
+      "step": 938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019617999478995483,
+      "loss": 0.0785,
+      "step": 939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019616906484250972,
+      "loss": 0.0778,
+      "step": 940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019615811958615314,
+      "loss": 0.0764,
+      "step": 941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001961471590226274,
+      "loss": 0.0794,
+      "step": 942
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019613618315367733,
+      "loss": 0.0774,
+      "step": 943
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019612519198105015,
+      "loss": 0.0765,
+      "step": 944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001961141855064955,
+      "loss": 0.0763,
+      "step": 945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019610316373176547,
+      "loss": 0.0788,
+      "step": 946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019609212665861462,
+      "loss": 0.0774,
+      "step": 947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019608107428879988,
+      "loss": 0.076,
+      "step": 948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019607000662408066,
+      "loss": 0.074,
+      "step": 949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019605892366621877,
+      "loss": 0.0763,
+      "step": 950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019604782541697856,
+      "loss": 0.0793,
+      "step": 951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00019603671187812664,
+      "loss": 0.0768,
+      "step": 952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001960255830514322,
+      "loss": 0.0753,
+      "step": 953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001960144389386668,
+      "loss": 0.0765,
+      "step": 954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019600327954160442,
+      "loss": 0.0761,
+      "step": 955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019599210486202152,
+      "loss": 0.0762,
+      "step": 956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019598091490169694,
+      "loss": 0.0752,
+      "step": 957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019596970966241203,
+      "loss": 0.077,
+      "step": 958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019595848914595045,
+      "loss": 0.0744,
+      "step": 959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019594725335409847,
+      "loss": 0.0768,
+      "step": 960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001959360022886446,
+      "loss": 0.0768,
+      "step": 961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019592473595137988,
+      "loss": 0.077,
+      "step": 962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019591345434409783,
+      "loss": 0.0749,
+      "step": 963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019590215746859428,
+      "loss": 0.0755,
+      "step": 964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019589084532666755,
+      "loss": 0.0755,
+      "step": 965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019587951792011845,
+      "loss": 0.076,
+      "step": 966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019586817525075009,
+      "loss": 0.0749,
+      "step": 967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019585681732036813,
+      "loss": 0.0768,
+      "step": 968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019584544413078062,
+      "loss": 0.0769,
+      "step": 969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019583405568379797,
+      "loss": 0.0756,
+      "step": 970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019582265198123312,
+      "loss": 0.0763,
+      "step": 971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019581123302490138,
+      "loss": 0.0761,
+      "step": 972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019579979881662052,
+      "loss": 0.0752,
+      "step": 973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001957883493582107,
+      "loss": 0.0721,
+      "step": 974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019577688465149457,
+      "loss": 0.0778,
+      "step": 975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019576540469829715,
+      "loss": 0.079,
+      "step": 976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019575390950044588,
+      "loss": 0.0736,
+      "step": 977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019574239905977069,
+      "loss": 0.0745,
+      "step": 978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019573087337810386,
+      "loss": 0.0764,
+      "step": 979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019571933245728017,
+      "loss": 0.0767,
+      "step": 980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019570777629913675,
+      "loss": 0.0767,
+      "step": 981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019569620490551324,
+      "loss": 0.079,
+      "step": 982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019568461827825165,
+      "loss": 0.0748,
+      "step": 983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001956730164191964,
+      "loss": 0.0762,
+      "step": 984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001956613993301944,
+      "loss": 0.0772,
+      "step": 985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019564976701309488,
+      "loss": 0.0775,
+      "step": 986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019563811946974964,
+      "loss": 0.0777,
+      "step": 987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019562645670201276,
+      "loss": 0.0798,
+      "step": 988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019561477871174083,
+      "loss": 0.0772,
+      "step": 989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019560308550079286,
+      "loss": 0.0773,
+      "step": 990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019559137707103025,
+      "loss": 0.0746,
+      "step": 991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001955796534243168,
+      "loss": 0.0759,
+      "step": 992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019556791456251885,
+      "loss": 0.0762,
+      "step": 993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019555616048750495,
+      "loss": 0.0774,
+      "step": 994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019554439120114634,
+      "loss": 0.0749,
+      "step": 995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001955326067053165,
+      "loss": 0.0755,
+      "step": 996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019552080700189128,
+      "loss": 0.0747,
+      "step": 997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019550899209274918,
+      "loss": 0.0759,
+      "step": 998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001954971619797709,
+      "loss": 0.0785,
+      "step": 999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019548531666483968,
+      "loss": 0.0749,
+      "step": 1000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019547345614984114,
+      "loss": 0.0784,
+      "step": 1001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019546158043666335,
+      "loss": 0.0736,
+      "step": 1002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001954496895271967,
+      "loss": 0.0757,
+      "step": 1003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019543778342333415,
+      "loss": 0.0781,
+      "step": 1004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000195425862126971,
+      "loss": 0.0758,
+      "step": 1005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019541392564000488,
+      "loss": 0.077,
+      "step": 1006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019540197396433607,
+      "loss": 0.0754,
+      "step": 1007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000195390007101867,
+      "loss": 0.0767,
+      "step": 1008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001953780250545027,
+      "loss": 0.0741,
+      "step": 1009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001953660278241506,
+      "loss": 0.0756,
+      "step": 1010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019535401541272046,
+      "loss": 0.0782,
+      "step": 1011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001953419878221245,
+      "loss": 0.0733,
+      "step": 1012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019532994505427737,
+      "loss": 0.0766,
+      "step": 1013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019531788711109614,
+      "loss": 0.0807,
+      "step": 1014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019530581399450033,
+      "loss": 0.0776,
+      "step": 1015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019529372570641171,
+      "loss": 0.0783,
+      "step": 1016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019528162224875467,
+      "loss": 0.0773,
+      "step": 1017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019526950362345593,
+      "loss": 0.078,
+      "step": 1018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001952573698324446,
+      "loss": 0.078,
+      "step": 1019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001952452208776522,
+      "loss": 0.0762,
+      "step": 1020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00019523305676101275,
+      "loss": 0.0745,
+      "step": 1021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001952208774844626,
+      "loss": 0.077,
+      "step": 1022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019520868304994052,
+      "loss": 0.0755,
+      "step": 1023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019519647345938774,
+      "loss": 0.0759,
+      "step": 1024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019518424871474785,
+      "loss": 0.075,
+      "step": 1025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001951720088179669,
+      "loss": 0.0777,
+      "step": 1026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001951597537709933,
+      "loss": 0.0783,
+      "step": 1027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019514748357577798,
+      "loss": 0.0747,
+      "step": 1028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019513519823427408,
+      "loss": 0.0781,
+      "step": 1029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019512289774843735,
+      "loss": 0.0779,
+      "step": 1030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019511058212022585,
+      "loss": 0.0746,
+      "step": 1031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019509825135160006,
+      "loss": 0.0731,
+      "step": 1032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001950859054445229,
+      "loss": 0.0764,
+      "step": 1033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001950735444009597,
+      "loss": 0.0765,
+      "step": 1034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00019506116822287818,
+      "loss": 0.0758,
+      "step": 1035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019504877691224848,
+      "loss": 0.0772,
+      "step": 1036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019503637047104308,
+      "loss": 0.075,
+      "step": 1037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000195023948901237,
+      "loss": 0.0776,
+      "step": 1038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019501151220480755,
+      "loss": 0.0755,
+      "step": 1039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019499906038373456,
+      "loss": 0.0758,
+      "step": 1040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019498659344000015,
+      "loss": 0.0762,
+      "step": 1041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019497411137558887,
+      "loss": 0.0765,
+      "step": 1042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001949616141924878,
+      "loss": 0.075,
+      "step": 1043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019494910189268627,
+      "loss": 0.0772,
+      "step": 1044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001949365744781761,
+      "loss": 0.0768,
+      "step": 1045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001949240319509515,
+      "loss": 0.0762,
+      "step": 1046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001949114743130091,
+      "loss": 0.0798,
+      "step": 1047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00019489890156634788,
+      "loss": 0.0755,
+      "step": 1048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019488631371296928,
+      "loss": 0.0751,
+      "step": 1049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019487371075487713,
+      "loss": 0.0758,
+      "step": 1050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019486109269407765,
+      "loss": 0.0756,
+      "step": 1051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001948484595325795,
+      "loss": 0.0751,
+      "step": 1052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019483581127239377,
+      "loss": 0.0752,
+      "step": 1053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001948231479155338,
+      "loss": 0.0788,
+      "step": 1054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019481046946401547,
+      "loss": 0.0767,
+      "step": 1055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019479777591985706,
+      "loss": 0.0784,
+      "step": 1056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019478506728507923,
+      "loss": 0.0768,
+      "step": 1057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000194772343561705,
+      "loss": 0.0763,
+      "step": 1058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019475960475175984,
+      "loss": 0.0739,
+      "step": 1059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001947468508572716,
+      "loss": 0.0755,
+      "step": 1060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001947340818802706,
+      "loss": 0.0766,
+      "step": 1061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00019472129782278943,
+      "loss": 0.0777,
+      "step": 1062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019470849868686316,
+      "loss": 0.0782,
+      "step": 1063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001946956844745293,
+      "loss": 0.0768,
+      "step": 1064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019468285518782765,
+      "loss": 0.0778,
+      "step": 1065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001946700108288005,
+      "loss": 0.0789,
+      "step": 1066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019465715139949254,
+      "loss": 0.0774,
+      "step": 1067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001946442769019508,
+      "loss": 0.0771,
+      "step": 1068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019463138733822474,
+      "loss": 0.0769,
+      "step": 1069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019461848271036623,
+      "loss": 0.0801,
+      "step": 1070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001946055630204295,
+      "loss": 0.0783,
+      "step": 1071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001945926282704712,
+      "loss": 0.0759,
+      "step": 1072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019457967846255042,
+      "loss": 0.0759,
+      "step": 1073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019456671359872858,
+      "loss": 0.0735,
+      "step": 1074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00019455373368106952,
+      "loss": 0.0768,
+      "step": 1075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001945407387116395,
+      "loss": 0.0763,
+      "step": 1076
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019452772869250716,
+      "loss": 0.0772,
+      "step": 1077
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019451470362574351,
+      "loss": 0.0758,
+      "step": 1078
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000194501663513422,
+      "loss": 0.0782,
+      "step": 1079
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019448860835761842,
+      "loss": 0.0756,
+      "step": 1080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000194475538160411,
+      "loss": 0.0759,
+      "step": 1081
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019446245292388038,
+      "loss": 0.0772,
+      "step": 1082
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019444935265010952,
+      "loss": 0.0746,
+      "step": 1083
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019443623734118388,
+      "loss": 0.0759,
+      "step": 1084
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019442310699919118,
+      "loss": 0.0775,
+      "step": 1085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019440996162622167,
+      "loss": 0.0755,
+      "step": 1086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019439680122436789,
+      "loss": 0.0784,
+      "step": 1087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019438362579572482,
+      "loss": 0.0785,
+      "step": 1088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00019437043534238985,
+      "loss": 0.0788,
+      "step": 1089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019435722986646268,
+      "loss": 0.075,
+      "step": 1090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019434400937004548,
+      "loss": 0.0729,
+      "step": 1091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019433077385524284,
+      "loss": 0.0768,
+      "step": 1092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001943175233241616,
+      "loss": 0.079,
+      "step": 1093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019430425777891116,
+      "loss": 0.0763,
+      "step": 1094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019429097722160318,
+      "loss": 0.0767,
+      "step": 1095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019427768165435175,
+      "loss": 0.0745,
+      "step": 1096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001942643710792734,
+      "loss": 0.077,
+      "step": 1097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019425104549848697,
+      "loss": 0.0776,
+      "step": 1098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019423770491411375,
+      "loss": 0.0771,
+      "step": 1099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019422434932827738,
+      "loss": 0.0769,
+      "step": 1100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00019421097874310392,
+      "loss": 0.0774,
+      "step": 1101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001941975931607218,
+      "loss": 0.0769,
+      "step": 1102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019418419258326178,
+      "loss": 0.0774,
+      "step": 1103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019417077701285712,
+      "loss": 0.0754,
+      "step": 1104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019415734645164342,
+      "loss": 0.0743,
+      "step": 1105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019414390090175863,
+      "loss": 0.0745,
+      "step": 1106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019413044036534308,
+      "loss": 0.077,
+      "step": 1107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019411696484453961,
+      "loss": 0.0774,
+      "step": 1108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001941034743414933,
+      "loss": 0.072,
+      "step": 1109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019408996885835164,
+      "loss": 0.0782,
+      "step": 1110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019407644839726458,
+      "loss": 0.0791,
+      "step": 1111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001940629129603844,
+      "loss": 0.0763,
+      "step": 1112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019404936254986576,
+      "loss": 0.078,
+      "step": 1113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019403579716786572,
+      "loss": 0.075,
+      "step": 1114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00019402221681654372,
+      "loss": 0.0741,
+      "step": 1115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001940086214980616,
+      "loss": 0.0761,
+      "step": 1116
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019399501121458352,
+      "loss": 0.0773,
+      "step": 1117
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001939813859682761,
+      "loss": 0.076,
+      "step": 1118
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001939677457613083,
+      "loss": 0.0764,
+      "step": 1119
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001939540905958515,
+      "loss": 0.0735,
+      "step": 1120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019394042047407939,
+      "loss": 0.0751,
+      "step": 1121
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019392673539816812,
+      "loss": 0.0786,
+      "step": 1122
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001939130353702961,
+      "loss": 0.0801,
+      "step": 1123
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019389932039264428,
+      "loss": 0.075,
+      "step": 1124
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019388559046739593,
+      "loss": 0.0779,
+      "step": 1125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019387184559673662,
+      "loss": 0.0774,
+      "step": 1126
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019385808578285444,
+      "loss": 0.0767,
+      "step": 1127
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001938443110279397,
+      "loss": 0.0805,
+      "step": 1128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00019383052133418516,
+      "loss": 0.0743,
+      "step": 1129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000193816716703786,
+      "loss": 0.0751,
+      "step": 1130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001938028971389398,
+      "loss": 0.0752,
+      "step": 1131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019378906264184637,
+      "loss": 0.0748,
+      "step": 1132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019377521321470805,
+      "loss": 0.0752,
+      "step": 1133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019376134885972947,
+      "loss": 0.0762,
+      "step": 1134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019374746957911767,
+      "loss": 0.0779,
+      "step": 1135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019373357537508205,
+      "loss": 0.076,
+      "step": 1136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001937196662498344,
+      "loss": 0.0777,
+      "step": 1137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001937057422055889,
+      "loss": 0.0788,
+      "step": 1138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019369180324456202,
+      "loss": 0.0784,
+      "step": 1139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019367784936897273,
+      "loss": 0.0759,
+      "step": 1140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019366388058104228,
+      "loss": 0.0773,
+      "step": 1141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019364989688299433,
+      "loss": 0.0753,
+      "step": 1142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00019363589827705492,
+      "loss": 0.0742,
+      "step": 1143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019362188476545247,
+      "loss": 0.0733,
+      "step": 1144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001936078563504177,
+      "loss": 0.0766,
+      "step": 1145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019359381303418386,
+      "loss": 0.0816,
+      "step": 1146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019357975481898632,
+      "loss": 0.0764,
+      "step": 1147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001935656817070631,
+      "loss": 0.0788,
+      "step": 1148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001935515937006544,
+      "loss": 0.0742,
+      "step": 1149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001935374908020029,
+      "loss": 0.0745,
+      "step": 1150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019352337301335355,
+      "loss": 0.0762,
+      "step": 1151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019350924033695376,
+      "loss": 0.0754,
+      "step": 1152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019349509277505325,
+      "loss": 0.0778,
+      "step": 1153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001934809303299042,
+      "loss": 0.0776,
+      "step": 1154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000193466753003761,
+      "loss": 0.0749,
+      "step": 1155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00019345256079888058,
+      "loss": 0.0735,
+      "step": 1156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019343835371752214,
+      "loss": 0.0765,
+      "step": 1157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019342413176194723,
+      "loss": 0.077,
+      "step": 1158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019340989493441986,
+      "loss": 0.0762,
+      "step": 1159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001933956432372063,
+      "loss": 0.0758,
+      "step": 1160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001933813766725753,
+      "loss": 0.0765,
+      "step": 1161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001933670952427979,
+      "loss": 0.0773,
+      "step": 1162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019335279895014748,
+      "loss": 0.078,
+      "step": 1163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019333848779689992,
+      "loss": 0.0772,
+      "step": 1164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019332416178533327,
+      "loss": 0.0732,
+      "step": 1165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019330982091772815,
+      "loss": 0.0777,
+      "step": 1166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001932954651963674,
+      "loss": 0.0754,
+      "step": 1167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019328109462353626,
+      "loss": 0.0759,
+      "step": 1168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00019326670920152237,
+      "loss": 0.0758,
+      "step": 1169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001932523089326157,
+      "loss": 0.0762,
+      "step": 1170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019323789381910856,
+      "loss": 0.0765,
+      "step": 1171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019322346386329572,
+      "loss": 0.0762,
+      "step": 1172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001932090190674742,
+      "loss": 0.0763,
+      "step": 1173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019319455943394347,
+      "loss": 0.0751,
+      "step": 1174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001931800849650053,
+      "loss": 0.0754,
+      "step": 1175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001931655956629638,
+      "loss": 0.0772,
+      "step": 1176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019315109153012557,
+      "loss": 0.0762,
+      "step": 1177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019313657256879942,
+      "loss": 0.0771,
+      "step": 1178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019312203878129664,
+      "loss": 0.0774,
+      "step": 1179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019310749016993077,
+      "loss": 0.0732,
+      "step": 1180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001930929267370178,
+      "loss": 0.0755,
+      "step": 1181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00019307834848487606,
+      "loss": 0.0749,
+      "step": 1182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001930637554158262,
+      "loss": 0.0739,
+      "step": 1183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019304914753219126,
+      "loss": 0.0756,
+      "step": 1184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019303452483629662,
+      "loss": 0.0753,
+      "step": 1185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019301988733047006,
+      "loss": 0.0755,
+      "step": 1186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001930052350170417,
+      "loss": 0.0778,
+      "step": 1187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019299056789834394,
+      "loss": 0.0772,
+      "step": 1188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019297588597671163,
+      "loss": 0.0725,
+      "step": 1189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000192961189254482,
+      "loss": 0.0784,
+      "step": 1190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001929464777339945,
+      "loss": 0.0778,
+      "step": 1191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019293175141759107,
+      "loss": 0.0756,
+      "step": 1192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019291701030761595,
+      "loss": 0.0764,
+      "step": 1193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019290225440641575,
+      "loss": 0.0722,
+      "step": 1194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019288748371633937,
+      "loss": 0.0736,
+      "step": 1195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019287269823973818,
+      "loss": 0.0742,
+      "step": 1196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00019285789797896584,
+      "loss": 0.0732,
+      "step": 1197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019284308293637836,
+      "loss": 0.0758,
+      "step": 1198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019282825311433408,
+      "loss": 0.0769,
+      "step": 1199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019281340851519372,
+      "loss": 0.0796,
+      "step": 1200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001927985491413204,
+      "loss": 0.077,
+      "step": 1201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001927836749950795,
+      "loss": 0.0773,
+      "step": 1202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019276878607883883,
+      "loss": 0.0771,
+      "step": 1203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019275388239496852,
+      "loss": 0.0757,
+      "step": 1204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019273896394584103,
+      "loss": 0.0746,
+      "step": 1205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019272403073383122,
+      "loss": 0.0774,
+      "step": 1206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019270908276131622,
+      "loss": 0.0764,
+      "step": 1207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001926941200306756,
+      "loss": 0.0775,
+      "step": 1208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019267914254429122,
+      "loss": 0.0765,
+      "step": 1209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00019266415030454732,
+      "loss": 0.078,
+      "step": 1210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019264914331383048,
+      "loss": 0.0768,
+      "step": 1211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019263412157452962,
+      "loss": 0.0772,
+      "step": 1212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019261908508903602,
+      "loss": 0.0771,
+      "step": 1213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019260403385974327,
+      "loss": 0.0745,
+      "step": 1214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019258896788904733,
+      "loss": 0.0753,
+      "step": 1215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019257388717934662,
+      "loss": 0.0775,
+      "step": 1216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019255879173304168,
+      "loss": 0.0759,
+      "step": 1217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019254368155253556,
+      "loss": 0.0769,
+      "step": 1218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001925285566402336,
+      "loss": 0.0774,
+      "step": 1219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019251341699854356,
+      "loss": 0.0769,
+      "step": 1220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019249826262987538,
+      "loss": 0.0771,
+      "step": 1221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001924830935366415,
+      "loss": 0.0767,
+      "step": 1222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019246790972125667,
+      "loss": 0.0753,
+      "step": 1223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00019245271118613791,
+      "loss": 0.0745,
+      "step": 1224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019243749793370469,
+      "loss": 0.076,
+      "step": 1225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019242226996637874,
+      "loss": 0.0775,
+      "step": 1226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019240702728658414,
+      "loss": 0.0784,
+      "step": 1227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001923917698967474,
+      "loss": 0.0756,
+      "step": 1228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019237649779929722,
+      "loss": 0.0767,
+      "step": 1229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019236121099666483,
+      "loss": 0.0769,
+      "step": 1230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001923459094912836,
+      "loss": 0.0768,
+      "step": 1231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001923305932855894,
+      "loss": 0.079,
+      "step": 1232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019231526238202033,
+      "loss": 0.0752,
+      "step": 1233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019229991678301693,
+      "loss": 0.0778,
+      "step": 1234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019228455649102197,
+      "loss": 0.0798,
+      "step": 1235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019226918150848068,
+      "loss": 0.0768,
+      "step": 1236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00019225379183784053,
+      "loss": 0.074,
+      "step": 1237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019223838748155133,
+      "loss": 0.0745,
+      "step": 1238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019222296844206534,
+      "loss": 0.0796,
+      "step": 1239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019220753472183702,
+      "loss": 0.0756,
+      "step": 1240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019219208632332324,
+      "loss": 0.0756,
+      "step": 1241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019217662324898317,
+      "loss": 0.0728,
+      "step": 1242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001921611455012784,
+      "loss": 0.0768,
+      "step": 1243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019214565308267272,
+      "loss": 0.0772,
+      "step": 1244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019213014599563238,
+      "loss": 0.0741,
+      "step": 1245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001921146242426259,
+      "loss": 0.0764,
+      "step": 1246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019209908782612413,
+      "loss": 0.0759,
+      "step": 1247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001920835367486003,
+      "loss": 0.0778,
+      "step": 1248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001920679710125299,
+      "loss": 0.075,
+      "step": 1249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019205239062039086,
+      "loss": 0.0767,
+      "step": 1250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00019203679557466336,
+      "loss": 0.0728,
+      "step": 1251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019202118587782996,
+      "loss": 0.0783,
+      "step": 1252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019200556153237548,
+      "loss": 0.0773,
+      "step": 1253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019198992254078717,
+      "loss": 0.076,
+      "step": 1254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019197426890555448,
+      "loss": 0.0741,
+      "step": 1255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019195860062916936,
+      "loss": 0.0759,
+      "step": 1256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019194291771412595,
+      "loss": 0.079,
+      "step": 1257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019192722016292085,
+      "loss": 0.0784,
+      "step": 1258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001919115079780528,
+      "loss": 0.0751,
+      "step": 1259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019189578116202307,
+      "loss": 0.0757,
+      "step": 1260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019188003971733515,
+      "loss": 0.0788,
+      "step": 1261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019186428364649486,
+      "loss": 0.0751,
+      "step": 1262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019184851295201037,
+      "loss": 0.0757,
+      "step": 1263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001918327276363922,
+      "loss": 0.0735,
+      "step": 1264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00019181692770215322,
+      "loss": 0.0733,
+      "step": 1265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001918011131518085,
+      "loss": 0.0785,
+      "step": 1266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00019178528398787551,
+      "loss": 0.0741,
+      "step": 1267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001917694402128741,
+      "loss": 0.0748,
+      "step": 1268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00019175358182932644,
+      "loss": 0.0758,
+      "step": 1269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001917377088397569,
+      "loss": 0.0752,
+      "step": 1270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001917218212466923,
+      "loss": 0.0767,
+      "step": 1271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00019170591905266174,
+      "loss": 0.0771,
+      "step": 1272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001916900022601967,
+      "loss": 0.0727,
+      "step": 1273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00019167407087183086,
+      "loss": 0.0746,
+      "step": 1274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001916581248901003,
+      "loss": 0.0793,
+      "step": 1275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001916421643175435,
+      "loss": 0.0727,
+      "step": 1276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001916261891567011,
+      "loss": 0.0736,
+      "step": 1277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00019161019941011618,
+      "loss": 0.0765,
+      "step": 1278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001915941950803341,
+      "loss": 0.0743,
+      "step": 1279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019157817616990256,
+      "loss": 0.0742,
+      "step": 1280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001915621426813716,
+      "loss": 0.0774,
+      "step": 1281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019154609461729348,
+      "loss": 0.0791,
+      "step": 1282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019153003198022286,
+      "loss": 0.075,
+      "step": 1283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019151395477271676,
+      "loss": 0.0747,
+      "step": 1284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019149786299733448,
+      "loss": 0.077,
+      "step": 1285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019148175665663754,
+      "loss": 0.0786,
+      "step": 1286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019146563575318996,
+      "loss": 0.0773,
+      "step": 1287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019144950028955794,
+      "loss": 0.0756,
+      "step": 1288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019143335026831007,
+      "loss": 0.0751,
+      "step": 1289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001914171856920172,
+      "loss": 0.0755,
+      "step": 1290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00019140100656325255,
+      "loss": 0.077,
+      "step": 1291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001913848128845916,
+      "loss": 0.0782,
+      "step": 1292
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019136860465861224,
+      "loss": 0.0767,
+      "step": 1293
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019135238188789456,
+      "loss": 0.0802,
+      "step": 1294
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019133614457502108,
+      "loss": 0.0748,
+      "step": 1295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019131989272257651,
+      "loss": 0.0748,
+      "step": 1296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019130362633314803,
+      "loss": 0.0796,
+      "step": 1297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019128734540932495,
+      "loss": 0.0774,
+      "step": 1298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019127104995369902,
+      "loss": 0.0745,
+      "step": 1299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001912547399688643,
+      "loss": 0.079,
+      "step": 1300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019123841545741713,
+      "loss": 0.0773,
+      "step": 1301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019122207642195615,
+      "loss": 0.0752,
+      "step": 1302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001912057228650823,
+      "loss": 0.0763,
+      "step": 1303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019118935478939898,
+      "loss": 0.0718,
+      "step": 1304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00019117297219751166,
+      "loss": 0.0761,
+      "step": 1305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019115657509202825,
+      "loss": 0.0742,
+      "step": 1306
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019114016347555906,
+      "loss": 0.0791,
+      "step": 1307
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001911237373507165,
+      "loss": 0.0756,
+      "step": 1308
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019110729672011548,
+      "loss": 0.0786,
+      "step": 1309
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019109084158637312,
+      "loss": 0.0754,
+      "step": 1310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019107437195210888,
+      "loss": 0.0742,
+      "step": 1311
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001910578878199445,
+      "loss": 0.0787,
+      "step": 1312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019104138919250402,
+      "loss": 0.0761,
+      "step": 1313
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001910248760724139,
+      "loss": 0.0781,
+      "step": 1314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019100834846230277,
+      "loss": 0.0738,
+      "step": 1315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019099180636480162,
+      "loss": 0.0743,
+      "step": 1316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019097524978254376,
+      "loss": 0.0821,
+      "step": 1317
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019095867871816476,
+      "loss": 0.0779,
+      "step": 1318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00019094209317430255,
+      "loss": 0.076,
+      "step": 1319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019092549315359732,
+      "loss": 0.0741,
+      "step": 1320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001909088786586916,
+      "loss": 0.0748,
+      "step": 1321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019089224969223023,
+      "loss": 0.0758,
+      "step": 1322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019087560625686032,
+      "loss": 0.0749,
+      "step": 1323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019085894835523125,
+      "loss": 0.0736,
+      "step": 1324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019084227598999484,
+      "loss": 0.0778,
+      "step": 1325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019082558916380505,
+      "loss": 0.0754,
+      "step": 1326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019080888787931827,
+      "loss": 0.0768,
+      "step": 1327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019079217213919305,
+      "loss": 0.076,
+      "step": 1328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019077544194609042,
+      "loss": 0.0755,
+      "step": 1329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019075869730267352,
+      "loss": 0.0781,
+      "step": 1330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00019074193821160798,
+      "loss": 0.0771,
+      "step": 1331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001907251646755616,
+      "loss": 0.0769,
+      "step": 1332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001907083766972045,
+      "loss": 0.0767,
+      "step": 1333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019069157427920917,
+      "loss": 0.0773,
+      "step": 1334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001906747574242503,
+      "loss": 0.0743,
+      "step": 1335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001906579261350049,
+      "loss": 0.0719,
+      "step": 1336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019064108041415235,
+      "loss": 0.0772,
+      "step": 1337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019062422026437428,
+      "loss": 0.0753,
+      "step": 1338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019060734568835457,
+      "loss": 0.0735,
+      "step": 1339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019059045668877946,
+      "loss": 0.075,
+      "step": 1340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001905735532683375,
+      "loss": 0.0784,
+      "step": 1341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019055663542971946,
+      "loss": 0.0752,
+      "step": 1342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019053970317561847,
+      "loss": 0.0765,
+      "step": 1343
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019052275650872992,
+      "loss": 0.0775,
+      "step": 1344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019050579543175152,
+      "loss": 0.0761,
+      "step": 1345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00019048881994738323,
+      "loss": 0.0744,
+      "step": 1346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001904718300583274,
+      "loss": 0.0761,
+      "step": 1347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019045482576728854,
+      "loss": 0.0777,
+      "step": 1348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001904378070769736,
+      "loss": 0.0775,
+      "step": 1349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019042077399009162,
+      "loss": 0.0752,
+      "step": 1350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019040372650935415,
+      "loss": 0.0773,
+      "step": 1351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019038666463747493,
+      "loss": 0.0727,
+      "step": 1352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019036958837716998,
+      "loss": 0.076,
+      "step": 1353
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001903524977311576,
+      "loss": 0.0735,
+      "step": 1354
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019033539270215844,
+      "loss": 0.0766,
+      "step": 1355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001903182732928954,
+      "loss": 0.0765,
+      "step": 1356
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019030113950609367,
+      "loss": 0.0775,
+      "step": 1357
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019028399134448073,
+      "loss": 0.0752,
+      "step": 1358
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00019026682881078636,
+      "loss": 0.0735,
+      "step": 1359
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019024965190774263,
+      "loss": 0.0753,
+      "step": 1360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001902324606380839,
+      "loss": 0.0763,
+      "step": 1361
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019021525500454676,
+      "loss": 0.0723,
+      "step": 1362
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019019803500987015,
+      "loss": 0.079,
+      "step": 1363
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019018080065679528,
+      "loss": 0.0743,
+      "step": 1364
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019016355194806565,
+      "loss": 0.0767,
+      "step": 1365
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019014628888642705,
+      "loss": 0.0753,
+      "step": 1366
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019012901147462752,
+      "loss": 0.0781,
+      "step": 1367
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019011171971541741,
+      "loss": 0.078,
+      "step": 1368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019009441361154936,
+      "loss": 0.0737,
+      "step": 1369
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019007709316577832,
+      "loss": 0.0764,
+      "step": 1370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019005975838086138,
+      "loss": 0.0757,
+      "step": 1371
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00019004240925955815,
+      "loss": 0.0767,
+      "step": 1372
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001900250458046303,
+      "loss": 0.076,
+      "step": 1373
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00019000766801884194,
+      "loss": 0.0737,
+      "step": 1374
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018999027590495932,
+      "loss": 0.0774,
+      "step": 1375
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018997286946575114,
+      "loss": 0.0773,
+      "step": 1376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001899554487039882,
+      "loss": 0.0765,
+      "step": 1377
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018993801362244373,
+      "loss": 0.0773,
+      "step": 1378
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018992056422389314,
+      "loss": 0.0734,
+      "step": 1379
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001899031005111142,
+      "loss": 0.0728,
+      "step": 1380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018988562248688684,
+      "loss": 0.0756,
+      "step": 1381
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018986813015399342,
+      "loss": 0.076,
+      "step": 1382
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018985062351521847,
+      "loss": 0.076,
+      "step": 1383
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001898331025733488,
+      "loss": 0.0761,
+      "step": 1384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00018981556733117355,
+      "loss": 0.075,
+      "step": 1385
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001897980177914841,
+      "loss": 0.0774,
+      "step": 1386
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018978045395707418,
+      "loss": 0.0776,
+      "step": 1387
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018976287583073964,
+      "loss": 0.0763,
+      "step": 1388
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018974528341527875,
+      "loss": 0.0764,
+      "step": 1389
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000189727676713492,
+      "loss": 0.0774,
+      "step": 1390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018971005572818213,
+      "loss": 0.0758,
+      "step": 1391
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018969242046215417,
+      "loss": 0.0736,
+      "step": 1392
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001896747709182155,
+      "loss": 0.073,
+      "step": 1393
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018965710709917567,
+      "loss": 0.0764,
+      "step": 1394
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018963942900784652,
+      "loss": 0.0765,
+      "step": 1395
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018962173664704223,
+      "loss": 0.0725,
+      "step": 1396
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018960403001957912,
+      "loss": 0.077,
+      "step": 1397
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001895863091282759,
+      "loss": 0.0778,
+      "step": 1398
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018956857397595358,
+      "loss": 0.0764,
+      "step": 1399
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001895508245654353,
+      "loss": 0.0736,
+      "step": 1400
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018953306089954656,
+      "loss": 0.0728,
+      "step": 1401
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018951528298111515,
+      "loss": 0.0771,
+      "step": 1402
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000189497490812971,
+      "loss": 0.0717,
+      "step": 1403
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001894796843979465,
+      "loss": 0.0786,
+      "step": 1404
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018946186373887617,
+      "loss": 0.0756,
+      "step": 1405
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018944402883859683,
+      "loss": 0.0737,
+      "step": 1406
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001894261796999476,
+      "loss": 0.075,
+      "step": 1407
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001894083163257698,
+      "loss": 0.0747,
+      "step": 1408
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018939043871890707,
+      "loss": 0.0747,
+      "step": 1409
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001893725468822053,
+      "loss": 0.0768,
+      "step": 1410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018935464081851265,
+      "loss": 0.0731,
+      "step": 1411
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018933672053067957,
+      "loss": 0.0776,
+      "step": 1412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001893187860215587,
+      "loss": 0.075,
+      "step": 1413
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000189300837294005,
+      "loss": 0.0774,
+      "step": 1414
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018928287435087567,
+      "loss": 0.0781,
+      "step": 1415
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018926489719503023,
+      "loss": 0.0731,
+      "step": 1416
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018924690582933042,
+      "loss": 0.0777,
+      "step": 1417
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018922890025664018,
+      "loss": 0.0769,
+      "step": 1418
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018921088047982583,
+      "loss": 0.0772,
+      "step": 1419
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018919284650175584,
+      "loss": 0.0747,
+      "step": 1420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018917479832530104,
+      "loss": 0.0739,
+      "step": 1421
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018915673595333444,
+      "loss": 0.0766,
+      "step": 1422
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018913865938873138,
+      "loss": 0.0781,
+      "step": 1423
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001891205686343694,
+      "loss": 0.0748,
+      "step": 1424
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018910246369312832,
+      "loss": 0.0764,
+      "step": 1425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001890843445678902,
+      "loss": 0.0752,
+      "step": 1426
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00018906621126153945,
+      "loss": 0.0749,
+      "step": 1427
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001890480637769626,
+      "loss": 0.0739,
+      "step": 1428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018902990211704852,
+      "loss": 0.0803,
+      "step": 1429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001890117262846883,
+      "loss": 0.0765,
+      "step": 1430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018899353628277533,
+      "loss": 0.0768,
+      "step": 1431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018897533211420525,
+      "loss": 0.0778,
+      "step": 1432
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001889571137818759,
+      "loss": 0.0767,
+      "step": 1433
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018893888128868738,
+      "loss": 0.0748,
+      "step": 1434
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018892063463754214,
+      "loss": 0.0743,
+      "step": 1435
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018890237383134483,
+      "loss": 0.0758,
+      "step": 1436
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001888840988730023,
+      "loss": 0.0758,
+      "step": 1437
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018886580976542366,
+      "loss": 0.0746,
+      "step": 1438
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018884750651152038,
+      "loss": 0.0756,
+      "step": 1439
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018882918911420607,
+      "loss": 0.0791,
+      "step": 1440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018881085757639662,
+      "loss": 0.0739,
+      "step": 1441
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001887925119010102,
+      "loss": 0.0734,
+      "step": 1442
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018877415209096724,
+      "loss": 0.0737,
+      "step": 1443
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018875577814919035,
+      "loss": 0.0773,
+      "step": 1444
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018873739007860442,
+      "loss": 0.0771,
+      "step": 1445
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018871898788213665,
+      "loss": 0.0741,
+      "step": 1446
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001887005715627164,
+      "loss": 0.0762,
+      "step": 1447
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018868214112327536,
+      "loss": 0.0782,
+      "step": 1448
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018866369656674736,
+      "loss": 0.0759,
+      "step": 1449
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001886452378960686,
+      "loss": 0.0771,
+      "step": 1450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018862676511417745,
+      "loss": 0.0794,
+      "step": 1451
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001886082782240145,
+      "loss": 0.0736,
+      "step": 1452
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018858977722852275,
+      "loss": 0.0755,
+      "step": 1453
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00018857126213064722,
+      "loss": 0.0761,
+      "step": 1454
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018855273293333531,
+      "loss": 0.0719,
+      "step": 1455
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018853418963953665,
+      "loss": 0.0732,
+      "step": 1456
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001885156322522031,
+      "loss": 0.0772,
+      "step": 1457
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001884970607742887,
+      "loss": 0.0765,
+      "step": 1458
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001884784752087499,
+      "loss": 0.073,
+      "step": 1459
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018845987555854524,
+      "loss": 0.0786,
+      "step": 1460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018844126182663553,
+      "loss": 0.0747,
+      "step": 1461
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001884226340159839,
+      "loss": 0.0758,
+      "step": 1462
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018840399212955559,
+      "loss": 0.0747,
+      "step": 1463
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018838533617031823,
+      "loss": 0.0771,
+      "step": 1464
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018836666614124157,
+      "loss": 0.0793,
+      "step": 1465
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018834798204529768,
+      "loss": 0.0772,
+      "step": 1466
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018832928388546076,
+      "loss": 0.0756,
+      "step": 1467
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018831057166470737,
+      "loss": 0.0737,
+      "step": 1468
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001882918453860163,
+      "loss": 0.0737,
+      "step": 1469
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001882731050523685,
+      "loss": 0.0752,
+      "step": 1470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001882543506667472,
+      "loss": 0.0787,
+      "step": 1471
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018823558223213786,
+      "loss": 0.0748,
+      "step": 1472
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018821679975152816,
+      "loss": 0.075,
+      "step": 1473
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018819800322790808,
+      "loss": 0.0747,
+      "step": 1474
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018817919266426978,
+      "loss": 0.0716,
+      "step": 1475
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018816036806360763,
+      "loss": 0.0744,
+      "step": 1476
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001881415294289183,
+      "loss": 0.0748,
+      "step": 1477
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001881226767632007,
+      "loss": 0.0761,
+      "step": 1478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001881038100694559,
+      "loss": 0.0754,
+      "step": 1479
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018808492935068718,
+      "loss": 0.077,
+      "step": 1480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00018806603460990023,
+      "loss": 0.0765,
+      "step": 1481
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018804712585010277,
+      "loss": 0.0776,
+      "step": 1482
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001880282030743049,
+      "loss": 0.0749,
+      "step": 1483
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018800926628551886,
+      "loss": 0.0762,
+      "step": 1484
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001879903154867591,
+      "loss": 0.0736,
+      "step": 1485
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018797135068104247,
+      "loss": 0.0742,
+      "step": 1486
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018795237187138781,
+      "loss": 0.0753,
+      "step": 1487
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001879333790608164,
+      "loss": 0.0764,
+      "step": 1488
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018791437225235155,
+      "loss": 0.077,
+      "step": 1489
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018789535144901902,
+      "loss": 0.0773,
+      "step": 1490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018787631665384663,
+      "loss": 0.0743,
+      "step": 1491
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018785726786986446,
+      "loss": 0.0752,
+      "step": 1492
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001878382051001049,
+      "loss": 0.0765,
+      "step": 1493
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018781912834760247,
+      "loss": 0.0766,
+      "step": 1494
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018780003761539394,
+      "loss": 0.0751,
+      "step": 1495
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001877809329065183,
+      "loss": 0.0766,
+      "step": 1496
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018776181422401682,
+      "loss": 0.076,
+      "step": 1497
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018774268157093296,
+      "loss": 0.0758,
+      "step": 1498
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018772353495031236,
+      "loss": 0.077,
+      "step": 1499
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018770437436520294,
+      "loss": 0.0739,
+      "step": 1500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018768519981865486,
+      "loss": 0.0766,
+      "step": 1501
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001876660113137204,
+      "loss": 0.0754,
+      "step": 1502
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018764680885345416,
+      "loss": 0.0765,
+      "step": 1503
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018762759244091295,
+      "loss": 0.0765,
+      "step": 1504
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018760836207915577,
+      "loss": 0.0749,
+      "step": 1505
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018758911777124384,
+      "loss": 0.0747,
+      "step": 1506
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018756985952024063,
+      "loss": 0.0789,
+      "step": 1507
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00018755058732921183,
+      "loss": 0.0782,
+      "step": 1508
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001875313012012253,
+      "loss": 0.0759,
+      "step": 1509
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018751200113935112,
+      "loss": 0.0777,
+      "step": 1510
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001874926871466617,
+      "loss": 0.0746,
+      "step": 1511
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018747335922623154,
+      "loss": 0.077,
+      "step": 1512
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018745401738113737,
+      "loss": 0.0758,
+      "step": 1513
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018743466161445823,
+      "loss": 0.0782,
+      "step": 1514
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018741529192927526,
+      "loss": 0.0747,
+      "step": 1515
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018739590832867197,
+      "loss": 0.0749,
+      "step": 1516
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018737651081573387,
+      "loss": 0.0758,
+      "step": 1517
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018735709939354887,
+      "loss": 0.0749,
+      "step": 1518
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018733767406520698,
+      "loss": 0.0759,
+      "step": 1519
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001873182348338005,
+      "loss": 0.0757,
+      "step": 1520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018729878170242392,
+      "loss": 0.0767,
+      "step": 1521
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018727931467417395,
+      "loss": 0.0737,
+      "step": 1522
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018725983375214944,
+      "loss": 0.0751,
+      "step": 1523
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018724033893945151,
+      "loss": 0.0762,
+      "step": 1524
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018722083023918357,
+      "loss": 0.0753,
+      "step": 1525
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018720130765445108,
+      "loss": 0.0752,
+      "step": 1526
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018718177118836184,
+      "loss": 0.0726,
+      "step": 1527
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001871622208440258,
+      "loss": 0.0765,
+      "step": 1528
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018714265662455508,
+      "loss": 0.0736,
+      "step": 1529
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018712307853306412,
+      "loss": 0.0755,
+      "step": 1530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001871034865726695,
+      "loss": 0.0735,
+      "step": 1531
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018708388074649,
+      "loss": 0.074,
+      "step": 1532
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001870642610576466,
+      "loss": 0.0734,
+      "step": 1533
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018704462750926258,
+      "loss": 0.0776,
+      "step": 1534
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018702498010446326,
+      "loss": 0.0756,
+      "step": 1535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018700531884637635,
+      "loss": 0.0722,
+      "step": 1536
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018698564373813162,
+      "loss": 0.0735,
+      "step": 1537
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018696595478286113,
+      "loss": 0.0766,
+      "step": 1538
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018694625198369908,
+      "loss": 0.0734,
+      "step": 1539
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018692653534378194,
+      "loss": 0.0765,
+      "step": 1540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018690680486624833,
+      "loss": 0.0751,
+      "step": 1541
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018688706055423914,
+      "loss": 0.0772,
+      "step": 1542
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001868673024108974,
+      "loss": 0.078,
+      "step": 1543
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001868475304393683,
+      "loss": 0.0766,
+      "step": 1544
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018682774464279933,
+      "loss": 0.0748,
+      "step": 1545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018680794502434018,
+      "loss": 0.0751,
+      "step": 1546
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018678813158714265,
+      "loss": 0.0769,
+      "step": 1547
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018676830433436082,
+      "loss": 0.0777,
+      "step": 1548
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00018674846326915091,
+      "loss": 0.0767,
+      "step": 1549
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018672860839467144,
+      "loss": 0.0779,
+      "step": 1550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018670873971408298,
+      "loss": 0.0733,
+      "step": 1551
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018668885723054839,
+      "loss": 0.0724,
+      "step": 1552
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001866689609472327,
+      "loss": 0.0744,
+      "step": 1553
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018664905086730323,
+      "loss": 0.0762,
+      "step": 1554
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018662912699392932,
+      "loss": 0.0777,
+      "step": 1555
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018660918933028267,
+      "loss": 0.0741,
+      "step": 1556
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018658923787953703,
+      "loss": 0.0746,
+      "step": 1557
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001865692726448685,
+      "loss": 0.0751,
+      "step": 1558
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018654929362945527,
+      "loss": 0.0742,
+      "step": 1559
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018652930083647775,
+      "loss": 0.0775,
+      "step": 1560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018650929426911855,
+      "loss": 0.076,
+      "step": 1561
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001864892739305624,
+      "loss": 0.0779,
+      "step": 1562
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018646923982399636,
+      "loss": 0.0746,
+      "step": 1563
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018644919195260959,
+      "loss": 0.0752,
+      "step": 1564
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018642913031959344,
+      "loss": 0.0754,
+      "step": 1565
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001864090549281415,
+      "loss": 0.077,
+      "step": 1566
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018638896578144953,
+      "loss": 0.0763,
+      "step": 1567
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001863688628827154,
+      "loss": 0.076,
+      "step": 1568
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018634874623513935,
+      "loss": 0.0714,
+      "step": 1569
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001863286158419236,
+      "loss": 0.0729,
+      "step": 1570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018630847170627272,
+      "loss": 0.0766,
+      "step": 1571
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018628831383139336,
+      "loss": 0.0756,
+      "step": 1572
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018626814222049443,
+      "loss": 0.0748,
+      "step": 1573
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.000186247956876787,
+      "loss": 0.0759,
+      "step": 1574
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001862277578034843,
+      "loss": 0.0759,
+      "step": 1575
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00018620754500380175,
+      "loss": 0.075,
+      "step": 1576
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018618731848095706,
+      "loss": 0.0736,
+      "step": 1577
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018616707823816995,
+      "loss": 0.0791,
+      "step": 1578
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018614682427866246,
+      "loss": 0.0733,
+      "step": 1579
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018612655660565875,
+      "loss": 0.0756,
+      "step": 1580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018610627522238523,
+      "loss": 0.0746,
+      "step": 1581
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018608598013207032,
+      "loss": 0.0783,
+      "step": 1582
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001860656713379449,
+      "loss": 0.0765,
+      "step": 1583
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018604534884324175,
+      "loss": 0.0767,
+      "step": 1584
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018602501265119602,
+      "loss": 0.0764,
+      "step": 1585
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018600466276504497,
+      "loss": 0.0731,
+      "step": 1586
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018598429918802802,
+      "loss": 0.0746,
+      "step": 1587
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018596392192338684,
+      "loss": 0.0725,
+      "step": 1588
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018594353097436522,
+      "loss": 0.0741,
+      "step": 1589
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018592312634420913,
+      "loss": 0.0733,
+      "step": 1590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018590270803616672,
+      "loss": 0.0796,
+      "step": 1591
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018588227605348837,
+      "loss": 0.0755,
+      "step": 1592
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018586183039942652,
+      "loss": 0.0767,
+      "step": 1593
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018584137107723598,
+      "loss": 0.0776,
+      "step": 1594
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018582089809017352,
+      "loss": 0.0761,
+      "step": 1595
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001858004114414982,
+      "loss": 0.075,
+      "step": 1596
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001857799111344713,
+      "loss": 0.0763,
+      "step": 1597
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018575939717235614,
+      "loss": 0.0743,
+      "step": 1598
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018573886955841832,
+      "loss": 0.0743,
+      "step": 1599
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018571832829592557,
+      "loss": 0.0764,
+      "step": 1600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001856977733881478,
+      "loss": 0.0765,
+      "step": 1601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018567720483835708,
+      "loss": 0.0765,
+      "step": 1602
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018565662264982772,
+      "loss": 0.0772,
+      "step": 1603
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001856360268258361,
+      "loss": 0.0749,
+      "step": 1604
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018561541736966084,
+      "loss": 0.0792,
+      "step": 1605
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018559479428458266,
+      "loss": 0.0796,
+      "step": 1606
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018557415757388456,
+      "loss": 0.0767,
+      "step": 1607
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018555350724085162,
+      "loss": 0.0745,
+      "step": 1608
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001855328432887711,
+      "loss": 0.0788,
+      "step": 1609
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018551216572093243,
+      "loss": 0.0764,
+      "step": 1610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018549147454062728,
+      "loss": 0.0736,
+      "step": 1611
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018547076975114937,
+      "loss": 0.0782,
+      "step": 1612
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001854500513557947,
+      "loss": 0.0779,
+      "step": 1613
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018542931935786134,
+      "loss": 0.0745,
+      "step": 1614
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018540857376064956,
+      "loss": 0.0765,
+      "step": 1615
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00018538781456746181,
+      "loss": 0.0726,
+      "step": 1616
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001853670417816027,
+      "loss": 0.0765,
+      "step": 1617
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018534625540637896,
+      "loss": 0.0741,
+      "step": 1618
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018532545544509957,
+      "loss": 0.0767,
+      "step": 1619
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001853046419010756,
+      "loss": 0.0751,
+      "step": 1620
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001852838147776203,
+      "loss": 0.0744,
+      "step": 1621
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018526297407804913,
+      "loss": 0.0754,
+      "step": 1622
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001852421198056796,
+      "loss": 0.0766,
+      "step": 1623
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018522125196383153,
+      "loss": 0.0741,
+      "step": 1624
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018520037055582672,
+      "loss": 0.0737,
+      "step": 1625
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018517947558498933,
+      "loss": 0.0755,
+      "step": 1626
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018515856705464553,
+      "loss": 0.0758,
+      "step": 1627
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018513764496812368,
+      "loss": 0.0769,
+      "step": 1628
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018511670932875432,
+      "loss": 0.0766,
+      "step": 1629
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00018509576013987016,
+      "loss": 0.074,
+      "step": 1630
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018507479740480606,
+      "loss": 0.0766,
+      "step": 1631
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018505382112689897,
+      "loss": 0.0752,
+      "step": 1632
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018503283130948814,
+      "loss": 0.0757,
+      "step": 1633
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001850118279559148,
+      "loss": 0.0766,
+      "step": 1634
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018499081106952247,
+      "loss": 0.0742,
+      "step": 1635
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018496978065365676,
+      "loss": 0.0772,
+      "step": 1636
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018494873671166543,
+      "loss": 0.0766,
+      "step": 1637
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018492767924689843,
+      "loss": 0.0736,
+      "step": 1638
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001849066082627079,
+      "loss": 0.078,
+      "step": 1639
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018488552376244797,
+      "loss": 0.0747,
+      "step": 1640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018486442574947511,
+      "loss": 0.0768,
+      "step": 1641
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018484331422714782,
+      "loss": 0.0772,
+      "step": 1642
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018482218919882679,
+      "loss": 0.0755,
+      "step": 1643
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001848010506678749,
+      "loss": 0.0726,
+      "step": 1644
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001847798986376571,
+      "loss": 0.0765,
+      "step": 1645
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018475873311154053,
+      "loss": 0.0772,
+      "step": 1646
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018473755409289448,
+      "loss": 0.0766,
+      "step": 1647
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018471636158509043,
+      "loss": 0.0785,
+      "step": 1648
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001846951555915019,
+      "loss": 0.0729,
+      "step": 1649
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018467393611550464,
+      "loss": 0.0744,
+      "step": 1650
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001846527031604765,
+      "loss": 0.0746,
+      "step": 1651
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018463145672979758,
+      "loss": 0.0756,
+      "step": 1652
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018461019682684997,
+      "loss": 0.0766,
+      "step": 1653
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000184588923455018,
+      "loss": 0.0749,
+      "step": 1654
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018456763661768815,
+      "loss": 0.0792,
+      "step": 1655
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018454633631824897,
+      "loss": 0.0767,
+      "step": 1656
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018452502256009125,
+      "loss": 0.0752,
+      "step": 1657
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018450369534660785,
+      "loss": 0.0747,
+      "step": 1658
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018448235468119378,
+      "loss": 0.077,
+      "step": 1659
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018446100056724625,
+      "loss": 0.0736,
+      "step": 1660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018443963300816452,
+      "loss": 0.0784,
+      "step": 1661
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018441825200735007,
+      "loss": 0.0751,
+      "step": 1662
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018439685756820647,
+      "loss": 0.0759,
+      "step": 1663
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018437544969413945,
+      "loss": 0.0754,
+      "step": 1664
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018435402838855688,
+      "loss": 0.0785,
+      "step": 1665
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018433259365486878,
+      "loss": 0.0775,
+      "step": 1666
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018431114549648727,
+      "loss": 0.0761,
+      "step": 1667
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018428968391682662,
+      "loss": 0.0761,
+      "step": 1668
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018426820891930326,
+      "loss": 0.0773,
+      "step": 1669
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018424672050733576,
+      "loss": 0.0771,
+      "step": 1670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018422521868434477,
+      "loss": 0.0771,
+      "step": 1671
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018420370345375312,
+      "loss": 0.0755,
+      "step": 1672
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018418217481898578,
+      "loss": 0.0745,
+      "step": 1673
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018416063278346983,
+      "loss": 0.0815,
+      "step": 1674
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001841390773506345,
+      "loss": 0.0754,
+      "step": 1675
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018411750852391113,
+      "loss": 0.0795,
+      "step": 1676
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018409592630673323,
+      "loss": 0.0754,
+      "step": 1677
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018407433070253637,
+      "loss": 0.0776,
+      "step": 1678
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018405272171475838,
+      "loss": 0.0804,
+      "step": 1679
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018403109934683905,
+      "loss": 0.0747,
+      "step": 1680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018400946360222048,
+      "loss": 0.0756,
+      "step": 1681
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018398781448434673,
+      "loss": 0.078,
+      "step": 1682
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018396615199666412,
+      "loss": 0.0731,
+      "step": 1683
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00018394447614262103,
+      "loss": 0.0731,
+      "step": 1684
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018392278692566798,
+      "loss": 0.0759,
+      "step": 1685
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018390108434925763,
+      "loss": 0.0758,
+      "step": 1686
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018387936841684477,
+      "loss": 0.0745,
+      "step": 1687
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018385763913188625,
+      "loss": 0.0782,
+      "step": 1688
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018383589649784116,
+      "loss": 0.0755,
+      "step": 1689
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018381414051817064,
+      "loss": 0.0763,
+      "step": 1690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000183792371196338,
+      "loss": 0.0765,
+      "step": 1691
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018377058853580856,
+      "loss": 0.0766,
+      "step": 1692
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001837487925400499,
+      "loss": 0.0724,
+      "step": 1693
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018372698321253168,
+      "loss": 0.0743,
+      "step": 1694
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018370516055672567,
+      "loss": 0.0761,
+      "step": 1695
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018368332457610573,
+      "loss": 0.0758,
+      "step": 1696
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001836614752741479,
+      "loss": 0.0795,
+      "step": 1697
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00018363961265433032,
+      "loss": 0.0775,
+      "step": 1698
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00018361773672013324,
+      "loss": 0.0762,
+      "step": 1699
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000183595847475039,
+      "loss": 0.077,
+      "step": 1700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00018357394492253215,
+      "loss": 0.0762,
+      "step": 1701
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00018355202906609927,
+      "loss": 0.0764,
+      "step": 1702
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001835300999092291,
+      "loss": 0.0782,
+      "step": 1703
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001835081574554125,
+      "loss": 0.0748,
+      "step": 1704
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00018348620170814241,
+      "loss": 0.0721,
+      "step": 1705
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001834642326709139,
+      "loss": 0.0759,
+      "step": 1706
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00018344225034722423,
+      "loss": 0.0762,
+      "step": 1707
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00018342025474057263,
+      "loss": 0.076,
+      "step": 1708
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00018339824585446058,
+      "loss": 0.0778,
+      "step": 1709
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00018337622369239158,
+      "loss": 0.0731,
+      "step": 1710
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001833541882578713,
+      "loss": 0.0751,
+      "step": 1711
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018333213955440753,
+      "loss": 0.0736,
+      "step": 1712
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018331007758551012,
+      "loss": 0.0761,
+      "step": 1713
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018328800235469108,
+      "loss": 0.072,
+      "step": 1714
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018326591386546448,
+      "loss": 0.0752,
+      "step": 1715
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018324381212134652,
+      "loss": 0.0762,
+      "step": 1716
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001832216971258556,
+      "loss": 0.0766,
+      "step": 1717
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018319956888251205,
+      "loss": 0.0781,
+      "step": 1718
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018317742739483849,
+      "loss": 0.0788,
+      "step": 1719
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018315527266635955,
+      "loss": 0.0802,
+      "step": 1720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018313310470060193,
+      "loss": 0.0755,
+      "step": 1721
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018311092350109455,
+      "loss": 0.0782,
+      "step": 1722
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018308872907136836,
+      "loss": 0.0773,
+      "step": 1723
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018306652141495647,
+      "loss": 0.0778,
+      "step": 1724
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00018304430053539398,
+      "loss": 0.0793,
+      "step": 1725
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018302206643621823,
+      "loss": 0.0791,
+      "step": 1726
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018299981912096866,
+      "loss": 0.0745,
+      "step": 1727
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018297755859318667,
+      "loss": 0.0746,
+      "step": 1728
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018295528485641588,
+      "loss": 0.0708,
+      "step": 1729
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018293299791420205,
+      "loss": 0.0748,
+      "step": 1730
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018291069777009292,
+      "loss": 0.0757,
+      "step": 1731
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018288838442763838,
+      "loss": 0.072,
+      "step": 1732
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018286605789039051,
+      "loss": 0.0769,
+      "step": 1733
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018284371816190337,
+      "loss": 0.0745,
+      "step": 1734
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018282136524573316,
+      "loss": 0.0814,
+      "step": 1735
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001827989991454382,
+      "loss": 0.0759,
+      "step": 1736
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018277661986457888,
+      "loss": 0.0719,
+      "step": 1737
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001827542274067177,
+      "loss": 0.077,
+      "step": 1738
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018273182177541927,
+      "loss": 0.0746,
+      "step": 1739
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001827094029742503,
+      "loss": 0.0755,
+      "step": 1740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018268697100677956,
+      "loss": 0.077,
+      "step": 1741
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018266452587657793,
+      "loss": 0.0776,
+      "step": 1742
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018264206758721843,
+      "loss": 0.075,
+      "step": 1743
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018261959614227608,
+      "loss": 0.0752,
+      "step": 1744
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018259711154532812,
+      "loss": 0.0789,
+      "step": 1745
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018257461379995378,
+      "loss": 0.0764,
+      "step": 1746
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018255210290973441,
+      "loss": 0.0738,
+      "step": 1747
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001825295788782535,
+      "loss": 0.0773,
+      "step": 1748
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018250704170909656,
+      "loss": 0.0767,
+      "step": 1749
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018248449140585121,
+      "loss": 0.0749,
+      "step": 1750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018246192797210725,
+      "loss": 0.0768,
+      "step": 1751
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018243935141145642,
+      "loss": 0.0744,
+      "step": 1752
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018241676172749266,
+      "loss": 0.0749,
+      "step": 1753
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018239415892381197,
+      "loss": 0.0757,
+      "step": 1754
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001823715430040124,
+      "loss": 0.0744,
+      "step": 1755
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018234891397169415,
+      "loss": 0.074,
+      "step": 1756
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018232627183045952,
+      "loss": 0.0743,
+      "step": 1757
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018230361658391276,
+      "loss": 0.0752,
+      "step": 1758
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018228094823566034,
+      "loss": 0.0734,
+      "step": 1759
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001822582667893108,
+      "loss": 0.0726,
+      "step": 1760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018223557224847475,
+      "loss": 0.0735,
+      "step": 1761
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018221286461676486,
+      "loss": 0.08,
+      "step": 1762
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018219014389779585,
+      "loss": 0.0732,
+      "step": 1763
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018216741009518465,
+      "loss": 0.0755,
+      "step": 1764
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00018214466321255017,
+      "loss": 0.0753,
+      "step": 1765
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018212190325351345,
+      "loss": 0.0742,
+      "step": 1766
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018209913022169754,
+      "loss": 0.0751,
+      "step": 1767
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018207634412072764,
+      "loss": 0.076,
+      "step": 1768
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018205354495423105,
+      "loss": 0.0744,
+      "step": 1769
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018203073272583703,
+      "loss": 0.0773,
+      "step": 1770
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018200790743917714,
+      "loss": 0.0763,
+      "step": 1771
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018198506909788474,
+      "loss": 0.0754,
+      "step": 1772
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018196221770559547,
+      "loss": 0.0759,
+      "step": 1773
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018193935326594698,
+      "loss": 0.0747,
+      "step": 1774
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000181916475782579,
+      "loss": 0.0739,
+      "step": 1775
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018189358525913334,
+      "loss": 0.0743,
+      "step": 1776
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018187068169925387,
+      "loss": 0.0746,
+      "step": 1777
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018184776510658658,
+      "loss": 0.0771,
+      "step": 1778
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018182483548477948,
+      "loss": 0.0746,
+      "step": 1779
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018180189283748273,
+      "loss": 0.0754,
+      "step": 1780
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018177893716834844,
+      "loss": 0.0784,
+      "step": 1781
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018175596848103088,
+      "loss": 0.0738,
+      "step": 1782
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018173298677918643,
+      "loss": 0.0768,
+      "step": 1783
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001817099920664734,
+      "loss": 0.0801,
+      "step": 1784
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018168698434655237,
+      "loss": 0.0781,
+      "step": 1785
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018166396362308582,
+      "loss": 0.0756,
+      "step": 1786
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018164092989973832,
+      "loss": 0.0771,
+      "step": 1787
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018161788318017664,
+      "loss": 0.0753,
+      "step": 1788
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018159482346806946,
+      "loss": 0.074,
+      "step": 1789
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018157175076708763,
+      "loss": 0.0762,
+      "step": 1790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018154866508090405,
+      "loss": 0.0732,
+      "step": 1791
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00018152556641319362,
+      "loss": 0.0746,
+      "step": 1792
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018150245476763338,
+      "loss": 0.0772,
+      "step": 1793
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018147933014790244,
+      "loss": 0.0746,
+      "step": 1794
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018145619255768196,
+      "loss": 0.0758,
+      "step": 1795
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001814330420006551,
+      "loss": 0.0723,
+      "step": 1796
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018140987848050717,
+      "loss": 0.0743,
+      "step": 1797
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001813867020009255,
+      "loss": 0.075,
+      "step": 1798
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001813635125655995,
+      "loss": 0.0766,
+      "step": 1799
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001813403101782207,
+      "loss": 0.0751,
+      "step": 1800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018131709484248254,
+      "loss": 0.0772,
+      "step": 1801
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001812938665620806,
+      "loss": 0.0755,
+      "step": 1802
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018127062534071265,
+      "loss": 0.0733,
+      "step": 1803
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001812473711820783,
+      "loss": 0.0778,
+      "step": 1804
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001812241040898793,
+      "loss": 0.0744,
+      "step": 1805
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018120082406781958,
+      "loss": 0.0714,
+      "step": 1806
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018117753111960495,
+      "loss": 0.0771,
+      "step": 1807
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001811542252489434,
+      "loss": 0.0767,
+      "step": 1808
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018113090645954493,
+      "loss": 0.0772,
+      "step": 1809
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018110757475512157,
+      "loss": 0.0726,
+      "step": 1810
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018108423013938745,
+      "loss": 0.0739,
+      "step": 1811
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001810608726160587,
+      "loss": 0.075,
+      "step": 1812
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018103750218885365,
+      "loss": 0.0753,
+      "step": 1813
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001810141188614925,
+      "loss": 0.0765,
+      "step": 1814
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018099072263769755,
+      "loss": 0.0792,
+      "step": 1815
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018096731352119328,
+      "loss": 0.0749,
+      "step": 1816
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018094389151570604,
+      "loss": 0.0757,
+      "step": 1817
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001809204566249644,
+      "loss": 0.0767,
+      "step": 1818
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00018089700885269886,
+      "loss": 0.0776,
+      "step": 1819
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018087354820264201,
+      "loss": 0.0755,
+      "step": 1820
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001808500746785285,
+      "loss": 0.0735,
+      "step": 1821
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018082658828409503,
+      "loss": 0.0775,
+      "step": 1822
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001808030890230803,
+      "loss": 0.075,
+      "step": 1823
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018077957689922517,
+      "loss": 0.0746,
+      "step": 1824
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001807560519162724,
+      "loss": 0.0736,
+      "step": 1825
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001807325140779669,
+      "loss": 0.0739,
+      "step": 1826
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018070896338805563,
+      "loss": 0.0738,
+      "step": 1827
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018068539985028755,
+      "loss": 0.0779,
+      "step": 1828
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018066182346841364,
+      "loss": 0.078,
+      "step": 1829
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018063823424618698,
+      "loss": 0.0761,
+      "step": 1830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001806146321873627,
+      "loss": 0.0765,
+      "step": 1831
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000180591017295698,
+      "loss": 0.0716,
+      "step": 1832
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.000180567389574952,
+      "loss": 0.0733,
+      "step": 1833
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018054374902888592,
+      "loss": 0.0763,
+      "step": 1834
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001805200956612631,
+      "loss": 0.0744,
+      "step": 1835
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018049642947584883,
+      "loss": 0.0755,
+      "step": 1836
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001804727504764105,
+      "loss": 0.0779,
+      "step": 1837
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001804490586667175,
+      "loss": 0.0726,
+      "step": 1838
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018042535405054125,
+      "loss": 0.0756,
+      "step": 1839
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018040163663165523,
+      "loss": 0.0733,
+      "step": 1840
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018037790641383495,
+      "loss": 0.0764,
+      "step": 1841
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018035416340085798,
+      "loss": 0.0762,
+      "step": 1842
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018033040759650391,
+      "loss": 0.075,
+      "step": 1843
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018030663900455438,
+      "loss": 0.0726,
+      "step": 1844
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018028285762879302,
+      "loss": 0.0771,
+      "step": 1845
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00018025906347300558,
+      "loss": 0.0764,
+      "step": 1846
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018023525654097967,
+      "loss": 0.0771,
+      "step": 1847
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018021143683650523,
+      "loss": 0.0786,
+      "step": 1848
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018018760436337394,
+      "loss": 0.0764,
+      "step": 1849
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018016375912537964,
+      "loss": 0.0749,
+      "step": 1850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018013990112631824,
+      "loss": 0.0746,
+      "step": 1851
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001801160303699876,
+      "loss": 0.0786,
+      "step": 1852
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018009214686018766,
+      "loss": 0.0775,
+      "step": 1853
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018006825060072038,
+      "loss": 0.0745,
+      "step": 1854
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018004434159538972,
+      "loss": 0.0733,
+      "step": 1855
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018002041984800174,
+      "loss": 0.0769,
+      "step": 1856
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001799964853623644,
+      "loss": 0.0777,
+      "step": 1857
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017997253814228787,
+      "loss": 0.078,
+      "step": 1858
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017994857819158416,
+      "loss": 0.0735,
+      "step": 1859
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001799246055140675,
+      "loss": 0.0749,
+      "step": 1860
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017990062011355394,
+      "loss": 0.0736,
+      "step": 1861
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017987662199386167,
+      "loss": 0.0743,
+      "step": 1862
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017985261115881094,
+      "loss": 0.0801,
+      "step": 1863
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017982858761222398,
+      "loss": 0.076,
+      "step": 1864
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017980455135792493,
+      "loss": 0.0754,
+      "step": 1865
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017978050239974018,
+      "loss": 0.0736,
+      "step": 1866
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017975644074149798,
+      "loss": 0.076,
+      "step": 1867
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017973236638702865,
+      "loss": 0.0742,
+      "step": 1868
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001797082793401645,
+      "loss": 0.0733,
+      "step": 1869
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001796841796047399,
+      "loss": 0.0727,
+      "step": 1870
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017966006718459125,
+      "loss": 0.0713,
+      "step": 1871
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017963594208355693,
+      "loss": 0.076,
+      "step": 1872
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017961180430547735,
+      "loss": 0.0751,
+      "step": 1873
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017958765385419492,
+      "loss": 0.0755,
+      "step": 1874
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017956349073355415,
+      "loss": 0.0787,
+      "step": 1875
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017953931494740143,
+      "loss": 0.0747,
+      "step": 1876
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001795151264995853,
+      "loss": 0.0763,
+      "step": 1877
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017949092539395624,
+      "loss": 0.0762,
+      "step": 1878
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017946671163436678,
+      "loss": 0.0773,
+      "step": 1879
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017944248522467143,
+      "loss": 0.0772,
+      "step": 1880
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017941824616872674,
+      "loss": 0.0758,
+      "step": 1881
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017939399447039123,
+      "loss": 0.0752,
+      "step": 1882
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001793697301335255,
+      "loss": 0.0788,
+      "step": 1883
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017934545316199211,
+      "loss": 0.0782,
+      "step": 1884
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001793211635596557,
+      "loss": 0.0761,
+      "step": 1885
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017929686133038282,
+      "loss": 0.0751,
+      "step": 1886
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017927254647804209,
+      "loss": 0.0777,
+      "step": 1887
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017924821900650414,
+      "loss": 0.0757,
+      "step": 1888
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017922387891964155,
+      "loss": 0.0754,
+      "step": 1889
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017919952622132905,
+      "loss": 0.0776,
+      "step": 1890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017917516091544322,
+      "loss": 0.0746,
+      "step": 1891
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017915078300586272,
+      "loss": 0.0745,
+      "step": 1892
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001791263924964682,
+      "loss": 0.0764,
+      "step": 1893
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017910198939114237,
+      "loss": 0.0743,
+      "step": 1894
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017907757369376985,
+      "loss": 0.0765,
+      "step": 1895
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017905314540823735,
+      "loss": 0.0753,
+      "step": 1896
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017902870453843353,
+      "loss": 0.0731,
+      "step": 1897
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017900425108824907,
+      "loss": 0.0757,
+      "step": 1898
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017897978506157663,
+      "loss": 0.0724,
+      "step": 1899
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017895530646231092,
+      "loss": 0.076,
+      "step": 1900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001789308152943486,
+      "loss": 0.0731,
+      "step": 1901
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001789063115615884,
+      "loss": 0.0773,
+      "step": 1902
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017888179526793103,
+      "loss": 0.075,
+      "step": 1903
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001788572664172791,
+      "loss": 0.0735,
+      "step": 1904
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017883272501353735,
+      "loss": 0.0756,
+      "step": 1905
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017880817106061242,
+      "loss": 0.0759,
+      "step": 1906
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017878360456241303,
+      "loss": 0.0766,
+      "step": 1907
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017875902552284985,
+      "loss": 0.0766,
+      "step": 1908
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001787344339458356,
+      "loss": 0.0768,
+      "step": 1909
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001787098298352848,
+      "loss": 0.0747,
+      "step": 1910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001786852131951143,
+      "loss": 0.0747,
+      "step": 1911
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017866058402924267,
+      "loss": 0.0773,
+      "step": 1912
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017863594234159057,
+      "loss": 0.0771,
+      "step": 1913
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017861128813608063,
+      "loss": 0.0748,
+      "step": 1914
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017858662141663752,
+      "loss": 0.0758,
+      "step": 1915
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001785619421871879,
+      "loss": 0.0768,
+      "step": 1916
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017853725045166035,
+      "loss": 0.0773,
+      "step": 1917
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001785125462139855,
+      "loss": 0.0789,
+      "step": 1918
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017848782947809595,
+      "loss": 0.0751,
+      "step": 1919
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017846310024792632,
+      "loss": 0.0755,
+      "step": 1920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017843835852741314,
+      "loss": 0.0773,
+      "step": 1921
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000178413604320495,
+      "loss": 0.0788,
+      "step": 1922
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001783888376311125,
+      "loss": 0.0753,
+      "step": 1923
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001783640584632082,
+      "loss": 0.0771,
+      "step": 1924
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017833926682072658,
+      "loss": 0.0757,
+      "step": 1925
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017831446270761414,
+      "loss": 0.076,
+      "step": 1926
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017828964612781942,
+      "loss": 0.0733,
+      "step": 1927
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017826481708529293,
+      "loss": 0.0753,
+      "step": 1928
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017823997558398712,
+      "loss": 0.0759,
+      "step": 1929
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017821512162785643,
+      "loss": 0.0761,
+      "step": 1930
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017819025522085733,
+      "loss": 0.0765,
+      "step": 1931
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001781653763669482,
+      "loss": 0.0754,
+      "step": 1932
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017814048507008948,
+      "loss": 0.0749,
+      "step": 1933
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017811558133424359,
+      "loss": 0.0765,
+      "step": 1934
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001780906651633748,
+      "loss": 0.0766,
+      "step": 1935
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017806573656144948,
+      "loss": 0.0735,
+      "step": 1936
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000178040795532436,
+      "loss": 0.0774,
+      "step": 1937
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017801584208030464,
+      "loss": 0.0736,
+      "step": 1938
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017799087620902764,
+      "loss": 0.0757,
+      "step": 1939
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017796589792257926,
+      "loss": 0.0744,
+      "step": 1940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001779409072249358,
+      "loss": 0.0761,
+      "step": 1941
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017791590412007536,
+      "loss": 0.077,
+      "step": 1942
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017789088861197824,
+      "loss": 0.0744,
+      "step": 1943
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001778658607046265,
+      "loss": 0.0747,
+      "step": 1944
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001778408204020043,
+      "loss": 0.0754,
+      "step": 1945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017781576770809773,
+      "loss": 0.0756,
+      "step": 1946
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017779070262689492,
+      "loss": 0.0771,
+      "step": 1947
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017776562516238586,
+      "loss": 0.0768,
+      "step": 1948
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017774053531856258,
+      "loss": 0.077,
+      "step": 1949
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001777154330994191,
+      "loss": 0.0782,
+      "step": 1950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017769031850895132,
+      "loss": 0.0741,
+      "step": 1951
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017766519155115723,
+      "loss": 0.0791,
+      "step": 1952
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001776400522300367,
+      "loss": 0.0755,
+      "step": 1953
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001776149005495916,
+      "loss": 0.0775,
+      "step": 1954
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017758973651382573,
+      "loss": 0.0748,
+      "step": 1955
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017756456012674494,
+      "loss": 0.0762,
+      "step": 1956
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000177539371392357,
+      "loss": 0.0781,
+      "step": 1957
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017751417031467158,
+      "loss": 0.0771,
+      "step": 1958
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001774889568977004,
+      "loss": 0.0759,
+      "step": 1959
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017746373114545713,
+      "loss": 0.0772,
+      "step": 1960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017743849306195742,
+      "loss": 0.0749,
+      "step": 1961
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017741324265121882,
+      "loss": 0.0772,
+      "step": 1962
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017738797991726093,
+      "loss": 0.077,
+      "step": 1963
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001773627048641052,
+      "loss": 0.0734,
+      "step": 1964
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001773374174957751,
+      "loss": 0.0759,
+      "step": 1965
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001773121178162961,
+      "loss": 0.0764,
+      "step": 1966
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001772868058296956,
+      "loss": 0.0745,
+      "step": 1967
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017726148154000292,
+      "loss": 0.0773,
+      "step": 1968
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017723614495124938,
+      "loss": 0.0768,
+      "step": 1969
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017721079606746826,
+      "loss": 0.0766,
+      "step": 1970
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017718543489269478,
+      "loss": 0.0746,
+      "step": 1971
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017716006143096607,
+      "loss": 0.0777,
+      "step": 1972
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017713467568632136,
+      "loss": 0.077,
+      "step": 1973
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001771092776628017,
+      "loss": 0.0751,
+      "step": 1974
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017708386736445012,
+      "loss": 0.0762,
+      "step": 1975
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001770584447953116,
+      "loss": 0.0783,
+      "step": 1976
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017703300995943318,
+      "loss": 0.0758,
+      "step": 1977
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001770075628608637,
+      "loss": 0.0774,
+      "step": 1978
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017698210350365403,
+      "loss": 0.0737,
+      "step": 1979
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000176956631891857,
+      "loss": 0.0769,
+      "step": 1980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017693114802952735,
+      "loss": 0.0745,
+      "step": 1981
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017690565192072182,
+      "loss": 0.0778,
+      "step": 1982
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017688014356949904,
+      "loss": 0.0774,
+      "step": 1983
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017685462297991963,
+      "loss": 0.0765,
+      "step": 1984
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017682909015604613,
+      "loss": 0.0767,
+      "step": 1985
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017680354510194314,
+      "loss": 0.0741,
+      "step": 1986
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017677798782167697,
+      "loss": 0.0766,
+      "step": 1987
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017675241831931614,
+      "loss": 0.0751,
+      "step": 1988
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017672683659893093,
+      "loss": 0.0782,
+      "step": 1989
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017670124266459363,
+      "loss": 0.0723,
+      "step": 1990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017667563652037848,
+      "loss": 0.0767,
+      "step": 1991
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017665001817036168,
+      "loss": 0.0734,
+      "step": 1992
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017662438761862137,
+      "loss": 0.0757,
+      "step": 1993
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017659874486923753,
+      "loss": 0.0754,
+      "step": 1994
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017657308992629223,
+      "loss": 0.0733,
+      "step": 1995
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001765474227938694,
+      "loss": 0.0782,
+      "step": 1996
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017652174347605496,
+      "loss": 0.0769,
+      "step": 1997
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017649605197693667,
+      "loss": 0.0746,
+      "step": 1998
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001764703483006043,
+      "loss": 0.0773,
+      "step": 1999
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017644463245114963,
+      "loss": 0.0757,
+      "step": 2000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017641890443266626,
+      "loss": 0.0768,
+      "step": 2001
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017639316424924973,
+      "loss": 0.0737,
+      "step": 2002
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001763674119049976,
+      "loss": 0.0762,
+      "step": 2003
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001763416474040093,
+      "loss": 0.075,
+      "step": 2004
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017631587075038626,
+      "loss": 0.077,
+      "step": 2005
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001762900819482317,
+      "loss": 0.0721,
+      "step": 2006
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000176264281001651,
+      "loss": 0.0755,
+      "step": 2007
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017623846791475125,
+      "loss": 0.0775,
+      "step": 2008
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001762126426916416,
+      "loss": 0.0699,
+      "step": 2009
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017618680533643314,
+      "loss": 0.0761,
+      "step": 2010
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017616095585323878,
+      "loss": 0.0737,
+      "step": 2011
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001761350942461735,
+      "loss": 0.0745,
+      "step": 2012
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017610922051935414,
+      "loss": 0.0774,
+      "step": 2013
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017608333467689945,
+      "loss": 0.0742,
+      "step": 2014
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017605743672293014,
+      "loss": 0.0767,
+      "step": 2015
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017603152666156882,
+      "loss": 0.077,
+      "step": 2016
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017600560449694005,
+      "loss": 0.0768,
+      "step": 2017
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017597967023317036,
+      "loss": 0.0744,
+      "step": 2018
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001759537238743881,
+      "loss": 0.0753,
+      "step": 2019
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017592776542472364,
+      "loss": 0.0735,
+      "step": 2020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017590179488830926,
+      "loss": 0.078,
+      "step": 2021
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001758758122692791,
+      "loss": 0.0731,
+      "step": 2022
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001758498175717693,
+      "loss": 0.0773,
+      "step": 2023
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017582381079991785,
+      "loss": 0.0741,
+      "step": 2024
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017579779195786474,
+      "loss": 0.0763,
+      "step": 2025
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017577176104975186,
+      "loss": 0.0749,
+      "step": 2026
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017574571807972295,
+      "loss": 0.0738,
+      "step": 2027
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017571966305192377,
+      "loss": 0.0771,
+      "step": 2028
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001756935959705019,
+      "loss": 0.0763,
+      "step": 2029
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017566751683960698,
+      "loss": 0.0771,
+      "step": 2030
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017564142566339042,
+      "loss": 0.0747,
+      "step": 2031
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001756153224460056,
+      "loss": 0.0761,
+      "step": 2032
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017558920719160787,
+      "loss": 0.076,
+      "step": 2033
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017556307990435443,
+      "loss": 0.0726,
+      "step": 2034
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001755369405884044,
+      "loss": 0.076,
+      "step": 2035
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017551078924791882,
+      "loss": 0.0762,
+      "step": 2036
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017548462588706074,
+      "loss": 0.0759,
+      "step": 2037
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017545845050999495,
+      "loss": 0.0754,
+      "step": 2038
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017543226312088827,
+      "loss": 0.0765,
+      "step": 2039
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017540606372390944,
+      "loss": 0.0746,
+      "step": 2040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017537985232322903,
+      "loss": 0.0763,
+      "step": 2041
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017535362892301954,
+      "loss": 0.0741,
+      "step": 2042
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017532739352745552,
+      "loss": 0.0763,
+      "step": 2043
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001753011461407132,
+      "loss": 0.0785,
+      "step": 2044
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001752748867669709,
+      "loss": 0.0757,
+      "step": 2045
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017524861541040875,
+      "loss": 0.0763,
+      "step": 2046
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017522233207520885,
+      "loss": 0.0754,
+      "step": 2047
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001751960367655552,
+      "loss": 0.0772,
+      "step": 2048
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001751697294856336,
+      "loss": 0.0759,
+      "step": 2049
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017514341023963187,
+      "loss": 0.0747,
+      "step": 2050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017511707903173975,
+      "loss": 0.0753,
+      "step": 2051
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017509073586614884,
+      "loss": 0.0738,
+      "step": 2052
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017506438074705257,
+      "loss": 0.0763,
+      "step": 2053
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017503801367864643,
+      "loss": 0.0755,
+      "step": 2054
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017501163466512764,
+      "loss": 0.0761,
+      "step": 2055
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017498524371069548,
+      "loss": 0.0743,
+      "step": 2056
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017495884081955104,
+      "loss": 0.0776,
+      "step": 2057
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017493242599589731,
+      "loss": 0.0738,
+      "step": 2058
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017490599924393924,
+      "loss": 0.0758,
+      "step": 2059
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001748795605678836,
+      "loss": 0.0763,
+      "step": 2060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001748531099719391,
+      "loss": 0.0769,
+      "step": 2061
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017482664746031637,
+      "loss": 0.0735,
+      "step": 2062
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00017480017303722788,
+      "loss": 0.0771,
+      "step": 2063
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000174773686706888,
+      "loss": 0.0754,
+      "step": 2064
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001747471884735131,
+      "loss": 0.0759,
+      "step": 2065
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00017472067834132135,
+      "loss": 0.0751,
+      "step": 2066
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00017469415631453277,
+      "loss": 0.0744,
+      "step": 2067
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00017466762239736941,
+      "loss": 0.0772,
+      "step": 2068
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001746410765940551,
+      "loss": 0.0726,
+      "step": 2069
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001746145189088156,
+      "loss": 0.0757,
+      "step": 2070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00017458794934587855,
+      "loss": 0.0748,
+      "step": 2071
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00017456136790947347,
+      "loss": 0.0743,
+      "step": 2072
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001745347746038319,
+      "loss": 0.0768,
+      "step": 2073
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00017450816943318708,
+      "loss": 0.078,
+      "step": 2074
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00017448155240177418,
+      "loss": 0.0746,
+      "step": 2075
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00017445492351383037,
+      "loss": 0.0751,
+      "step": 2076
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017442828277359462,
+      "loss": 0.0774,
+      "step": 2077
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017440163018530777,
+      "loss": 0.0732,
+      "step": 2078
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017437496575321262,
+      "loss": 0.0738,
+      "step": 2079
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001743482894815538,
+      "loss": 0.0749,
+      "step": 2080
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017432160137457783,
+      "loss": 0.075,
+      "step": 2081
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017429490143653316,
+      "loss": 0.0804,
+      "step": 2082
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017426818967167002,
+      "loss": 0.0752,
+      "step": 2083
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017424146608424065,
+      "loss": 0.0774,
+      "step": 2084
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017421473067849905,
+      "loss": 0.0761,
+      "step": 2085
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017418798345870119,
+      "loss": 0.0732,
+      "step": 2086
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017416122442910495,
+      "loss": 0.0751,
+      "step": 2087
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017413445359396994,
+      "loss": 0.0783,
+      "step": 2088
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001741076709575578,
+      "loss": 0.0786,
+      "step": 2089
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00017408087652413196,
+      "loss": 0.0789,
+      "step": 2090
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001740540702979578,
+      "loss": 0.0757,
+      "step": 2091
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017402725228330248,
+      "loss": 0.0733,
+      "step": 2092
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001740004224844351,
+      "loss": 0.0751,
+      "step": 2093
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001739735809056267,
+      "loss": 0.0745,
+      "step": 2094
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017394672755115001,
+      "loss": 0.0738,
+      "step": 2095
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001739198624252799,
+      "loss": 0.0783,
+      "step": 2096
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017389298553229277,
+      "loss": 0.0766,
+      "step": 2097
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017386609687646725,
+      "loss": 0.076,
+      "step": 2098
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017383919646208362,
+      "loss": 0.0747,
+      "step": 2099
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017381228429342406,
+      "loss": 0.0769,
+      "step": 2100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017378536037477264,
+      "loss": 0.0725,
+      "step": 2101
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017375842471041542,
+      "loss": 0.0711,
+      "step": 2102
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001737314773046401,
+      "loss": 0.0747,
+      "step": 2103
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001737045181617364,
+      "loss": 0.0733,
+      "step": 2104
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001736775472859959,
+      "loss": 0.0732,
+      "step": 2105
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017365056468171203,
+      "loss": 0.0782,
+      "step": 2106
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017362357035318008,
+      "loss": 0.0767,
+      "step": 2107
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001735965643046972,
+      "loss": 0.0758,
+      "step": 2108
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001735695465405624,
+      "loss": 0.0748,
+      "step": 2109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017354251706507657,
+      "loss": 0.078,
+      "step": 2110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017351547588254253,
+      "loss": 0.0788,
+      "step": 2111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001734884229972648,
+      "loss": 0.0732,
+      "step": 2112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017346135841354992,
+      "loss": 0.0755,
+      "step": 2113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017343428213570622,
+      "loss": 0.0716,
+      "step": 2114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017340719416804394,
+      "loss": 0.0753,
+      "step": 2115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001733800945148751,
+      "loss": 0.0748,
+      "step": 2116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001733529831805136,
+      "loss": 0.0765,
+      "step": 2117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017332586016927527,
+      "loss": 0.0732,
+      "step": 2118
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001732987254854778,
+      "loss": 0.0783,
+      "step": 2119
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001732715791334406,
+      "loss": 0.0768,
+      "step": 2120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017324442111748504,
+      "loss": 0.074,
+      "step": 2121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001732172514419344,
+      "loss": 0.0752,
+      "step": 2122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001731900701111137,
+      "loss": 0.0736,
+      "step": 2123
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017316287712934987,
+      "loss": 0.076,
+      "step": 2124
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017313567250097173,
+      "loss": 0.0759,
+      "step": 2125
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001731084562303099,
+      "loss": 0.0734,
+      "step": 2126
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017308122832169684,
+      "loss": 0.0737,
+      "step": 2127
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001730539887794669,
+      "loss": 0.0741,
+      "step": 2128
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017302673760795636,
+      "loss": 0.0772,
+      "step": 2129
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00017299947481150315,
+      "loss": 0.074,
+      "step": 2130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017297220039444718,
+      "loss": 0.0731,
+      "step": 2131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017294491436113026,
+      "loss": 0.0767,
+      "step": 2132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017291761671589593,
+      "loss": 0.0757,
+      "step": 2133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017289030746308966,
+      "loss": 0.0764,
+      "step": 2134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017286298660705875,
+      "loss": 0.0738,
+      "step": 2135
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001728356541521523,
+      "loss": 0.0739,
+      "step": 2136
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017280831010272135,
+      "loss": 0.0732,
+      "step": 2137
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017278095446311866,
+      "loss": 0.0768,
+      "step": 2138
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000172753587237699,
+      "loss": 0.0808,
+      "step": 2139
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017272620843081877,
+      "loss": 0.0743,
+      "step": 2140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017269881804683643,
+      "loss": 0.0755,
+      "step": 2141
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017267141609011215,
+      "loss": 0.0786,
+      "step": 2142
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000172644002565008,
+      "loss": 0.0747,
+      "step": 2143
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017261657747588783,
+      "loss": 0.0741,
+      "step": 2144
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001725891408271174,
+      "loss": 0.0773,
+      "step": 2145
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017256169262306427,
+      "loss": 0.0728,
+      "step": 2146
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017253423286809784,
+      "loss": 0.0737,
+      "step": 2147
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001725067615665894,
+      "loss": 0.0769,
+      "step": 2148
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000172479278722912,
+      "loss": 0.0784,
+      "step": 2149
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001724517843414406,
+      "loss": 0.0759,
+      "step": 2150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001724242784265519,
+      "loss": 0.0741,
+      "step": 2151
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017239676098262455,
+      "loss": 0.072,
+      "step": 2152
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000172369232014039,
+      "loss": 0.0765,
+      "step": 2153
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017234169152517743,
+      "loss": 0.0731,
+      "step": 2154
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000172314139520424,
+      "loss": 0.0728,
+      "step": 2155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017228657600416468,
+      "loss": 0.078,
+      "step": 2156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017225900098078717,
+      "loss": 0.0736,
+      "step": 2157
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00017223141445468112,
+      "loss": 0.0728,
+      "step": 2158
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017220381643023793,
+      "loss": 0.0761,
+      "step": 2159
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017217620691185084,
+      "loss": 0.0776,
+      "step": 2160
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000172148585903915,
+      "loss": 0.0746,
+      "step": 2161
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017212095341082728,
+      "loss": 0.0758,
+      "step": 2162
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017209330943698643,
+      "loss": 0.0782,
+      "step": 2163
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017206565398679305,
+      "loss": 0.072,
+      "step": 2164
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017203798706464952,
+      "loss": 0.075,
+      "step": 2165
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017201030867496005,
+      "loss": 0.077,
+      "step": 2166
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017198261882213074,
+      "loss": 0.0774,
+      "step": 2167
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017195491751056946,
+      "loss": 0.0776,
+      "step": 2168
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001719272047446859,
+      "loss": 0.0749,
+      "step": 2169
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017189948052889154,
+      "loss": 0.0753,
+      "step": 2170
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017187174486759982,
+      "loss": 0.0769,
+      "step": 2171
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017184399776522583,
+      "loss": 0.0773,
+      "step": 2172
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017181623922618664,
+      "loss": 0.0764,
+      "step": 2173
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000171788469254901,
+      "loss": 0.079,
+      "step": 2174
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017176068785578955,
+      "loss": 0.0768,
+      "step": 2175
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017173289503327472,
+      "loss": 0.0745,
+      "step": 2176
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017170509079178083,
+      "loss": 0.0761,
+      "step": 2177
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017167727513573393,
+      "loss": 0.0726,
+      "step": 2178
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017164944806956199,
+      "loss": 0.0779,
+      "step": 2179
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017162160959769462,
+      "loss": 0.0772,
+      "step": 2180
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017159375972456345,
+      "loss": 0.0743,
+      "step": 2181
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017156589845460177,
+      "loss": 0.0746,
+      "step": 2182
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001715380257922448,
+      "loss": 0.0764,
+      "step": 2183
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001715101417419295,
+      "loss": 0.0758,
+      "step": 2184
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00017148224630809462,
+      "loss": 0.0747,
+      "step": 2185
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001714543394951808,
+      "loss": 0.0737,
+      "step": 2186
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001714264213076305,
+      "loss": 0.0758,
+      "step": 2187
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017139849174988787,
+      "loss": 0.0771,
+      "step": 2188
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017137055082639896,
+      "loss": 0.0737,
+      "step": 2189
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017134259854161162,
+      "loss": 0.0748,
+      "step": 2190
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017131463489997558,
+      "loss": 0.0777,
+      "step": 2191
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017128665990594216,
+      "loss": 0.0766,
+      "step": 2192
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017125867356396476,
+      "loss": 0.0759,
+      "step": 2193
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001712306758784984,
+      "loss": 0.0744,
+      "step": 2194
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017120266685399994,
+      "loss": 0.0757,
+      "step": 2195
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017117464649492808,
+      "loss": 0.0752,
+      "step": 2196
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017114661480574332,
+      "loss": 0.0762,
+      "step": 2197
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017111857179090797,
+      "loss": 0.0779,
+      "step": 2198
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001710905174548861,
+      "loss": 0.0763,
+      "step": 2199
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001710624518021436,
+      "loss": 0.0776,
+      "step": 2200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017103437483714822,
+      "loss": 0.0767,
+      "step": 2201
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017100628656436942,
+      "loss": 0.0758,
+      "step": 2202
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017097818698827853,
+      "loss": 0.0779,
+      "step": 2203
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001709500761133486,
+      "loss": 0.0771,
+      "step": 2204
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017092195394405457,
+      "loss": 0.0757,
+      "step": 2205
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017089382048487314,
+      "loss": 0.0748,
+      "step": 2206
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001708656757402828,
+      "loss": 0.0757,
+      "step": 2207
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001708375197147638,
+      "loss": 0.0772,
+      "step": 2208
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001708093524127983,
+      "loss": 0.077,
+      "step": 2209
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017078117383887014,
+      "loss": 0.0723,
+      "step": 2210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000170752983997465,
+      "loss": 0.0761,
+      "step": 2211
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00017072478289307033,
+      "loss": 0.0756,
+      "step": 2212
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017069657053017542,
+      "loss": 0.0747,
+      "step": 2213
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017066834691327132,
+      "loss": 0.0765,
+      "step": 2214
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017064011204685092,
+      "loss": 0.0796,
+      "step": 2215
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017061186593540876,
+      "loss": 0.0749,
+      "step": 2216
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017058360858344132,
+      "loss": 0.0776,
+      "step": 2217
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001705553399954468,
+      "loss": 0.0755,
+      "step": 2218
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017052706017592527,
+      "loss": 0.0739,
+      "step": 2219
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017049876912937845,
+      "loss": 0.0771,
+      "step": 2220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017047046686030997,
+      "loss": 0.0765,
+      "step": 2221
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017044215337322512,
+      "loss": 0.075,
+      "step": 2222
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017041382867263115,
+      "loss": 0.0753,
+      "step": 2223
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017038549276303692,
+      "loss": 0.0749,
+      "step": 2224
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001703571456489532,
+      "loss": 0.0769,
+      "step": 2225
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001703287873348925,
+      "loss": 0.0715,
+      "step": 2226
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001703004178253691,
+      "loss": 0.0764,
+      "step": 2227
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000170272037124899,
+      "loss": 0.0753,
+      "step": 2228
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017024364523800017,
+      "loss": 0.0774,
+      "step": 2229
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017021524216919218,
+      "loss": 0.0745,
+      "step": 2230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017018682792299642,
+      "loss": 0.077,
+      "step": 2231
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017015840250393614,
+      "loss": 0.077,
+      "step": 2232
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017012996591653625,
+      "loss": 0.0774,
+      "step": 2233
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017010151816532356,
+      "loss": 0.0735,
+      "step": 2234
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017007305925482657,
+      "loss": 0.0748,
+      "step": 2235
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017004458918957555,
+      "loss": 0.0744,
+      "step": 2236
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017001610797410263,
+      "loss": 0.0764,
+      "step": 2237
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016998761561294164,
+      "loss": 0.0751,
+      "step": 2238
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001699591121106282,
+      "loss": 0.0772,
+      "step": 2239
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001699305974716997,
+      "loss": 0.0764,
+      "step": 2240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016990207170069536,
+      "loss": 0.076,
+      "step": 2241
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001698735348021561,
+      "loss": 0.0758,
+      "step": 2242
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016984498678062463,
+      "loss": 0.0773,
+      "step": 2243
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016981642764064546,
+      "loss": 0.0773,
+      "step": 2244
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001697878573867648,
+      "loss": 0.0731,
+      "step": 2245
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001697592760235307,
+      "loss": 0.0783,
+      "step": 2246
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.000169730683555493,
+      "loss": 0.0777,
+      "step": 2247
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001697020799872032,
+      "loss": 0.0736,
+      "step": 2248
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016967346532321465,
+      "loss": 0.078,
+      "step": 2249
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016964483956808247,
+      "loss": 0.0742,
+      "step": 2250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016961620272636348,
+      "loss": 0.0757,
+      "step": 2251
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016958755480261638,
+      "loss": 0.0758,
+      "step": 2252
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016955889580140145,
+      "loss": 0.0767,
+      "step": 2253
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016953022572728095,
+      "loss": 0.0747,
+      "step": 2254
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016950154458481872,
+      "loss": 0.076,
+      "step": 2255
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001694728523785805,
+      "loss": 0.0774,
+      "step": 2256
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001694441491131337,
+      "loss": 0.0777,
+      "step": 2257
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001694154347930475,
+      "loss": 0.0755,
+      "step": 2258
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016938670942289293,
+      "loss": 0.0773,
+      "step": 2259
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001693579730072426,
+      "loss": 0.0729,
+      "step": 2260
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001693292255506711,
+      "loss": 0.0745,
+      "step": 2261
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016930046705775458,
+      "loss": 0.0733,
+      "step": 2262
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016927169753307107,
+      "loss": 0.0761,
+      "step": 2263
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016924291698120034,
+      "loss": 0.0724,
+      "step": 2264
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016921412540672385,
+      "loss": 0.0775,
+      "step": 2265
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001691853228142249,
+      "loss": 0.0769,
+      "step": 2266
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016915650920828847,
+      "loss": 0.0765,
+      "step": 2267
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016912768459350134,
+      "loss": 0.0768,
+      "step": 2268
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000169098848974452,
+      "loss": 0.0782,
+      "step": 2269
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016907000235573074,
+      "loss": 0.0735,
+      "step": 2270
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001690411447419296,
+      "loss": 0.0772,
+      "step": 2271
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016901227613764233,
+      "loss": 0.0762,
+      "step": 2272
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016898339654746448,
+      "loss": 0.0767,
+      "step": 2273
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016895450597599328,
+      "loss": 0.0761,
+      "step": 2274
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016892560442782775,
+      "loss": 0.0769,
+      "step": 2275
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016889669190756868,
+      "loss": 0.0768,
+      "step": 2276
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016886776841981856,
+      "loss": 0.0762,
+      "step": 2277
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016883883396918165,
+      "loss": 0.0742,
+      "step": 2278
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016880988856026393,
+      "loss": 0.0776,
+      "step": 2279
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016878093219767324,
+      "loss": 0.0756,
+      "step": 2280
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016875196488601895,
+      "loss": 0.0791,
+      "step": 2281
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016872298662991237,
+      "loss": 0.0745,
+      "step": 2282
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016869399743396648,
+      "loss": 0.0771,
+      "step": 2283
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001686649973027959,
+      "loss": 0.0805,
+      "step": 2284
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016863598624101723,
+      "loss": 0.0765,
+      "step": 2285
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016860696425324858,
+      "loss": 0.0758,
+      "step": 2286
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016857793134410987,
+      "loss": 0.0757,
+      "step": 2287
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016854888751822284,
+      "loss": 0.074,
+      "step": 2288
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001685198327802109,
+      "loss": 0.0764,
+      "step": 2289
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016849076713469914,
+      "loss": 0.0736,
+      "step": 2290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016846169058631446,
+      "loss": 0.0731,
+      "step": 2291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016843260313968554,
+      "loss": 0.0771,
+      "step": 2292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001684035047994427,
+      "loss": 0.0745,
+      "step": 2293
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000168374395570218,
+      "loss": 0.0741,
+      "step": 2294
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016834527545664532,
+      "loss": 0.0754,
+      "step": 2295
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016831614446336017,
+      "loss": 0.0757,
+      "step": 2296
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001682870025949999,
+      "loss": 0.0737,
+      "step": 2297
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016825784985620348,
+      "loss": 0.0757,
+      "step": 2298
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016822868625161166,
+      "loss": 0.0756,
+      "step": 2299
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016819951178586697,
+      "loss": 0.0772,
+      "step": 2300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016817032646361359,
+      "loss": 0.0731,
+      "step": 2301
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016814113028949744,
+      "loss": 0.0754,
+      "step": 2302
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001681119232681662,
+      "loss": 0.0735,
+      "step": 2303
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016808270540426924,
+      "loss": 0.0756,
+      "step": 2304
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016805347670245774,
+      "loss": 0.0721,
+      "step": 2305
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001680242371673845,
+      "loss": 0.0803,
+      "step": 2306
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016799498680370408,
+      "loss": 0.0784,
+      "step": 2307
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016796572561607277,
+      "loss": 0.0774,
+      "step": 2308
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016793645360914863,
+      "loss": 0.0745,
+      "step": 2309
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016790717078759136,
+      "loss": 0.0747,
+      "step": 2310
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001678778771560624,
+      "loss": 0.0738,
+      "step": 2311
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016784857271922497,
+      "loss": 0.0745,
+      "step": 2312
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016781925748174397,
+      "loss": 0.0796,
+      "step": 2313
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000167789931448286,
+      "loss": 0.0758,
+      "step": 2314
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016776059462351938,
+      "loss": 0.0768,
+      "step": 2315
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016773124701211417,
+      "loss": 0.075,
+      "step": 2316
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001677018886187422,
+      "loss": 0.0758,
+      "step": 2317
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016767251944807692,
+      "loss": 0.0757,
+      "step": 2318
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001676431395047935,
+      "loss": 0.0733,
+      "step": 2319
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001676137487935689,
+      "loss": 0.0752,
+      "step": 2320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016758434731908178,
+      "loss": 0.0755,
+      "step": 2321
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016755493508601238,
+      "loss": 0.0753,
+      "step": 2322
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016752551209904288,
+      "loss": 0.0785,
+      "step": 2323
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000167496078362857,
+      "loss": 0.0702,
+      "step": 2324
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016746663388214018,
+      "loss": 0.0739,
+      "step": 2325
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016743717866157974,
+      "loss": 0.0761,
+      "step": 2326
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016740771270586444,
+      "loss": 0.076,
+      "step": 2327
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016737823601968495,
+      "loss": 0.0735,
+      "step": 2328
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001673487486077336,
+      "loss": 0.077,
+      "step": 2329
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001673192504747044,
+      "loss": 0.0768,
+      "step": 2330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016728974162529312,
+      "loss": 0.0769,
+      "step": 2331
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016726022206419713,
+      "loss": 0.0749,
+      "step": 2332
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016723069179611566,
+      "loss": 0.0769,
+      "step": 2333
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016720115082574947,
+      "loss": 0.0769,
+      "step": 2334
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016717159915780118,
+      "loss": 0.0731,
+      "step": 2335
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016714203679697503,
+      "loss": 0.0734,
+      "step": 2336
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016711246374797695,
+      "loss": 0.0776,
+      "step": 2337
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016708288001551465,
+      "loss": 0.0719,
+      "step": 2338
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016705328560429742,
+      "loss": 0.0754,
+      "step": 2339
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016702368051903636,
+      "loss": 0.073,
+      "step": 2340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016699406476444425,
+      "loss": 0.0743,
+      "step": 2341
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001669644383452355,
+      "loss": 0.0761,
+      "step": 2342
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016693480126612635,
+      "loss": 0.0729,
+      "step": 2343
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016690515353183453,
+      "loss": 0.077,
+      "step": 2344
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016687549514707967,
+      "loss": 0.0766,
+      "step": 2345
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000166845826116583,
+      "loss": 0.0766,
+      "step": 2346
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016681614644506747,
+      "loss": 0.076,
+      "step": 2347
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016678645613725767,
+      "loss": 0.0749,
+      "step": 2348
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016675675519787995,
+      "loss": 0.0754,
+      "step": 2349
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016672704363166236,
+      "loss": 0.077,
+      "step": 2350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016669732144333454,
+      "loss": 0.0758,
+      "step": 2351
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016666758863762793,
+      "loss": 0.0737,
+      "step": 2352
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016663784521927567,
+      "loss": 0.0759,
+      "step": 2353
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016660809119301243,
+      "loss": 0.0767,
+      "step": 2354
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016657832656357476,
+      "loss": 0.0757,
+      "step": 2355
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016654855133570079,
+      "loss": 0.0773,
+      "step": 2356
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016651876551413038,
+      "loss": 0.0757,
+      "step": 2357
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016648896910360502,
+      "loss": 0.077,
+      "step": 2358
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016645916210886798,
+      "loss": 0.0755,
+      "step": 2359
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016642934453466412,
+      "loss": 0.0779,
+      "step": 2360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016639951638574004,
+      "loss": 0.0805,
+      "step": 2361
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000166369677666844,
+      "loss": 0.0773,
+      "step": 2362
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016633982838272596,
+      "loss": 0.075,
+      "step": 2363
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016630996853813756,
+      "loss": 0.0741,
+      "step": 2364
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016628009813783213,
+      "loss": 0.0801,
+      "step": 2365
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016625021718656458,
+      "loss": 0.0752,
+      "step": 2366
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016622032568909168,
+      "loss": 0.078,
+      "step": 2367
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016619042365017173,
+      "loss": 0.0747,
+      "step": 2368
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016616051107456477,
+      "loss": 0.0787,
+      "step": 2369
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001661305879670325,
+      "loss": 0.0741,
+      "step": 2370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001661006543323383,
+      "loss": 0.0761,
+      "step": 2371
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001660707101752473,
+      "loss": 0.0749,
+      "step": 2372
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016604075550052613,
+      "loss": 0.0783,
+      "step": 2373
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016601079031294325,
+      "loss": 0.0774,
+      "step": 2374
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016598081461726878,
+      "loss": 0.0799,
+      "step": 2375
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001659508284182744,
+      "loss": 0.0763,
+      "step": 2376
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001659208317207336,
+      "loss": 0.0743,
+      "step": 2377
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016589082452942146,
+      "loss": 0.0762,
+      "step": 2378
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016586080684911472,
+      "loss": 0.0756,
+      "step": 2379
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016583077868459185,
+      "loss": 0.0764,
+      "step": 2380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016580074004063296,
+      "loss": 0.0743,
+      "step": 2381
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001657706909220198,
+      "loss": 0.0777,
+      "step": 2382
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016574063133353582,
+      "loss": 0.0771,
+      "step": 2383
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016571056127996617,
+      "loss": 0.075,
+      "step": 2384
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016568048076609756,
+      "loss": 0.0745,
+      "step": 2385
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001656503897967185,
+      "loss": 0.0734,
+      "step": 2386
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016562028837661906,
+      "loss": 0.0801,
+      "step": 2387
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000165590176510591,
+      "loss": 0.0758,
+      "step": 2388
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016556005420342778,
+      "loss": 0.0781,
+      "step": 2389
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016552992145992446,
+      "loss": 0.074,
+      "step": 2390
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001654997782848778,
+      "loss": 0.0754,
+      "step": 2391
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016546962468308628,
+      "loss": 0.0769,
+      "step": 2392
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001654394606593499,
+      "loss": 0.0764,
+      "step": 2393
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016540928621847043,
+      "loss": 0.0748,
+      "step": 2394
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016537910136525124,
+      "loss": 0.0796,
+      "step": 2395
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001653489061044974,
+      "loss": 0.0729,
+      "step": 2396
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016531870044101562,
+      "loss": 0.0744,
+      "step": 2397
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016528848437961425,
+      "loss": 0.0735,
+      "step": 2398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001652582579251033,
+      "loss": 0.0765,
+      "step": 2399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001652280210822945,
+      "loss": 0.0769,
+      "step": 2400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016519777385600113,
+      "loss": 0.0761,
+      "step": 2401
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016516751625103815,
+      "loss": 0.0716,
+      "step": 2402
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016513724827222227,
+      "loss": 0.0762,
+      "step": 2403
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016510696992437165,
+      "loss": 0.0755,
+      "step": 2404
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001650766812123063,
+      "loss": 0.08,
+      "step": 2405
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016504638214084785,
+      "loss": 0.0752,
+      "step": 2406
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016501607271481944,
+      "loss": 0.0766,
+      "step": 2407
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016498575293904603,
+      "loss": 0.0746,
+      "step": 2408
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016495542281835408,
+      "loss": 0.0768,
+      "step": 2409
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016492508235757184,
+      "loss": 0.0752,
+      "step": 2410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016489473156152903,
+      "loss": 0.0762,
+      "step": 2411
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001648643704350572,
+      "loss": 0.0743,
+      "step": 2412
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016483399898298946,
+      "loss": 0.0769,
+      "step": 2413
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016480361721016054,
+      "loss": 0.0726,
+      "step": 2414
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016477322512140682,
+      "loss": 0.0763,
+      "step": 2415
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016474282272156639,
+      "loss": 0.0731,
+      "step": 2416
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016471241001547886,
+      "loss": 0.0748,
+      "step": 2417
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016468198700798561,
+      "loss": 0.0785,
+      "step": 2418
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016465155370392958,
+      "loss": 0.0788,
+      "step": 2419
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016462111010815542,
+      "loss": 0.0763,
+      "step": 2420
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016459065622550928,
+      "loss": 0.0769,
+      "step": 2421
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001645601920608391,
+      "loss": 0.0766,
+      "step": 2422
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016452971761899438,
+      "loss": 0.0784,
+      "step": 2423
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016449923290482628,
+      "loss": 0.075,
+      "step": 2424
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016446873792318756,
+      "loss": 0.0763,
+      "step": 2425
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016443823267893267,
+      "loss": 0.0758,
+      "step": 2426
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001644077171769176,
+      "loss": 0.0737,
+      "step": 2427
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016437719142200013,
+      "loss": 0.077,
+      "step": 2428
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001643466554190395,
+      "loss": 0.0744,
+      "step": 2429
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016431610917289668,
+      "loss": 0.0794,
+      "step": 2430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001642855526884343,
+      "loss": 0.0766,
+      "step": 2431
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016425498597051648,
+      "loss": 0.0758,
+      "step": 2432
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016422440902400914,
+      "loss": 0.075,
+      "step": 2433
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001641938218537797,
+      "loss": 0.0767,
+      "step": 2434
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016416322446469728,
+      "loss": 0.077,
+      "step": 2435
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016413261686163257,
+      "loss": 0.0753,
+      "step": 2436
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016410199904945795,
+      "loss": 0.0761,
+      "step": 2437
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001640713710330474,
+      "loss": 0.0747,
+      "step": 2438
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001640407328172765,
+      "loss": 0.0774,
+      "step": 2439
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016401008440702244,
+      "loss": 0.0728,
+      "step": 2440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016397942580716407,
+      "loss": 0.0733,
+      "step": 2441
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001639487570225819,
+      "loss": 0.0734,
+      "step": 2442
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016391807805815799,
+      "loss": 0.0787,
+      "step": 2443
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016388738891877607,
+      "loss": 0.0781,
+      "step": 2444
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016385668960932143,
+      "loss": 0.0765,
+      "step": 2445
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016382598013468106,
+      "loss": 0.0771,
+      "step": 2446
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016379526049974347,
+      "loss": 0.0745,
+      "step": 2447
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016376453070939887,
+      "loss": 0.0769,
+      "step": 2448
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016373379076853908,
+      "loss": 0.074,
+      "step": 2449
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016370304068205746,
+      "loss": 0.0736,
+      "step": 2450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016367228045484908,
+      "loss": 0.0784,
+      "step": 2451
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001636415100918106,
+      "loss": 0.0749,
+      "step": 2452
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001636107295978402,
+      "loss": 0.0739,
+      "step": 2453
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001635799389778378,
+      "loss": 0.0766,
+      "step": 2454
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016354913823670494,
+      "loss": 0.0777,
+      "step": 2455
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016351832737934458,
+      "loss": 0.0742,
+      "step": 2456
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016348750641066154,
+      "loss": 0.0723,
+      "step": 2457
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016345667533556205,
+      "loss": 0.0766,
+      "step": 2458
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016342583415895412,
+      "loss": 0.0778,
+      "step": 2459
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016339498288574719,
+      "loss": 0.0717,
+      "step": 2460
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016336412152085248,
+      "loss": 0.0768,
+      "step": 2461
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016333325006918266,
+      "loss": 0.0735,
+      "step": 2462
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001633023685356521,
+      "loss": 0.0741,
+      "step": 2463
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016327147692517675,
+      "loss": 0.0775,
+      "step": 2464
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016324057524267418,
+      "loss": 0.0734,
+      "step": 2465
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016320966349306357,
+      "loss": 0.0716,
+      "step": 2466
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016317874168126565,
+      "loss": 0.0733,
+      "step": 2467
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001631478098122028,
+      "loss": 0.0751,
+      "step": 2468
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016311686789079898,
+      "loss": 0.0749,
+      "step": 2469
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016308591592197974,
+      "loss": 0.0745,
+      "step": 2470
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001630549539106723,
+      "loss": 0.0752,
+      "step": 2471
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001630239818618054,
+      "loss": 0.0772,
+      "step": 2472
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016299299978030936,
+      "loss": 0.0777,
+      "step": 2473
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001629620076711162,
+      "loss": 0.0771,
+      "step": 2474
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016293100553915947,
+      "loss": 0.074,
+      "step": 2475
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016289999338937427,
+      "loss": 0.078,
+      "step": 2476
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016286897122669736,
+      "loss": 0.0798,
+      "step": 2477
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016283793905606713,
+      "loss": 0.0728,
+      "step": 2478
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016280689688242346,
+      "loss": 0.0745,
+      "step": 2479
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016277584471070792,
+      "loss": 0.078,
+      "step": 2480
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016274478254586357,
+      "loss": 0.0728,
+      "step": 2481
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016271371039283517,
+      "loss": 0.0774,
+      "step": 2482
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016268262825656901,
+      "loss": 0.0749,
+      "step": 2483
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016265153614201293,
+      "loss": 0.0733,
+      "step": 2484
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001626204340541165,
+      "loss": 0.0771,
+      "step": 2485
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016258932199783067,
+      "loss": 0.0748,
+      "step": 2486
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016255819997810814,
+      "loss": 0.0773,
+      "step": 2487
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001625270679999032,
+      "loss": 0.0729,
+      "step": 2488
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016249592606817158,
+      "loss": 0.0761,
+      "step": 2489
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016246477418787076,
+      "loss": 0.0742,
+      "step": 2490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001624336123639597,
+      "loss": 0.0775,
+      "step": 2491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016240244060139895,
+      "loss": 0.0782,
+      "step": 2492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016237125890515065,
+      "loss": 0.0745,
+      "step": 2493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001623400672801786,
+      "loss": 0.0776,
+      "step": 2494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016230886573144812,
+      "loss": 0.0777,
+      "step": 2495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016227765426392603,
+      "loss": 0.0773,
+      "step": 2496
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001622464328825809,
+      "loss": 0.0758,
+      "step": 2497
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016221520159238266,
+      "loss": 0.0731,
+      "step": 2498
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016218396039830303,
+      "loss": 0.0758,
+      "step": 2499
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016215270930531518,
+      "loss": 0.076,
+      "step": 2500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016212144831839395,
+      "loss": 0.0757,
+      "step": 2501
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016209017744251564,
+      "loss": 0.0778,
+      "step": 2502
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016205889668265822,
+      "loss": 0.0759,
+      "step": 2503
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016202760604380116,
+      "loss": 0.0757,
+      "step": 2504
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016199630553092556,
+      "loss": 0.0726,
+      "step": 2505
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016196499514901404,
+      "loss": 0.0775,
+      "step": 2506
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016193367490305088,
+      "loss": 0.0761,
+      "step": 2507
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016190234479802182,
+      "loss": 0.0786,
+      "step": 2508
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016187100483891422,
+      "loss": 0.0747,
+      "step": 2509
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016183965503071705,
+      "loss": 0.0762,
+      "step": 2510
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016180829537842077,
+      "loss": 0.076,
+      "step": 2511
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016177692588701748,
+      "loss": 0.0751,
+      "step": 2512
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016174554656150076,
+      "loss": 0.0771,
+      "step": 2513
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016171415740686584,
+      "loss": 0.0748,
+      "step": 2514
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016168275842810947,
+      "loss": 0.0727,
+      "step": 2515
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016165134963023,
+      "loss": 0.0781,
+      "step": 2516
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016161993101822727,
+      "loss": 0.0773,
+      "step": 2517
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016158850259710277,
+      "loss": 0.0777,
+      "step": 2518
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016155706437185949,
+      "loss": 0.0714,
+      "step": 2519
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016152561634750202,
+      "loss": 0.0755,
+      "step": 2520
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016149415852903648,
+      "loss": 0.0748,
+      "step": 2521
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016146269092147053,
+      "loss": 0.0752,
+      "step": 2522
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016143121352981348,
+      "loss": 0.0733,
+      "step": 2523
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001613997263590761,
+      "loss": 0.0733,
+      "step": 2524
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016136822941427076,
+      "loss": 0.0752,
+      "step": 2525
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001613367227004114,
+      "loss": 0.0776,
+      "step": 2526
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016130520622251347,
+      "loss": 0.0737,
+      "step": 2527
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016127367998559398,
+      "loss": 0.0751,
+      "step": 2528
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016124214399467155,
+      "loss": 0.0768,
+      "step": 2529
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001612105982547663,
+      "loss": 0.0724,
+      "step": 2530
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016117904277089992,
+      "loss": 0.0774,
+      "step": 2531
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016114747754809563,
+      "loss": 0.0765,
+      "step": 2532
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016111590259137824,
+      "loss": 0.0766,
+      "step": 2533
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001610843179057741,
+      "loss": 0.0738,
+      "step": 2534
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016105272349631105,
+      "loss": 0.0761,
+      "step": 2535
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016102111936801863,
+      "loss": 0.0771,
+      "step": 2536
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016098950552592765,
+      "loss": 0.0739,
+      "step": 2537
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016095788197507081,
+      "loss": 0.075,
+      "step": 2538
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016092624872048205,
+      "loss": 0.0795,
+      "step": 2539
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001608946057671971,
+      "loss": 0.0766,
+      "step": 2540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000160862953120253,
+      "loss": 0.0747,
+      "step": 2541
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016083129078468857,
+      "loss": 0.0737,
+      "step": 2542
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000160799618765544,
+      "loss": 0.076,
+      "step": 2543
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016076793706786108,
+      "loss": 0.0705,
+      "step": 2544
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001607362456966832,
+      "loss": 0.0748,
+      "step": 2545
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016070454465705512,
+      "loss": 0.0759,
+      "step": 2546
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001606728339540233,
+      "loss": 0.0748,
+      "step": 2547
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016064111359263573,
+      "loss": 0.0752,
+      "step": 2548
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016060938357794182,
+      "loss": 0.0738,
+      "step": 2549
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001605776439149927,
+      "loss": 0.0736,
+      "step": 2550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016054589460884078,
+      "loss": 0.0749,
+      "step": 2551
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016051413566454026,
+      "loss": 0.0762,
+      "step": 2552
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016048236708714674,
+      "loss": 0.0755,
+      "step": 2553
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016045058888171737,
+      "loss": 0.0742,
+      "step": 2554
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016041880105331085,
+      "loss": 0.074,
+      "step": 2555
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016038700360698738,
+      "loss": 0.0762,
+      "step": 2556
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016035519654780878,
+      "loss": 0.0755,
+      "step": 2557
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016032337988083827,
+      "loss": 0.0747,
+      "step": 2558
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016029155361114067,
+      "loss": 0.0736,
+      "step": 2559
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001602597177437824,
+      "loss": 0.0723,
+      "step": 2560
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016022787228383125,
+      "loss": 0.0747,
+      "step": 2561
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001601960172363566,
+      "loss": 0.0744,
+      "step": 2562
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016016415260642947,
+      "loss": 0.0741,
+      "step": 2563
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016013227839912223,
+      "loss": 0.0772,
+      "step": 2564
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016010039461950893,
+      "loss": 0.0778,
+      "step": 2565
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016006850127266498,
+      "loss": 0.0769,
+      "step": 2566
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016003659836366744,
+      "loss": 0.076,
+      "step": 2567
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016000468589759485,
+      "loss": 0.0778,
+      "step": 2568
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015997276387952732,
+      "loss": 0.0744,
+      "step": 2569
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015994083231454638,
+      "loss": 0.0738,
+      "step": 2570
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015990889120773513,
+      "loss": 0.0732,
+      "step": 2571
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015987694056417823,
+      "loss": 0.0804,
+      "step": 2572
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015984498038896182,
+      "loss": 0.0757,
+      "step": 2573
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015981301068717358,
+      "loss": 0.0731,
+      "step": 2574
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015978103146390262,
+      "loss": 0.0738,
+      "step": 2575
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015974904272423966,
+      "loss": 0.0755,
+      "step": 2576
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00015971704447327695,
+      "loss": 0.0765,
+      "step": 2577
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015968503671610814,
+      "loss": 0.076,
+      "step": 2578
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015965301945782852,
+      "loss": 0.0752,
+      "step": 2579
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015962099270353485,
+      "loss": 0.0722,
+      "step": 2580
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015958895645832533,
+      "loss": 0.075,
+      "step": 2581
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015955691072729971,
+      "loss": 0.075,
+      "step": 2582
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015952485551555938,
+      "loss": 0.071,
+      "step": 2583
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015949279082820702,
+      "loss": 0.077,
+      "step": 2584
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.000159460716670347,
+      "loss": 0.0785,
+      "step": 2585
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015942863304708508,
+      "loss": 0.0796,
+      "step": 2586
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015939653996352858,
+      "loss": 0.0784,
+      "step": 2587
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015936443742478633,
+      "loss": 0.0795,
+      "step": 2588
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015933232543596869,
+      "loss": 0.0772,
+      "step": 2589
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00015930020400218737,
+      "loss": 0.0745,
+      "step": 2590
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015926807312855582,
+      "loss": 0.0746,
+      "step": 2591
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001592359328201888,
+      "loss": 0.0767,
+      "step": 2592
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015920378308220272,
+      "loss": 0.0724,
+      "step": 2593
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015917162391971532,
+      "loss": 0.0773,
+      "step": 2594
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015913945533784597,
+      "loss": 0.074,
+      "step": 2595
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015910727734171553,
+      "loss": 0.0774,
+      "step": 2596
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015907508993644634,
+      "loss": 0.0756,
+      "step": 2597
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015904289312716223,
+      "loss": 0.075,
+      "step": 2598
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015901068691898847,
+      "loss": 0.0731,
+      "step": 2599
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015897847131705195,
+      "loss": 0.0736,
+      "step": 2600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015894624632648097,
+      "loss": 0.074,
+      "step": 2601
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00015891401195240534,
+      "loss": 0.0767,
+      "step": 2602
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001588817681999564,
+      "loss": 0.0756,
+      "step": 2603
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001588495150742669,
+      "loss": 0.0748,
+      "step": 2604
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015881725258047117,
+      "loss": 0.0753,
+      "step": 2605
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015878498072370497,
+      "loss": 0.0724,
+      "step": 2606
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015875269950910557,
+      "loss": 0.076,
+      "step": 2607
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001587204089418118,
+      "loss": 0.0743,
+      "step": 2608
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015868810902696384,
+      "loss": 0.0754,
+      "step": 2609
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015865579976970348,
+      "loss": 0.0766,
+      "step": 2610
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015862348117517394,
+      "loss": 0.0751,
+      "step": 2611
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015859115324851989,
+      "loss": 0.0754,
+      "step": 2612
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015855881599488762,
+      "loss": 0.074,
+      "step": 2613
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015852646941942472,
+      "loss": 0.0747,
+      "step": 2614
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015849411352728042,
+      "loss": 0.0741,
+      "step": 2615
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015846174832360535,
+      "loss": 0.0753,
+      "step": 2616
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00015842937381355164,
+      "loss": 0.0763,
+      "step": 2617
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.000158396990002273,
+      "loss": 0.0781,
+      "step": 2618
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001583645968949244,
+      "loss": 0.0713,
+      "step": 2619
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00015833219449666242,
+      "loss": 0.0737,
+      "step": 2620
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00015829978281264517,
+      "loss": 0.0761,
+      "step": 2621
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00015826736184803224,
+      "loss": 0.0757,
+      "step": 2622
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001582349316079845,
+      "loss": 0.0745,
+      "step": 2623
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00015820249209766457,
+      "loss": 0.0754,
+      "step": 2624
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00015817004332223633,
+      "loss": 0.0745,
+      "step": 2625
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00015813758528686523,
+      "loss": 0.0772,
+      "step": 2626
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00015810511799671826,
+      "loss": 0.0787,
+      "step": 2627
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001580726414569637,
+      "loss": 0.0762,
+      "step": 2628
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001580401556727715,
+      "loss": 0.0757,
+      "step": 2629
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00015800766064931286,
+      "loss": 0.078,
+      "step": 2630
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00015797515639176074,
+      "loss": 0.0748,
+      "step": 2631
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015794264290528936,
+      "loss": 0.0779,
+      "step": 2632
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001579101201950744,
+      "loss": 0.0752,
+      "step": 2633
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015787758826629308,
+      "loss": 0.0757,
+      "step": 2634
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015784504712412415,
+      "loss": 0.0745,
+      "step": 2635
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015781249677374767,
+      "loss": 0.0755,
+      "step": 2636
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015777993722034529,
+      "loss": 0.0749,
+      "step": 2637
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015774736846910009,
+      "loss": 0.0752,
+      "step": 2638
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001577147905251966,
+      "loss": 0.0765,
+      "step": 2639
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015768220339382077,
+      "loss": 0.0745,
+      "step": 2640
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015764960708016017,
+      "loss": 0.0726,
+      "step": 2641
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015761700158940364,
+      "loss": 0.0726,
+      "step": 2642
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015758438692674157,
+      "loss": 0.0767,
+      "step": 2643
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015755176309736586,
+      "loss": 0.0769,
+      "step": 2644
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00015751913010646974,
+      "loss": 0.0737,
+      "step": 2645
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015748648795924804,
+      "loss": 0.0762,
+      "step": 2646
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000157453836660897,
+      "loss": 0.075,
+      "step": 2647
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015742117621661417,
+      "loss": 0.0752,
+      "step": 2648
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015738850663159882,
+      "loss": 0.0739,
+      "step": 2649
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015735582791105148,
+      "loss": 0.0758,
+      "step": 2650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015732314006017422,
+      "loss": 0.0746,
+      "step": 2651
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001572904430841705,
+      "loss": 0.0751,
+      "step": 2652
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015725773698824525,
+      "loss": 0.0766,
+      "step": 2653
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015722502177760496,
+      "loss": 0.0725,
+      "step": 2654
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015719229745745738,
+      "loss": 0.0783,
+      "step": 2655
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001571595640330119,
+      "loss": 0.0753,
+      "step": 2656
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015712682150947923,
+      "loss": 0.0725,
+      "step": 2657
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00015709406989207158,
+      "loss": 0.0741,
+      "step": 2658
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015706130918600257,
+      "loss": 0.0756,
+      "step": 2659
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015702853939648733,
+      "loss": 0.0766,
+      "step": 2660
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015699576052874238,
+      "loss": 0.0743,
+      "step": 2661
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001569629725879857,
+      "loss": 0.0746,
+      "step": 2662
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001569301755794368,
+      "loss": 0.0761,
+      "step": 2663
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001568973695083164,
+      "loss": 0.0757,
+      "step": 2664
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015686455437984693,
+      "loss": 0.077,
+      "step": 2665
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015683173019925213,
+      "loss": 0.0774,
+      "step": 2666
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001567988969717572,
+      "loss": 0.0771,
+      "step": 2667
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015676605470258873,
+      "loss": 0.0753,
+      "step": 2668
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015673320339697485,
+      "loss": 0.0741,
+      "step": 2669
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015670034306014503,
+      "loss": 0.0759,
+      "step": 2670
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015666747369733029,
+      "loss": 0.0758,
+      "step": 2671
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00015663459531376297,
+      "loss": 0.0772,
+      "step": 2672
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001566017079146769,
+      "loss": 0.0727,
+      "step": 2673
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001565688115053074,
+      "loss": 0.0734,
+      "step": 2674
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001565359060908911,
+      "loss": 0.0735,
+      "step": 2675
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00015650299167666615,
+      "loss": 0.0754,
+      "step": 2676
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00015647006826787212,
+      "loss": 0.0786,
+      "step": 2677
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00015643713586975,
+      "loss": 0.0788,
+      "step": 2678
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001564041944875422,
+      "loss": 0.077,
+      "step": 2679
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00015637124412649263,
+      "loss": 0.0758,
+      "step": 2680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001563382847918465,
+      "loss": 0.0754,
+      "step": 2681
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00015630531648885056,
+      "loss": 0.0738,
+      "step": 2682
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00015627233922275297,
+      "loss": 0.0749,
+      "step": 2683
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00015623935299880322,
+      "loss": 0.074,
+      "step": 2684
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00015620635782225245,
+      "loss": 0.074,
+      "step": 2685
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015617335369835295,
+      "loss": 0.0739,
+      "step": 2686
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015614034063235862,
+      "loss": 0.0748,
+      "step": 2687
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015610731862952468,
+      "loss": 0.0735,
+      "step": 2688
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001560742876951079,
+      "loss": 0.0773,
+      "step": 2689
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001560412478343663,
+      "loss": 0.0753,
+      "step": 2690
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015600819905255947,
+      "loss": 0.0751,
+      "step": 2691
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015597514135494837,
+      "loss": 0.0759,
+      "step": 2692
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015594207474679532,
+      "loss": 0.0757,
+      "step": 2693
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015590899923336414,
+      "loss": 0.0751,
+      "step": 2694
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015587591481992007,
+      "loss": 0.0761,
+      "step": 2695
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015584282151172969,
+      "loss": 0.0724,
+      "step": 2696
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015580971931406103,
+      "loss": 0.0721,
+      "step": 2697
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015577660823218358,
+      "loss": 0.0753,
+      "step": 2698
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00015574348827136823,
+      "loss": 0.0763,
+      "step": 2699
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001557103594368872,
+      "loss": 0.0756,
+      "step": 2700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001556772217340142,
+      "loss": 0.0728,
+      "step": 2701
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015564407516802437,
+      "loss": 0.0751,
+      "step": 2702
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015561091974419423,
+      "loss": 0.074,
+      "step": 2703
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015557775546780168,
+      "loss": 0.0773,
+      "step": 2704
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015554458234412607,
+      "loss": 0.0769,
+      "step": 2705
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015551140037844814,
+      "loss": 0.0762,
+      "step": 2706
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015547820957605005,
+      "loss": 0.0744,
+      "step": 2707
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015544500994221539,
+      "loss": 0.0747,
+      "step": 2708
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015541180148222905,
+      "loss": 0.077,
+      "step": 2709
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015537858420137747,
+      "loss": 0.0727,
+      "step": 2710
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001553453581049484,
+      "loss": 0.0781,
+      "step": 2711
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015531212319823103,
+      "loss": 0.076,
+      "step": 2712
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001552788794865159,
+      "loss": 0.076,
+      "step": 2713
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015524562697509506,
+      "loss": 0.0786,
+      "step": 2714
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015521236566926186,
+      "loss": 0.0727,
+      "step": 2715
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015517909557431105,
+      "loss": 0.0758,
+      "step": 2716
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015514581669553884,
+      "loss": 0.0744,
+      "step": 2717
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015511252903824285,
+      "loss": 0.0733,
+      "step": 2718
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015507923260772203,
+      "loss": 0.0753,
+      "step": 2719
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015504592740927674,
+      "loss": 0.08,
+      "step": 2720
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015501261344820873,
+      "loss": 0.0736,
+      "step": 2721
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015497929072982123,
+      "loss": 0.0729,
+      "step": 2722
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015494595925941875,
+      "loss": 0.0745,
+      "step": 2723
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015491261904230727,
+      "loss": 0.074,
+      "step": 2724
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015487927008379408,
+      "loss": 0.074,
+      "step": 2725
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015484591238918802,
+      "loss": 0.0752,
+      "step": 2726
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015481254596379912,
+      "loss": 0.0756,
+      "step": 2727
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015477917081293895,
+      "loss": 0.0783,
+      "step": 2728
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015474578694192042,
+      "loss": 0.0736,
+      "step": 2729
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015471239435605776,
+      "loss": 0.078,
+      "step": 2730
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015467899306066674,
+      "loss": 0.0739,
+      "step": 2731
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015464558306106435,
+      "loss": 0.0779,
+      "step": 2732
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015461216436256912,
+      "loss": 0.0758,
+      "step": 2733
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015457873697050082,
+      "loss": 0.0751,
+      "step": 2734
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015454530089018068,
+      "loss": 0.0748,
+      "step": 2735
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015451185612693136,
+      "loss": 0.0751,
+      "step": 2736
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015447840268607683,
+      "loss": 0.0752,
+      "step": 2737
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015444494057294242,
+      "loss": 0.0753,
+      "step": 2738
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015441146979285487,
+      "loss": 0.0752,
+      "step": 2739
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001543779903511424,
+      "loss": 0.0711,
+      "step": 2740
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015434450225313443,
+      "loss": 0.075,
+      "step": 2741
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015431100550416187,
+      "loss": 0.0758,
+      "step": 2742
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000154277500109557,
+      "loss": 0.0733,
+      "step": 2743
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015424398607465344,
+      "loss": 0.0734,
+      "step": 2744
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015421046340478625,
+      "loss": 0.0753,
+      "step": 2745
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015417693210529172,
+      "loss": 0.0739,
+      "step": 2746
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015414339218150773,
+      "loss": 0.0747,
+      "step": 2747
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015410984363877335,
+      "loss": 0.074,
+      "step": 2748
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015407628648242907,
+      "loss": 0.0754,
+      "step": 2749
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015404272071781682,
+      "loss": 0.0733,
+      "step": 2750
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015400914635027983,
+      "loss": 0.0764,
+      "step": 2751
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015397556338516273,
+      "loss": 0.0757,
+      "step": 2752
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015394197182781147,
+      "loss": 0.0751,
+      "step": 2753
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015390837168357345,
+      "loss": 0.0744,
+      "step": 2754
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015387476295779736,
+      "loss": 0.0714,
+      "step": 2755
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001538411456558333,
+      "loss": 0.0746,
+      "step": 2756
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015380751978303278,
+      "loss": 0.0791,
+      "step": 2757
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001537738853447485,
+      "loss": 0.0745,
+      "step": 2758
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015374024234633473,
+      "loss": 0.0767,
+      "step": 2759
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.000153706590793147,
+      "loss": 0.0781,
+      "step": 2760
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015367293069054217,
+      "loss": 0.0795,
+      "step": 2761
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0001536392620438786,
+      "loss": 0.0737,
+      "step": 2762
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015360558485851583,
+      "loss": 0.0776,
+      "step": 2763
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015357189913981487,
+      "loss": 0.0754,
+      "step": 2764
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015353820489313807,
+      "loss": 0.0751,
+      "step": 2765
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015350450212384912,
+      "loss": 0.0741,
+      "step": 2766
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015347079083731315,
+      "loss": 0.0743,
+      "step": 2767
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015343707103889647,
+      "loss": 0.0744,
+      "step": 2768
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015340334273396692,
+      "loss": 0.0747,
+      "step": 2769
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015336960592789357,
+      "loss": 0.0781,
+      "step": 2770
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015333586062604696,
+      "loss": 0.0805,
+      "step": 2771
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015330210683379888,
+      "loss": 0.0774,
+      "step": 2772
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001532683445565225,
+      "loss": 0.0751,
+      "step": 2773
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001532345737995924,
+      "loss": 0.0754,
+      "step": 2774
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015320079456838443,
+      "loss": 0.0737,
+      "step": 2775
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015316700686827584,
+      "loss": 0.0726,
+      "step": 2776
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015313321070464518,
+      "loss": 0.0779,
+      "step": 2777
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015309940608287234,
+      "loss": 0.0739,
+      "step": 2778
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001530655930083387,
+      "loss": 0.0751,
+      "step": 2779
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001530317714864268,
+      "loss": 0.0751,
+      "step": 2780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015299794152252063,
+      "loss": 0.0752,
+      "step": 2781
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015296410312200552,
+      "loss": 0.0761,
+      "step": 2782
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015293025629026805,
+      "loss": 0.0762,
+      "step": 2783
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015289640103269625,
+      "loss": 0.0712,
+      "step": 2784
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015286253735467947,
+      "loss": 0.0789,
+      "step": 2785
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015282866526160837,
+      "loss": 0.0739,
+      "step": 2786
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015279478475887498,
+      "loss": 0.0766,
+      "step": 2787
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015276089585187258,
+      "loss": 0.0736,
+      "step": 2788
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015272699854599597,
+      "loss": 0.0741,
+      "step": 2789
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001526930928466411,
+      "loss": 0.0738,
+      "step": 2790
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015265917875920536,
+      "loss": 0.074,
+      "step": 2791
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015262525628908743,
+      "loss": 0.0751,
+      "step": 2792
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001525913254416874,
+      "loss": 0.0768,
+      "step": 2793
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001525573862224065,
+      "loss": 0.0718,
+      "step": 2794
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001525234386366476,
+      "loss": 0.0776,
+      "step": 2795
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015248948268981461,
+      "loss": 0.0798,
+      "step": 2796
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.000152455518387313,
+      "loss": 0.0771,
+      "step": 2797
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015242154573454934,
+      "loss": 0.0746,
+      "step": 2798
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015238756473693167,
+      "loss": 0.0776,
+      "step": 2799
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015235357539986944,
+      "loss": 0.0736,
+      "step": 2800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015231957772877323,
+      "loss": 0.0745,
+      "step": 2801
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001522855717290551,
+      "loss": 0.0768,
+      "step": 2802
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015225155740612834,
+      "loss": 0.0756,
+      "step": 2803
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001522175347654076,
+      "loss": 0.0775,
+      "step": 2804
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015218350381230895,
+      "loss": 0.0766,
+      "step": 2805
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015214946455224956,
+      "loss": 0.0741,
+      "step": 2806
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001521154169906481,
+      "loss": 0.08,
+      "step": 2807
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015208136113292457,
+      "loss": 0.0735,
+      "step": 2808
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015204729698450016,
+      "loss": 0.0774,
+      "step": 2809
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015201322455079755,
+      "loss": 0.0747,
+      "step": 2810
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015197914383724052,
+      "loss": 0.0744,
+      "step": 2811
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015194505484925443,
+      "loss": 0.0786,
+      "step": 2812
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001519109575922657,
+      "loss": 0.0727,
+      "step": 2813
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015187685207170227,
+      "loss": 0.0739,
+      "step": 2814
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015184273829299327,
+      "loss": 0.0757,
+      "step": 2815
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015180861626156914,
+      "loss": 0.0773,
+      "step": 2816
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001517744859828618,
+      "loss": 0.0764,
+      "step": 2817
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001517403474623043,
+      "loss": 0.0728,
+      "step": 2818
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015170620070533103,
+      "loss": 0.075,
+      "step": 2819
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015167204571737782,
+      "loss": 0.0767,
+      "step": 2820
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001516378825038816,
+      "loss": 0.0737,
+      "step": 2821
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015160371107028083,
+      "loss": 0.0687,
+      "step": 2822
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001515695314220151,
+      "loss": 0.0744,
+      "step": 2823
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015153534356452539,
+      "loss": 0.0713,
+      "step": 2824
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015150114750325403,
+      "loss": 0.0748,
+      "step": 2825
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015146694324364454,
+      "loss": 0.0794,
+      "step": 2826
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015143273079114188,
+      "loss": 0.0753,
+      "step": 2827
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015139851015119222,
+      "loss": 0.0723,
+      "step": 2828
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015136428132924302,
+      "loss": 0.0741,
+      "step": 2829
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015133004433074313,
+      "loss": 0.0751,
+      "step": 2830
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001512957991611426,
+      "loss": 0.0764,
+      "step": 2831
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015126154582589287,
+      "loss": 0.0741,
+      "step": 2832
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015122728433044667,
+      "loss": 0.0739,
+      "step": 2833
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001511930146802579,
+      "loss": 0.0749,
+      "step": 2834
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015115873688078196,
+      "loss": 0.0763,
+      "step": 2835
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015112445093747541,
+      "loss": 0.0739,
+      "step": 2836
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015109015685579614,
+      "loss": 0.0759,
+      "step": 2837
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015105585464120333,
+      "loss": 0.0755,
+      "step": 2838
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001510215442991575,
+      "loss": 0.072,
+      "step": 2839
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015098722583512038,
+      "loss": 0.0768,
+      "step": 2840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015095289925455506,
+      "loss": 0.0759,
+      "step": 2841
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001509185645629259,
+      "loss": 0.0737,
+      "step": 2842
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015088422176569858,
+      "loss": 0.0774,
+      "step": 2843
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015084987086834003,
+      "loss": 0.076,
+      "step": 2844
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015081551187631843,
+      "loss": 0.0767,
+      "step": 2845
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001507811447951034,
+      "loss": 0.0739,
+      "step": 2846
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015074676963016564,
+      "loss": 0.0762,
+      "step": 2847
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015071238638697732,
+      "loss": 0.0729,
+      "step": 2848
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001506779950710118,
+      "loss": 0.0748,
+      "step": 2849
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015064359568774376,
+      "loss": 0.0738,
+      "step": 2850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015060918824264915,
+      "loss": 0.0761,
+      "step": 2851
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015057477274120514,
+      "loss": 0.0747,
+      "step": 2852
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015054034918889037,
+      "loss": 0.0759,
+      "step": 2853
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015050591759118454,
+      "loss": 0.0775,
+      "step": 2854
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015047147795356876,
+      "loss": 0.0761,
+      "step": 2855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001504370302815254,
+      "loss": 0.0759,
+      "step": 2856
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015040257458053807,
+      "loss": 0.0746,
+      "step": 2857
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015036811085609176,
+      "loss": 0.0718,
+      "step": 2858
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015033363911367252,
+      "loss": 0.0745,
+      "step": 2859
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015029915935876797,
+      "loss": 0.0768,
+      "step": 2860
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001502646715968668,
+      "loss": 0.0789,
+      "step": 2861
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015023017583345896,
+      "loss": 0.0758,
+      "step": 2862
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015019567207403588,
+      "loss": 0.0714,
+      "step": 2863
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015016116032409,
+      "loss": 0.0736,
+      "step": 2864
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015012664058911522,
+      "loss": 0.0744,
+      "step": 2865
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015009211287460666,
+      "loss": 0.0733,
+      "step": 2866
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015005757718606065,
+      "loss": 0.0737,
+      "step": 2867
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015002303352897492,
+      "loss": 0.0738,
+      "step": 2868
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00014998848190884832,
+      "loss": 0.0748,
+      "step": 2869
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00014995392233118103,
+      "loss": 0.0749,
+      "step": 2870
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00014991935480147456,
+      "loss": 0.076,
+      "step": 2871
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00014988477932523158,
+      "loss": 0.0731,
+      "step": 2872
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001498501959079561,
+      "loss": 0.074,
+      "step": 2873
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00014981560455515336,
+      "loss": 0.072,
+      "step": 2874
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00014978100527232984,
+      "loss": 0.0764,
+      "step": 2875
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014974639806499335,
+      "loss": 0.0714,
+      "step": 2876
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014971178293865292,
+      "loss": 0.0746,
+      "step": 2877
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014967715989881882,
+      "loss": 0.0765,
+      "step": 2878
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014964252895100264,
+      "loss": 0.077,
+      "step": 2879
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014960789010071715,
+      "loss": 0.0777,
+      "step": 2880
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001495732433534765,
+      "loss": 0.0709,
+      "step": 2881
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014953858871479594,
+      "loss": 0.0743,
+      "step": 2882
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014950392619019205,
+      "loss": 0.0748,
+      "step": 2883
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014946925578518272,
+      "loss": 0.0746,
+      "step": 2884
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014943457750528705,
+      "loss": 0.0726,
+      "step": 2885
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014939989135602532,
+      "loss": 0.0734,
+      "step": 2886
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001493651973429192,
+      "loss": 0.0779,
+      "step": 2887
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00014933049547149153,
+      "loss": 0.0749,
+      "step": 2888
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014929578574726638,
+      "loss": 0.0745,
+      "step": 2889
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014926106817576913,
+      "loss": 0.074,
+      "step": 2890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014922634276252637,
+      "loss": 0.0761,
+      "step": 2891
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014919160951306596,
+      "loss": 0.0715,
+      "step": 2892
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.000149156868432917,
+      "loss": 0.0763,
+      "step": 2893
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014912211952760983,
+      "loss": 0.076,
+      "step": 2894
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014908736280267604,
+      "loss": 0.077,
+      "step": 2895
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014905259826364847,
+      "loss": 0.0743,
+      "step": 2896
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001490178259160612,
+      "loss": 0.0771,
+      "step": 2897
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014898304576544956,
+      "loss": 0.0754,
+      "step": 2898
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014894825781735008,
+      "loss": 0.0738,
+      "step": 2899
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014891346207730058,
+      "loss": 0.0752,
+      "step": 2900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014887865855084016,
+      "loss": 0.0741,
+      "step": 2901
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00014884384724350906,
+      "loss": 0.0713,
+      "step": 2902
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014880902816084878,
+      "loss": 0.0758,
+      "step": 2903
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014877420130840214,
+      "loss": 0.0728,
+      "step": 2904
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014873936669171307,
+      "loss": 0.076,
+      "step": 2905
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001487045243163269,
+      "loss": 0.0762,
+      "step": 2906
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014866967418778997,
+      "loss": 0.0781,
+      "step": 2907
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001486348163116501,
+      "loss": 0.069,
+      "step": 2908
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001485999506934562,
+      "loss": 0.0736,
+      "step": 2909
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014856507733875836,
+      "loss": 0.0743,
+      "step": 2910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014853019625310813,
+      "loss": 0.0775,
+      "step": 2911
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014849530744205798,
+      "loss": 0.0742,
+      "step": 2912
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014846041091116192,
+      "loss": 0.0703,
+      "step": 2913
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00014842550666597495,
+      "loss": 0.0777,
+      "step": 2914
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0001483905947120534,
+      "loss": 0.0712,
+      "step": 2915
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00014835567505495486,
+      "loss": 0.0754,
+      "step": 2916
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.000148320747700238,
+      "loss": 0.072,
+      "step": 2917
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00014828581265346295,
+      "loss": 0.0755,
+      "step": 2918
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00014825086992019087,
+      "loss": 0.0796,
+      "step": 2919
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00014821591950598415,
+      "loss": 0.0721,
+      "step": 2920
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001481809614164066,
+      "loss": 0.0735,
+      "step": 2921
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00014814599565702295,
+      "loss": 0.0737,
+      "step": 2922
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001481110222333994,
+      "loss": 0.076,
+      "step": 2923
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001480760411511033,
+      "loss": 0.0749,
+      "step": 2924
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001480410524157031,
+      "loss": 0.0754,
+      "step": 2925
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001480060560327687,
+      "loss": 0.0761,
+      "step": 2926
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00014797105200787102,
+      "loss": 0.0779,
+      "step": 2927
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00014793604034658224,
+      "loss": 0.0743,
+      "step": 2928
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001479010210544758,
+      "loss": 0.0757,
+      "step": 2929
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014786599413712633,
+      "loss": 0.0772,
+      "step": 2930
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014783095960010974,
+      "loss": 0.0738,
+      "step": 2931
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.000147795917449003,
+      "loss": 0.0751,
+      "step": 2932
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014776086768938438,
+      "loss": 0.0774,
+      "step": 2933
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014772581032683344,
+      "loss": 0.0726,
+      "step": 2934
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014769074536693082,
+      "loss": 0.0732,
+      "step": 2935
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014765567281525843,
+      "loss": 0.0748,
+      "step": 2936
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014762059267739937,
+      "loss": 0.073,
+      "step": 2937
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.000147585504958938,
+      "loss": 0.077,
+      "step": 2938
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014755040966545982,
+      "loss": 0.0751,
+      "step": 2939
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014751530680255154,
+      "loss": 0.0781,
+      "step": 2940
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014748019637580114,
+      "loss": 0.074,
+      "step": 2941
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00014744507839079771,
+      "loss": 0.0748,
+      "step": 2942
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014740995285313164,
+      "loss": 0.0756,
+      "step": 2943
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014737481976839447,
+      "loss": 0.0736,
+      "step": 2944
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014733967914217893,
+      "loss": 0.0746,
+      "step": 2945
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014730453098007895,
+      "loss": 0.0723,
+      "step": 2946
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001472693752876897,
+      "loss": 0.0758,
+      "step": 2947
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014723421207060754,
+      "loss": 0.0714,
+      "step": 2948
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014719904133443,
+      "loss": 0.0768,
+      "step": 2949
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014716386308475583,
+      "loss": 0.073,
+      "step": 2950
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014712867732718496,
+      "loss": 0.0752,
+      "step": 2951
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001470934840673185,
+      "loss": 0.0781,
+      "step": 2952
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014705828331075882,
+      "loss": 0.0779,
+      "step": 2953
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.0001470230750631094,
+      "loss": 0.0773,
+      "step": 2954
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.000146987859329975,
+      "loss": 0.0744,
+      "step": 2955
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00014695263611696146,
+      "loss": 0.075,
+      "step": 2956
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014691740542967593,
+      "loss": 0.075,
+      "step": 2957
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014688216727372664,
+      "loss": 0.0764,
+      "step": 2958
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014684692165472315,
+      "loss": 0.0751,
+      "step": 2959
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014681166857827602,
+      "loss": 0.0764,
+      "step": 2960
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014677640804999716,
+      "loss": 0.0773,
+      "step": 2961
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001467411400754996,
+      "loss": 0.0748,
+      "step": 2962
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014670586466039752,
+      "loss": 0.0776,
+      "step": 2963
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014667058181030642,
+      "loss": 0.0747,
+      "step": 2964
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014663529153084276,
+      "loss": 0.0728,
+      "step": 2965
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001465999938276244,
+      "loss": 0.0739,
+      "step": 2966
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014656468870627028,
+      "loss": 0.0739,
+      "step": 2967
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00014652937617240047,
+      "loss": 0.0769,
+      "step": 2968
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0001464940562316364,
+      "loss": 0.0762,
+      "step": 2969
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014645872888960045,
+      "loss": 0.0757,
+      "step": 2970
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014642339415191634,
+      "loss": 0.0758,
+      "step": 2971
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014638805202420895,
+      "loss": 0.0784,
+      "step": 2972
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014635270251210422,
+      "loss": 0.076,
+      "step": 2973
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014631734562122945,
+      "loss": 0.0754,
+      "step": 2974
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014628198135721294,
+      "loss": 0.0772,
+      "step": 2975
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014624660972568427,
+      "loss": 0.073,
+      "step": 2976
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014621123073227415,
+      "loss": 0.0782,
+      "step": 2977
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014617584438261444,
+      "loss": 0.0758,
+      "step": 2978
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001461404506823383,
+      "loss": 0.0768,
+      "step": 2979
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014610504963707987,
+      "loss": 0.074,
+      "step": 2980
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001460696412524746,
+      "loss": 0.0757,
+      "step": 2981
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014603422553415905,
+      "loss": 0.0763,
+      "step": 2982
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00014599880248777094,
+      "loss": 0.075,
+      "step": 2983
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014596337211894922,
+      "loss": 0.0748,
+      "step": 2984
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001459279344333339,
+      "loss": 0.0754,
+      "step": 2985
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014589248943656629,
+      "loss": 0.0755,
+      "step": 2986
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014585703713428872,
+      "loss": 0.0755,
+      "step": 2987
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001458215775321448,
+      "loss": 0.0724,
+      "step": 2988
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014578611063577925,
+      "loss": 0.0742,
+      "step": 2989
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001457506364508379,
+      "loss": 0.0747,
+      "step": 2990
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014571515498296788,
+      "loss": 0.0726,
+      "step": 2991
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014567966623781734,
+      "loss": 0.0738,
+      "step": 2992
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001456441702210357,
+      "loss": 0.0774,
+      "step": 2993
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014560866693827339,
+      "loss": 0.0757,
+      "step": 2994
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014557315639518217,
+      "loss": 0.0734,
+      "step": 2995
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00014553763859741483,
+      "loss": 0.0756,
+      "step": 2996
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014550211355062538,
+      "loss": 0.0757,
+      "step": 2997
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014546658126046898,
+      "loss": 0.0761,
+      "step": 2998
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014543104173260187,
+      "loss": 0.0716,
+      "step": 2999
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014539549497268153,
+      "loss": 0.0757,
+      "step": 3000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014535994098636656,
+      "loss": 0.0769,
+      "step": 3001
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014532437977931673,
+      "loss": 0.078,
+      "step": 3002
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001452888113571929,
+      "loss": 0.0739,
+      "step": 3003
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014525323572565712,
+      "loss": 0.0759,
+      "step": 3004
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014521765289037263,
+      "loss": 0.0756,
+      "step": 3005
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014518206285700372,
+      "loss": 0.0709,
+      "step": 3006
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014514646563121593,
+      "loss": 0.0757,
+      "step": 3007
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014511086121867583,
+      "loss": 0.0731,
+      "step": 3008
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00014507524962505128,
+      "loss": 0.0722,
+      "step": 3009
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001450396308560111,
+      "loss": 0.0802,
+      "step": 3010
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001450040049172254,
+      "loss": 0.0759,
+      "step": 3011
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014496837181436544,
+      "loss": 0.0705,
+      "step": 3012
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014493273155310348,
+      "loss": 0.073,
+      "step": 3013
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014489708413911303,
+      "loss": 0.0735,
+      "step": 3014
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001448614295780687,
+      "loss": 0.077,
+      "step": 3015
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014482576787564627,
+      "loss": 0.0755,
+      "step": 3016
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014479009903752268,
+      "loss": 0.0746,
+      "step": 3017
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014475442306937587,
+      "loss": 0.0737,
+      "step": 3018
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014471873997688506,
+      "loss": 0.0762,
+      "step": 3019
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014468304976573054,
+      "loss": 0.0735,
+      "step": 3020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00014464735244159376,
+      "loss": 0.0755,
+      "step": 3021
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001446116480101573,
+      "loss": 0.0768,
+      "step": 3022
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001445759364771048,
+      "loss": 0.075,
+      "step": 3023
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00014454021784812113,
+      "loss": 0.0715,
+      "step": 3024
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00014450449212889227,
+      "loss": 0.0773,
+      "step": 3025
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001444687593251052,
+      "loss": 0.0773,
+      "step": 3026
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00014443301944244832,
+      "loss": 0.0755,
+      "step": 3027
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001443972724866108,
+      "loss": 0.0737,
+      "step": 3028
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001443615184632832,
+      "loss": 0.0756,
+      "step": 3029
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00014432575737815707,
+      "loss": 0.0773,
+      "step": 3030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00014428998923692515,
+      "loss": 0.0762,
+      "step": 3031
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00014425421404528133,
+      "loss": 0.0755,
+      "step": 3032
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00014421843180892046,
+      "loss": 0.0763,
+      "step": 3033
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001441826425335387,
+      "loss": 0.0763,
+      "step": 3034
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001441468462248332,
+      "loss": 0.0735,
+      "step": 3035
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00014411104288850236,
+      "loss": 0.0749,
+      "step": 3036
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001440752325302456,
+      "loss": 0.0761,
+      "step": 3037
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00014403941515576344,
+      "loss": 0.0756,
+      "step": 3038
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001440035907707576,
+      "loss": 0.0753,
+      "step": 3039
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00014396775938093083,
+      "loss": 0.0769,
+      "step": 3040
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001439319209919871,
+      "loss": 0.0732,
+      "step": 3041
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001438960756096314,
+      "loss": 0.0751,
+      "step": 3042
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00014386022323956984,
+      "loss": 0.0806,
+      "step": 3043
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001438243638875097,
+      "loss": 0.0741,
+      "step": 3044
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00014378849755915932,
+      "loss": 0.0764,
+      "step": 3045
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001437526242602282,
+      "loss": 0.0728,
+      "step": 3046
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00014371674399642692,
+      "loss": 0.0759,
+      "step": 3047
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00014368085677346712,
+      "loss": 0.0759,
+      "step": 3048
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00014364496259706164,
+      "loss": 0.0737,
+      "step": 3049
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00014360906147292438,
+      "loss": 0.0715,
+      "step": 3050
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00014357315340677036,
+      "loss": 0.0778,
+      "step": 3051
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001435372384043157,
+      "loss": 0.0719,
+      "step": 3052
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00014350131647127755,
+      "loss": 0.0747,
+      "step": 3053
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00014346538761337427,
+      "loss": 0.0733,
+      "step": 3054
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001434294518363253,
+      "loss": 0.0767,
+      "step": 3055
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001433935091458512,
+      "loss": 0.0742,
+      "step": 3056
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001433575595476735,
+      "loss": 0.0751,
+      "step": 3057
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00014332160304751502,
+      "loss": 0.0759,
+      "step": 3058
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00014328563965109952,
+      "loss": 0.0731,
+      "step": 3059
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000143249669364152,
+      "loss": 0.0785,
+      "step": 3060
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001432136921923984,
+      "loss": 0.0761,
+      "step": 3061
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00014317770814156587,
+      "loss": 0.0758,
+      "step": 3062
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001431417172173826,
+      "loss": 0.0725,
+      "step": 3063
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001431057194255779,
+      "loss": 0.0794,
+      "step": 3064
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014306971477188223,
+      "loss": 0.0739,
+      "step": 3065
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014303370326202697,
+      "loss": 0.0733,
+      "step": 3066
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001429976849017448,
+      "loss": 0.077,
+      "step": 3067
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014296165969676934,
+      "loss": 0.0741,
+      "step": 3068
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014292562765283532,
+      "loss": 0.0755,
+      "step": 3069
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001428895887756787,
+      "loss": 0.0731,
+      "step": 3070
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001428535430710363,
+      "loss": 0.0739,
+      "step": 3071
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014281749054464624,
+      "loss": 0.076,
+      "step": 3072
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014278143120224757,
+      "loss": 0.0763,
+      "step": 3073
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014274536504958047,
+      "loss": 0.0769,
+      "step": 3074
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014270929209238633,
+      "loss": 0.0744,
+      "step": 3075
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014267321233640736,
+      "loss": 0.0805,
+      "step": 3076
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00014263712578738714,
+      "loss": 0.0766,
+      "step": 3077
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001426010324510701,
+      "loss": 0.0723,
+      "step": 3078
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001425649323332019,
+      "loss": 0.0755,
+      "step": 3079
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014252882543952923,
+      "loss": 0.073,
+      "step": 3080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014249271177579984,
+      "loss": 0.0773,
+      "step": 3081
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014245659134776254,
+      "loss": 0.074,
+      "step": 3082
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014242046416116733,
+      "loss": 0.074,
+      "step": 3083
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014238433022176512,
+      "loss": 0.0738,
+      "step": 3084
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014234818953530803,
+      "loss": 0.0723,
+      "step": 3085
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001423120421075492,
+      "loss": 0.0714,
+      "step": 3086
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014227588794424285,
+      "loss": 0.0739,
+      "step": 3087
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014223972705114426,
+      "loss": 0.0767,
+      "step": 3088
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014220355943400978,
+      "loss": 0.0752,
+      "step": 3089
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00014216738509859687,
+      "loss": 0.0755,
+      "step": 3090
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.000142131204050664,
+      "loss": 0.0754,
+      "step": 3091
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001420950162959708,
+      "loss": 0.0748,
+      "step": 3092
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00014205882184027783,
+      "loss": 0.0767,
+      "step": 3093
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00014202262068934685,
+      "loss": 0.0753,
+      "step": 3094
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001419864128489406,
+      "loss": 0.0769,
+      "step": 3095
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001419501983248229,
+      "loss": 0.0728,
+      "step": 3096
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001419139771227587,
+      "loss": 0.0749,
+      "step": 3097
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00014187774924851392,
+      "loss": 0.0714,
+      "step": 3098
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00014184151470785565,
+      "loss": 0.0753,
+      "step": 3099
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001418052735065519,
+      "loss": 0.0736,
+      "step": 3100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00014176902565037183,
+      "loss": 0.0732,
+      "step": 3101
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00014173277114508566,
+      "loss": 0.0767,
+      "step": 3102
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00014169650999646465,
+      "loss": 0.0734,
+      "step": 3103
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00014166024221028112,
+      "loss": 0.0756,
+      "step": 3104
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00014162396779230843,
+      "loss": 0.0772,
+      "step": 3105
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001415876867483211,
+      "loss": 0.0724,
+      "step": 3106
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00014155139908409448,
+      "loss": 0.0738,
+      "step": 3107
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001415151048054052,
+      "loss": 0.0753,
+      "step": 3108
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00014147880391803086,
+      "loss": 0.0729,
+      "step": 3109
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00014144249642775004,
+      "loss": 0.0737,
+      "step": 3110
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00014140618234034254,
+      "loss": 0.073,
+      "step": 3111
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.000141369861661589,
+      "loss": 0.0739,
+      "step": 3112
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00014133353439727128,
+      "loss": 0.0756,
+      "step": 3113
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00014129720055317224,
+      "loss": 0.0732,
+      "step": 3114
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001412608601350757,
+      "loss": 0.0753,
+      "step": 3115
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001412245131487667,
+      "loss": 0.0768,
+      "step": 3116
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00014118815960003113,
+      "loss": 0.0788,
+      "step": 3117
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0001411517994946561,
+      "loss": 0.0747,
+      "step": 3118
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001411154328384296,
+      "loss": 0.0734,
+      "step": 3119
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001410790596371408,
+      "loss": 0.0769,
+      "step": 3120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014104267989657992,
+      "loss": 0.0756,
+      "step": 3121
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014100629362253798,
+      "loss": 0.0784,
+      "step": 3122
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014096990082080742,
+      "loss": 0.0744,
+      "step": 3123
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001409335014971814,
+      "loss": 0.0749,
+      "step": 3124
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014089709565745422,
+      "loss": 0.078,
+      "step": 3125
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014086068330742134,
+      "loss": 0.0757,
+      "step": 3126
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014082426445287903,
+      "loss": 0.0735,
+      "step": 3127
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014078783909962482,
+      "loss": 0.0748,
+      "step": 3128
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014075140725345713,
+      "loss": 0.0702,
+      "step": 3129
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014071496892017544,
+      "loss": 0.0773,
+      "step": 3130
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00014067852410558026,
+      "loss": 0.076,
+      "step": 3131
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001406420728154732,
+      "loss": 0.074,
+      "step": 3132
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014060561505565682,
+      "loss": 0.0742,
+      "step": 3133
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014056915083193473,
+      "loss": 0.0768,
+      "step": 3134
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001405326801501116,
+      "loss": 0.076,
+      "step": 3135
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001404962030159931,
+      "loss": 0.0717,
+      "step": 3136
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001404597194353859,
+      "loss": 0.0754,
+      "step": 3137
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014042322941409778,
+      "loss": 0.0746,
+      "step": 3138
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014038673295793747,
+      "loss": 0.0767,
+      "step": 3139
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014035023007271478,
+      "loss": 0.0734,
+      "step": 3140
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014031372076424046,
+      "loss": 0.0736,
+      "step": 3141
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014027720503832636,
+      "loss": 0.0737,
+      "step": 3142
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014024068290078532,
+      "loss": 0.0737,
+      "step": 3143
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014020415435743122,
+      "loss": 0.0757,
+      "step": 3144
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00014016761941407894,
+      "loss": 0.0744,
+      "step": 3145
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001401310780765444,
+      "loss": 0.0768,
+      "step": 3146
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001400945303506445,
+      "loss": 0.0698,
+      "step": 3147
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00014005797624219718,
+      "loss": 0.0743,
+      "step": 3148
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00014002141575702141,
+      "loss": 0.0766,
+      "step": 3149
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00013998484890093718,
+      "loss": 0.0723,
+      "step": 3150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00013994827567976543,
+      "loss": 0.0741,
+      "step": 3151
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001399116960993282,
+      "loss": 0.0763,
+      "step": 3152
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00013987511016544848,
+      "loss": 0.0776,
+      "step": 3153
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00013983851788395028,
+      "loss": 0.0765,
+      "step": 3154
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001398019192606587,
+      "loss": 0.0769,
+      "step": 3155
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00013976531430139969,
+      "loss": 0.0747,
+      "step": 3156
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00013972870301200033,
+      "loss": 0.0781,
+      "step": 3157
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00013969208539828872,
+      "loss": 0.0751,
+      "step": 3158
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001396554614660939,
+      "loss": 0.0759,
+      "step": 3159
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013961883122124596,
+      "loss": 0.0737,
+      "step": 3160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013958219466957594,
+      "loss": 0.0747,
+      "step": 3161
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013954555181691593,
+      "loss": 0.076,
+      "step": 3162
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013950890266909902,
+      "loss": 0.0742,
+      "step": 3163
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013947224723195932,
+      "loss": 0.0749,
+      "step": 3164
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013943558551133186,
+      "loss": 0.0742,
+      "step": 3165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013939891751305278,
+      "loss": 0.0767,
+      "step": 3166
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013936224324295917,
+      "loss": 0.0762,
+      "step": 3167
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013932556270688904,
+      "loss": 0.0727,
+      "step": 3168
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013928887591068158,
+      "loss": 0.0749,
+      "step": 3169
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001392521828601768,
+      "loss": 0.0777,
+      "step": 3170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013921548356121577,
+      "loss": 0.075,
+      "step": 3171
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00013917877801964058,
+      "loss": 0.0751,
+      "step": 3172
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001391420662412943,
+      "loss": 0.0756,
+      "step": 3173
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013910534823202098,
+      "loss": 0.0737,
+      "step": 3174
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013906862399766564,
+      "loss": 0.0749,
+      "step": 3175
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001390318935440744,
+      "loss": 0.0771,
+      "step": 3176
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013899515687709415,
+      "loss": 0.0755,
+      "step": 3177
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000138958414002573,
+      "loss": 0.0763,
+      "step": 3178
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013892166492636,
+      "loss": 0.0749,
+      "step": 3179
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013888490965430504,
+      "loss": 0.0736,
+      "step": 3180
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013884814819225917,
+      "loss": 0.0737,
+      "step": 3181
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013881138054607433,
+      "loss": 0.0751,
+      "step": 3182
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013877460672160344,
+      "loss": 0.0743,
+      "step": 3183
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001387378267247005,
+      "loss": 0.0739,
+      "step": 3184
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013870104056122034,
+      "loss": 0.0754,
+      "step": 3185
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00013866424823701893,
+      "loss": 0.0738,
+      "step": 3186
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00013862744975795316,
+      "loss": 0.074,
+      "step": 3187
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001385906451298808,
+      "loss": 0.0749,
+      "step": 3188
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00013855383435866077,
+      "loss": 0.0735,
+      "step": 3189
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001385170174501528,
+      "loss": 0.0773,
+      "step": 3190
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00013848019441021776,
+      "loss": 0.0731,
+      "step": 3191
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00013844336524471737,
+      "loss": 0.0742,
+      "step": 3192
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001384065299595144,
+      "loss": 0.0732,
+      "step": 3193
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00013836968856047259,
+      "loss": 0.076,
+      "step": 3194
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00013833284105345655,
+      "loss": 0.0747,
+      "step": 3195
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.000138295987444332,
+      "loss": 0.075,
+      "step": 3196
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00013825912773896557,
+      "loss": 0.0739,
+      "step": 3197
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00013822226194322486,
+      "loss": 0.0748,
+      "step": 3198
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00013818539006297842,
+      "loss": 0.0783,
+      "step": 3199
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001381485121040958,
+      "loss": 0.0772,
+      "step": 3200
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00013811162807244752,
+      "loss": 0.0757,
+      "step": 3201
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001380747379739051,
+      "loss": 0.0755,
+      "step": 3202
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001380378418143409,
+      "loss": 0.0754,
+      "step": 3203
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00013800093959962837,
+      "loss": 0.0757,
+      "step": 3204
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00013796403133564187,
+      "loss": 0.0732,
+      "step": 3205
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00013792711702825672,
+      "loss": 0.0757,
+      "step": 3206
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001378901966833493,
+      "loss": 0.0778,
+      "step": 3207
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00013785327030679673,
+      "loss": 0.0749,
+      "step": 3208
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00013781633790447732,
+      "loss": 0.0745,
+      "step": 3209
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00013777939948227024,
+      "loss": 0.0769,
+      "step": 3210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00013774245504605557,
+      "loss": 0.0718,
+      "step": 3211
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001377055046017145,
+      "loss": 0.0737,
+      "step": 3212
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00013766854815512898,
+      "loss": 0.0722,
+      "step": 3213
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013763158571218205,
+      "loss": 0.074,
+      "step": 3214
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013759461727875767,
+      "loss": 0.0724,
+      "step": 3215
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013755764286074075,
+      "loss": 0.0785,
+      "step": 3216
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001375206624640172,
+      "loss": 0.0743,
+      "step": 3217
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013748367609447374,
+      "loss": 0.0756,
+      "step": 3218
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013744668375799822,
+      "loss": 0.0735,
+      "step": 3219
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013740968546047935,
+      "loss": 0.0765,
+      "step": 3220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013737268120780671,
+      "loss": 0.0756,
+      "step": 3221
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013733567100587102,
+      "loss": 0.0752,
+      "step": 3222
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013729865486056378,
+      "loss": 0.0759,
+      "step": 3223
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013726163277777754,
+      "loss": 0.0774,
+      "step": 3224
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001372246047634057,
+      "loss": 0.0752,
+      "step": 3225
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00013718757082334268,
+      "loss": 0.075,
+      "step": 3226
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013715053096348387,
+      "loss": 0.0764,
+      "step": 3227
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013711348518972546,
+      "loss": 0.0725,
+      "step": 3228
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013707643350796474,
+      "loss": 0.0746,
+      "step": 3229
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013703937592409986,
+      "loss": 0.074,
+      "step": 3230
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013700231244402988,
+      "loss": 0.0734,
+      "step": 3231
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013696524307365494,
+      "loss": 0.0767,
+      "step": 3232
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001369281678188759,
+      "loss": 0.0764,
+      "step": 3233
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013689108668559475,
+      "loss": 0.075,
+      "step": 3234
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013685399967971435,
+      "loss": 0.0749,
+      "step": 3235
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013681690680713847,
+      "loss": 0.0761,
+      "step": 3236
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001367798080737718,
+      "loss": 0.0759,
+      "step": 3237
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013674270348552,
+      "loss": 0.0727,
+      "step": 3238
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013670559304828972,
+      "loss": 0.0732,
+      "step": 3239
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00013666847676798842,
+      "loss": 0.0742,
+      "step": 3240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00013663135465052456,
+      "loss": 0.0728,
+      "step": 3241
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00013659422670180753,
+      "loss": 0.0723,
+      "step": 3242
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001365570929277476,
+      "loss": 0.0764,
+      "step": 3243
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.000136519953334256,
+      "loss": 0.0747,
+      "step": 3244
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00013648280792724496,
+      "loss": 0.0777,
+      "step": 3245
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00013644565671262751,
+      "loss": 0.0736,
+      "step": 3246
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00013640849969631763,
+      "loss": 0.0741,
+      "step": 3247
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00013637133688423032,
+      "loss": 0.0761,
+      "step": 3248
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00013633416828228137,
+      "loss": 0.073,
+      "step": 3249
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001362969938963876,
+      "loss": 0.0723,
+      "step": 3250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001362598137324667,
+      "loss": 0.0754,
+      "step": 3251
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001362226277964373,
+      "loss": 0.0733,
+      "step": 3252
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001361854360942189,
+      "loss": 0.0744,
+      "step": 3253
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013614823863173194,
+      "loss": 0.0743,
+      "step": 3254
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013611103541489786,
+      "loss": 0.0767,
+      "step": 3255
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013607382644963888,
+      "loss": 0.0772,
+      "step": 3256
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013603661174187826,
+      "loss": 0.0741,
+      "step": 3257
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013599939129754006,
+      "loss": 0.0714,
+      "step": 3258
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013596216512254934,
+      "loss": 0.0759,
+      "step": 3259
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013592493322283207,
+      "loss": 0.0753,
+      "step": 3260
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.000135887695604315,
+      "loss": 0.0732,
+      "step": 3261
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013585045227292597,
+      "loss": 0.0722,
+      "step": 3262
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001358132032345937,
+      "loss": 0.0708,
+      "step": 3263
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013577594849524763,
+      "loss": 0.0723,
+      "step": 3264
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013573868806081838,
+      "loss": 0.0757,
+      "step": 3265
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0001357014219372373,
+      "loss": 0.0756,
+      "step": 3266
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00013566415013043667,
+      "loss": 0.0743,
+      "step": 3267
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013562687264634973,
+      "loss": 0.0784,
+      "step": 3268
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013558958949091055,
+      "loss": 0.079,
+      "step": 3269
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013555230067005417,
+      "loss": 0.0745,
+      "step": 3270
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001355150061897165,
+      "loss": 0.0781,
+      "step": 3271
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013547770605583433,
+      "loss": 0.0762,
+      "step": 3272
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013544040027434543,
+      "loss": 0.0786,
+      "step": 3273
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013540308885118832,
+      "loss": 0.0739,
+      "step": 3274
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001353657717923026,
+      "loss": 0.0744,
+      "step": 3275
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013532844910362865,
+      "loss": 0.074,
+      "step": 3276
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013529112079110778,
+      "loss": 0.0735,
+      "step": 3277
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013525378686068216,
+      "loss": 0.0737,
+      "step": 3278
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013521644731829491,
+      "loss": 0.0764,
+      "step": 3279
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013517910216989006,
+      "loss": 0.0749,
+      "step": 3280
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00013514175142141242,
+      "loss": 0.0744,
+      "step": 3281
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00013510439507880776,
+      "loss": 0.0753,
+      "step": 3282
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001350670331480228,
+      "loss": 0.0735,
+      "step": 3283
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00013502966563500504,
+      "loss": 0.0789,
+      "step": 3284
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00013499229254570298,
+      "loss": 0.0737,
+      "step": 3285
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00013495491388606587,
+      "loss": 0.0761,
+      "step": 3286
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00013491752966204396,
+      "loss": 0.0761,
+      "step": 3287
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001348801398795884,
+      "loss": 0.0766,
+      "step": 3288
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001348427445446511,
+      "loss": 0.077,
+      "step": 3289
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00013480534366318495,
+      "loss": 0.0752,
+      "step": 3290
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001347679372411437,
+      "loss": 0.0747,
+      "step": 3291
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00013473052528448201,
+      "loss": 0.0786,
+      "step": 3292
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00013469310779915542,
+      "loss": 0.0736,
+      "step": 3293
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00013465568479112026,
+      "loss": 0.0746,
+      "step": 3294
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013461825626633384,
+      "loss": 0.0775,
+      "step": 3295
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013458082223075424,
+      "loss": 0.0719,
+      "step": 3296
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013454338269034064,
+      "loss": 0.0758,
+      "step": 3297
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013450593765105282,
+      "loss": 0.078,
+      "step": 3298
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001344684871188516,
+      "loss": 0.0791,
+      "step": 3299
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013443103109969865,
+      "loss": 0.0765,
+      "step": 3300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001343935695995565,
+      "loss": 0.073,
+      "step": 3301
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013435610262438848,
+      "loss": 0.0772,
+      "step": 3302
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013431863018015898,
+      "loss": 0.0781,
+      "step": 3303
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013428115227283308,
+      "loss": 0.0724,
+      "step": 3304
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013424366890837678,
+      "loss": 0.0779,
+      "step": 3305
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.000134206180092757,
+      "loss": 0.0778,
+      "step": 3306
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00013416868583194144,
+      "loss": 0.0751,
+      "step": 3307
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0001341311861318988,
+      "loss": 0.0768,
+      "step": 3308
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013409368099859847,
+      "loss": 0.0733,
+      "step": 3309
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013405617043801086,
+      "loss": 0.0739,
+      "step": 3310
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013401865445610716,
+      "loss": 0.0738,
+      "step": 3311
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013398113305885945,
+      "loss": 0.0739,
+      "step": 3312
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013394360625224068,
+      "loss": 0.0753,
+      "step": 3313
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001339060740422246,
+      "loss": 0.073,
+      "step": 3314
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013386853643478593,
+      "loss": 0.0749,
+      "step": 3315
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013383099343590012,
+      "loss": 0.0731,
+      "step": 3316
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013379344505154358,
+      "loss": 0.0743,
+      "step": 3317
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001337558912876936,
+      "loss": 0.075,
+      "step": 3318
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001337183321503282,
+      "loss": 0.0783,
+      "step": 3319
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013368076764542632,
+      "loss": 0.0758,
+      "step": 3320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00013364319777896779,
+      "loss": 0.0749,
+      "step": 3321
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013360562255693322,
+      "loss": 0.0782,
+      "step": 3322
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013356804198530418,
+      "loss": 0.0752,
+      "step": 3323
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013353045607006299,
+      "loss": 0.0773,
+      "step": 3324
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013349286481719282,
+      "loss": 0.0762,
+      "step": 3325
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001334552682326778,
+      "loss": 0.0822,
+      "step": 3326
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001334176663225028,
+      "loss": 0.0754,
+      "step": 3327
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013338005909265362,
+      "loss": 0.0727,
+      "step": 3328
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013334244654911677,
+      "loss": 0.0732,
+      "step": 3329
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013330482869787974,
+      "loss": 0.0738,
+      "step": 3330
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013326720554493083,
+      "loss": 0.0732,
+      "step": 3331
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013322957709625916,
+      "loss": 0.0748,
+      "step": 3332
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013319194335785476,
+      "loss": 0.0726,
+      "step": 3333
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013315430433570835,
+      "loss": 0.0732,
+      "step": 3334
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00013311666003581167,
+      "loss": 0.0774,
+      "step": 3335
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00013307901046415721,
+      "loss": 0.0753,
+      "step": 3336
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001330413556267383,
+      "loss": 0.0745,
+      "step": 3337
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00013300369552954913,
+      "loss": 0.0754,
+      "step": 3338
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00013296603017858466,
+      "loss": 0.0763,
+      "step": 3339
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00013292835957984082,
+      "loss": 0.0767,
+      "step": 3340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00013289068373931425,
+      "loss": 0.077,
+      "step": 3341
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00013285300266300252,
+      "loss": 0.0749,
+      "step": 3342
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001328153163569039,
+      "loss": 0.074,
+      "step": 3343
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00013277762482701767,
+      "loss": 0.0738,
+      "step": 3344
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001327399280793438,
+      "loss": 0.0736,
+      "step": 3345
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00013270222611988318,
+      "loss": 0.079,
+      "step": 3346
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00013266451895463743,
+      "loss": 0.075,
+      "step": 3347
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001326268065896091,
+      "loss": 0.0766,
+      "step": 3348
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013258908903080152,
+      "loss": 0.0769,
+      "step": 3349
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013255136628421884,
+      "loss": 0.0723,
+      "step": 3350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001325136383558661,
+      "loss": 0.0737,
+      "step": 3351
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013247590525174903,
+      "loss": 0.0738,
+      "step": 3352
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001324381669778743,
+      "loss": 0.0775,
+      "step": 3353
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013240042354024944,
+      "loss": 0.0729,
+      "step": 3354
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001323626749448826,
+      "loss": 0.074,
+      "step": 3355
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000132324921197783,
+      "loss": 0.0759,
+      "step": 3356
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013228716230496054,
+      "loss": 0.0784,
+      "step": 3357
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013224939827242594,
+      "loss": 0.0761,
+      "step": 3358
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013221162910619075,
+      "loss": 0.077,
+      "step": 3359
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013217385481226735,
+      "loss": 0.0788,
+      "step": 3360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013213607539666898,
+      "loss": 0.0766,
+      "step": 3361
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00013209829086540965,
+      "loss": 0.0776,
+      "step": 3362
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001320605012245041,
+      "loss": 0.0733,
+      "step": 3363
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013202270647996806,
+      "loss": 0.0739,
+      "step": 3364
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001319849066378179,
+      "loss": 0.0757,
+      "step": 3365
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000131947101704071,
+      "loss": 0.076,
+      "step": 3366
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013190929168474527,
+      "loss": 0.0723,
+      "step": 3367
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013187147658585974,
+      "loss": 0.0769,
+      "step": 3368
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013183365641343403,
+      "loss": 0.0761,
+      "step": 3369
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013179583117348863,
+      "loss": 0.0763,
+      "step": 3370
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001317580008720449,
+      "loss": 0.0741,
+      "step": 3371
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013172016551512487,
+      "loss": 0.0743,
+      "step": 3372
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013168232510875152,
+      "loss": 0.0714,
+      "step": 3373
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013164447965894856,
+      "loss": 0.0729,
+      "step": 3374
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00013160662917174044,
+      "loss": 0.0762,
+      "step": 3375
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00013156877365315262,
+      "loss": 0.0748,
+      "step": 3376
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001315309131092111,
+      "loss": 0.0765,
+      "step": 3377
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00013149304754594287,
+      "loss": 0.0773,
+      "step": 3378
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00013145517696937567,
+      "loss": 0.0772,
+      "step": 3379
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001314173013855379,
+      "loss": 0.0728,
+      "step": 3380
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00013137942080045906,
+      "loss": 0.0794,
+      "step": 3381
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00013134153522016912,
+      "loss": 0.0738,
+      "step": 3382
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00013130364465069907,
+      "loss": 0.0731,
+      "step": 3383
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001312657490980806,
+      "loss": 0.0754,
+      "step": 3384
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00013122784856834613,
+      "loss": 0.0746,
+      "step": 3385
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001311899430675291,
+      "loss": 0.073,
+      "step": 3386
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00013115203260166344,
+      "loss": 0.0774,
+      "step": 3387
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001311141171767841,
+      "loss": 0.0741,
+      "step": 3388
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00013107619679892675,
+      "loss": 0.0733,
+      "step": 3389
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001310382714741278,
+      "loss": 0.0743,
+      "step": 3390
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013100034120842454,
+      "loss": 0.0729,
+      "step": 3391
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001309624060078549,
+      "loss": 0.0735,
+      "step": 3392
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013092446587845778,
+      "loss": 0.0818,
+      "step": 3393
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013088652082627274,
+      "loss": 0.0722,
+      "step": 3394
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001308485708573402,
+      "loss": 0.074,
+      "step": 3395
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013081061597770122,
+      "loss": 0.076,
+      "step": 3396
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013077265619339782,
+      "loss": 0.0728,
+      "step": 3397
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013073469151047272,
+      "loss": 0.0769,
+      "step": 3398
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013069672193496938,
+      "loss": 0.0768,
+      "step": 3399
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00013065874747293212,
+      "loss": 0.0742,
+      "step": 3400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.000130620768130406,
+      "loss": 0.0729,
+      "step": 3401
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001305827839134368,
+      "loss": 0.0763,
+      "step": 3402
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013054479482807121,
+      "loss": 0.0758,
+      "step": 3403
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001305068008803566,
+      "loss": 0.0749,
+      "step": 3404
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001304688020763411,
+      "loss": 0.0734,
+      "step": 3405
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001304307984220736,
+      "loss": 0.0774,
+      "step": 3406
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013039278992360394,
+      "loss": 0.0767,
+      "step": 3407
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013035477658698247,
+      "loss": 0.079,
+      "step": 3408
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013031675841826052,
+      "loss": 0.073,
+      "step": 3409
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013027873542349004,
+      "loss": 0.0765,
+      "step": 3410
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001302407076087239,
+      "loss": 0.0755,
+      "step": 3411
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013020267498001554,
+      "loss": 0.0764,
+      "step": 3412
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013016463754341934,
+      "loss": 0.0778,
+      "step": 3413
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013012659530499044,
+      "loss": 0.0724,
+      "step": 3414
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013008854827078459,
+      "loss": 0.0749,
+      "step": 3415
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00013005049644685847,
+      "loss": 0.0791,
+      "step": 3416
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001300124398392694,
+      "loss": 0.0758,
+      "step": 3417
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00012997437845407555,
+      "loss": 0.0779,
+      "step": 3418
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00012993631229733582,
+      "loss": 0.0764,
+      "step": 3419
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00012989824137510985,
+      "loss": 0.0744,
+      "step": 3420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00012986016569345804,
+      "loss": 0.0731,
+      "step": 3421
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001298220852584416,
+      "loss": 0.0732,
+      "step": 3422
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001297840000761224,
+      "loss": 0.0785,
+      "step": 3423
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00012974591015256325,
+      "loss": 0.0739,
+      "step": 3424
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00012970781549382742,
+      "loss": 0.0746,
+      "step": 3425
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001296697161059792,
+      "loss": 0.0729,
+      "step": 3426
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00012963161199508355,
+      "loss": 0.0743,
+      "step": 3427
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00012959350316720612,
+      "loss": 0.0737,
+      "step": 3428
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001295553896284134,
+      "loss": 0.0762,
+      "step": 3429
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012951727138477254,
+      "loss": 0.0774,
+      "step": 3430
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012947914844235154,
+      "loss": 0.0772,
+      "step": 3431
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012944102080721906,
+      "loss": 0.076,
+      "step": 3432
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001294028884854445,
+      "loss": 0.0728,
+      "step": 3433
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001293647514830982,
+      "loss": 0.0735,
+      "step": 3434
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012932660980625088,
+      "loss": 0.0767,
+      "step": 3435
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001292884634609744,
+      "loss": 0.0763,
+      "step": 3436
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0001292503124533411,
+      "loss": 0.073,
+      "step": 3437
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012921215678942413,
+      "loss": 0.0715,
+      "step": 3438
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012917399647529746,
+      "loss": 0.0749,
+      "step": 3439
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012913583151703565,
+      "loss": 0.0754,
+      "step": 3440
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012909766192071415,
+      "loss": 0.0714,
+      "step": 3441
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012905948769240905,
+      "loss": 0.078,
+      "step": 3442
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00012902130883819723,
+      "loss": 0.0728,
+      "step": 3443
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012898312536415625,
+      "loss": 0.0783,
+      "step": 3444
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012894493727636446,
+      "loss": 0.0714,
+      "step": 3445
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012890674458090098,
+      "loss": 0.0748,
+      "step": 3446
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012886854728384552,
+      "loss": 0.0759,
+      "step": 3447
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012883034539127865,
+      "loss": 0.0752,
+      "step": 3448
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012879213890928167,
+      "loss": 0.074,
+      "step": 3449
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012875392784393648,
+      "loss": 0.0769,
+      "step": 3450
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012871571220132589,
+      "loss": 0.0742,
+      "step": 3451
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012867749198753332,
+      "loss": 0.0762,
+      "step": 3452
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012863926720864293,
+      "loss": 0.074,
+      "step": 3453
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0001286010378707397,
+      "loss": 0.0754,
+      "step": 3454
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012856280397990916,
+      "loss": 0.075,
+      "step": 3455
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00012852456554223775,
+      "loss": 0.0764,
+      "step": 3456
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001284863225638125,
+      "loss": 0.0726,
+      "step": 3457
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012844807505072125,
+      "loss": 0.0741,
+      "step": 3458
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012840982300905247,
+      "loss": 0.0745,
+      "step": 3459
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001283715664448955,
+      "loss": 0.0785,
+      "step": 3460
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001283333053643402,
+      "loss": 0.0764,
+      "step": 3461
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012829503977347735,
+      "loss": 0.0716,
+      "step": 3462
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012825676967839829,
+      "loss": 0.0753,
+      "step": 3463
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012821849508519513,
+      "loss": 0.0725,
+      "step": 3464
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012818021599996077,
+      "loss": 0.0763,
+      "step": 3465
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012814193242878869,
+      "loss": 0.074,
+      "step": 3466
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012810364437777322,
+      "loss": 0.0733,
+      "step": 3467
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0001280653518530093,
+      "loss": 0.0768,
+      "step": 3468
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012802705486059264,
+      "loss": 0.0748,
+      "step": 3469
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00012798875340661964,
+      "loss": 0.0731,
+      "step": 3470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012795044749718736,
+      "loss": 0.0741,
+      "step": 3471
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012791213713839373,
+      "loss": 0.0762,
+      "step": 3472
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012787382233633718,
+      "loss": 0.0742,
+      "step": 3473
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012783550309711697,
+      "loss": 0.0738,
+      "step": 3474
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0001277971794268331,
+      "loss": 0.0733,
+      "step": 3475
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012775885133158612,
+      "loss": 0.0746,
+      "step": 3476
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012772051881747748,
+      "loss": 0.076,
+      "step": 3477
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012768218189060913,
+      "loss": 0.0735,
+      "step": 3478
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012764384055708394,
+      "loss": 0.0726,
+      "step": 3479
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012760549482300533,
+      "loss": 0.0756,
+      "step": 3480
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012756714469447744,
+      "loss": 0.0749,
+      "step": 3481
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00012752879017760514,
+      "loss": 0.0749,
+      "step": 3482
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000127490431278494,
+      "loss": 0.0748,
+      "step": 3483
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001274520680032503,
+      "loss": 0.0755,
+      "step": 3484
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012741370035798091,
+      "loss": 0.0762,
+      "step": 3485
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012737532834879354,
+      "loss": 0.0769,
+      "step": 3486
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001273369519817966,
+      "loss": 0.0743,
+      "step": 3487
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012729857126309897,
+      "loss": 0.0773,
+      "step": 3488
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001272601861988105,
+      "loss": 0.0783,
+      "step": 3489
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012722179679504155,
+      "loss": 0.0739,
+      "step": 3490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012718340305790325,
+      "loss": 0.0753,
+      "step": 3491
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012714500499350746,
+      "loss": 0.0733,
+      "step": 3492
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012710660260796658,
+      "loss": 0.0715,
+      "step": 3493
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012706819590739386,
+      "loss": 0.0761,
+      "step": 3494
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012702978489790312,
+      "loss": 0.0745,
+      "step": 3495
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00012699136958560893,
+      "loss": 0.0753,
+      "step": 3496
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001269529499766265,
+      "loss": 0.076,
+      "step": 3497
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012691452607707182,
+      "loss": 0.0756,
+      "step": 3498
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012687609789306144,
+      "loss": 0.0739,
+      "step": 3499
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012683766543071263,
+      "loss": 0.0751,
+      "step": 3500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001267992286961434,
+      "loss": 0.0749,
+      "step": 3501
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012676078769547237,
+      "loss": 0.074,
+      "step": 3502
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012672234243481887,
+      "loss": 0.0761,
+      "step": 3503
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012668389292030296,
+      "loss": 0.0771,
+      "step": 3504
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012664543915804523,
+      "loss": 0.0794,
+      "step": 3505
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012660698115416708,
+      "loss": 0.075,
+      "step": 3506
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012656851891479053,
+      "loss": 0.077,
+      "step": 3507
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012653005244603834,
+      "loss": 0.0736,
+      "step": 3508
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00012649158175403384,
+      "loss": 0.0743,
+      "step": 3509
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001264531068449011,
+      "loss": 0.0754,
+      "step": 3510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001264146277247648,
+      "loss": 0.0759,
+      "step": 3511
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001263761443997504,
+      "loss": 0.0754,
+      "step": 3512
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00012633765687598394,
+      "loss": 0.0758,
+      "step": 3513
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001262991651595921,
+      "loss": 0.076,
+      "step": 3514
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00012626066925670236,
+      "loss": 0.0747,
+      "step": 3515
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00012622216917344275,
+      "loss": 0.0721,
+      "step": 3516
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.000126183664915942,
+      "loss": 0.0734,
+      "step": 3517
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00012614515649032954,
+      "loss": 0.0746,
+      "step": 3518
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00012610664390273537,
+      "loss": 0.074,
+      "step": 3519
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00012606812715929025,
+      "loss": 0.0755,
+      "step": 3520
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00012602960626612554,
+      "loss": 0.0723,
+      "step": 3521
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001259910812293733,
+      "loss": 0.0742,
+      "step": 3522
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00012595255205516626,
+      "loss": 0.0745,
+      "step": 3523
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00012591401874963772,
+      "loss": 0.0761,
+      "step": 3524
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012587548131892174,
+      "loss": 0.0771,
+      "step": 3525
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.000125836939769153,
+      "loss": 0.0737,
+      "step": 3526
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001257983941064668,
+      "loss": 0.0733,
+      "step": 3527
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012575984433699922,
+      "loss": 0.0753,
+      "step": 3528
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012572129046688673,
+      "loss": 0.0766,
+      "step": 3529
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001256827325022668,
+      "loss": 0.0737,
+      "step": 3530
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012564417044927728,
+      "loss": 0.0772,
+      "step": 3531
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012560560431405676,
+      "loss": 0.0761,
+      "step": 3532
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012556703410274457,
+      "loss": 0.0761,
+      "step": 3533
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001255284598214805,
+      "loss": 0.0758,
+      "step": 3534
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012548988147640516,
+      "loss": 0.0723,
+      "step": 3535
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012545129907365975,
+      "loss": 0.0754,
+      "step": 3536
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012541271261938604,
+      "loss": 0.078,
+      "step": 3537
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00012537412211972653,
+      "loss": 0.0732,
+      "step": 3538
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012533552758082434,
+      "loss": 0.0759,
+      "step": 3539
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001252969290088233,
+      "loss": 0.0737,
+      "step": 3540
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012525832640986778,
+      "loss": 0.0745,
+      "step": 3541
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012521971979010283,
+      "loss": 0.0726,
+      "step": 3542
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012518110915567414,
+      "loss": 0.078,
+      "step": 3543
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.000125142494512728,
+      "loss": 0.0722,
+      "step": 3544
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012510387586741145,
+      "loss": 0.0754,
+      "step": 3545
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012506525322587207,
+      "loss": 0.0748,
+      "step": 3546
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012502662659425807,
+      "loss": 0.0754,
+      "step": 3547
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012498799597871835,
+      "loss": 0.0764,
+      "step": 3548
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012494936138540246,
+      "loss": 0.0756,
+      "step": 3549
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012491072282046044,
+      "loss": 0.074,
+      "step": 3550
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00012487208029004316,
+      "loss": 0.0762,
+      "step": 3551
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000124833433800302,
+      "loss": 0.0723,
+      "step": 3552
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00012479478335738898,
+      "loss": 0.0751,
+      "step": 3553
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00012475612896745681,
+      "loss": 0.0757,
+      "step": 3554
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001247174706366587,
+      "loss": 0.0749,
+      "step": 3555
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00012467880837114867,
+      "loss": 0.0788,
+      "step": 3556
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00012464014217708122,
+      "loss": 0.0738,
+      "step": 3557
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00012460147206061156,
+      "loss": 0.0785,
+      "step": 3558
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001245627980278954,
+      "loss": 0.0765,
+      "step": 3559
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00012452412008508924,
+      "loss": 0.0751,
+      "step": 3560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00012448543823835015,
+      "loss": 0.075,
+      "step": 3561
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001244467524938357,
+      "loss": 0.0768,
+      "step": 3562
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00012440806285770426,
+      "loss": 0.077,
+      "step": 3563
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00012436936933611469,
+      "loss": 0.073,
+      "step": 3564
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001243306719352265,
+      "loss": 0.0748,
+      "step": 3565
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012429197066119992,
+      "loss": 0.0753,
+      "step": 3566
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012425326552019557,
+      "loss": 0.0766,
+      "step": 3567
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012421455651837496,
+      "loss": 0.0758,
+      "step": 3568
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012417584366190002,
+      "loss": 0.0748,
+      "step": 3569
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012413712695693332,
+      "loss": 0.0719,
+      "step": 3570
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012409840640963818,
+      "loss": 0.0751,
+      "step": 3571
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012405968202617828,
+      "loss": 0.0744,
+      "step": 3572
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012402095381271817,
+      "loss": 0.0755,
+      "step": 3573
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012398222177542283,
+      "loss": 0.0769,
+      "step": 3574
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012394348592045798,
+      "loss": 0.077,
+      "step": 3575
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012390474625398982,
+      "loss": 0.0728,
+      "step": 3576
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012386600278218524,
+      "loss": 0.0781,
+      "step": 3577
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00012382725551121174,
+      "loss": 0.0731,
+      "step": 3578
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001237885044472374,
+      "loss": 0.0736,
+      "step": 3579
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012374974959643086,
+      "loss": 0.0738,
+      "step": 3580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012371099096496145,
+      "loss": 0.0735,
+      "step": 3581
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012367222855899906,
+      "loss": 0.074,
+      "step": 3582
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012363346238471415,
+      "loss": 0.0745,
+      "step": 3583
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012359469244827782,
+      "loss": 0.0742,
+      "step": 3584
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012355591875586175,
+      "loss": 0.0745,
+      "step": 3585
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012351714131363827,
+      "loss": 0.0755,
+      "step": 3586
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001234783601277802,
+      "loss": 0.0745,
+      "step": 3587
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012343957520446106,
+      "loss": 0.0743,
+      "step": 3588
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012340078654985494,
+      "loss": 0.0758,
+      "step": 3589
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001233619941701365,
+      "loss": 0.0742,
+      "step": 3590
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012332319807148093,
+      "loss": 0.0777,
+      "step": 3591
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00012328439826006415,
+      "loss": 0.078,
+      "step": 3592
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001232455947420626,
+      "loss": 0.0761,
+      "step": 3593
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012320678752365333,
+      "loss": 0.0767,
+      "step": 3594
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012316797661101394,
+      "loss": 0.0769,
+      "step": 3595
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012312916201032262,
+      "loss": 0.0719,
+      "step": 3596
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001230903437277582,
+      "loss": 0.0752,
+      "step": 3597
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012305152176950008,
+      "loss": 0.0735,
+      "step": 3598
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001230126961417282,
+      "loss": 0.0768,
+      "step": 3599
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001229738668506231,
+      "loss": 0.0747,
+      "step": 3600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012293503390236593,
+      "loss": 0.0755,
+      "step": 3601
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00012289619730313845,
+      "loss": 0.0734,
+      "step": 3602
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001228573570591229,
+      "loss": 0.0754,
+      "step": 3603
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001228185131765022,
+      "loss": 0.0747,
+      "step": 3604
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.0001227796656614598,
+      "loss": 0.0747,
+      "step": 3605
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012274081452017975,
+      "loss": 0.0764,
+      "step": 3606
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012270195975884663,
+      "loss": 0.0735,
+      "step": 3607
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012266310138364566,
+      "loss": 0.0739,
+      "step": 3608
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012262423940076258,
+      "loss": 0.0741,
+      "step": 3609
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0001225853738163838,
+      "loss": 0.0713,
+      "step": 3610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012254650463669613,
+      "loss": 0.0744,
+      "step": 3611
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012250763186788717,
+      "loss": 0.077,
+      "step": 3612
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012246875551614488,
+      "loss": 0.0754,
+      "step": 3613
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012242987558765798,
+      "loss": 0.0747,
+      "step": 3614
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012239099208861555,
+      "loss": 0.0731,
+      "step": 3615
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012235210502520747,
+      "loss": 0.0749,
+      "step": 3616
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.000122313214403624,
+      "loss": 0.0765,
+      "step": 3617
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012227432023005608,
+      "loss": 0.0731,
+      "step": 3618
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00012223542251069518,
+      "loss": 0.0776,
+      "step": 3619
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001221965212517333,
+      "loss": 0.0752,
+      "step": 3620
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012215761645936307,
+      "loss": 0.0758,
+      "step": 3621
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012211870813977758,
+      "loss": 0.0726,
+      "step": 3622
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0001220797962991706,
+      "loss": 0.0755,
+      "step": 3623
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012204088094373647,
+      "loss": 0.0771,
+      "step": 3624
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012200196207966988,
+      "loss": 0.0745,
+      "step": 3625
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012196303971316632,
+      "loss": 0.0757,
+      "step": 3626
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012192411385042175,
+      "loss": 0.0719,
+      "step": 3627
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012188518449763261,
+      "loss": 0.0753,
+      "step": 3628
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012184625166099608,
+      "loss": 0.077,
+      "step": 3629
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012180731534670964,
+      "loss": 0.0746,
+      "step": 3630
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012176837556097158,
+      "loss": 0.0755,
+      "step": 3631
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00012172943230998058,
+      "loss": 0.0782,
+      "step": 3632
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001216904855999359,
+      "loss": 0.0751,
+      "step": 3633
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012165153543703742,
+      "loss": 0.0727,
+      "step": 3634
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012161258182748547,
+      "loss": 0.0755,
+      "step": 3635
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.000121573624777481,
+      "loss": 0.0774,
+      "step": 3636
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012153466429322548,
+      "loss": 0.0742,
+      "step": 3637
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001214957003809209,
+      "loss": 0.0757,
+      "step": 3638
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012145673304676994,
+      "loss": 0.0732,
+      "step": 3639
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012141776229697556,
+      "loss": 0.0763,
+      "step": 3640
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001213787881377415,
+      "loss": 0.0749,
+      "step": 3641
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012133981057527197,
+      "loss": 0.0778,
+      "step": 3642
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012130082961577168,
+      "loss": 0.0742,
+      "step": 3643
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012126184526544591,
+      "loss": 0.077,
+      "step": 3644
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012122285753050047,
+      "loss": 0.0726,
+      "step": 3645
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00012118386641714175,
+      "loss": 0.074,
+      "step": 3646
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012114487193157661,
+      "loss": 0.0758,
+      "step": 3647
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012110587408001255,
+      "loss": 0.0772,
+      "step": 3648
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012106687286865748,
+      "loss": 0.0763,
+      "step": 3649
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012102786830371988,
+      "loss": 0.0745,
+      "step": 3650
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012098886039140889,
+      "loss": 0.0744,
+      "step": 3651
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.000120949849137934,
+      "loss": 0.0731,
+      "step": 3652
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012091083454950534,
+      "loss": 0.0758,
+      "step": 3653
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012087181663233354,
+      "loss": 0.0744,
+      "step": 3654
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012083279539262976,
+      "loss": 0.0776,
+      "step": 3655
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012079377083660565,
+      "loss": 0.0758,
+      "step": 3656
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012075474297047353,
+      "loss": 0.0747,
+      "step": 3657
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0001207157118004461,
+      "loss": 0.0784,
+      "step": 3658
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00012067667733273662,
+      "loss": 0.0745,
+      "step": 3659
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012063763957355889,
+      "loss": 0.0749,
+      "step": 3660
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012059859852912722,
+      "loss": 0.0747,
+      "step": 3661
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001205595542056565,
+      "loss": 0.0754,
+      "step": 3662
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012052050660936208,
+      "loss": 0.0734,
+      "step": 3663
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012048145574645983,
+      "loss": 0.0741,
+      "step": 3664
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012044240162316618,
+      "loss": 0.0738,
+      "step": 3665
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012040334424569801,
+      "loss": 0.0754,
+      "step": 3666
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012036428362027287,
+      "loss": 0.0762,
+      "step": 3667
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001203252197531086,
+      "loss": 0.0768,
+      "step": 3668
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012028615265042376,
+      "loss": 0.0749,
+      "step": 3669
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012024708231843731,
+      "loss": 0.078,
+      "step": 3670
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012020800876336877,
+      "loss": 0.0763,
+      "step": 3671
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012016893199143817,
+      "loss": 0.0722,
+      "step": 3672
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00012012985200886602,
+      "loss": 0.0745,
+      "step": 3673
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00012009076882187338,
+      "loss": 0.0758,
+      "step": 3674
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00012005168243668178,
+      "loss": 0.0746,
+      "step": 3675
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00012001259285951332,
+      "loss": 0.0727,
+      "step": 3676
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011997350009659056,
+      "loss": 0.0736,
+      "step": 3677
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011993440415413654,
+      "loss": 0.0796,
+      "step": 3678
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011989530503837492,
+      "loss": 0.0765,
+      "step": 3679
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011985620275552973,
+      "loss": 0.0749,
+      "step": 3680
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011981709731182557,
+      "loss": 0.0771,
+      "step": 3681
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011977798871348758,
+      "loss": 0.0762,
+      "step": 3682
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011973887696674129,
+      "loss": 0.0713,
+      "step": 3683
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011969976207781286,
+      "loss": 0.0723,
+      "step": 3684
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011966064405292887,
+      "loss": 0.0745,
+      "step": 3685
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00011962152289831639,
+      "loss": 0.0737,
+      "step": 3686
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011958239862020311,
+      "loss": 0.077,
+      "step": 3687
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011954327122481698,
+      "loss": 0.073,
+      "step": 3688
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011950414071838673,
+      "loss": 0.0776,
+      "step": 3689
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011946500710714137,
+      "loss": 0.0769,
+      "step": 3690
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011942587039731053,
+      "loss": 0.0776,
+      "step": 3691
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011938673059512422,
+      "loss": 0.0739,
+      "step": 3692
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011934758770681305,
+      "loss": 0.0739,
+      "step": 3693
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011930844173860808,
+      "loss": 0.0738,
+      "step": 3694
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011926929269674085,
+      "loss": 0.0735,
+      "step": 3695
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011923014058744342,
+      "loss": 0.0747,
+      "step": 3696
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011919098541694827,
+      "loss": 0.0744,
+      "step": 3697
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011915182719148841,
+      "loss": 0.0727,
+      "step": 3698
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001191126659172974,
+      "loss": 0.0736,
+      "step": 3699
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00011907350160060917,
+      "loss": 0.0749,
+      "step": 3700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011903433424765822,
+      "loss": 0.0734,
+      "step": 3701
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011899516386467948,
+      "loss": 0.0778,
+      "step": 3702
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001189559904579084,
+      "loss": 0.0764,
+      "step": 3703
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001189168140335809,
+      "loss": 0.0767,
+      "step": 3704
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011887763459793335,
+      "loss": 0.0752,
+      "step": 3705
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011883845215720266,
+      "loss": 0.0768,
+      "step": 3706
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011879926671762617,
+      "loss": 0.0773,
+      "step": 3707
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011876007828544171,
+      "loss": 0.0772,
+      "step": 3708
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011872088686688759,
+      "loss": 0.0755,
+      "step": 3709
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011868169246820258,
+      "loss": 0.0768,
+      "step": 3710
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011864249509562595,
+      "loss": 0.0742,
+      "step": 3711
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011860329475539745,
+      "loss": 0.0732,
+      "step": 3712
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00011856409145375724,
+      "loss": 0.0773,
+      "step": 3713
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.000118524885196946,
+      "loss": 0.0711,
+      "step": 3714
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011848567599120492,
+      "loss": 0.0725,
+      "step": 3715
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001184464638427756,
+      "loss": 0.0728,
+      "step": 3716
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011840724875790011,
+      "loss": 0.0742,
+      "step": 3717
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011836803074282098,
+      "loss": 0.0779,
+      "step": 3718
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011832880980378125,
+      "loss": 0.0746,
+      "step": 3719
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011828958594702444,
+      "loss": 0.0735,
+      "step": 3720
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011825035917879441,
+      "loss": 0.0776,
+      "step": 3721
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011821112950533564,
+      "loss": 0.076,
+      "step": 3722
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011817189693289299,
+      "loss": 0.0766,
+      "step": 3723
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011813266146771178,
+      "loss": 0.0733,
+      "step": 3724
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011809342311603783,
+      "loss": 0.0728,
+      "step": 3725
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00011805418188411734,
+      "loss": 0.0735,
+      "step": 3726
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001180149377781971,
+      "loss": 0.0743,
+      "step": 3727
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011797569080452421,
+      "loss": 0.0743,
+      "step": 3728
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011793644096934633,
+      "loss": 0.0754,
+      "step": 3729
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011789718827891156,
+      "loss": 0.0744,
+      "step": 3730
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011785793273946839,
+      "loss": 0.0706,
+      "step": 3731
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011781867435726586,
+      "loss": 0.074,
+      "step": 3732
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011777941313855338,
+      "loss": 0.0749,
+      "step": 3733
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011774014908958084,
+      "loss": 0.0744,
+      "step": 3734
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011770088221659864,
+      "loss": 0.0773,
+      "step": 3735
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001176616125258575,
+      "loss": 0.0753,
+      "step": 3736
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011762234002360872,
+      "loss": 0.0765,
+      "step": 3737
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011758306471610397,
+      "loss": 0.0755,
+      "step": 3738
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011754378660959535,
+      "loss": 0.0747,
+      "step": 3739
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011750450571033554,
+      "loss": 0.0736,
+      "step": 3740
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00011746522202457746,
+      "loss": 0.0716,
+      "step": 3741
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011742593555857464,
+      "loss": 0.0738,
+      "step": 3742
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011738664631858099,
+      "loss": 0.076,
+      "step": 3743
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011734735431085084,
+      "loss": 0.0724,
+      "step": 3744
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011730805954163902,
+      "loss": 0.0785,
+      "step": 3745
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011726876201720074,
+      "loss": 0.073,
+      "step": 3746
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011722946174379168,
+      "loss": 0.0763,
+      "step": 3747
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011719015872766797,
+      "loss": 0.0729,
+      "step": 3748
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011715085297508613,
+      "loss": 0.0735,
+      "step": 3749
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011711154449230314,
+      "loss": 0.0771,
+      "step": 3750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011707223328557644,
+      "loss": 0.0781,
+      "step": 3751
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001170329193611639,
+      "loss": 0.0762,
+      "step": 3752
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011699360272532376,
+      "loss": 0.0724,
+      "step": 3753
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00011695428338431477,
+      "loss": 0.0752,
+      "step": 3754
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011691496134439606,
+      "loss": 0.0792,
+      "step": 3755
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011687563661182723,
+      "loss": 0.0745,
+      "step": 3756
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011683630919286823,
+      "loss": 0.0745,
+      "step": 3757
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011679697909377953,
+      "loss": 0.0755,
+      "step": 3758
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011675764632082202,
+      "loss": 0.0771,
+      "step": 3759
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011671831088025695,
+      "loss": 0.0719,
+      "step": 3760
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011667897277834603,
+      "loss": 0.0759,
+      "step": 3761
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011663963202135137,
+      "loss": 0.0748,
+      "step": 3762
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011660028861553558,
+      "loss": 0.0782,
+      "step": 3763
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011656094256716162,
+      "loss": 0.0741,
+      "step": 3764
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011652159388249286,
+      "loss": 0.0801,
+      "step": 3765
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011648224256779314,
+      "loss": 0.076,
+      "step": 3766
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011644288862932668,
+      "loss": 0.0768,
+      "step": 3767
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00011640353207335817,
+      "loss": 0.073,
+      "step": 3768
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011636417290615267,
+      "loss": 0.0742,
+      "step": 3769
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011632481113397564,
+      "loss": 0.0768,
+      "step": 3770
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011628544676309301,
+      "loss": 0.0764,
+      "step": 3771
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011624607979977105,
+      "loss": 0.0773,
+      "step": 3772
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011620671025027658,
+      "loss": 0.078,
+      "step": 3773
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011616733812087664,
+      "loss": 0.077,
+      "step": 3774
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011612796341783883,
+      "loss": 0.0735,
+      "step": 3775
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001160885861474311,
+      "loss": 0.0741,
+      "step": 3776
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0001160492063159218,
+      "loss": 0.0753,
+      "step": 3777
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011600982392957978,
+      "loss": 0.0759,
+      "step": 3778
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011597043899467412,
+      "loss": 0.0725,
+      "step": 3779
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011593105151747448,
+      "loss": 0.073,
+      "step": 3780
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00011589166150425081,
+      "loss": 0.0754,
+      "step": 3781
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011585226896127353,
+      "loss": 0.0747,
+      "step": 3782
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011581287389481348,
+      "loss": 0.0779,
+      "step": 3783
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011577347631114176,
+      "loss": 0.0745,
+      "step": 3784
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011573407621653008,
+      "loss": 0.0746,
+      "step": 3785
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011569467361725037,
+      "loss": 0.0754,
+      "step": 3786
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011565526851957504,
+      "loss": 0.0755,
+      "step": 3787
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011561586092977697,
+      "loss": 0.0772,
+      "step": 3788
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011557645085412921,
+      "loss": 0.0806,
+      "step": 3789
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011553703829890546,
+      "loss": 0.0742,
+      "step": 3790
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011549762327037968,
+      "loss": 0.0755,
+      "step": 3791
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011545820577482622,
+      "loss": 0.0734,
+      "step": 3792
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011541878581851993,
+      "loss": 0.0718,
+      "step": 3793
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011537936340773585,
+      "loss": 0.0765,
+      "step": 3794
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00011533993854874962,
+      "loss": 0.0733,
+      "step": 3795
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00011530051124783721,
+      "loss": 0.0735,
+      "step": 3796
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00011526108151127488,
+      "loss": 0.0769,
+      "step": 3797
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001152216493453394,
+      "loss": 0.0745,
+      "step": 3798
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001151822147563078,
+      "loss": 0.0746,
+      "step": 3799
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00011514277775045768,
+      "loss": 0.0729,
+      "step": 3800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00011510333833406686,
+      "loss": 0.0748,
+      "step": 3801
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001150638965134136,
+      "loss": 0.0744,
+      "step": 3802
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00011502445229477658,
+      "loss": 0.0725,
+      "step": 3803
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00011498500568443477,
+      "loss": 0.0766,
+      "step": 3804
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00011494555668866762,
+      "loss": 0.0761,
+      "step": 3805
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00011490610531375492,
+      "loss": 0.0771,
+      "step": 3806
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0001148666515659768,
+      "loss": 0.0766,
+      "step": 3807
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00011482719545161383,
+      "loss": 0.076,
+      "step": 3808
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011478773697694691,
+      "loss": 0.0741,
+      "step": 3809
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011474827614825734,
+      "loss": 0.0771,
+      "step": 3810
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011470881297182681,
+      "loss": 0.0726,
+      "step": 3811
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011466934745393736,
+      "loss": 0.0731,
+      "step": 3812
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011462987960087139,
+      "loss": 0.0772,
+      "step": 3813
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011459040941891168,
+      "loss": 0.0751,
+      "step": 3814
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001145509369143414,
+      "loss": 0.0701,
+      "step": 3815
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001145114620934441,
+      "loss": 0.0733,
+      "step": 3816
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011447198496250367,
+      "loss": 0.0723,
+      "step": 3817
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011443250552780433,
+      "loss": 0.0747,
+      "step": 3818
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011439302379563079,
+      "loss": 0.0728,
+      "step": 3819
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011435353977226797,
+      "loss": 0.0752,
+      "step": 3820
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011431405346400128,
+      "loss": 0.0749,
+      "step": 3821
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00011427456487711643,
+      "loss": 0.0719,
+      "step": 3822
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001142350740178995,
+      "loss": 0.0764,
+      "step": 3823
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.000114195580892637,
+      "loss": 0.0759,
+      "step": 3824
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00011415608550761563,
+      "loss": 0.077,
+      "step": 3825
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00011411658786912269,
+      "loss": 0.0754,
+      "step": 3826
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00011407708798344559,
+      "loss": 0.0721,
+      "step": 3827
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00011403758585687226,
+      "loss": 0.0737,
+      "step": 3828
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00011399808149569102,
+      "loss": 0.0755,
+      "step": 3829
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00011395857490619036,
+      "loss": 0.0754,
+      "step": 3830
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001139190660946593,
+      "loss": 0.0735,
+      "step": 3831
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001138795550673871,
+      "loss": 0.0756,
+      "step": 3832
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00011384004183066345,
+      "loss": 0.0738,
+      "step": 3833
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001138005263907784,
+      "loss": 0.0764,
+      "step": 3834
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00011376100875402224,
+      "loss": 0.0803,
+      "step": 3835
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011372148892668578,
+      "loss": 0.0732,
+      "step": 3836
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011368196691505994,
+      "loss": 0.0743,
+      "step": 3837
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011364244272543627,
+      "loss": 0.0721,
+      "step": 3838
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011360291636410645,
+      "loss": 0.0737,
+      "step": 3839
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011356338783736255,
+      "loss": 0.0736,
+      "step": 3840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011352385715149711,
+      "loss": 0.0738,
+      "step": 3841
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011348432431280289,
+      "loss": 0.0743,
+      "step": 3842
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011344478932757301,
+      "loss": 0.073,
+      "step": 3843
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011340525220210092,
+      "loss": 0.0743,
+      "step": 3844
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011336571294268043,
+      "loss": 0.0735,
+      "step": 3845
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011332617155560576,
+      "loss": 0.0762,
+      "step": 3846
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011328662804717138,
+      "loss": 0.078,
+      "step": 3847
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001132470824236721,
+      "loss": 0.0722,
+      "step": 3848
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00011320753469140311,
+      "loss": 0.0751,
+      "step": 3849
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001131679848566599,
+      "loss": 0.0754,
+      "step": 3850
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011312843292573827,
+      "loss": 0.0757,
+      "step": 3851
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011308887890493447,
+      "loss": 0.0758,
+      "step": 3852
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011304932280054497,
+      "loss": 0.0758,
+      "step": 3853
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011300976461886658,
+      "loss": 0.0781,
+      "step": 3854
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011297020436619652,
+      "loss": 0.0747,
+      "step": 3855
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011293064204883224,
+      "loss": 0.0734,
+      "step": 3856
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011289107767307159,
+      "loss": 0.0779,
+      "step": 3857
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011285151124521274,
+      "loss": 0.0725,
+      "step": 3858
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011281194277155413,
+      "loss": 0.0744,
+      "step": 3859
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011277237225839458,
+      "loss": 0.0775,
+      "step": 3860
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011273279971203324,
+      "loss": 0.0757,
+      "step": 3861
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00011269322513876954,
+      "loss": 0.0776,
+      "step": 3862
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011265364854490325,
+      "loss": 0.0791,
+      "step": 3863
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011261406993673451,
+      "loss": 0.0742,
+      "step": 3864
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011257448932056372,
+      "loss": 0.0724,
+      "step": 3865
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011253490670269158,
+      "loss": 0.0773,
+      "step": 3866
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011249532208941922,
+      "loss": 0.0764,
+      "step": 3867
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011245573548704793,
+      "loss": 0.0781,
+      "step": 3868
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011241614690187947,
+      "loss": 0.0767,
+      "step": 3869
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011237655634021582,
+      "loss": 0.0759,
+      "step": 3870
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001123369638083593,
+      "loss": 0.0742,
+      "step": 3871
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011229736931261258,
+      "loss": 0.077,
+      "step": 3872
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011225777285927855,
+      "loss": 0.0769,
+      "step": 3873
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011221817445466049,
+      "loss": 0.0787,
+      "step": 3874
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.000112178574105062,
+      "loss": 0.0732,
+      "step": 3875
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00011213897181678692,
+      "loss": 0.0755,
+      "step": 3876
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011209936759613951,
+      "loss": 0.0752,
+      "step": 3877
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011205976144942414,
+      "loss": 0.0736,
+      "step": 3878
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011202015338294574,
+      "loss": 0.0724,
+      "step": 3879
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011198054340300934,
+      "loss": 0.0746,
+      "step": 3880
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011194093151592038,
+      "loss": 0.0753,
+      "step": 3881
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001119013177279846,
+      "loss": 0.0753,
+      "step": 3882
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011186170204550794,
+      "loss": 0.0759,
+      "step": 3883
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011182208447479683,
+      "loss": 0.0738,
+      "step": 3884
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001117824650221578,
+      "loss": 0.0757,
+      "step": 3885
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011174284369389783,
+      "loss": 0.075,
+      "step": 3886
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011170322049632413,
+      "loss": 0.0745,
+      "step": 3887
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011166359543574416,
+      "loss": 0.0792,
+      "step": 3888
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00011162396851846581,
+      "loss": 0.0749,
+      "step": 3889
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011158433975079717,
+      "loss": 0.0734,
+      "step": 3890
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011154470913904663,
+      "loss": 0.0754,
+      "step": 3891
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011150507668952285,
+      "loss": 0.0744,
+      "step": 3892
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011146544240853487,
+      "loss": 0.0749,
+      "step": 3893
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011142580630239197,
+      "loss": 0.0756,
+      "step": 3894
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011138616837740372,
+      "loss": 0.0753,
+      "step": 3895
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011134652863987996,
+      "loss": 0.0737,
+      "step": 3896
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011130688709613087,
+      "loss": 0.0733,
+      "step": 3897
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011126724375246684,
+      "loss": 0.076,
+      "step": 3898
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011122759861519864,
+      "loss": 0.0748,
+      "step": 3899
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011118795169063727,
+      "loss": 0.0746,
+      "step": 3900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011114830298509402,
+      "loss": 0.0754,
+      "step": 3901
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011110865250488047,
+      "loss": 0.0735,
+      "step": 3902
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00011106900025630849,
+      "loss": 0.0774,
+      "step": 3903
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011102934624569017,
+      "loss": 0.0745,
+      "step": 3904
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011098969047933798,
+      "loss": 0.0777,
+      "step": 3905
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011095003296356463,
+      "loss": 0.0723,
+      "step": 3906
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011091037370468306,
+      "loss": 0.0761,
+      "step": 3907
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001108707127090066,
+      "loss": 0.0761,
+      "step": 3908
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011083104998284868,
+      "loss": 0.0728,
+      "step": 3909
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011079138553252317,
+      "loss": 0.0718,
+      "step": 3910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011075171936434416,
+      "loss": 0.0732,
+      "step": 3911
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011071205148462602,
+      "loss": 0.0757,
+      "step": 3912
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011067238189968332,
+      "loss": 0.074,
+      "step": 3913
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.000110632710615831,
+      "loss": 0.0747,
+      "step": 3914
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011059303763938426,
+      "loss": 0.0767,
+      "step": 3915
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00011055336297665848,
+      "loss": 0.071,
+      "step": 3916
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011051368663396941,
+      "loss": 0.0707,
+      "step": 3917
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011047400861763301,
+      "loss": 0.0748,
+      "step": 3918
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011043432893396552,
+      "loss": 0.0761,
+      "step": 3919
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011039464758928351,
+      "loss": 0.073,
+      "step": 3920
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011035496458990365,
+      "loss": 0.0763,
+      "step": 3921
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011031527994214304,
+      "loss": 0.0752,
+      "step": 3922
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011027559365231899,
+      "loss": 0.0753,
+      "step": 3923
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.000110235905726749,
+      "loss": 0.0737,
+      "step": 3924
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011019621617175098,
+      "loss": 0.0759,
+      "step": 3925
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011015652499364293,
+      "loss": 0.0772,
+      "step": 3926
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011011683219874323,
+      "loss": 0.0734,
+      "step": 3927
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011007713779337046,
+      "loss": 0.0756,
+      "step": 3928
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00011003744178384348,
+      "loss": 0.0721,
+      "step": 3929
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00010999774417648141,
+      "loss": 0.0751,
+      "step": 3930
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010995804497760358,
+      "loss": 0.0761,
+      "step": 3931
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010991834419352962,
+      "loss": 0.0749,
+      "step": 3932
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010987864183057943,
+      "loss": 0.0772,
+      "step": 3933
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010983893789507306,
+      "loss": 0.0768,
+      "step": 3934
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010979923239333099,
+      "loss": 0.0752,
+      "step": 3935
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010975952533167369,
+      "loss": 0.0779,
+      "step": 3936
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010971981671642215,
+      "loss": 0.0773,
+      "step": 3937
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010968010655389744,
+      "loss": 0.0717,
+      "step": 3938
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001096403948504209,
+      "loss": 0.0765,
+      "step": 3939
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001096006816123142,
+      "loss": 0.076,
+      "step": 3940
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010956096684589912,
+      "loss": 0.0734,
+      "step": 3941
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010952125055749777,
+      "loss": 0.0747,
+      "step": 3942
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00010948153275343253,
+      "loss": 0.0745,
+      "step": 3943
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010944181344002595,
+      "loss": 0.0735,
+      "step": 3944
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010940209262360081,
+      "loss": 0.0727,
+      "step": 3945
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010936237031048022,
+      "loss": 0.073,
+      "step": 3946
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010932264650698745,
+      "loss": 0.076,
+      "step": 3947
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010928292121944605,
+      "loss": 0.0755,
+      "step": 3948
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010924319445417977,
+      "loss": 0.0731,
+      "step": 3949
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010920346621751263,
+      "loss": 0.0763,
+      "step": 3950
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010916373651576884,
+      "loss": 0.0753,
+      "step": 3951
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001091240053552729,
+      "loss": 0.0751,
+      "step": 3952
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010908427274234949,
+      "loss": 0.0765,
+      "step": 3953
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010904453868332358,
+      "loss": 0.0766,
+      "step": 3954
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0001090048031845203,
+      "loss": 0.0761,
+      "step": 3955
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010896506625226505,
+      "loss": 0.0754,
+      "step": 3956
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00010892532789288345,
+      "loss": 0.0741,
+      "step": 3957
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001088855881127014,
+      "loss": 0.0731,
+      "step": 3958
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010884584691804491,
+      "loss": 0.0767,
+      "step": 3959
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010880610431524033,
+      "loss": 0.0781,
+      "step": 3960
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010876636031061412,
+      "loss": 0.0763,
+      "step": 3961
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010872661491049308,
+      "loss": 0.0732,
+      "step": 3962
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010868686812120416,
+      "loss": 0.0744,
+      "step": 3963
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010864711994907458,
+      "loss": 0.0785,
+      "step": 3964
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010860737040043175,
+      "loss": 0.0732,
+      "step": 3965
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010856761948160323,
+      "loss": 0.0759,
+      "step": 3966
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010852786719891694,
+      "loss": 0.0733,
+      "step": 3967
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010848811355870097,
+      "loss": 0.0798,
+      "step": 3968
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010844835856728348,
+      "loss": 0.0777,
+      "step": 3969
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0001084086022309931,
+      "loss": 0.0793,
+      "step": 3970
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00010836884455615848,
+      "loss": 0.0745,
+      "step": 3971
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00010832908554910853,
+      "loss": 0.075,
+      "step": 3972
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00010828932521617244,
+      "loss": 0.073,
+      "step": 3973
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001082495635636795,
+      "loss": 0.0716,
+      "step": 3974
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00010820980059795929,
+      "loss": 0.0759,
+      "step": 3975
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001081700363253416,
+      "loss": 0.0758,
+      "step": 3976
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00010813027075215634,
+      "loss": 0.0747,
+      "step": 3977
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001080905038847338,
+      "loss": 0.0762,
+      "step": 3978
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00010805073572940425,
+      "loss": 0.0741,
+      "step": 3979
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00010801096629249836,
+      "loss": 0.0786,
+      "step": 3980
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0001079711955803469,
+      "loss": 0.0754,
+      "step": 3981
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00010793142359928084,
+      "loss": 0.077,
+      "step": 3982
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00010789165035563145,
+      "loss": 0.0753,
+      "step": 3983
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00010785187585573007,
+      "loss": 0.076,
+      "step": 3984
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010781210010590835,
+      "loss": 0.0739,
+      "step": 3985
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010777232311249806,
+      "loss": 0.0751,
+      "step": 3986
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010773254488183117,
+      "loss": 0.074,
+      "step": 3987
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010769276542023998,
+      "loss": 0.074,
+      "step": 3988
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010765298473405678,
+      "loss": 0.0706,
+      "step": 3989
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001076132028296142,
+      "loss": 0.0787,
+      "step": 3990
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010757341971324504,
+      "loss": 0.0749,
+      "step": 3991
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010753363539128223,
+      "loss": 0.0775,
+      "step": 3992
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010749384987005895,
+      "loss": 0.0769,
+      "step": 3993
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010745406315590855,
+      "loss": 0.0748,
+      "step": 3994
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010741427525516463,
+      "loss": 0.0749,
+      "step": 3995
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010737448617416086,
+      "loss": 0.0748,
+      "step": 3996
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010733469591923122,
+      "loss": 0.0752,
+      "step": 3997
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00010729490449670976,
+      "loss": 0.0736,
+      "step": 3998
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001072551119129308,
+      "loss": 0.0725,
+      "step": 3999
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010721531817422885,
+      "loss": 0.0753,
+      "step": 4000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010717552328693854,
+      "loss": 0.0741,
+      "step": 4001
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010713572725739475,
+      "loss": 0.0727,
+      "step": 4002
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010709593009193248,
+      "loss": 0.0763,
+      "step": 4003
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010705613179688693,
+      "loss": 0.0764,
+      "step": 4004
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010701633237859354,
+      "loss": 0.0763,
+      "step": 4005
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010697653184338785,
+      "loss": 0.0763,
+      "step": 4006
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010693673019760562,
+      "loss": 0.076,
+      "step": 4007
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010689692744758275,
+      "loss": 0.077,
+      "step": 4008
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010685712359965534,
+      "loss": 0.0751,
+      "step": 4009
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00010681731866015966,
+      "loss": 0.0763,
+      "step": 4010
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001067775126354322,
+      "loss": 0.0735,
+      "step": 4011
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010673770553180957,
+      "loss": 0.0741,
+      "step": 4012
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010669789735562854,
+      "loss": 0.0743,
+      "step": 4013
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010665808811322609,
+      "loss": 0.0763,
+      "step": 4014
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001066182778109393,
+      "loss": 0.0779,
+      "step": 4015
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010657846645510556,
+      "loss": 0.0712,
+      "step": 4016
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010653865405206227,
+      "loss": 0.0729,
+      "step": 4017
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010649884060814712,
+      "loss": 0.074,
+      "step": 4018
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010645902612969787,
+      "loss": 0.0754,
+      "step": 4019
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010641921062305249,
+      "loss": 0.0716,
+      "step": 4020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010637939409454916,
+      "loss": 0.0748,
+      "step": 4021
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010633957655052608,
+      "loss": 0.0733,
+      "step": 4022
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010629975799732179,
+      "loss": 0.0759,
+      "step": 4023
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00010625993844127488,
+      "loss": 0.0706,
+      "step": 4024
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001062201178887241,
+      "loss": 0.0751,
+      "step": 4025
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010618029634600843,
+      "loss": 0.077,
+      "step": 4026
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010614047381946691,
+      "loss": 0.0741,
+      "step": 4027
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001061006503154388,
+      "loss": 0.0751,
+      "step": 4028
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010606082584026355,
+      "loss": 0.0753,
+      "step": 4029
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010602100040028067,
+      "loss": 0.0744,
+      "step": 4030
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001059811740018299,
+      "loss": 0.0768,
+      "step": 4031
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010594134665125106,
+      "loss": 0.0779,
+      "step": 4032
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001059015183548842,
+      "loss": 0.0747,
+      "step": 4033
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001058616891190695,
+      "loss": 0.0741,
+      "step": 4034
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010582185895014722,
+      "loss": 0.0747,
+      "step": 4035
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010578202785445793,
+      "loss": 0.0761,
+      "step": 4036
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001057421958383421,
+      "loss": 0.0747,
+      "step": 4037
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00010570236290814059,
+      "loss": 0.0735,
+      "step": 4038
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010566252907019429,
+      "loss": 0.0753,
+      "step": 4039
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001056226943308442,
+      "loss": 0.074,
+      "step": 4040
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010558285869643158,
+      "loss": 0.0776,
+      "step": 4041
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010554302217329772,
+      "loss": 0.0732,
+      "step": 4042
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010550318476778412,
+      "loss": 0.0723,
+      "step": 4043
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010546334648623235,
+      "loss": 0.074,
+      "step": 4044
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010542350733498423,
+      "loss": 0.0756,
+      "step": 4045
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001053836673203816,
+      "loss": 0.0752,
+      "step": 4046
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001053438264487665,
+      "loss": 0.0763,
+      "step": 4047
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010530398472648115,
+      "loss": 0.0735,
+      "step": 4048
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010526414215986782,
+      "loss": 0.0743,
+      "step": 4049
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010522429875526892,
+      "loss": 0.0749,
+      "step": 4050
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00010518445451902706,
+      "loss": 0.0764,
+      "step": 4051
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001051446094574849,
+      "loss": 0.0739,
+      "step": 4052
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010510476357698533,
+      "loss": 0.0773,
+      "step": 4053
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010506491688387127,
+      "loss": 0.0783,
+      "step": 4054
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010502506938448586,
+      "loss": 0.074,
+      "step": 4055
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001049852210851723,
+      "loss": 0.0749,
+      "step": 4056
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010494537199227392,
+      "loss": 0.076,
+      "step": 4057
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001049055221121342,
+      "loss": 0.0755,
+      "step": 4058
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010486567145109678,
+      "loss": 0.0741,
+      "step": 4059
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010482582001550535,
+      "loss": 0.0745,
+      "step": 4060
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001047859678117038,
+      "loss": 0.0729,
+      "step": 4061
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010474611484603607,
+      "loss": 0.0746,
+      "step": 4062
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010470626112484623,
+      "loss": 0.0751,
+      "step": 4063
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010466640665447853,
+      "loss": 0.0744,
+      "step": 4064
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00010462655144127734,
+      "loss": 0.0766,
+      "step": 4065
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010458669549158702,
+      "loss": 0.0736,
+      "step": 4066
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001045468388117522,
+      "loss": 0.0717,
+      "step": 4067
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010450698140811753,
+      "loss": 0.0717,
+      "step": 4068
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010446712328702783,
+      "loss": 0.0727,
+      "step": 4069
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010442726445482804,
+      "loss": 0.0758,
+      "step": 4070
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010438740491786316,
+      "loss": 0.0746,
+      "step": 4071
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010434754468247832,
+      "loss": 0.0768,
+      "step": 4072
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010430768375501875,
+      "loss": 0.0762,
+      "step": 4073
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010426782214182991,
+      "loss": 0.0731,
+      "step": 4074
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010422795984925713,
+      "loss": 0.0764,
+      "step": 4075
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001041880968836461,
+      "loss": 0.0689,
+      "step": 4076
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010414823325134247,
+      "loss": 0.0722,
+      "step": 4077
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010410836895869198,
+      "loss": 0.0708,
+      "step": 4078
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00010406850401204062,
+      "loss": 0.0745,
+      "step": 4079
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010402863841773431,
+      "loss": 0.0743,
+      "step": 4080
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001039887721821192,
+      "loss": 0.0731,
+      "step": 4081
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010394890531154149,
+      "loss": 0.0729,
+      "step": 4082
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010390903781234746,
+      "loss": 0.076,
+      "step": 4083
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010386916969088356,
+      "loss": 0.0711,
+      "step": 4084
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010382930095349625,
+      "loss": 0.0739,
+      "step": 4085
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010378943160653216,
+      "loss": 0.0725,
+      "step": 4086
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000103749561656338,
+      "loss": 0.071,
+      "step": 4087
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010370969110926052,
+      "loss": 0.0742,
+      "step": 4088
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001036698199716467,
+      "loss": 0.0759,
+      "step": 4089
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010362994824984343,
+      "loss": 0.0739,
+      "step": 4090
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00010359007595019785,
+      "loss": 0.0697,
+      "step": 4091
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001035502030790571,
+      "loss": 0.0749,
+      "step": 4092
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010351032964276846,
+      "loss": 0.0765,
+      "step": 4093
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010347045564767928,
+      "loss": 0.0753,
+      "step": 4094
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000103430581100137,
+      "loss": 0.0711,
+      "step": 4095
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010339070600648912,
+      "loss": 0.0742,
+      "step": 4096
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001033508303730833,
+      "loss": 0.0731,
+      "step": 4097
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010331095420626724,
+      "loss": 0.0741,
+      "step": 4098
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010327107751238869,
+      "loss": 0.0728,
+      "step": 4099
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010323120029779554,
+      "loss": 0.0737,
+      "step": 4100
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010319132256883575,
+      "loss": 0.0739,
+      "step": 4101
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010315144433185735,
+      "loss": 0.0757,
+      "step": 4102
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010311156559320845,
+      "loss": 0.0756,
+      "step": 4103
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010307168635923726,
+      "loss": 0.0749,
+      "step": 4104
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010303180663629202,
+      "loss": 0.073,
+      "step": 4105
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00010299192643072115,
+      "loss": 0.0742,
+      "step": 4106
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010295204574887303,
+      "loss": 0.0746,
+      "step": 4107
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010291216459709616,
+      "loss": 0.0714,
+      "step": 4108
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010287228298173914,
+      "loss": 0.0747,
+      "step": 4109
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010283240090915063,
+      "loss": 0.073,
+      "step": 4110
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010279251838567931,
+      "loss": 0.0731,
+      "step": 4111
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010275263541767404,
+      "loss": 0.0757,
+      "step": 4112
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010271275201148367,
+      "loss": 0.0728,
+      "step": 4113
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010267286817345714,
+      "loss": 0.0758,
+      "step": 4114
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010263298390994343,
+      "loss": 0.0703,
+      "step": 4115
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010259309922729161,
+      "loss": 0.0721,
+      "step": 4116
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001025532141318509,
+      "loss": 0.0781,
+      "step": 4117
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00010251332862997043,
+      "loss": 0.0748,
+      "step": 4118
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001024734427279995,
+      "loss": 0.0739,
+      "step": 4119
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010243355643228747,
+      "loss": 0.0718,
+      "step": 4120
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010239366974918368,
+      "loss": 0.0723,
+      "step": 4121
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010235378268503763,
+      "loss": 0.0715,
+      "step": 4122
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010231389524619885,
+      "loss": 0.0754,
+      "step": 4123
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010227400743901692,
+      "loss": 0.0741,
+      "step": 4124
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010223411926984145,
+      "loss": 0.0745,
+      "step": 4125
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010219423074502212,
+      "loss": 0.0764,
+      "step": 4126
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001021543418709088,
+      "loss": 0.0752,
+      "step": 4127
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010211445265385113,
+      "loss": 0.0749,
+      "step": 4128
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010207456310019912,
+      "loss": 0.0707,
+      "step": 4129
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010203467321630263,
+      "loss": 0.0779,
+      "step": 4130
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010199478300851157,
+      "loss": 0.0706,
+      "step": 4131
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010195489248317609,
+      "loss": 0.0706,
+      "step": 4132
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00010191500164664616,
+      "loss": 0.0723,
+      "step": 4133
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010187511050527195,
+      "loss": 0.0715,
+      "step": 4134
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010183521906540361,
+      "loss": 0.0751,
+      "step": 4135
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010179532733339135,
+      "loss": 0.0743,
+      "step": 4136
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010175543531558549,
+      "loss": 0.0725,
+      "step": 4137
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010171554301833625,
+      "loss": 0.0724,
+      "step": 4138
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010167565044799405,
+      "loss": 0.0749,
+      "step": 4139
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010163575761090928,
+      "loss": 0.0755,
+      "step": 4140
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010159586451343235,
+      "loss": 0.0752,
+      "step": 4141
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010155597116191383,
+      "loss": 0.0731,
+      "step": 4142
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0001015160775627041,
+      "loss": 0.0727,
+      "step": 4143
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010147618372215381,
+      "loss": 0.0729,
+      "step": 4144
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010143628964661357,
+      "loss": 0.0725,
+      "step": 4145
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00010139639534243396,
+      "loss": 0.0736,
+      "step": 4146
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010135650081596575,
+      "loss": 0.0705,
+      "step": 4147
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010131660607355955,
+      "loss": 0.0736,
+      "step": 4148
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010127671112156613,
+      "loss": 0.0732,
+      "step": 4149
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010123681596633629,
+      "loss": 0.0748,
+      "step": 4150
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010119692061422085,
+      "loss": 0.0753,
+      "step": 4151
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010115702507157061,
+      "loss": 0.0754,
+      "step": 4152
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010111712934473645,
+      "loss": 0.0735,
+      "step": 4153
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0001010772334400693,
+      "loss": 0.0707,
+      "step": 4154
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010103733736392005,
+      "loss": 0.0743,
+      "step": 4155
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010099744112263969,
+      "loss": 0.0762,
+      "step": 4156
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010095754472257919,
+      "loss": 0.0746,
+      "step": 4157
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010091764817008952,
+      "loss": 0.0744,
+      "step": 4158
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010087775147152179,
+      "loss": 0.0761,
+      "step": 4159
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00010083785463322699,
+      "loss": 0.073,
+      "step": 4160
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010079795766155622,
+      "loss": 0.0736,
+      "step": 4161
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0001007580605628606,
+      "loss": 0.0744,
+      "step": 4162
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0001007181633434912,
+      "loss": 0.0746,
+      "step": 4163
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010067826600979918,
+      "loss": 0.0732,
+      "step": 4164
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010063836856813571,
+      "loss": 0.0734,
+      "step": 4165
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010059847102485196,
+      "loss": 0.0754,
+      "step": 4166
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0001005585733862991,
+      "loss": 0.0734,
+      "step": 4167
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010051867565882838,
+      "loss": 0.0764,
+      "step": 4168
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010047877784879094,
+      "loss": 0.0706,
+      "step": 4169
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010043887996253809,
+      "loss": 0.0724,
+      "step": 4170
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010039898200642104,
+      "loss": 0.0721,
+      "step": 4171
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010035908398679101,
+      "loss": 0.0745,
+      "step": 4172
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010031918590999938,
+      "loss": 0.0719,
+      "step": 4173
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00010027928778239728,
+      "loss": 0.0745,
+      "step": 4174
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00010023938961033613,
+      "loss": 0.0704,
+      "step": 4175
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00010019949140016707,
+      "loss": 0.0751,
+      "step": 4176
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0001001595931582415,
+      "loss": 0.0762,
+      "step": 4177
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00010011969489091073,
+      "loss": 0.075,
+      "step": 4178
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.000100079796604526,
+      "loss": 0.0738,
+      "step": 4179
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00010003989830543867,
+      "loss": 0.0762,
+      "step": 4180
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0001,
+      "loss": 0.0719,
+      "step": 4181
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.996010169456135e-05,
+      "loss": 0.075,
+      "step": 4182
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.992020339547401e-05,
+      "loss": 0.0725,
+      "step": 4183
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.988030510908929e-05,
+      "loss": 0.0724,
+      "step": 4184
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.984040684175853e-05,
+      "loss": 0.075,
+      "step": 4185
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.980050859983297e-05,
+      "loss": 0.0748,
+      "step": 4186
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.976061038966391e-05,
+      "loss": 0.0707,
+      "step": 4187
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.972071221760274e-05,
+      "loss": 0.0721,
+      "step": 4188
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.968081409000066e-05,
+      "loss": 0.0728,
+      "step": 4189
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.964091601320897e-05,
+      "loss": 0.0734,
+      "step": 4190
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.960101799357898e-05,
+      "loss": 0.0714,
+      "step": 4191
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.956112003746194e-05,
+      "loss": 0.0752,
+      "step": 4192
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.952122215120906e-05,
+      "loss": 0.0744,
+      "step": 4193
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.948132434117164e-05,
+      "loss": 0.0722,
+      "step": 4194
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.944142661370091e-05,
+      "loss": 0.0741,
+      "step": 4195
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.940152897514809e-05,
+      "loss": 0.0759,
+      "step": 4196
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.936163143186429e-05,
+      "loss": 0.0774,
+      "step": 4197
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.932173399020085e-05,
+      "loss": 0.0735,
+      "step": 4198
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.928183665650884e-05,
+      "loss": 0.0722,
+      "step": 4199
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.924193943713943e-05,
+      "loss": 0.0768,
+      "step": 4200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.920204233844379e-05,
+      "loss": 0.0763,
+      "step": 4201
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.916214536677303e-05,
+      "loss": 0.0734,
+      "step": 4202
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.912224852847825e-05,
+      "loss": 0.0751,
+      "step": 4203
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.908235182991047e-05,
+      "loss": 0.0727,
+      "step": 4204
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.904245527742083e-05,
+      "loss": 0.0768,
+      "step": 4205
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.900255887736035e-05,
+      "loss": 0.0734,
+      "step": 4206
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.896266263607995e-05,
+      "loss": 0.0751,
+      "step": 4207
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.892276655993074e-05,
+      "loss": 0.0713,
+      "step": 4208
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.888287065526358e-05,
+      "loss": 0.0772,
+      "step": 4209
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.884297492842944e-05,
+      "loss": 0.0726,
+      "step": 4210
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.880307938577917e-05,
+      "loss": 0.0746,
+      "step": 4211
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.876318403366372e-05,
+      "loss": 0.0759,
+      "step": 4212
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.87232888784339e-05,
+      "loss": 0.0729,
+      "step": 4213
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.868339392644046e-05,
+      "loss": 0.0715,
+      "step": 4214
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.864349918403427e-05,
+      "loss": 0.0718,
+      "step": 4215
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.860360465756605e-05,
+      "loss": 0.0765,
+      "step": 4216
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.856371035338642e-05,
+      "loss": 0.0728,
+      "step": 4217
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.85238162778462e-05,
+      "loss": 0.0755,
+      "step": 4218
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.848392243729593e-05,
+      "loss": 0.0732,
+      "step": 4219
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.844402883808624e-05,
+      "loss": 0.076,
+      "step": 4220
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.840413548656765e-05,
+      "loss": 0.0747,
+      "step": 4221
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.836424238909073e-05,
+      "loss": 0.0732,
+      "step": 4222
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.832434955200597e-05,
+      "loss": 0.0714,
+      "step": 4223
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.828445698166376e-05,
+      "loss": 0.0743,
+      "step": 4224
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.824456468441455e-05,
+      "loss": 0.0746,
+      "step": 4225
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.820467266660868e-05,
+      "loss": 0.0708,
+      "step": 4226
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.816478093459644e-05,
+      "loss": 0.072,
+      "step": 4227
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.812488949472807e-05,
+      "loss": 0.0743,
+      "step": 4228
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.808499835335386e-05,
+      "loss": 0.072,
+      "step": 4229
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.804510751682394e-05,
+      "loss": 0.0766,
+      "step": 4230
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.800521699148842e-05,
+      "loss": 0.0732,
+      "step": 4231
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.796532678369741e-05,
+      "loss": 0.0744,
+      "step": 4232
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.79254368998009e-05,
+      "loss": 0.0728,
+      "step": 4233
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.78855473461489e-05,
+      "loss": 0.0717,
+      "step": 4234
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.784565812909123e-05,
+      "loss": 0.0734,
+      "step": 4235
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.780576925497789e-05,
+      "loss": 0.0717,
+      "step": 4236
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.77658807301586e-05,
+      "loss": 0.0743,
+      "step": 4237
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.772599256098311e-05,
+      "loss": 0.0759,
+      "step": 4238
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.768610475380116e-05,
+      "loss": 0.0743,
+      "step": 4239
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.764621731496239e-05,
+      "loss": 0.0754,
+      "step": 4240
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.760633025081632e-05,
+      "loss": 0.076,
+      "step": 4241
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.756644356771256e-05,
+      "loss": 0.0723,
+      "step": 4242
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.75265572720005e-05,
+      "loss": 0.0764,
+      "step": 4243
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.748667137002962e-05,
+      "loss": 0.0753,
+      "step": 4244
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.744678586814911e-05,
+      "loss": 0.0726,
+      "step": 4245
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.74069007727084e-05,
+      "loss": 0.0749,
+      "step": 4246
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.736701609005662e-05,
+      "loss": 0.0767,
+      "step": 4247
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.73271318265429e-05,
+      "loss": 0.0723,
+      "step": 4248
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.728724798851635e-05,
+      "loss": 0.0714,
+      "step": 4249
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.724736458232597e-05,
+      "loss": 0.0766,
+      "step": 4250
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.720748161432071e-05,
+      "loss": 0.0743,
+      "step": 4251
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.71675990908494e-05,
+      "loss": 0.0733,
+      "step": 4252
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.712771701826088e-05,
+      "loss": 0.076,
+      "step": 4253
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.708783540290388e-05,
+      "loss": 0.0758,
+      "step": 4254
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.704795425112698e-05,
+      "loss": 0.0718,
+      "step": 4255
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.700807356927887e-05,
+      "loss": 0.0726,
+      "step": 4256
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.696819336370799e-05,
+      "loss": 0.0735,
+      "step": 4257
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.692831364076275e-05,
+      "loss": 0.0786,
+      "step": 4258
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.688843440679157e-05,
+      "loss": 0.0727,
+      "step": 4259
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.684855566814268e-05,
+      "loss": 0.077,
+      "step": 4260
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.680867743116429e-05,
+      "loss": 0.0763,
+      "step": 4261
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.676879970220447e-05,
+      "loss": 0.073,
+      "step": 4262
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.672892248761132e-05,
+      "loss": 0.076,
+      "step": 4263
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.668904579373281e-05,
+      "loss": 0.0768,
+      "step": 4264
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.66491696269167e-05,
+      "loss": 0.0718,
+      "step": 4265
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.66092939935109e-05,
+      "loss": 0.0731,
+      "step": 4266
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.656941889986304e-05,
+      "loss": 0.0704,
+      "step": 4267
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.652954435232076e-05,
+      "loss": 0.0731,
+      "step": 4268
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.648967035723155e-05,
+      "loss": 0.0756,
+      "step": 4269
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.644979692094292e-05,
+      "loss": 0.0735,
+      "step": 4270
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.640992404980218e-05,
+      "loss": 0.0731,
+      "step": 4271
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.637005175015657e-05,
+      "loss": 0.0735,
+      "step": 4272
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.633018002835332e-05,
+      "loss": 0.0754,
+      "step": 4273
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.629030889073949e-05,
+      "loss": 0.0745,
+      "step": 4274
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.625043834366204e-05,
+      "loss": 0.0762,
+      "step": 4275
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.621056839346785e-05,
+      "loss": 0.0733,
+      "step": 4276
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.617069904650377e-05,
+      "loss": 0.074,
+      "step": 4277
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.613083030911647e-05,
+      "loss": 0.0721,
+      "step": 4278
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.609096218765255e-05,
+      "loss": 0.0738,
+      "step": 4279
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.605109468845854e-05,
+      "loss": 0.0733,
+      "step": 4280
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.601122781788081e-05,
+      "loss": 0.0726,
+      "step": 4281
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.597136158226568e-05,
+      "loss": 0.0722,
+      "step": 4282
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.59314959879594e-05,
+      "loss": 0.078,
+      "step": 4283
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.589163104130804e-05,
+      "loss": 0.0735,
+      "step": 4284
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.58517667486576e-05,
+      "loss": 0.0706,
+      "step": 4285
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.581190311635391e-05,
+      "loss": 0.074,
+      "step": 4286
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.577204015074288e-05,
+      "loss": 0.0741,
+      "step": 4287
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.573217785817014e-05,
+      "loss": 0.0728,
+      "step": 4288
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.569231624498124e-05,
+      "loss": 0.073,
+      "step": 4289
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.565245531752171e-05,
+      "loss": 0.0739,
+      "step": 4290
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.561259508213687e-05,
+      "loss": 0.074,
+      "step": 4291
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.5572735545172e-05,
+      "loss": 0.0782,
+      "step": 4292
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.553287671297215e-05,
+      "loss": 0.0727,
+      "step": 4293
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.54930185918825e-05,
+      "loss": 0.0747,
+      "step": 4294
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.545316118824784e-05,
+      "loss": 0.0755,
+      "step": 4295
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.541330450841301e-05,
+      "loss": 0.0746,
+      "step": 4296
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.53734485587227e-05,
+      "loss": 0.0752,
+      "step": 4297
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.533359334552148e-05,
+      "loss": 0.0758,
+      "step": 4298
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.52937388751538e-05,
+      "loss": 0.0756,
+      "step": 4299
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.525388515396395e-05,
+      "loss": 0.0724,
+      "step": 4300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.521403218829622e-05,
+      "loss": 0.0764,
+      "step": 4301
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.517417998449467e-05,
+      "loss": 0.0726,
+      "step": 4302
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.513432854890321e-05,
+      "loss": 0.0756,
+      "step": 4303
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.509447788786582e-05,
+      "loss": 0.0743,
+      "step": 4304
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.505462800772612e-05,
+      "loss": 0.0781,
+      "step": 4305
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.501477891482773e-05,
+      "loss": 0.0697,
+      "step": 4306
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.497493061551416e-05,
+      "loss": 0.0726,
+      "step": 4307
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.493508311612874e-05,
+      "loss": 0.0751,
+      "step": 4308
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.48952364230147e-05,
+      "loss": 0.0777,
+      "step": 4309
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.485539054251512e-05,
+      "loss": 0.0755,
+      "step": 4310
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.481554548097297e-05,
+      "loss": 0.0763,
+      "step": 4311
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.477570124473113e-05,
+      "loss": 0.0754,
+      "step": 4312
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.473585784013219e-05,
+      "loss": 0.0697,
+      "step": 4313
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.469601527351886e-05,
+      "loss": 0.075,
+      "step": 4314
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.465617355123351e-05,
+      "loss": 0.0722,
+      "step": 4315
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.461633267961844e-05,
+      "loss": 0.0737,
+      "step": 4316
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.45764926650158e-05,
+      "loss": 0.0722,
+      "step": 4317
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.453665351376767e-05,
+      "loss": 0.0754,
+      "step": 4318
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.449681523221592e-05,
+      "loss": 0.0725,
+      "step": 4319
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.445697782670229e-05,
+      "loss": 0.0764,
+      "step": 4320
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.441714130356843e-05,
+      "loss": 0.0746,
+      "step": 4321
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.437730566915582e-05,
+      "loss": 0.0742,
+      "step": 4322
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.433747092980572e-05,
+      "loss": 0.073,
+      "step": 4323
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.429763709185942e-05,
+      "loss": 0.0724,
+      "step": 4324
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.425780416165792e-05,
+      "loss": 0.0768,
+      "step": 4325
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.421797214554214e-05,
+      "loss": 0.0734,
+      "step": 4326
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.417814104985278e-05,
+      "loss": 0.0728,
+      "step": 4327
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.413831088093052e-05,
+      "loss": 0.0787,
+      "step": 4328
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.409848164511582e-05,
+      "loss": 0.0727,
+      "step": 4329
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.405865334874895e-05,
+      "loss": 0.0745,
+      "step": 4330
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.401882599817012e-05,
+      "loss": 0.0734,
+      "step": 4331
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.397899959971936e-05,
+      "loss": 0.074,
+      "step": 4332
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.393917415973649e-05,
+      "loss": 0.0735,
+      "step": 4333
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.38993496845612e-05,
+      "loss": 0.0778,
+      "step": 4334
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.385952618053312e-05,
+      "loss": 0.0751,
+      "step": 4335
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.38197036539916e-05,
+      "loss": 0.0707,
+      "step": 4336
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.377988211127591e-05,
+      "loss": 0.073,
+      "step": 4337
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.374006155872514e-05,
+      "loss": 0.0741,
+      "step": 4338
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.370024200267822e-05,
+      "loss": 0.0712,
+      "step": 4339
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.366042344947395e-05,
+      "loss": 0.0743,
+      "step": 4340
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.362060590545087e-05,
+      "loss": 0.0707,
+      "step": 4341
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.358078937694754e-05,
+      "loss": 0.0773,
+      "step": 4342
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.354097387030217e-05,
+      "loss": 0.0745,
+      "step": 4343
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.35011593918529e-05,
+      "loss": 0.0737,
+      "step": 4344
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.346134594793774e-05,
+      "loss": 0.074,
+      "step": 4345
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.342153354489446e-05,
+      "loss": 0.0705,
+      "step": 4346
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.338172218906069e-05,
+      "loss": 0.0727,
+      "step": 4347
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.334191188677394e-05,
+      "loss": 0.0759,
+      "step": 4348
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.330210264437149e-05,
+      "loss": 0.0759,
+      "step": 4349
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.326229446819048e-05,
+      "loss": 0.0761,
+      "step": 4350
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.322248736456778e-05,
+      "loss": 0.0761,
+      "step": 4351
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.318268133984034e-05,
+      "loss": 0.0733,
+      "step": 4352
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.314287640034471e-05,
+      "loss": 0.0746,
+      "step": 4353
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.310307255241729e-05,
+      "loss": 0.0743,
+      "step": 4354
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.306326980239442e-05,
+      "loss": 0.0732,
+      "step": 4355
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.302346815661217e-05,
+      "loss": 0.0722,
+      "step": 4356
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.298366762140648e-05,
+      "loss": 0.0734,
+      "step": 4357
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.294386820311306e-05,
+      "loss": 0.0747,
+      "step": 4358
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.290406990806754e-05,
+      "loss": 0.0736,
+      "step": 4359
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.28642727426053e-05,
+      "loss": 0.0759,
+      "step": 4360
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.282447671306145e-05,
+      "loss": 0.0754,
+      "step": 4361
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.278468182577117e-05,
+      "loss": 0.0747,
+      "step": 4362
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.274488808706922e-05,
+      "loss": 0.0746,
+      "step": 4363
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.270509550329027e-05,
+      "loss": 0.0755,
+      "step": 4364
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.26653040807688e-05,
+      "loss": 0.0719,
+      "step": 4365
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.262551382583915e-05,
+      "loss": 0.0734,
+      "step": 4366
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.25857247448354e-05,
+      "loss": 0.0746,
+      "step": 4367
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.254593684409144e-05,
+      "loss": 0.0736,
+      "step": 4368
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.250615012994106e-05,
+      "loss": 0.0748,
+      "step": 4369
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.246636460871782e-05,
+      "loss": 0.076,
+      "step": 4370
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.242658028675497e-05,
+      "loss": 0.0777,
+      "step": 4371
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.23867971703858e-05,
+      "loss": 0.0733,
+      "step": 4372
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.234701526594324e-05,
+      "loss": 0.0765,
+      "step": 4373
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.230723457976006e-05,
+      "loss": 0.0755,
+      "step": 4374
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.226745511816882e-05,
+      "loss": 0.075,
+      "step": 4375
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.222767688750196e-05,
+      "loss": 0.0709,
+      "step": 4376
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.218789989409167e-05,
+      "loss": 0.0736,
+      "step": 4377
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.214812414426994e-05,
+      "loss": 0.0743,
+      "step": 4378
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.210834964436856e-05,
+      "loss": 0.0732,
+      "step": 4379
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.206857640071919e-05,
+      "loss": 0.0775,
+      "step": 4380
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.202880441965316e-05,
+      "loss": 0.0735,
+      "step": 4381
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.198903370750168e-05,
+      "loss": 0.0743,
+      "step": 4382
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.194926427059579e-05,
+      "loss": 0.0756,
+      "step": 4383
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.190949611526625e-05,
+      "loss": 0.073,
+      "step": 4384
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.186972924784365e-05,
+      "loss": 0.0716,
+      "step": 4385
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.182996367465843e-05,
+      "loss": 0.0754,
+      "step": 4386
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.179019940204072e-05,
+      "loss": 0.0748,
+      "step": 4387
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.175043643632051e-05,
+      "loss": 0.0757,
+      "step": 4388
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.171067478382757e-05,
+      "loss": 0.0731,
+      "step": 4389
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.16709144508915e-05,
+      "loss": 0.076,
+      "step": 4390
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.163115544384156e-05,
+      "loss": 0.0723,
+      "step": 4391
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.159139776900691e-05,
+      "loss": 0.0747,
+      "step": 4392
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.155164143271653e-05,
+      "loss": 0.0738,
+      "step": 4393
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.151188644129909e-05,
+      "loss": 0.0733,
+      "step": 4394
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.147213280108307e-05,
+      "loss": 0.074,
+      "step": 4395
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.143238051839678e-05,
+      "loss": 0.0749,
+      "step": 4396
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.139262959956829e-05,
+      "loss": 0.0757,
+      "step": 4397
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.135288005092546e-05,
+      "loss": 0.0738,
+      "step": 4398
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.131313187879583e-05,
+      "loss": 0.0773,
+      "step": 4399
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.127338508950696e-05,
+      "loss": 0.0734,
+      "step": 4400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.123363968938592e-05,
+      "loss": 0.0741,
+      "step": 4401
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.119389568475971e-05,
+      "loss": 0.0774,
+      "step": 4402
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.115415308195511e-05,
+      "loss": 0.0727,
+      "step": 4403
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.111441188729862e-05,
+      "loss": 0.075,
+      "step": 4404
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.107467210711656e-05,
+      "loss": 0.0758,
+      "step": 4405
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.103493374773496e-05,
+      "loss": 0.0752,
+      "step": 4406
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.099519681547972e-05,
+      "loss": 0.0735,
+      "step": 4407
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.095546131667647e-05,
+      "loss": 0.0713,
+      "step": 4408
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.091572725765051e-05,
+      "loss": 0.0729,
+      "step": 4409
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.087599464472714e-05,
+      "loss": 0.0757,
+      "step": 4410
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.08362634842312e-05,
+      "loss": 0.0752,
+      "step": 4411
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.07965337824874e-05,
+      "loss": 0.0741,
+      "step": 4412
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.075680554582024e-05,
+      "loss": 0.072,
+      "step": 4413
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.071707878055397e-05,
+      "loss": 0.0779,
+      "step": 4414
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.067735349301257e-05,
+      "loss": 0.0724,
+      "step": 4415
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.063762968951978e-05,
+      "loss": 0.0726,
+      "step": 4416
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.05979073763992e-05,
+      "loss": 0.0754,
+      "step": 4417
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.05581865599741e-05,
+      "loss": 0.0727,
+      "step": 4418
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.051846724656746e-05,
+      "loss": 0.0734,
+      "step": 4419
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.047874944250225e-05,
+      "loss": 0.075,
+      "step": 4420
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.043903315410092e-05,
+      "loss": 0.0766,
+      "step": 4421
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.039931838768584e-05,
+      "loss": 0.0767,
+      "step": 4422
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.03596051495791e-05,
+      "loss": 0.0718,
+      "step": 4423
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.031989344610258e-05,
+      "loss": 0.0748,
+      "step": 4424
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.028018328357787e-05,
+      "loss": 0.0736,
+      "step": 4425
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.024047466832631e-05,
+      "loss": 0.0719,
+      "step": 4426
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.020076760666905e-05,
+      "loss": 0.0728,
+      "step": 4427
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.016106210492696e-05,
+      "loss": 0.0749,
+      "step": 4428
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.012135816942058e-05,
+      "loss": 0.0761,
+      "step": 4429
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.008165580647039e-05,
+      "loss": 0.0739,
+      "step": 4430
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.004195502239645e-05,
+      "loss": 0.0741,
+      "step": 4431
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.000225582351863e-05,
+      "loss": 0.0745,
+      "step": 4432
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.996255821615653e-05,
+      "loss": 0.073,
+      "step": 4433
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.992286220662955e-05,
+      "loss": 0.0744,
+      "step": 4434
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.98831678012568e-05,
+      "loss": 0.0734,
+      "step": 4435
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.984347500635707e-05,
+      "loss": 0.0745,
+      "step": 4436
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.980378382824903e-05,
+      "loss": 0.077,
+      "step": 4437
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.976409427325102e-05,
+      "loss": 0.0752,
+      "step": 4438
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.972440634768106e-05,
+      "loss": 0.0735,
+      "step": 4439
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.968472005785697e-05,
+      "loss": 0.0738,
+      "step": 4440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.964503541009638e-05,
+      "loss": 0.0732,
+      "step": 4441
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.960535241071654e-05,
+      "loss": 0.0734,
+      "step": 4442
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.956567106603447e-05,
+      "loss": 0.0778,
+      "step": 4443
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 8.952599138236701e-05,
+      "loss": 0.0758,
+      "step": 4444
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.948631336603061e-05,
+      "loss": 0.0757,
+      "step": 4445
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.944663702334157e-05,
+      "loss": 0.0764,
+      "step": 4446
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.940696236061575e-05,
+      "loss": 0.077,
+      "step": 4447
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.936728938416901e-05,
+      "loss": 0.0748,
+      "step": 4448
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.932761810031672e-05,
+      "loss": 0.0722,
+      "step": 4449
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.9287948515374e-05,
+      "loss": 0.0718,
+      "step": 4450
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.924828063565585e-05,
+      "loss": 0.0722,
+      "step": 4451
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.920861446747685e-05,
+      "loss": 0.0739,
+      "step": 4452
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.916895001715133e-05,
+      "loss": 0.0732,
+      "step": 4453
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.912928729099344e-05,
+      "loss": 0.0765,
+      "step": 4454
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.908962629531695e-05,
+      "loss": 0.0707,
+      "step": 4455
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.90499670364354e-05,
+      "loss": 0.0743,
+      "step": 4456
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.901030952066202e-05,
+      "loss": 0.0713,
+      "step": 4457
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 8.897065375430987e-05,
+      "loss": 0.0754,
+      "step": 4458
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.893099974369156e-05,
+      "loss": 0.0765,
+      "step": 4459
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.889134749511955e-05,
+      "loss": 0.0757,
+      "step": 4460
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.8851697014906e-05,
+      "loss": 0.0725,
+      "step": 4461
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.881204830936275e-05,
+      "loss": 0.075,
+      "step": 4462
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.877240138480138e-05,
+      "loss": 0.0734,
+      "step": 4463
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.873275624753317e-05,
+      "loss": 0.0715,
+      "step": 4464
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.869311290386916e-05,
+      "loss": 0.074,
+      "step": 4465
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.865347136012008e-05,
+      "loss": 0.077,
+      "step": 4466
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.86138316225963e-05,
+      "loss": 0.0745,
+      "step": 4467
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.857419369760805e-05,
+      "loss": 0.0744,
+      "step": 4468
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.853455759146515e-05,
+      "loss": 0.0726,
+      "step": 4469
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.849492331047718e-05,
+      "loss": 0.0765,
+      "step": 4470
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 8.845529086095341e-05,
+      "loss": 0.077,
+      "step": 4471
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.841566024920286e-05,
+      "loss": 0.0736,
+      "step": 4472
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.83760314815342e-05,
+      "loss": 0.0734,
+      "step": 4473
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.833640456425582e-05,
+      "loss": 0.0738,
+      "step": 4474
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.829677950367589e-05,
+      "loss": 0.0719,
+      "step": 4475
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.82571563061022e-05,
+      "loss": 0.0746,
+      "step": 4476
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.821753497784218e-05,
+      "loss": 0.0746,
+      "step": 4477
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.81779155252032e-05,
+      "loss": 0.0736,
+      "step": 4478
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.813829795449207e-05,
+      "loss": 0.0746,
+      "step": 4479
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.809868227201545e-05,
+      "loss": 0.073,
+      "step": 4480
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.805906848407963e-05,
+      "loss": 0.0752,
+      "step": 4481
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.801945659699068e-05,
+      "loss": 0.0766,
+      "step": 4482
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.79798466170543e-05,
+      "loss": 0.0726,
+      "step": 4483
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.794023855057586e-05,
+      "loss": 0.0714,
+      "step": 4484
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 8.790063240386053e-05,
+      "loss": 0.0737,
+      "step": 4485
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.78610281832131e-05,
+      "loss": 0.076,
+      "step": 4486
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.782142589493805e-05,
+      "loss": 0.0739,
+      "step": 4487
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.778182554533952e-05,
+      "loss": 0.0729,
+      "step": 4488
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.774222714072149e-05,
+      "loss": 0.0725,
+      "step": 4489
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.770263068738747e-05,
+      "loss": 0.0745,
+      "step": 4490
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.76630361916407e-05,
+      "loss": 0.0734,
+      "step": 4491
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.76234436597842e-05,
+      "loss": 0.0726,
+      "step": 4492
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.758385309812055e-05,
+      "loss": 0.0712,
+      "step": 4493
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.754426451295207e-05,
+      "loss": 0.0725,
+      "step": 4494
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.75046779105808e-05,
+      "loss": 0.0747,
+      "step": 4495
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.746509329730845e-05,
+      "loss": 0.073,
+      "step": 4496
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.742551067943633e-05,
+      "loss": 0.0778,
+      "step": 4497
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 8.73859300632655e-05,
+      "loss": 0.0729,
+      "step": 4498
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.734635145509677e-05,
+      "loss": 0.0765,
+      "step": 4499
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.730677486123048e-05,
+      "loss": 0.0765,
+      "step": 4500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.726720028796677e-05,
+      "loss": 0.0738,
+      "step": 4501
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.722762774160544e-05,
+      "loss": 0.0745,
+      "step": 4502
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.718805722844589e-05,
+      "loss": 0.0736,
+      "step": 4503
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.714848875478732e-05,
+      "loss": 0.0737,
+      "step": 4504
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.71089223269284e-05,
+      "loss": 0.0739,
+      "step": 4505
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.706935795116779e-05,
+      "loss": 0.0775,
+      "step": 4506
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.702979563380351e-05,
+      "loss": 0.0728,
+      "step": 4507
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.69902353811334e-05,
+      "loss": 0.0741,
+      "step": 4508
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.695067719945505e-05,
+      "loss": 0.0729,
+      "step": 4509
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.691112109506556e-05,
+      "loss": 0.0759,
+      "step": 4510
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.687156707426175e-05,
+      "loss": 0.0737,
+      "step": 4511
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 8.683201514334013e-05,
+      "loss": 0.0721,
+      "step": 4512
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.679246530859692e-05,
+      "loss": 0.0739,
+      "step": 4513
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.675291757632793e-05,
+      "loss": 0.0717,
+      "step": 4514
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.671337195282861e-05,
+      "loss": 0.0762,
+      "step": 4515
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.667382844439425e-05,
+      "loss": 0.0762,
+      "step": 4516
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.663428705731958e-05,
+      "loss": 0.0708,
+      "step": 4517
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.659474779789909e-05,
+      "loss": 0.0728,
+      "step": 4518
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.655521067242703e-05,
+      "loss": 0.0752,
+      "step": 4519
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.651567568719713e-05,
+      "loss": 0.0778,
+      "step": 4520
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.64761428485029e-05,
+      "loss": 0.0736,
+      "step": 4521
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.643661216263743e-05,
+      "loss": 0.0766,
+      "step": 4522
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.639708363589359e-05,
+      "loss": 0.0741,
+      "step": 4523
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.635755727456379e-05,
+      "loss": 0.0737,
+      "step": 4524
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 8.631803308494004e-05,
+      "loss": 0.0736,
+      "step": 4525
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.627851107331425e-05,
+      "loss": 0.0735,
+      "step": 4526
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.623899124597777e-05,
+      "loss": 0.0727,
+      "step": 4527
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.619947360922163e-05,
+      "loss": 0.0748,
+      "step": 4528
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.615995816933654e-05,
+      "loss": 0.078,
+      "step": 4529
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.612044493261291e-05,
+      "loss": 0.0745,
+      "step": 4530
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.608093390534073e-05,
+      "loss": 0.0741,
+      "step": 4531
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.604142509380965e-05,
+      "loss": 0.0741,
+      "step": 4532
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.600191850430901e-05,
+      "loss": 0.0756,
+      "step": 4533
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.596241414312775e-05,
+      "loss": 0.0748,
+      "step": 4534
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.592291201655446e-05,
+      "loss": 0.0749,
+      "step": 4535
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.588341213087734e-05,
+      "loss": 0.0766,
+      "step": 4536
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.584391449238439e-05,
+      "loss": 0.0754,
+      "step": 4537
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.580441910736305e-05,
+      "loss": 0.076,
+      "step": 4538
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.57649259821005e-05,
+      "loss": 0.0749,
+      "step": 4539
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.572543512288359e-05,
+      "loss": 0.0729,
+      "step": 4540
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.568594653599874e-05,
+      "loss": 0.0751,
+      "step": 4541
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.564646022773204e-05,
+      "loss": 0.0705,
+      "step": 4542
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.560697620436923e-05,
+      "loss": 0.074,
+      "step": 4543
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.55674944721957e-05,
+      "loss": 0.0731,
+      "step": 4544
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.552801503749638e-05,
+      "loss": 0.0735,
+      "step": 4545
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.548853790655589e-05,
+      "loss": 0.0723,
+      "step": 4546
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.54490630856586e-05,
+      "loss": 0.0732,
+      "step": 4547
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.540959058108835e-05,
+      "loss": 0.0727,
+      "step": 4548
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.537012039912865e-05,
+      "loss": 0.075,
+      "step": 4549
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.533065254606266e-05,
+      "loss": 0.073,
+      "step": 4550
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.52911870281732e-05,
+      "loss": 0.0721,
+      "step": 4551
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.52517238517427e-05,
+      "loss": 0.0731,
+      "step": 4552
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.521226302305311e-05,
+      "loss": 0.0749,
+      "step": 4553
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.517280454838618e-05,
+      "loss": 0.073,
+      "step": 4554
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.513334843402324e-05,
+      "loss": 0.0734,
+      "step": 4555
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.509389468624509e-05,
+      "loss": 0.074,
+      "step": 4556
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.50544433113324e-05,
+      "loss": 0.0771,
+      "step": 4557
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.501499431556525e-05,
+      "loss": 0.0763,
+      "step": 4558
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.497554770522345e-05,
+      "loss": 0.073,
+      "step": 4559
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.493610348658641e-05,
+      "loss": 0.0729,
+      "step": 4560
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.489666166593317e-05,
+      "loss": 0.0747,
+      "step": 4561
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.485722224954237e-05,
+      "loss": 0.0725,
+      "step": 4562
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.48177852436922e-05,
+      "loss": 0.0724,
+      "step": 4563
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.477835065466064e-05,
+      "loss": 0.0755,
+      "step": 4564
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.473891848872517e-05,
+      "loss": 0.0753,
+      "step": 4565
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.46994887521628e-05,
+      "loss": 0.0755,
+      "step": 4566
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.466006145125038e-05,
+      "loss": 0.0736,
+      "step": 4567
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.462063659226419e-05,
+      "loss": 0.0728,
+      "step": 4568
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.458121418148012e-05,
+      "loss": 0.0742,
+      "step": 4569
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.454179422517379e-05,
+      "loss": 0.0738,
+      "step": 4570
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.450237672962033e-05,
+      "loss": 0.0756,
+      "step": 4571
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.446296170109455e-05,
+      "loss": 0.0769,
+      "step": 4572
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.442354914587079e-05,
+      "loss": 0.0747,
+      "step": 4573
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.438413907022307e-05,
+      "loss": 0.0758,
+      "step": 4574
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.434473148042498e-05,
+      "loss": 0.0742,
+      "step": 4575
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.430532638274966e-05,
+      "loss": 0.073,
+      "step": 4576
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.426592378346995e-05,
+      "loss": 0.0708,
+      "step": 4577
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.422652368885825e-05,
+      "loss": 0.0758,
+      "step": 4578
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.418712610518657e-05,
+      "loss": 0.0771,
+      "step": 4579
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.414773103872647e-05,
+      "loss": 0.0704,
+      "step": 4580
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.41083384957492e-05,
+      "loss": 0.075,
+      "step": 4581
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.406894848252556e-05,
+      "loss": 0.0744,
+      "step": 4582
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.402956100532588e-05,
+      "loss": 0.0725,
+      "step": 4583
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.399017607042025e-05,
+      "loss": 0.0737,
+      "step": 4584
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.395079368407822e-05,
+      "loss": 0.0727,
+      "step": 4585
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.391141385256894e-05,
+      "loss": 0.0732,
+      "step": 4586
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.38720365821612e-05,
+      "loss": 0.0741,
+      "step": 4587
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.383266187912339e-05,
+      "loss": 0.0725,
+      "step": 4588
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.379328974972347e-05,
+      "loss": 0.0745,
+      "step": 4589
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.375392020022895e-05,
+      "loss": 0.0746,
+      "step": 4590
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.371455323690702e-05,
+      "loss": 0.0736,
+      "step": 4591
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.367518886602438e-05,
+      "loss": 0.0736,
+      "step": 4592
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.363582709384738e-05,
+      "loss": 0.0763,
+      "step": 4593
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.359646792664181e-05,
+      "loss": 0.0751,
+      "step": 4594
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.355711137067333e-05,
+      "loss": 0.0767,
+      "step": 4595
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.35177574322069e-05,
+      "loss": 0.0736,
+      "step": 4596
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.347840611750717e-05,
+      "loss": 0.073,
+      "step": 4597
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.343905743283842e-05,
+      "loss": 0.072,
+      "step": 4598
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.339971138446446e-05,
+      "loss": 0.0752,
+      "step": 4599
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.336036797864865e-05,
+      "loss": 0.0758,
+      "step": 4600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.3321027221654e-05,
+      "loss": 0.074,
+      "step": 4601
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.328168911974307e-05,
+      "loss": 0.0735,
+      "step": 4602
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.324235367917803e-05,
+      "loss": 0.0744,
+      "step": 4603
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.320302090622045e-05,
+      "loss": 0.0741,
+      "step": 4604
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.31636908071318e-05,
+      "loss": 0.0761,
+      "step": 4605
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.312436338817282e-05,
+      "loss": 0.0777,
+      "step": 4606
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.308503865560395e-05,
+      "loss": 0.0759,
+      "step": 4607
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.304571661568525e-05,
+      "loss": 0.0721,
+      "step": 4608
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.300639727467626e-05,
+      "loss": 0.0777,
+      "step": 4609
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.296708063883614e-05,
+      "loss": 0.0744,
+      "step": 4610
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.292776671442356e-05,
+      "loss": 0.0745,
+      "step": 4611
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.288845550769687e-05,
+      "loss": 0.0748,
+      "step": 4612
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.284914702491392e-05,
+      "loss": 0.0745,
+      "step": 4613
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.280984127233204e-05,
+      "loss": 0.0758,
+      "step": 4614
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.277053825620836e-05,
+      "loss": 0.0752,
+      "step": 4615
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.27312379827993e-05,
+      "loss": 0.0721,
+      "step": 4616
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.269194045836102e-05,
+      "loss": 0.0726,
+      "step": 4617
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.265264568914917e-05,
+      "loss": 0.0714,
+      "step": 4618
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.261335368141903e-05,
+      "loss": 0.0775,
+      "step": 4619
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.257406444142538e-05,
+      "loss": 0.0753,
+      "step": 4620
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.253477797542255e-05,
+      "loss": 0.0705,
+      "step": 4621
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.249549428966447e-05,
+      "loss": 0.0742,
+      "step": 4622
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.245621339040467e-05,
+      "loss": 0.0701,
+      "step": 4623
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.241693528389603e-05,
+      "loss": 0.0737,
+      "step": 4624
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.23776599763913e-05,
+      "loss": 0.0737,
+      "step": 4625
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.23383874741425e-05,
+      "loss": 0.0733,
+      "step": 4626
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.229911778340139e-05,
+      "loss": 0.0749,
+      "step": 4627
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.225985091041915e-05,
+      "loss": 0.0761,
+      "step": 4628
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.222058686144664e-05,
+      "loss": 0.077,
+      "step": 4629
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.218132564273415e-05,
+      "loss": 0.0744,
+      "step": 4630
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.214206726053159e-05,
+      "loss": 0.0736,
+      "step": 4631
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.210281172108845e-05,
+      "loss": 0.0785,
+      "step": 4632
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.20635590306537e-05,
+      "loss": 0.079,
+      "step": 4633
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.202430919547584e-05,
+      "loss": 0.0728,
+      "step": 4634
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.198506222180293e-05,
+      "loss": 0.0708,
+      "step": 4635
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.194581811588267e-05,
+      "loss": 0.076,
+      "step": 4636
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.190657688396222e-05,
+      "loss": 0.0762,
+      "step": 4637
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.186733853228823e-05,
+      "loss": 0.0738,
+      "step": 4638
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.182810306710702e-05,
+      "loss": 0.0742,
+      "step": 4639
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.178887049466437e-05,
+      "loss": 0.0731,
+      "step": 4640
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.174964082120564e-05,
+      "loss": 0.0744,
+      "step": 4641
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.171041405297558e-05,
+      "loss": 0.0746,
+      "step": 4642
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.167119019621877e-05,
+      "loss": 0.0745,
+      "step": 4643
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.163196925717905e-05,
+      "loss": 0.0745,
+      "step": 4644
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.159275124209991e-05,
+      "loss": 0.0728,
+      "step": 4645
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.155353615722442e-05,
+      "loss": 0.0741,
+      "step": 4646
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 8.151432400879509e-05,
+      "loss": 0.0738,
+      "step": 4647
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.147511480305399e-05,
+      "loss": 0.072,
+      "step": 4648
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.143590854624278e-05,
+      "loss": 0.0713,
+      "step": 4649
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.139670524460258e-05,
+      "loss": 0.0746,
+      "step": 4650
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.135750490437409e-05,
+      "loss": 0.0767,
+      "step": 4651
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.131830753179743e-05,
+      "loss": 0.0754,
+      "step": 4652
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.127911313311244e-05,
+      "loss": 0.0758,
+      "step": 4653
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.123992171455831e-05,
+      "loss": 0.0752,
+      "step": 4654
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.120073328237384e-05,
+      "loss": 0.073,
+      "step": 4655
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.116154784279735e-05,
+      "loss": 0.072,
+      "step": 4656
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.112236540206666e-05,
+      "loss": 0.0731,
+      "step": 4657
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.108318596641913e-05,
+      "loss": 0.0743,
+      "step": 4658
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.104400954209161e-05,
+      "loss": 0.0785,
+      "step": 4659
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.100483613532053e-05,
+      "loss": 0.0766,
+      "step": 4660
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 8.096566575234183e-05,
+      "loss": 0.0748,
+      "step": 4661
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.092649839939084e-05,
+      "loss": 0.0734,
+      "step": 4662
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.088733408270264e-05,
+      "loss": 0.0712,
+      "step": 4663
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.084817280851162e-05,
+      "loss": 0.0714,
+      "step": 4664
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.080901458305179e-05,
+      "loss": 0.0726,
+      "step": 4665
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.076985941255661e-05,
+      "loss": 0.0734,
+      "step": 4666
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.073070730325917e-05,
+      "loss": 0.0713,
+      "step": 4667
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.069155826139194e-05,
+      "loss": 0.0719,
+      "step": 4668
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.065241229318695e-05,
+      "loss": 0.0745,
+      "step": 4669
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.061326940487579e-05,
+      "loss": 0.0761,
+      "step": 4670
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.057412960268952e-05,
+      "loss": 0.0761,
+      "step": 4671
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.053499289285861e-05,
+      "loss": 0.0699,
+      "step": 4672
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.049585928161329e-05,
+      "loss": 0.0756,
+      "step": 4673
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 8.045672877518303e-05,
+      "loss": 0.0757,
+      "step": 4674
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.041760137979695e-05,
+      "loss": 0.0746,
+      "step": 4675
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.037847710168361e-05,
+      "loss": 0.076,
+      "step": 4676
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.033935594707117e-05,
+      "loss": 0.0759,
+      "step": 4677
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.030023792218716e-05,
+      "loss": 0.0731,
+      "step": 4678
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.026112303325871e-05,
+      "loss": 0.0758,
+      "step": 4679
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.022201128651243e-05,
+      "loss": 0.0731,
+      "step": 4680
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.018290268817445e-05,
+      "loss": 0.075,
+      "step": 4681
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.01437972444703e-05,
+      "loss": 0.0728,
+      "step": 4682
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.010469496162509e-05,
+      "loss": 0.0714,
+      "step": 4683
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.006559584586347e-05,
+      "loss": 0.0755,
+      "step": 4684
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 8.002649990340946e-05,
+      "loss": 0.0737,
+      "step": 4685
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.998740714048668e-05,
+      "loss": 0.0724,
+      "step": 4686
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.994831756331823e-05,
+      "loss": 0.0737,
+      "step": 4687
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.990923117812665e-05,
+      "loss": 0.074,
+      "step": 4688
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.987014799113397e-05,
+      "loss": 0.0752,
+      "step": 4689
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.983106800856184e-05,
+      "loss": 0.0745,
+      "step": 4690
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.979199123663126e-05,
+      "loss": 0.0762,
+      "step": 4691
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.975291768156272e-05,
+      "loss": 0.0711,
+      "step": 4692
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.971384734957625e-05,
+      "loss": 0.0704,
+      "step": 4693
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.967478024689143e-05,
+      "loss": 0.074,
+      "step": 4694
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.963571637972718e-05,
+      "loss": 0.075,
+      "step": 4695
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.959665575430199e-05,
+      "loss": 0.074,
+      "step": 4696
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.955759837683385e-05,
+      "loss": 0.0749,
+      "step": 4697
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.951854425354019e-05,
+      "loss": 0.0758,
+      "step": 4698
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.947949339063797e-05,
+      "loss": 0.0753,
+      "step": 4699
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.94404457943435e-05,
+      "loss": 0.0732,
+      "step": 4700
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.94014014708728e-05,
+      "loss": 0.0759,
+      "step": 4701
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.936236042644116e-05,
+      "loss": 0.0751,
+      "step": 4702
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.93233226672634e-05,
+      "loss": 0.0746,
+      "step": 4703
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.928428819955394e-05,
+      "loss": 0.0735,
+      "step": 4704
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.924525702952649e-05,
+      "loss": 0.0735,
+      "step": 4705
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.920622916339437e-05,
+      "loss": 0.0755,
+      "step": 4706
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.916720460737028e-05,
+      "loss": 0.0729,
+      "step": 4707
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.91281833676665e-05,
+      "loss": 0.0712,
+      "step": 4708
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.908916545049472e-05,
+      "loss": 0.0743,
+      "step": 4709
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.905015086206601e-05,
+      "loss": 0.0744,
+      "step": 4710
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.901113960859115e-05,
+      "loss": 0.075,
+      "step": 4711
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.897213169628013e-05,
+      "loss": 0.073,
+      "step": 4712
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.893312713134256e-05,
+      "loss": 0.0741,
+      "step": 4713
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.889412591998748e-05,
+      "loss": 0.0763,
+      "step": 4714
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.88551280684234e-05,
+      "loss": 0.0758,
+      "step": 4715
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.881613358285829e-05,
+      "loss": 0.0727,
+      "step": 4716
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.877714246949954e-05,
+      "loss": 0.0751,
+      "step": 4717
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.873815473455413e-05,
+      "loss": 0.0737,
+      "step": 4718
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.869917038422837e-05,
+      "loss": 0.0742,
+      "step": 4719
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.866018942472803e-05,
+      "loss": 0.0719,
+      "step": 4720
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.862121186225851e-05,
+      "loss": 0.0731,
+      "step": 4721
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.858223770302446e-05,
+      "loss": 0.0771,
+      "step": 4722
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.854326695323011e-05,
+      "loss": 0.0752,
+      "step": 4723
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.850429961907908e-05,
+      "loss": 0.0738,
+      "step": 4724
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.846533570677453e-05,
+      "loss": 0.0768,
+      "step": 4725
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.842637522251903e-05,
+      "loss": 0.075,
+      "step": 4726
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.838741817251454e-05,
+      "loss": 0.073,
+      "step": 4727
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.834846456296259e-05,
+      "loss": 0.0741,
+      "step": 4728
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.830951440006411e-05,
+      "loss": 0.0733,
+      "step": 4729
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.827056769001942e-05,
+      "loss": 0.0738,
+      "step": 4730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.823162443902844e-05,
+      "loss": 0.0709,
+      "step": 4731
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.819268465329037e-05,
+      "loss": 0.0776,
+      "step": 4732
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.815374833900397e-05,
+      "loss": 0.0755,
+      "step": 4733
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.81148155023674e-05,
+      "loss": 0.0755,
+      "step": 4734
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.807588614957828e-05,
+      "loss": 0.0744,
+      "step": 4735
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.803696028683369e-05,
+      "loss": 0.0744,
+      "step": 4736
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.799803792033012e-05,
+      "loss": 0.0777,
+      "step": 4737
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.795911905626356e-05,
+      "loss": 0.0742,
+      "step": 4738
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.79202037008294e-05,
+      "loss": 0.076,
+      "step": 4739
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.788129186022244e-05,
+      "loss": 0.0755,
+      "step": 4740
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.784238354063697e-05,
+      "loss": 0.075,
+      "step": 4741
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.780347874826672e-05,
+      "loss": 0.0729,
+      "step": 4742
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.776457748930484e-05,
+      "loss": 0.0755,
+      "step": 4743
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.772567976994391e-05,
+      "loss": 0.0719,
+      "step": 4744
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.768678559637601e-05,
+      "loss": 0.0759,
+      "step": 4745
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.764789497479256e-05,
+      "loss": 0.0772,
+      "step": 4746
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.760900791138449e-05,
+      "loss": 0.0757,
+      "step": 4747
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.757012441234205e-05,
+      "loss": 0.0744,
+      "step": 4748
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.753124448385513e-05,
+      "loss": 0.0757,
+      "step": 4749
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.749236813211287e-05,
+      "loss": 0.0741,
+      "step": 4750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.745349536330386e-05,
+      "loss": 0.0756,
+      "step": 4751
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.741462618361623e-05,
+      "loss": 0.0748,
+      "step": 4752
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.737576059923743e-05,
+      "loss": 0.0751,
+      "step": 4753
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.733689861635435e-05,
+      "loss": 0.0769,
+      "step": 4754
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.729804024115339e-05,
+      "loss": 0.0747,
+      "step": 4755
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.725918547982028e-05,
+      "loss": 0.0764,
+      "step": 4756
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.722033433854023e-05,
+      "loss": 0.0765,
+      "step": 4757
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.718148682349779e-05,
+      "loss": 0.0787,
+      "step": 4758
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.714264294087712e-05,
+      "loss": 0.0723,
+      "step": 4759
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.71038026968616e-05,
+      "loss": 0.0724,
+      "step": 4760
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.706496609763407e-05,
+      "loss": 0.0759,
+      "step": 4761
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.702613314937692e-05,
+      "loss": 0.0736,
+      "step": 4762
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.698730385827184e-05,
+      "loss": 0.0773,
+      "step": 4763
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.694847823049996e-05,
+      "loss": 0.0778,
+      "step": 4764
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.690965627224181e-05,
+      "loss": 0.0728,
+      "step": 4765
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.687083798967739e-05,
+      "loss": 0.0775,
+      "step": 4766
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.683202338898611e-05,
+      "loss": 0.0721,
+      "step": 4767
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.679321247634667e-05,
+      "loss": 0.0743,
+      "step": 4768
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.675440525793741e-05,
+      "loss": 0.0741,
+      "step": 4769
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.671560173993587e-05,
+      "loss": 0.075,
+      "step": 4770
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.667680192851911e-05,
+      "loss": 0.0724,
+      "step": 4771
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.663800582986356e-05,
+      "loss": 0.0769,
+      "step": 4772
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.65992134501451e-05,
+      "loss": 0.0751,
+      "step": 4773
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.656042479553895e-05,
+      "loss": 0.0765,
+      "step": 4774
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.65216398722198e-05,
+      "loss": 0.073,
+      "step": 4775
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.648285868636175e-05,
+      "loss": 0.0726,
+      "step": 4776
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.644408124413829e-05,
+      "loss": 0.0712,
+      "step": 4777
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.640530755172219e-05,
+      "loss": 0.0753,
+      "step": 4778
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.636653761528589e-05,
+      "loss": 0.0706,
+      "step": 4779
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.632777144100098e-05,
+      "loss": 0.0718,
+      "step": 4780
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.628900903503857e-05,
+      "loss": 0.0755,
+      "step": 4781
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.625025040356915e-05,
+      "loss": 0.0764,
+      "step": 4782
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.621149555276262e-05,
+      "loss": 0.0749,
+      "step": 4783
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.617274448878827e-05,
+      "loss": 0.0761,
+      "step": 4784
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.613399721781475e-05,
+      "loss": 0.0732,
+      "step": 4785
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.609525374601019e-05,
+      "loss": 0.0757,
+      "step": 4786
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.605651407954206e-05,
+      "loss": 0.0776,
+      "step": 4787
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.60177782245772e-05,
+      "loss": 0.0739,
+      "step": 4788
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.597904618728186e-05,
+      "loss": 0.0761,
+      "step": 4789
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.594031797382173e-05,
+      "loss": 0.0738,
+      "step": 4790
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.590159359036187e-05,
+      "loss": 0.0746,
+      "step": 4791
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.586287304306666e-05,
+      "loss": 0.0751,
+      "step": 4792
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.582415633809999e-05,
+      "loss": 0.0722,
+      "step": 4793
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.578544348162505e-05,
+      "loss": 0.0733,
+      "step": 4794
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.57467344798044e-05,
+      "loss": 0.075,
+      "step": 4795
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.570802933880011e-05,
+      "loss": 0.0765,
+      "step": 4796
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.566932806477352e-05,
+      "loss": 0.0741,
+      "step": 4797
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.563063066388537e-05,
+      "loss": 0.0756,
+      "step": 4798
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.559193714229577e-05,
+      "loss": 0.075,
+      "step": 4799
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.555324750616433e-05,
+      "loss": 0.0736,
+      "step": 4800
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.551456176164989e-05,
+      "loss": 0.0725,
+      "step": 4801
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.547587991491076e-05,
+      "loss": 0.0742,
+      "step": 4802
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.543720197210461e-05,
+      "loss": 0.073,
+      "step": 4803
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.539852793938848e-05,
+      "loss": 0.0759,
+      "step": 4804
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.535985782291881e-05,
+      "loss": 0.0766,
+      "step": 4805
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.532119162885133e-05,
+      "loss": 0.0738,
+      "step": 4806
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.528252936334132e-05,
+      "loss": 0.0777,
+      "step": 4807
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.524387103254325e-05,
+      "loss": 0.0736,
+      "step": 4808
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 7.520521664261103e-05,
+      "loss": 0.0761,
+      "step": 4809
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.516656619969801e-05,
+      "loss": 0.072,
+      "step": 4810
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.512791970995686e-05,
+      "loss": 0.075,
+      "step": 4811
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.508927717953959e-05,
+      "loss": 0.0744,
+      "step": 4812
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.505063861459759e-05,
+      "loss": 0.0759,
+      "step": 4813
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.501200402128166e-05,
+      "loss": 0.0713,
+      "step": 4814
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.497337340574197e-05,
+      "loss": 0.0766,
+      "step": 4815
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.493474677412794e-05,
+      "loss": 0.0748,
+      "step": 4816
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.489612413258858e-05,
+      "loss": 0.0774,
+      "step": 4817
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.485750548727201e-05,
+      "loss": 0.072,
+      "step": 4818
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.481889084432588e-05,
+      "loss": 0.0732,
+      "step": 4819
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.47802802098972e-05,
+      "loss": 0.0722,
+      "step": 4820
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.474167359013223e-05,
+      "loss": 0.0732,
+      "step": 4821
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.47030709911767e-05,
+      "loss": 0.0755,
+      "step": 4822
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 7.466447241917564e-05,
+      "loss": 0.073,
+      "step": 4823
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.462587788027349e-05,
+      "loss": 0.0726,
+      "step": 4824
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.458728738061403e-05,
+      "loss": 0.0725,
+      "step": 4825
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.454870092634028e-05,
+      "loss": 0.0719,
+      "step": 4826
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.451011852359486e-05,
+      "loss": 0.0753,
+      "step": 4827
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.447154017851952e-05,
+      "loss": 0.0758,
+      "step": 4828
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.443296589725546e-05,
+      "loss": 0.0723,
+      "step": 4829
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.439439568594322e-05,
+      "loss": 0.0723,
+      "step": 4830
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.435582955072274e-05,
+      "loss": 0.0731,
+      "step": 4831
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.431726749773322e-05,
+      "loss": 0.0729,
+      "step": 4832
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.427870953311325e-05,
+      "loss": 0.0751,
+      "step": 4833
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.424015566300082e-05,
+      "loss": 0.0702,
+      "step": 4834
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.420160589353321e-05,
+      "loss": 0.0737,
+      "step": 4835
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 7.416306023084703e-05,
+      "loss": 0.0749,
+      "step": 4836
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.412451868107828e-05,
+      "loss": 0.0764,
+      "step": 4837
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.408598125036232e-05,
+      "loss": 0.0754,
+      "step": 4838
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.404744794483378e-05,
+      "loss": 0.0735,
+      "step": 4839
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.400891877062672e-05,
+      "loss": 0.0728,
+      "step": 4840
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.397039373387448e-05,
+      "loss": 0.0746,
+      "step": 4841
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.393187284070978e-05,
+      "loss": 0.0747,
+      "step": 4842
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.389335609726464e-05,
+      "loss": 0.0723,
+      "step": 4843
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.385484350967048e-05,
+      "loss": 0.0757,
+      "step": 4844
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.381633508405802e-05,
+      "loss": 0.0718,
+      "step": 4845
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.377783082655727e-05,
+      "loss": 0.0732,
+      "step": 4846
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.373933074329765e-05,
+      "loss": 0.0741,
+      "step": 4847
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.370083484040792e-05,
+      "loss": 0.0759,
+      "step": 4848
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.366234312401611e-05,
+      "loss": 0.0759,
+      "step": 4849
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 7.362385560024962e-05,
+      "loss": 0.0752,
+      "step": 4850
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.358537227523521e-05,
+      "loss": 0.0732,
+      "step": 4851
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.354689315509894e-05,
+      "loss": 0.0744,
+      "step": 4852
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.350841824596621e-05,
+      "loss": 0.0745,
+      "step": 4853
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.346994755396166e-05,
+      "loss": 0.0737,
+      "step": 4854
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.343148108520948e-05,
+      "loss": 0.0731,
+      "step": 4855
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.339301884583296e-05,
+      "loss": 0.0722,
+      "step": 4856
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.335456084195479e-05,
+      "loss": 0.0764,
+      "step": 4857
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.331610707969707e-05,
+      "loss": 0.0752,
+      "step": 4858
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.327765756518114e-05,
+      "loss": 0.0709,
+      "step": 4859
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.323921230452764e-05,
+      "loss": 0.0725,
+      "step": 4860
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.32007713038566e-05,
+      "loss": 0.074,
+      "step": 4861
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.316233456928738e-05,
+      "loss": 0.0719,
+      "step": 4862
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.312390210693863e-05,
+      "loss": 0.0748,
+      "step": 4863
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 7.30854739229282e-05,
+      "loss": 0.0719,
+      "step": 4864
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.304705002337351e-05,
+      "loss": 0.0763,
+      "step": 4865
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.300863041439112e-05,
+      "loss": 0.0747,
+      "step": 4866
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.297021510209689e-05,
+      "loss": 0.0707,
+      "step": 4867
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.293180409260618e-05,
+      "loss": 0.0766,
+      "step": 4868
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.289339739203343e-05,
+      "loss": 0.0772,
+      "step": 4869
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.285499500649258e-05,
+      "loss": 0.0726,
+      "step": 4870
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.281659694209673e-05,
+      "loss": 0.0729,
+      "step": 4871
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.277820320495846e-05,
+      "loss": 0.0774,
+      "step": 4872
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.273981380118955e-05,
+      "loss": 0.0738,
+      "step": 4873
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.270142873690104e-05,
+      "loss": 0.0775,
+      "step": 4874
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.266304801820345e-05,
+      "loss": 0.0739,
+      "step": 4875
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.262467165120645e-05,
+      "loss": 0.0763,
+      "step": 4876
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 7.258629964201911e-05,
+      "loss": 0.0768,
+      "step": 4877
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.254793199674974e-05,
+      "loss": 0.0758,
+      "step": 4878
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.2509568721506e-05,
+      "loss": 0.0721,
+      "step": 4879
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.247120982239487e-05,
+      "loss": 0.0758,
+      "step": 4880
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.243285530552255e-05,
+      "loss": 0.074,
+      "step": 4881
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.239450517699468e-05,
+      "loss": 0.0762,
+      "step": 4882
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.235615944291609e-05,
+      "loss": 0.0769,
+      "step": 4883
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.231781810939084e-05,
+      "loss": 0.0719,
+      "step": 4884
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.227948118252254e-05,
+      "loss": 0.0713,
+      "step": 4885
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.224114866841391e-05,
+      "loss": 0.0747,
+      "step": 4886
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.220282057316696e-05,
+      "loss": 0.0735,
+      "step": 4887
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.216449690288304e-05,
+      "loss": 0.0769,
+      "step": 4888
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.212617766366285e-05,
+      "loss": 0.072,
+      "step": 4889
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.20878628616063e-05,
+      "loss": 0.0753,
+      "step": 4890
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 7.204955250281263e-05,
+      "loss": 0.0746,
+      "step": 4891
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.201124659338037e-05,
+      "loss": 0.0742,
+      "step": 4892
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.197294513940739e-05,
+      "loss": 0.0741,
+      "step": 4893
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.193464814699073e-05,
+      "loss": 0.0747,
+      "step": 4894
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.189635562222677e-05,
+      "loss": 0.0744,
+      "step": 4895
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.185806757121132e-05,
+      "loss": 0.0729,
+      "step": 4896
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.181978400003926e-05,
+      "loss": 0.0745,
+      "step": 4897
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.178150491480488e-05,
+      "loss": 0.0769,
+      "step": 4898
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.174323032160175e-05,
+      "loss": 0.0758,
+      "step": 4899
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.170496022652269e-05,
+      "loss": 0.0779,
+      "step": 4900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.166669463565983e-05,
+      "loss": 0.0728,
+      "step": 4901
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.162843355510451e-05,
+      "loss": 0.0742,
+      "step": 4902
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.159017699094754e-05,
+      "loss": 0.0735,
+      "step": 4903
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 7.15519249492788e-05,
+      "loss": 0.0746,
+      "step": 4904
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.15136774361875e-05,
+      "loss": 0.0756,
+      "step": 4905
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.147543445776228e-05,
+      "loss": 0.0742,
+      "step": 4906
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.143719602009087e-05,
+      "loss": 0.0729,
+      "step": 4907
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.139896212926033e-05,
+      "loss": 0.0768,
+      "step": 4908
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.136073279135706e-05,
+      "loss": 0.075,
+      "step": 4909
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.132250801246672e-05,
+      "loss": 0.0745,
+      "step": 4910
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.128428779867416e-05,
+      "loss": 0.0712,
+      "step": 4911
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.124607215606354e-05,
+      "loss": 0.0746,
+      "step": 4912
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.120786109071838e-05,
+      "loss": 0.0754,
+      "step": 4913
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.116965460872138e-05,
+      "loss": 0.073,
+      "step": 4914
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.113145271615449e-05,
+      "loss": 0.0728,
+      "step": 4915
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.109325541909906e-05,
+      "loss": 0.0752,
+      "step": 4916
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.105506272363555e-05,
+      "loss": 0.074,
+      "step": 4917
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 7.101687463584377e-05,
+      "loss": 0.0735,
+      "step": 4918
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.09786911618028e-05,
+      "loss": 0.0728,
+      "step": 4919
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.094051230759096e-05,
+      "loss": 0.0737,
+      "step": 4920
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.090233807928589e-05,
+      "loss": 0.0745,
+      "step": 4921
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.086416848296435e-05,
+      "loss": 0.0727,
+      "step": 4922
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.082600352470256e-05,
+      "loss": 0.0721,
+      "step": 4923
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.078784321057589e-05,
+      "loss": 0.0748,
+      "step": 4924
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.07496875466589e-05,
+      "loss": 0.0715,
+      "step": 4925
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.071153653902562e-05,
+      "loss": 0.0748,
+      "step": 4926
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.067339019374912e-05,
+      "loss": 0.0762,
+      "step": 4927
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.063524851690186e-05,
+      "loss": 0.0746,
+      "step": 4928
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.059711151455549e-05,
+      "loss": 0.0752,
+      "step": 4929
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.055897919278097e-05,
+      "loss": 0.0755,
+      "step": 4930
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 7.052085155764849e-05,
+      "loss": 0.0752,
+      "step": 4931
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.048272861522746e-05,
+      "loss": 0.0749,
+      "step": 4932
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.044461037158661e-05,
+      "loss": 0.0743,
+      "step": 4933
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.04064968327939e-05,
+      "loss": 0.0742,
+      "step": 4934
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.036838800491647e-05,
+      "loss": 0.0762,
+      "step": 4935
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.03302838940208e-05,
+      "loss": 0.0762,
+      "step": 4936
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.02921845061726e-05,
+      "loss": 0.0793,
+      "step": 4937
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.02540898474368e-05,
+      "loss": 0.0731,
+      "step": 4938
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.021599992387759e-05,
+      "loss": 0.0749,
+      "step": 4939
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.017791474155841e-05,
+      "loss": 0.0781,
+      "step": 4940
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.013983430654198e-05,
+      "loss": 0.0761,
+      "step": 4941
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.010175862489021e-05,
+      "loss": 0.0765,
+      "step": 4942
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.006368770266421e-05,
+      "loss": 0.0714,
+      "step": 4943
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 7.002562154592449e-05,
+      "loss": 0.0757,
+      "step": 4944
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 6.998756016073065e-05,
+      "loss": 0.0709,
+      "step": 4945
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.994950355314156e-05,
+      "loss": 0.0744,
+      "step": 4946
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.991145172921542e-05,
+      "loss": 0.0722,
+      "step": 4947
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.987340469500959e-05,
+      "loss": 0.0725,
+      "step": 4948
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.983536245658064e-05,
+      "loss": 0.0731,
+      "step": 4949
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.979732501998447e-05,
+      "loss": 0.0733,
+      "step": 4950
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.975929239127614e-05,
+      "loss": 0.0749,
+      "step": 4951
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.972126457651e-05,
+      "loss": 0.0757,
+      "step": 4952
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.96832415817395e-05,
+      "loss": 0.0739,
+      "step": 4953
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.964522341301755e-05,
+      "loss": 0.0744,
+      "step": 4954
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.96072100763961e-05,
+      "loss": 0.074,
+      "step": 4955
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.956920157792639e-05,
+      "loss": 0.0764,
+      "step": 4956
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.953119792365894e-05,
+      "loss": 0.0729,
+      "step": 4957
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 6.949319911964343e-05,
+      "loss": 0.0708,
+      "step": 4958
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.945520517192881e-05,
+      "loss": 0.0789,
+      "step": 4959
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.941721608656318e-05,
+      "loss": 0.0723,
+      "step": 4960
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.937923186959401e-05,
+      "loss": 0.076,
+      "step": 4961
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.934125252706791e-05,
+      "loss": 0.074,
+      "step": 4962
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.93032780650306e-05,
+      "loss": 0.0764,
+      "step": 4963
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.926530848952732e-05,
+      "loss": 0.0762,
+      "step": 4964
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.922734380660221e-05,
+      "loss": 0.0732,
+      "step": 4965
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.918938402229881e-05,
+      "loss": 0.0728,
+      "step": 4966
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.915142914265985e-05,
+      "loss": 0.0725,
+      "step": 4967
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.911347917372725e-05,
+      "loss": 0.0757,
+      "step": 4968
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.907553412154223e-05,
+      "loss": 0.0722,
+      "step": 4969
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.903759399214509e-05,
+      "loss": 0.0738,
+      "step": 4970
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.89996587915755e-05,
+      "loss": 0.0732,
+      "step": 4971
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 6.896172852587224e-05,
+      "loss": 0.0699,
+      "step": 4972
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.892380320107326e-05,
+      "loss": 0.0777,
+      "step": 4973
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.888588282321591e-05,
+      "loss": 0.077,
+      "step": 4974
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.884796739833658e-05,
+      "loss": 0.0708,
+      "step": 4975
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.881005693247096e-05,
+      "loss": 0.0739,
+      "step": 4976
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.877215143165386e-05,
+      "loss": 0.0752,
+      "step": 4977
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.873425090191944e-05,
+      "loss": 0.0729,
+      "step": 4978
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.869635534930095e-05,
+      "loss": 0.0773,
+      "step": 4979
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.865846477983087e-05,
+      "loss": 0.0729,
+      "step": 4980
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.862057919954095e-05,
+      "loss": 0.0753,
+      "step": 4981
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.85826986144621e-05,
+      "loss": 0.076,
+      "step": 4982
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.854482303062439e-05,
+      "loss": 0.0716,
+      "step": 4983
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.850695245405714e-05,
+      "loss": 0.0715,
+      "step": 4984
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 6.846908689078891e-05,
+      "loss": 0.0761,
+      "step": 4985
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.843122634684742e-05,
+      "loss": 0.0761,
+      "step": 4986
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.839337082825955e-05,
+      "loss": 0.0742,
+      "step": 4987
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.835552034105147e-05,
+      "loss": 0.0768,
+      "step": 4988
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.83176748912485e-05,
+      "loss": 0.0721,
+      "step": 4989
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.827983448487514e-05,
+      "loss": 0.0783,
+      "step": 4990
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.824199912795512e-05,
+      "loss": 0.0736,
+      "step": 4991
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.820416882651139e-05,
+      "loss": 0.0747,
+      "step": 4992
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.8166343586566e-05,
+      "loss": 0.0757,
+      "step": 4993
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.812852341414027e-05,
+      "loss": 0.0747,
+      "step": 4994
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.809070831525472e-05,
+      "loss": 0.0709,
+      "step": 4995
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.805289829592905e-05,
+      "loss": 0.0723,
+      "step": 4996
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.801509336218209e-05,
+      "loss": 0.0723,
+      "step": 4997
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.797729352003196e-05,
+      "loss": 0.0749,
+      "step": 4998
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 6.79394987754959e-05,
+      "loss": 0.0739,
+      "step": 4999
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.79017091345904e-05,
+      "loss": 0.0759,
+      "step": 5000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.7863924603331e-05,
+      "loss": 0.0724,
+      "step": 5001
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.782614518773264e-05,
+      "loss": 0.0718,
+      "step": 5002
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.778837089380927e-05,
+      "loss": 0.0714,
+      "step": 5003
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.775060172757408e-05,
+      "loss": 0.074,
+      "step": 5004
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.771283769503948e-05,
+      "loss": 0.0726,
+      "step": 5005
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.7675078802217e-05,
+      "loss": 0.0697,
+      "step": 5006
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.76373250551174e-05,
+      "loss": 0.0728,
+      "step": 5007
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.75995764597506e-05,
+      "loss": 0.0742,
+      "step": 5008
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.75618330221257e-05,
+      "loss": 0.0722,
+      "step": 5009
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.7524094748251e-05,
+      "loss": 0.0725,
+      "step": 5010
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.748636164413392e-05,
+      "loss": 0.0772,
+      "step": 5011
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 6.744863371578117e-05,
+      "loss": 0.0755,
+      "step": 5012
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.74109109691985e-05,
+      "loss": 0.0755,
+      "step": 5013
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.73731934103909e-05,
+      "loss": 0.0756,
+      "step": 5014
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.733548104536258e-05,
+      "loss": 0.0735,
+      "step": 5015
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.729777388011684e-05,
+      "loss": 0.0782,
+      "step": 5016
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.72600719206562e-05,
+      "loss": 0.078,
+      "step": 5017
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.722237517298232e-05,
+      "loss": 0.0753,
+      "step": 5018
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.718468364309608e-05,
+      "loss": 0.0765,
+      "step": 5019
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.714699733699752e-05,
+      "loss": 0.0759,
+      "step": 5020
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.710931626068573e-05,
+      "loss": 0.0698,
+      "step": 5021
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.707164042015921e-05,
+      "loss": 0.0721,
+      "step": 5022
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.703396982141536e-05,
+      "loss": 0.0755,
+      "step": 5023
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.699630447045092e-05,
+      "loss": 0.0728,
+      "step": 5024
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.695864437326172e-05,
+      "loss": 0.0736,
+      "step": 5025
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 6.692098953584281e-05,
+      "loss": 0.077,
+      "step": 5026
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.688333996418834e-05,
+      "loss": 0.0735,
+      "step": 5027
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.684569566429164e-05,
+      "loss": 0.0758,
+      "step": 5028
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.680805664214527e-05,
+      "loss": 0.0719,
+      "step": 5029
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.677042290374086e-05,
+      "loss": 0.0744,
+      "step": 5030
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.673279445506917e-05,
+      "loss": 0.0741,
+      "step": 5031
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.669517130212029e-05,
+      "loss": 0.0721,
+      "step": 5032
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.665755345088328e-05,
+      "loss": 0.0763,
+      "step": 5033
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.661994090734643e-05,
+      "loss": 0.0775,
+      "step": 5034
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.65823336774972e-05,
+      "loss": 0.0735,
+      "step": 5035
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.654473176732218e-05,
+      "loss": 0.073,
+      "step": 5036
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.650713518280717e-05,
+      "loss": 0.0739,
+      "step": 5037
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.646954392993702e-05,
+      "loss": 0.0739,
+      "step": 5038
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 6.643195801469584e-05,
+      "loss": 0.0747,
+      "step": 5039
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.63943774430668e-05,
+      "loss": 0.0723,
+      "step": 5040
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.635680222103227e-05,
+      "loss": 0.0737,
+      "step": 5041
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.631923235457372e-05,
+      "loss": 0.0752,
+      "step": 5042
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.628166784967184e-05,
+      "loss": 0.074,
+      "step": 5043
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.624410871230643e-05,
+      "loss": 0.0742,
+      "step": 5044
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.62065549484564e-05,
+      "loss": 0.0748,
+      "step": 5045
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.616900656409989e-05,
+      "loss": 0.0768,
+      "step": 5046
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.613146356521411e-05,
+      "loss": 0.0753,
+      "step": 5047
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.609392595777544e-05,
+      "loss": 0.0757,
+      "step": 5048
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.605639374775934e-05,
+      "loss": 0.0748,
+      "step": 5049
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.601886694114058e-05,
+      "loss": 0.0757,
+      "step": 5050
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.598134554389287e-05,
+      "loss": 0.0718,
+      "step": 5051
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.594382956198915e-05,
+      "loss": 0.0692,
+      "step": 5052
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 6.590631900140154e-05,
+      "loss": 0.075,
+      "step": 5053
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.586881386810124e-05,
+      "loss": 0.078,
+      "step": 5054
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.583131416805855e-05,
+      "loss": 0.0728,
+      "step": 5055
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.579381990724302e-05,
+      "loss": 0.0723,
+      "step": 5056
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.575633109162323e-05,
+      "loss": 0.075,
+      "step": 5057
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.571884772716697e-05,
+      "loss": 0.0741,
+      "step": 5058
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.568136981984102e-05,
+      "loss": 0.0775,
+      "step": 5059
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.564389737561153e-05,
+      "loss": 0.0776,
+      "step": 5060
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.560643040044355e-05,
+      "loss": 0.0767,
+      "step": 5061
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.556896890030136e-05,
+      "loss": 0.0713,
+      "step": 5062
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.55315128811484e-05,
+      "loss": 0.0757,
+      "step": 5063
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.549406234894719e-05,
+      "loss": 0.077,
+      "step": 5064
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.54566173096594e-05,
+      "loss": 0.0749,
+      "step": 5065
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 6.541917776924573e-05,
+      "loss": 0.0732,
+      "step": 5066
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.53817437336662e-05,
+      "loss": 0.072,
+      "step": 5067
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.534431520887978e-05,
+      "loss": 0.0725,
+      "step": 5068
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.530689220084458e-05,
+      "loss": 0.0719,
+      "step": 5069
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.526947471551798e-05,
+      "loss": 0.071,
+      "step": 5070
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.523206275885631e-05,
+      "loss": 0.0741,
+      "step": 5071
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.519465633681508e-05,
+      "loss": 0.0735,
+      "step": 5072
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.515725545534893e-05,
+      "loss": 0.0706,
+      "step": 5073
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.511986012041163e-05,
+      "loss": 0.0748,
+      "step": 5074
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.508247033795605e-05,
+      "loss": 0.0755,
+      "step": 5075
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.504508611393414e-05,
+      "loss": 0.0766,
+      "step": 5076
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.500770745429704e-05,
+      "loss": 0.075,
+      "step": 5077
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.497033436499498e-05,
+      "loss": 0.0751,
+      "step": 5078
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.493296685197719e-05,
+      "loss": 0.0709,
+      "step": 5079
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 6.489560492119225e-05,
+      "loss": 0.0721,
+      "step": 5080
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.485824857858763e-05,
+      "loss": 0.0761,
+      "step": 5081
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.482089783010997e-05,
+      "loss": 0.0713,
+      "step": 5082
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.478355268170507e-05,
+      "loss": 0.0746,
+      "step": 5083
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.474621313931783e-05,
+      "loss": 0.0706,
+      "step": 5084
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.470887920889223e-05,
+      "loss": 0.0728,
+      "step": 5085
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.467155089637135e-05,
+      "loss": 0.0755,
+      "step": 5086
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.46342282076974e-05,
+      "loss": 0.0735,
+      "step": 5087
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.459691114881172e-05,
+      "loss": 0.0754,
+      "step": 5088
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.455959972565464e-05,
+      "loss": 0.0749,
+      "step": 5089
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.45222939441657e-05,
+      "loss": 0.0745,
+      "step": 5090
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.448499381028354e-05,
+      "loss": 0.0741,
+      "step": 5091
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.444769932994587e-05,
+      "loss": 0.0748,
+      "step": 5092
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.441041050908946e-05,
+      "loss": 0.0739,
+      "step": 5093
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 6.437312735365029e-05,
+      "loss": 0.0738,
+      "step": 5094
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.433584986956334e-05,
+      "loss": 0.0732,
+      "step": 5095
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.429857806276271e-05,
+      "loss": 0.077,
+      "step": 5096
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.426131193918161e-05,
+      "loss": 0.0728,
+      "step": 5097
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.42240515047524e-05,
+      "loss": 0.0724,
+      "step": 5098
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.418679676540635e-05,
+      "loss": 0.0731,
+      "step": 5099
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.414954772707404e-05,
+      "loss": 0.0744,
+      "step": 5100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.411230439568503e-05,
+      "loss": 0.0734,
+      "step": 5101
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.4075066777168e-05,
+      "loss": 0.078,
+      "step": 5102
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.403783487745067e-05,
+      "loss": 0.0736,
+      "step": 5103
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.400060870245995e-05,
+      "loss": 0.0735,
+      "step": 5104
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.396338825812176e-05,
+      "loss": 0.0725,
+      "step": 5105
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.392617355036115e-05,
+      "loss": 0.0752,
+      "step": 5106
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 6.388896458510214e-05,
+      "loss": 0.0744,
+      "step": 5107
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.385176136826807e-05,
+      "loss": 0.0733,
+      "step": 5108
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.381456390578114e-05,
+      "loss": 0.073,
+      "step": 5109
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.377737220356273e-05,
+      "loss": 0.0739,
+      "step": 5110
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.374018626753331e-05,
+      "loss": 0.0726,
+      "step": 5111
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.370300610361242e-05,
+      "loss": 0.0749,
+      "step": 5112
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.366583171771864e-05,
+      "loss": 0.074,
+      "step": 5113
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.36286631157697e-05,
+      "loss": 0.0729,
+      "step": 5114
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.359150030368239e-05,
+      "loss": 0.0728,
+      "step": 5115
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.355434328737255e-05,
+      "loss": 0.0716,
+      "step": 5116
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.351719207275505e-05,
+      "loss": 0.0757,
+      "step": 5117
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.3480046665744e-05,
+      "loss": 0.0749,
+      "step": 5118
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.344290707225244e-05,
+      "loss": 0.072,
+      "step": 5119
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.340577329819251e-05,
+      "loss": 0.0768,
+      "step": 5120
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 6.336864534947546e-05,
+      "loss": 0.0758,
+      "step": 5121
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.33315232320116e-05,
+      "loss": 0.0745,
+      "step": 5122
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.32944069517103e-05,
+      "loss": 0.0751,
+      "step": 5123
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.325729651447998e-05,
+      "loss": 0.0756,
+      "step": 5124
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.322019192622821e-05,
+      "loss": 0.0735,
+      "step": 5125
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.318309319286157e-05,
+      "loss": 0.0759,
+      "step": 5126
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.314600032028564e-05,
+      "loss": 0.0723,
+      "step": 5127
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.310891331440524e-05,
+      "loss": 0.0729,
+      "step": 5128
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.307183218112411e-05,
+      "loss": 0.0745,
+      "step": 5129
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.303475692634511e-05,
+      "loss": 0.0754,
+      "step": 5130
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.29976875559701e-05,
+      "loss": 0.0732,
+      "step": 5131
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.296062407590016e-05,
+      "loss": 0.0747,
+      "step": 5132
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.292356649203528e-05,
+      "loss": 0.074,
+      "step": 5133
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 6.288651481027454e-05,
+      "loss": 0.0751,
+      "step": 5134
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.284946903651614e-05,
+      "loss": 0.0762,
+      "step": 5135
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.281242917665733e-05,
+      "loss": 0.073,
+      "step": 5136
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.277539523659433e-05,
+      "loss": 0.0719,
+      "step": 5137
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.273836722222249e-05,
+      "loss": 0.0784,
+      "step": 5138
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.270134513943624e-05,
+      "loss": 0.0739,
+      "step": 5139
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.266432899412901e-05,
+      "loss": 0.072,
+      "step": 5140
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.26273187921933e-05,
+      "loss": 0.0718,
+      "step": 5141
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.259031453952069e-05,
+      "loss": 0.0731,
+      "step": 5142
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.25533162420018e-05,
+      "loss": 0.0726,
+      "step": 5143
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.251632390552626e-05,
+      "loss": 0.0758,
+      "step": 5144
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.247933753598282e-05,
+      "loss": 0.0768,
+      "step": 5145
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.244235713925926e-05,
+      "loss": 0.0762,
+      "step": 5146
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.240538272124236e-05,
+      "loss": 0.0729,
+      "step": 5147
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 6.236841428781797e-05,
+      "loss": 0.0735,
+      "step": 5148
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.233145184487106e-05,
+      "loss": 0.0746,
+      "step": 5149
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.229449539828554e-05,
+      "loss": 0.0764,
+      "step": 5150
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.225754495394442e-05,
+      "loss": 0.0766,
+      "step": 5151
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.222060051772978e-05,
+      "loss": 0.0738,
+      "step": 5152
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.218366209552269e-05,
+      "loss": 0.0752,
+      "step": 5153
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.214672969320332e-05,
+      "loss": 0.0749,
+      "step": 5154
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.210980331665073e-05,
+      "loss": 0.0744,
+      "step": 5155
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.20728829717433e-05,
+      "loss": 0.0772,
+      "step": 5156
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.203596866435817e-05,
+      "loss": 0.0741,
+      "step": 5157
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.199906040037166e-05,
+      "loss": 0.0723,
+      "step": 5158
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.196215818565914e-05,
+      "loss": 0.0715,
+      "step": 5159
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.192526202609494e-05,
+      "loss": 0.0743,
+      "step": 5160
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 6.188837192755248e-05,
+      "loss": 0.0716,
+      "step": 5161
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.185148789590421e-05,
+      "loss": 0.0732,
+      "step": 5162
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.18146099370216e-05,
+      "loss": 0.0734,
+      "step": 5163
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.17777380567752e-05,
+      "loss": 0.0742,
+      "step": 5164
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.174087226103443e-05,
+      "loss": 0.0725,
+      "step": 5165
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.170401255566802e-05,
+      "loss": 0.0755,
+      "step": 5166
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.166715894654349e-05,
+      "loss": 0.0754,
+      "step": 5167
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.163031143952744e-05,
+      "loss": 0.0737,
+      "step": 5168
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.15934700404856e-05,
+      "loss": 0.0742,
+      "step": 5169
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.155663475528264e-05,
+      "loss": 0.0751,
+      "step": 5170
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.151980558978227e-05,
+      "loss": 0.0718,
+      "step": 5171
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.148298254984721e-05,
+      "loss": 0.0727,
+      "step": 5172
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.144616564133928e-05,
+      "loss": 0.0752,
+      "step": 5173
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.140935487011925e-05,
+      "loss": 0.0736,
+      "step": 5174
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.137255024204685e-05,
+      "loss": 0.0774,
+      "step": 5175
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.133575176298107e-05,
+      "loss": 0.0722,
+      "step": 5176
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.129895943877967e-05,
+      "loss": 0.0715,
+      "step": 5177
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.126217327529956e-05,
+      "loss": 0.0737,
+      "step": 5178
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.122539327839657e-05,
+      "loss": 0.0739,
+      "step": 5179
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.11886194539257e-05,
+      "loss": 0.0742,
+      "step": 5180
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.115185180774085e-05,
+      "loss": 0.0761,
+      "step": 5181
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.111509034569496e-05,
+      "loss": 0.0727,
+      "step": 5182
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.107833507364002e-05,
+      "loss": 0.0729,
+      "step": 5183
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.1041585997427e-05,
+      "loss": 0.0763,
+      "step": 5184
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.100484312290584e-05,
+      "loss": 0.0747,
+      "step": 5185
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.0968106455925655e-05,
+      "loss": 0.0754,
+      "step": 5186
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.093137600233438e-05,
+      "loss": 0.0724,
+      "step": 5187
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.089465176797906e-05,
+      "loss": 0.0759,
+      "step": 5188
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.085793375870571e-05,
+      "loss": 0.0747,
+      "step": 5189
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.0821221980359434e-05,
+      "loss": 0.0713,
+      "step": 5190
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.078451643878425e-05,
+      "loss": 0.0771,
+      "step": 5191
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.074781713982321e-05,
+      "loss": 0.0736,
+      "step": 5192
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.0711124089318426e-05,
+      "loss": 0.0718,
+      "step": 5193
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.067443729311097e-05,
+      "loss": 0.0735,
+      "step": 5194
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.063775675704088e-05,
+      "loss": 0.074,
+      "step": 5195
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.060108248694722e-05,
+      "loss": 0.0732,
+      "step": 5196
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.0564414488668165e-05,
+      "loss": 0.0757,
+      "step": 5197
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.0527752768040724e-05,
+      "loss": 0.0731,
+      "step": 5198
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.049109733090099e-05,
+      "loss": 0.0743,
+      "step": 5199
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.04544481830841e-05,
+      "loss": 0.0753,
+      "step": 5200
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.041780533042409e-05,
+      "loss": 0.0736,
+      "step": 5201
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.038116877875409e-05,
+      "loss": 0.0722,
+      "step": 5202
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.034453853390609e-05,
+      "loss": 0.0726,
+      "step": 5203
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.0307914601711305e-05,
+      "loss": 0.0732,
+      "step": 5204
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.027129698799969e-05,
+      "loss": 0.0764,
+      "step": 5205
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.023468569860034e-05,
+      "loss": 0.0733,
+      "step": 5206
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.019808073934134e-05,
+      "loss": 0.0771,
+      "step": 5207
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.016148211604974e-05,
+      "loss": 0.0765,
+      "step": 5208
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.012488983455153e-05,
+      "loss": 0.0725,
+      "step": 5209
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.008830390067182e-05,
+      "loss": 0.0763,
+      "step": 5210
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.005172432023458e-05,
+      "loss": 0.074,
+      "step": 5211
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.001515109906286e-05,
+      "loss": 0.0752,
+      "step": 5212
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.997858424297859e-05,
+      "loss": 0.0752,
+      "step": 5213
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.9942023757802843e-05,
+      "loss": 0.0772,
+      "step": 5214
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.990546964935554e-05,
+      "loss": 0.0749,
+      "step": 5215
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.98689219234556e-05,
+      "loss": 0.0729,
+      "step": 5216
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.983238058592107e-05,
+      "loss": 0.0746,
+      "step": 5217
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.979584564256879e-05,
+      "loss": 0.0754,
+      "step": 5218
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9759317099214706e-05,
+      "loss": 0.074,
+      "step": 5219
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.972279496167365e-05,
+      "loss": 0.0762,
+      "step": 5220
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.968627923575956e-05,
+      "loss": 0.0763,
+      "step": 5221
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.964976992728527e-05,
+      "loss": 0.0739,
+      "step": 5222
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9613267042062515e-05,
+      "loss": 0.0722,
+      "step": 5223
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9576770585902244e-05,
+      "loss": 0.0741,
+      "step": 5224
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.9540280564614126e-05,
+      "loss": 0.0738,
+      "step": 5225
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.950379698400691e-05,
+      "loss": 0.0741,
+      "step": 5226
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.946731984988843e-05,
+      "loss": 0.075,
+      "step": 5227
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.94308491680653e-05,
+      "loss": 0.0713,
+      "step": 5228
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.939438494434322e-05,
+      "loss": 0.0758,
+      "step": 5229
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.9357927184526816e-05,
+      "loss": 0.0742,
+      "step": 5230
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.9321475894419754e-05,
+      "loss": 0.0761,
+      "step": 5231
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.928503107982461e-05,
+      "loss": 0.0757,
+      "step": 5232
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.924859274654289e-05,
+      "loss": 0.0697,
+      "step": 5233
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.92121609003752e-05,
+      "loss": 0.0728,
+      "step": 5234
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.917573554712097e-05,
+      "loss": 0.0731,
+      "step": 5235
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.913931669257871e-05,
+      "loss": 0.0754,
+      "step": 5236
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.9102904342545786e-05,
+      "loss": 0.0751,
+      "step": 5237
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.906649850281865e-05,
+      "loss": 0.0777,
+      "step": 5238
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.903009917919262e-05,
+      "loss": 0.0744,
+      "step": 5239
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.8993706377462e-05,
+      "loss": 0.0749,
+      "step": 5240
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.895732010342012e-05,
+      "loss": 0.0774,
+      "step": 5241
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.8920940362859214e-05,
+      "loss": 0.0736,
+      "step": 5242
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.888456716157043e-05,
+      "loss": 0.0726,
+      "step": 5243
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.884820050534392e-05,
+      "loss": 0.074,
+      "step": 5244
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.8811840399968884e-05,
+      "loss": 0.0736,
+      "step": 5245
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.877548685123334e-05,
+      "loss": 0.0763,
+      "step": 5246
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.87391398649243e-05,
+      "loss": 0.0737,
+      "step": 5247
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.8702799446827797e-05,
+      "loss": 0.0748,
+      "step": 5248
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.8666465602728725e-05,
+      "loss": 0.0751,
+      "step": 5249
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.8630138338411e-05,
+      "loss": 0.0717,
+      "step": 5250
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.8593817659657477e-05,
+      "loss": 0.0756,
+      "step": 5251
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.855750357224998e-05,
+      "loss": 0.0762,
+      "step": 5252
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.85211960819692e-05,
+      "loss": 0.0731,
+      "step": 5253
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.8484895194594794e-05,
+      "loss": 0.0734,
+      "step": 5254
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.844860091590555e-05,
+      "loss": 0.0739,
+      "step": 5255
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.841231325167895e-05,
+      "loss": 0.0745,
+      "step": 5256
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.837603220769157e-05,
+      "loss": 0.0774,
+      "step": 5257
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.833975778971887e-05,
+      "loss": 0.0728,
+      "step": 5258
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.830349000353537e-05,
+      "loss": 0.0747,
+      "step": 5259
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.826722885491439e-05,
+      "loss": 0.0771,
+      "step": 5260
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.823097434962821e-05,
+      "loss": 0.0758,
+      "step": 5261
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.819472649344813e-05,
+      "loss": 0.0712,
+      "step": 5262
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.815848529214438e-05,
+      "loss": 0.0757,
+      "step": 5263
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.812225075148607e-05,
+      "loss": 0.0741,
+      "step": 5264
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.80860228772413e-05,
+      "loss": 0.0727,
+      "step": 5265
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.8049801675177115e-05,
+      "loss": 0.073,
+      "step": 5266
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.801358715105947e-05,
+      "loss": 0.0748,
+      "step": 5267
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.7977379310653156e-05,
+      "loss": 0.0774,
+      "step": 5268
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.79411781597222e-05,
+      "loss": 0.0779,
+      "step": 5269
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.790498370402926e-05,
+      "loss": 0.074,
+      "step": 5270
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.7868795949336005e-05,
+      "loss": 0.0753,
+      "step": 5271
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.783261490140315e-05,
+      "loss": 0.0732,
+      "step": 5272
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.779644056599025e-05,
+      "loss": 0.0756,
+      "step": 5273
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.7760272948855774e-05,
+      "loss": 0.0738,
+      "step": 5274
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.772411205575716e-05,
+      "loss": 0.0734,
+      "step": 5275
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.768795789245083e-05,
+      "loss": 0.0779,
+      "step": 5276
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.7651810464692016e-05,
+      "loss": 0.0721,
+      "step": 5277
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.761566977823487e-05,
+      "loss": 0.0748,
+      "step": 5278
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.757953583883271e-05,
+      "loss": 0.075,
+      "step": 5279
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.7543408652237484e-05,
+      "loss": 0.0742,
+      "step": 5280
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.750728822420019e-05,
+      "loss": 0.0719,
+      "step": 5281
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.747117456047077e-05,
+      "loss": 0.0749,
+      "step": 5282
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.743506766679812e-05,
+      "loss": 0.0753,
+      "step": 5283
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.739896754892995e-05,
+      "loss": 0.0762,
+      "step": 5284
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.7362874212612874e-05,
+      "loss": 0.0721,
+      "step": 5285
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.7326787663592654e-05,
+      "loss": 0.0738,
+      "step": 5286
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.7290707907613736e-05,
+      "loss": 0.0719,
+      "step": 5287
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.72546349504195e-05,
+      "loss": 0.0737,
+      "step": 5288
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.721856879775245e-05,
+      "loss": 0.0699,
+      "step": 5289
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.7182509455353814e-05,
+      "loss": 0.0737,
+      "step": 5290
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.714645692896372e-05,
+      "loss": 0.073,
+      "step": 5291
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.7110411224321326e-05,
+      "loss": 0.0737,
+      "step": 5292
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.7074372347164694e-05,
+      "loss": 0.074,
+      "step": 5293
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.703834030323074e-05,
+      "loss": 0.0754,
+      "step": 5294
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.700231509825522e-05,
+      "loss": 0.0756,
+      "step": 5295
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.696629673797304e-05,
+      "loss": 0.074,
+      "step": 5296
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.693028522811783e-05,
+      "loss": 0.0741,
+      "step": 5297
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.6894280574422075e-05,
+      "loss": 0.074,
+      "step": 5298
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.685828278261742e-05,
+      "loss": 0.0764,
+      "step": 5299
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.6822291858434175e-05,
+      "loss": 0.0763,
+      "step": 5300
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.678630780760163e-05,
+      "loss": 0.0774,
+      "step": 5301
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.6750330635848e-05,
+      "loss": 0.0741,
+      "step": 5302
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.671436034890048e-05,
+      "loss": 0.0747,
+      "step": 5303
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.667839695248498e-05,
+      "loss": 0.0747,
+      "step": 5304
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.664244045232647e-05,
+      "loss": 0.0724,
+      "step": 5305
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.6606490854148844e-05,
+      "loss": 0.0716,
+      "step": 5306
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.657054816367473e-05,
+      "loss": 0.0747,
+      "step": 5307
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.653461238662576e-05,
+      "loss": 0.0714,
+      "step": 5308
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.649868352872249e-05,
+      "loss": 0.076,
+      "step": 5309
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.646276159568437e-05,
+      "loss": 0.0738,
+      "step": 5310
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.6426846593229654e-05,
+      "loss": 0.0757,
+      "step": 5311
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.6390938527075596e-05,
+      "loss": 0.0737,
+      "step": 5312
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.635503740293837e-05,
+      "loss": 0.071,
+      "step": 5313
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.631914322653289e-05,
+      "loss": 0.0747,
+      "step": 5314
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.628325600357309e-05,
+      "loss": 0.0757,
+      "step": 5315
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.6247375739771814e-05,
+      "loss": 0.0748,
+      "step": 5316
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.6211502440840724e-05,
+      "loss": 0.0759,
+      "step": 5317
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.6175636112490336e-05,
+      "loss": 0.0752,
+      "step": 5318
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.6139776760430186e-05,
+      "loss": 0.0751,
+      "step": 5319
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.6103924390368655e-05,
+      "loss": 0.0756,
+      "step": 5320
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.606807900801292e-05,
+      "loss": 0.0744,
+      "step": 5321
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.603224061906915e-05,
+      "loss": 0.0732,
+      "step": 5322
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.599640922924243e-05,
+      "loss": 0.0743,
+      "step": 5323
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.596058484423656e-05,
+      "loss": 0.0717,
+      "step": 5324
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.592476746975444e-05,
+      "loss": 0.0744,
+      "step": 5325
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.588895711149764e-05,
+      "loss": 0.0738,
+      "step": 5326
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5853153775166824e-05,
+      "loss": 0.072,
+      "step": 5327
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5817357466461336e-05,
+      "loss": 0.0737,
+      "step": 5328
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.578156819107956e-05,
+      "loss": 0.0717,
+      "step": 5329
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.574578595471873e-05,
+      "loss": 0.0759,
+      "step": 5330
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5710010763074846e-05,
+      "loss": 0.0772,
+      "step": 5331
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.567424262184292e-05,
+      "loss": 0.0775,
+      "step": 5332
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.5638481536716824e-05,
+      "loss": 0.0717,
+      "step": 5333
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.56027275133892e-05,
+      "loss": 0.0756,
+      "step": 5334
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.556698055755173e-05,
+      "loss": 0.0734,
+      "step": 5335
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.553124067489479e-05,
+      "loss": 0.0779,
+      "step": 5336
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.54955078711078e-05,
+      "loss": 0.0776,
+      "step": 5337
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5459782151878894e-05,
+      "loss": 0.0734,
+      "step": 5338
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5424063522895206e-05,
+      "loss": 0.0761,
+      "step": 5339
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.538835198984275e-05,
+      "loss": 0.0753,
+      "step": 5340
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.535264755840623e-05,
+      "loss": 0.0717,
+      "step": 5341
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.531695023426948e-05,
+      "loss": 0.0749,
+      "step": 5342
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.528126002311495e-05,
+      "loss": 0.0745,
+      "step": 5343
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5245576930624135e-05,
+      "loss": 0.0722,
+      "step": 5344
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.520990096247736e-05,
+      "loss": 0.0775,
+      "step": 5345
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.517423212435372e-05,
+      "loss": 0.0778,
+      "step": 5346
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.513857042193132e-05,
+      "loss": 0.0703,
+      "step": 5347
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5102915860886996e-05,
+      "loss": 0.0768,
+      "step": 5348
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.506726844689658e-05,
+      "loss": 0.0733,
+      "step": 5349
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.503162818563459e-05,
+      "loss": 0.0752,
+      "step": 5350
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.499599508277459e-05,
+      "loss": 0.0732,
+      "step": 5351
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.496036914398893e-05,
+      "loss": 0.0737,
+      "step": 5352
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.492475037494875e-05,
+      "loss": 0.0775,
+      "step": 5353
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4889138781324156e-05,
+      "loss": 0.0753,
+      "step": 5354
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.48535343687841e-05,
+      "loss": 0.0723,
+      "step": 5355
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4817937142996275e-05,
+      "loss": 0.073,
+      "step": 5356
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4782347109627396e-05,
+      "loss": 0.0735,
+      "step": 5357
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.474676427434289e-05,
+      "loss": 0.0731,
+      "step": 5358
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.471118864280716e-05,
+      "loss": 0.0724,
+      "step": 5359
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.467562022068331e-05,
+      "loss": 0.0732,
+      "step": 5360
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.464005901363345e-05,
+      "loss": 0.073,
+      "step": 5361
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.460450502731851e-05,
+      "loss": 0.0796,
+      "step": 5362
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.456895826739816e-05,
+      "loss": 0.0787,
+      "step": 5363
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.453341873953104e-05,
+      "loss": 0.0734,
+      "step": 5364
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.4497886449374635e-05,
+      "loss": 0.0745,
+      "step": 5365
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.446236140258521e-05,
+      "loss": 0.0737,
+      "step": 5366
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.442684360481787e-05,
+      "loss": 0.0738,
+      "step": 5367
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.439133306172661e-05,
+      "loss": 0.0765,
+      "step": 5368
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.435582977896435e-05,
+      "loss": 0.0724,
+      "step": 5369
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.432033376218266e-05,
+      "loss": 0.0745,
+      "step": 5370
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.4284845017032114e-05,
+      "loss": 0.0728,
+      "step": 5371
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.424936354916211e-05,
+      "loss": 0.0721,
+      "step": 5372
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.4213889364220816e-05,
+      "loss": 0.0732,
+      "step": 5373
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.4178422467855196e-05,
+      "loss": 0.0696,
+      "step": 5374
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.41429628657113e-05,
+      "loss": 0.0748,
+      "step": 5375
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.4107510563433764e-05,
+      "loss": 0.0713,
+      "step": 5376
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.407206556666612e-05,
+      "loss": 0.0755,
+      "step": 5377
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.4036627881050804e-05,
+      "loss": 0.0727,
+      "step": 5378
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.400119751222909e-05,
+      "loss": 0.0795,
+      "step": 5379
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.396577446584098e-05,
+      "loss": 0.0719,
+      "step": 5380
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.3930358747525414e-05,
+      "loss": 0.0785,
+      "step": 5381
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.389495036292016e-05,
+      "loss": 0.0735,
+      "step": 5382
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.385954931766175e-05,
+      "loss": 0.073,
+      "step": 5383
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.382415561738554e-05,
+      "loss": 0.0744,
+      "step": 5384
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.378876926772588e-05,
+      "loss": 0.0726,
+      "step": 5385
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.3753390274315786e-05,
+      "loss": 0.0742,
+      "step": 5386
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.371801864278709e-05,
+      "loss": 0.077,
+      "step": 5387
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.3682654378770557e-05,
+      "loss": 0.0711,
+      "step": 5388
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.3647297487895784e-05,
+      "loss": 0.0733,
+      "step": 5389
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.361194797579108e-05,
+      "loss": 0.0747,
+      "step": 5390
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.357660584808364e-05,
+      "loss": 0.0772,
+      "step": 5391
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.354127111039957e-05,
+      "loss": 0.0759,
+      "step": 5392
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.350594376836367e-05,
+      "loss": 0.0765,
+      "step": 5393
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.347062382759951e-05,
+      "loss": 0.0733,
+      "step": 5394
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.343531129372976e-05,
+      "loss": 0.072,
+      "step": 5395
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.3400006172375637e-05,
+      "loss": 0.0732,
+      "step": 5396
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.336470846915727e-05,
+      "loss": 0.0743,
+      "step": 5397
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.332941818969361e-05,
+      "loss": 0.0697,
+      "step": 5398
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.32941353396025e-05,
+      "loss": 0.076,
+      "step": 5399
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.3258859924500426e-05,
+      "loss": 0.0675,
+      "step": 5400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.322359195000284e-05,
+      "loss": 0.0723,
+      "step": 5401
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.3188331421724015e-05,
+      "loss": 0.0762,
+      "step": 5402
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.315307834527692e-05,
+      "loss": 0.0746,
+      "step": 5403
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.311783272627333e-05,
+      "loss": 0.0746,
+      "step": 5404
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.30825945703241e-05,
+      "loss": 0.0735,
+      "step": 5405
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.304736388303857e-05,
+      "loss": 0.0752,
+      "step": 5406
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.301214067002503e-05,
+      "loss": 0.0746,
+      "step": 5407
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.297692493689059e-05,
+      "loss": 0.0714,
+      "step": 5408
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.2941716689241204e-05,
+      "loss": 0.0761,
+      "step": 5409
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.29065159326815e-05,
+      "loss": 0.0759,
+      "step": 5410
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.287132267281504e-05,
+      "loss": 0.078,
+      "step": 5411
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.2836136915244184e-05,
+      "loss": 0.0767,
+      "step": 5412
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.280095866557002e-05,
+      "loss": 0.0695,
+      "step": 5413
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.2765787929392475e-05,
+      "loss": 0.0724,
+      "step": 5414
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.2730624712310296e-05,
+      "loss": 0.0713,
+      "step": 5415
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.269546901992108e-05,
+      "loss": 0.0678,
+      "step": 5416
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.266032085782111e-05,
+      "loss": 0.0765,
+      "step": 5417
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.2625180231605544e-05,
+      "loss": 0.0678,
+      "step": 5418
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.259004714686838e-05,
+      "loss": 0.0726,
+      "step": 5419
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.255492160920229e-05,
+      "loss": 0.0726,
+      "step": 5420
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.251980362419886e-05,
+      "loss": 0.0714,
+      "step": 5421
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.248469319744847e-05,
+      "loss": 0.0695,
+      "step": 5422
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.244959033454022e-05,
+      "loss": 0.073,
+      "step": 5423
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.241449504106202e-05,
+      "loss": 0.0693,
+      "step": 5424
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.2379407322600624e-05,
+      "loss": 0.0721,
+      "step": 5425
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.2344327184741606e-05,
+      "loss": 0.0731,
+      "step": 5426
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.230925463306919e-05,
+      "loss": 0.0733,
+      "step": 5427
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.2274189673166565e-05,
+      "loss": 0.0733,
+      "step": 5428
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.2239132310615635e-05,
+      "loss": 0.0696,
+      "step": 5429
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.220408255099703e-05,
+      "loss": 0.0709,
+      "step": 5430
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.21690403998903e-05,
+      "loss": 0.071,
+      "step": 5431
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.2134005862873656e-05,
+      "loss": 0.0689,
+      "step": 5432
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.209897894552422e-05,
+      "loss": 0.0705,
+      "step": 5433
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.206395965341778e-05,
+      "loss": 0.0724,
+      "step": 5434
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.2028947992128994e-05,
+      "loss": 0.0711,
+      "step": 5435
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.1993943967231315e-05,
+      "loss": 0.0739,
+      "step": 5436
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.195894758429689e-05,
+      "loss": 0.0699,
+      "step": 5437
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.1923958848896757e-05,
+      "loss": 0.0686,
+      "step": 5438
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.1888977766600624e-05,
+      "loss": 0.0719,
+      "step": 5439
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.185400434297707e-05,
+      "loss": 0.0704,
+      "step": 5440
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.181903858359346e-05,
+      "loss": 0.0713,
+      "step": 5441
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.178408049401584e-05,
+      "loss": 0.0749,
+      "step": 5442
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.1749130079809183e-05,
+      "loss": 0.0709,
+      "step": 5443
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.171418734653707e-05,
+      "loss": 0.0762,
+      "step": 5444
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 5.167925229976198e-05,
+      "loss": 0.0716,
+      "step": 5445
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.1644324945045185e-05,
+      "loss": 0.0709,
+      "step": 5446
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.160940528794661e-05,
+      "loss": 0.0711,
+      "step": 5447
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.1574493334025084e-05,
+      "loss": 0.0712,
+      "step": 5448
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.15395890888381e-05,
+      "loss": 0.0711,
+      "step": 5449
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.150469255794199e-05,
+      "loss": 0.0688,
+      "step": 5450
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.146980374689192e-05,
+      "loss": 0.0731,
+      "step": 5451
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.1434922661241635e-05,
+      "loss": 0.0715,
+      "step": 5452
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.140004930654385e-05,
+      "loss": 0.0719,
+      "step": 5453
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.1365183688349925e-05,
+      "loss": 0.0728,
+      "step": 5454
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.133032581221007e-05,
+      "loss": 0.0705,
+      "step": 5455
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.1295475683673165e-05,
+      "loss": 0.0693,
+      "step": 5456
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.1260633308286944e-05,
+      "loss": 0.075,
+      "step": 5457
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.122579869159792e-05,
+      "loss": 0.074,
+      "step": 5458
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 5.119097183915124e-05,
+      "loss": 0.0744,
+      "step": 5459
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.115615275649095e-05,
+      "loss": 0.0722,
+      "step": 5460
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.1121341449159854e-05,
+      "loss": 0.0723,
+      "step": 5461
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.1086537922699404e-05,
+      "loss": 0.0705,
+      "step": 5462
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.1051742182649956e-05,
+      "loss": 0.0702,
+      "step": 5463
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.101695423455046e-05,
+      "loss": 0.0714,
+      "step": 5464
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.098217408393884e-05,
+      "loss": 0.0729,
+      "step": 5465
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.0947401736351554e-05,
+      "loss": 0.0717,
+      "step": 5466
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.0912637197323974e-05,
+      "loss": 0.0704,
+      "step": 5467
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.08778804723902e-05,
+      "loss": 0.0726,
+      "step": 5468
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.084313156708302e-05,
+      "loss": 0.0731,
+      "step": 5469
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.0808390486934046e-05,
+      "loss": 0.0715,
+      "step": 5470
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.077365723747366e-05,
+      "loss": 0.072,
+      "step": 5471
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 5.073893182423093e-05,
+      "loss": 0.0708,
+      "step": 5472
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.070421425273366e-05,
+      "loss": 0.0718,
+      "step": 5473
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0669504528508495e-05,
+      "loss": 0.0718,
+      "step": 5474
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0634802657080825e-05,
+      "loss": 0.0738,
+      "step": 5475
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0600108643974686e-05,
+      "loss": 0.0712,
+      "step": 5476
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.056542249471296e-05,
+      "loss": 0.0689,
+      "step": 5477
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0530744214817294e-05,
+      "loss": 0.072,
+      "step": 5478
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.049607380980799e-05,
+      "loss": 0.0762,
+      "step": 5479
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.046141128520407e-05,
+      "loss": 0.071,
+      "step": 5480
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.0426756646523534e-05,
+      "loss": 0.0703,
+      "step": 5481
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.039210989928287e-05,
+      "loss": 0.0725,
+      "step": 5482
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.035747104899739e-05,
+      "loss": 0.0718,
+      "step": 5483
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.032284010118118e-05,
+      "loss": 0.0715,
+      "step": 5484
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.028821706134712e-05,
+      "loss": 0.0703,
+      "step": 5485
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 5.025360193500667e-05,
+      "loss": 0.0709,
+      "step": 5486
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.0218994727670154e-05,
+      "loss": 0.0722,
+      "step": 5487
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.0184395444846674e-05,
+      "loss": 0.0723,
+      "step": 5488
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.014980409204395e-05,
+      "loss": 0.076,
+      "step": 5489
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.0115220674768407e-05,
+      "loss": 0.0734,
+      "step": 5490
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.008064519852545e-05,
+      "loss": 0.0723,
+      "step": 5491
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.004607766881899e-05,
+      "loss": 0.07,
+      "step": 5492
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 5.001151809115171e-05,
+      "loss": 0.0737,
+      "step": 5493
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.997696647102509e-05,
+      "loss": 0.0703,
+      "step": 5494
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.994242281393936e-05,
+      "loss": 0.0731,
+      "step": 5495
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.990788712539336e-05,
+      "loss": 0.0735,
+      "step": 5496
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.987335941088478e-05,
+      "loss": 0.0715,
+      "step": 5497
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.983883967591003e-05,
+      "loss": 0.0732,
+      "step": 5498
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.980432792596419e-05,
+      "loss": 0.072,
+      "step": 5499
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.976982416654102e-05,
+      "loss": 0.0706,
+      "step": 5500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.9735328403133244e-05,
+      "loss": 0.0726,
+      "step": 5501
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.970084064123207e-05,
+      "loss": 0.074,
+      "step": 5502
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.966636088632749e-05,
+      "loss": 0.0711,
+      "step": 5503
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.9631889143908273e-05,
+      "loss": 0.0754,
+      "step": 5504
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.959742541946195e-05,
+      "loss": 0.0739,
+      "step": 5505
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.956296971847462e-05,
+      "loss": 0.0735,
+      "step": 5506
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.9528522046431235e-05,
+      "loss": 0.0734,
+      "step": 5507
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.949408240881548e-05,
+      "loss": 0.0721,
+      "step": 5508
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.945965081110967e-05,
+      "loss": 0.0721,
+      "step": 5509
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.9425227258794826e-05,
+      "loss": 0.0713,
+      "step": 5510
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.939081175735087e-05,
+      "loss": 0.0735,
+      "step": 5511
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.9356404312256275e-05,
+      "loss": 0.0712,
+      "step": 5512
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.932200492898822e-05,
+      "loss": 0.0724,
+      "step": 5513
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.928761361302269e-05,
+      "loss": 0.0706,
+      "step": 5514
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.925323036983439e-05,
+      "loss": 0.0715,
+      "step": 5515
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.921885520489664e-05,
+      "loss": 0.0717,
+      "step": 5516
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.918448812368156e-05,
+      "loss": 0.0687,
+      "step": 5517
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.9150129131660004e-05,
+      "loss": 0.0712,
+      "step": 5518
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.911577823430146e-05,
+      "loss": 0.0725,
+      "step": 5519
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.908143543707412e-05,
+      "loss": 0.07,
+      "step": 5520
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.904710074544495e-05,
+      "loss": 0.0708,
+      "step": 5521
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.901277416487966e-05,
+      "loss": 0.0712,
+      "step": 5522
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.897845570084253e-05,
+      "loss": 0.0752,
+      "step": 5523
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.894414535879668e-05,
+      "loss": 0.0732,
+      "step": 5524
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.890984314420389e-05,
+      "loss": 0.0717,
+      "step": 5525
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.887554906252461e-05,
+      "loss": 0.073,
+      "step": 5526
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.884126311921804e-05,
+      "loss": 0.0693,
+      "step": 5527
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.880698531974212e-05,
+      "loss": 0.0734,
+      "step": 5528
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.877271566955339e-05,
+      "loss": 0.0725,
+      "step": 5529
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.873845417410714e-05,
+      "loss": 0.0724,
+      "step": 5530
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8704200838857406e-05,
+      "loss": 0.07,
+      "step": 5531
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.86699556692569e-05,
+      "loss": 0.07,
+      "step": 5532
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.863571867075699e-05,
+      "loss": 0.0744,
+      "step": 5533
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.860148984880778e-05,
+      "loss": 0.0705,
+      "step": 5534
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8567269208858125e-05,
+      "loss": 0.0693,
+      "step": 5535
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8533056756355445e-05,
+      "loss": 0.0725,
+      "step": 5536
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.849885249674601e-05,
+      "loss": 0.0737,
+      "step": 5537
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.846465643547462e-05,
+      "loss": 0.071,
+      "step": 5538
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8430468577984955e-05,
+      "loss": 0.0722,
+      "step": 5539
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8396288929719216e-05,
+      "loss": 0.0702,
+      "step": 5540
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.83621174961184e-05,
+      "loss": 0.0714,
+      "step": 5541
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.832795428262222e-05,
+      "loss": 0.0696,
+      "step": 5542
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.8293799294668965e-05,
+      "loss": 0.0708,
+      "step": 5543
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.825965253769573e-05,
+      "loss": 0.0723,
+      "step": 5544
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.822551401713821e-05,
+      "loss": 0.0694,
+      "step": 5545
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.819138373843084e-05,
+      "loss": 0.0751,
+      "step": 5546
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.815726170700679e-05,
+      "loss": 0.068,
+      "step": 5547
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.8123147928297765e-05,
+      "loss": 0.0712,
+      "step": 5548
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.8089042407734334e-05,
+      "loss": 0.0714,
+      "step": 5549
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.80549451507456e-05,
+      "loss": 0.0687,
+      "step": 5550
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.802085616275947e-05,
+      "loss": 0.0734,
+      "step": 5551
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.7986775449202495e-05,
+      "loss": 0.072,
+      "step": 5552
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.7952703015499836e-05,
+      "loss": 0.0738,
+      "step": 5553
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.791863886707547e-05,
+      "loss": 0.0747,
+      "step": 5554
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7884583009351914e-05,
+      "loss": 0.0702,
+      "step": 5555
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.785053544775045e-05,
+      "loss": 0.074,
+      "step": 5556
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.7816496187691105e-05,
+      "loss": 0.073,
+      "step": 5557
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.778246523459239e-05,
+      "loss": 0.0727,
+      "step": 5558
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.774844259387169e-05,
+      "loss": 0.0721,
+      "step": 5559
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.771442827094492e-05,
+      "loss": 0.075,
+      "step": 5560
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.76804222712268e-05,
+      "loss": 0.0734,
+      "step": 5561
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.764642460013058e-05,
+      "loss": 0.0712,
+      "step": 5562
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.761243526306831e-05,
+      "loss": 0.0687,
+      "step": 5563
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.757845426545071e-05,
+      "loss": 0.0726,
+      "step": 5564
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.754448161268704e-05,
+      "loss": 0.0739,
+      "step": 5565
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.751051731018537e-05,
+      "loss": 0.0732,
+      "step": 5566
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.747656136335241e-05,
+      "loss": 0.0722,
+      "step": 5567
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.744261377759351e-05,
+      "loss": 0.0696,
+      "step": 5568
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.740867455831265e-05,
+      "loss": 0.0706,
+      "step": 5569
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.7374743710912574e-05,
+      "loss": 0.0722,
+      "step": 5570
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.734082124079467e-05,
+      "loss": 0.0691,
+      "step": 5571
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.730690715335891e-05,
+      "loss": 0.0695,
+      "step": 5572
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.727300145400403e-05,
+      "loss": 0.0733,
+      "step": 5573
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.7239104148127425e-05,
+      "loss": 0.0671,
+      "step": 5574
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.7205215241125046e-05,
+      "loss": 0.073,
+      "step": 5575
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.717133473839163e-05,
+      "loss": 0.0712,
+      "step": 5576
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.713746264532055e-05,
+      "loss": 0.0722,
+      "step": 5577
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.710359896730379e-05,
+      "loss": 0.0719,
+      "step": 5578
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.706974370973198e-05,
+      "loss": 0.07,
+      "step": 5579
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.703589687799451e-05,
+      "loss": 0.0717,
+      "step": 5580
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.700205847747939e-05,
+      "loss": 0.0746,
+      "step": 5581
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.696822851357321e-05,
+      "loss": 0.071,
+      "step": 5582
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.693440699166131e-05,
+      "loss": 0.072,
+      "step": 5583
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.6900593917127665e-05,
+      "loss": 0.074,
+      "step": 5584
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.686678929535489e-05,
+      "loss": 0.0724,
+      "step": 5585
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.683299313172418e-05,
+      "loss": 0.0775,
+      "step": 5586
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.679920543161559e-05,
+      "loss": 0.0711,
+      "step": 5587
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.676542620040764e-05,
+      "loss": 0.0702,
+      "step": 5588
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.673165544347747e-05,
+      "loss": 0.0698,
+      "step": 5589
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.669789316620113e-05,
+      "loss": 0.0711,
+      "step": 5590
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.666413937395307e-05,
+      "loss": 0.0773,
+      "step": 5591
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.663039407210644e-05,
+      "loss": 0.074,
+      "step": 5592
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.6596657266033096e-05,
+      "loss": 0.0723,
+      "step": 5593
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.6562928961103556e-05,
+      "loss": 0.0699,
+      "step": 5594
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.6529209162686914e-05,
+      "loss": 0.0717,
+      "step": 5595
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.6495497876150864e-05,
+      "loss": 0.0719,
+      "step": 5596
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.646179510686195e-05,
+      "loss": 0.0735,
+      "step": 5597
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.642810086018518e-05,
+      "loss": 0.0763,
+      "step": 5598
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.6394415141484174e-05,
+      "loss": 0.0752,
+      "step": 5599
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.6360737956121425e-05,
+      "loss": 0.0714,
+      "step": 5600
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.632706930945784e-05,
+      "loss": 0.0722,
+      "step": 5601
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.629340920685302e-05,
+      "loss": 0.074,
+      "step": 5602
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.625975765366527e-05,
+      "loss": 0.0728,
+      "step": 5603
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.622611465525152e-05,
+      "loss": 0.0728,
+      "step": 5604
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.619248021696728e-05,
+      "loss": 0.069,
+      "step": 5605
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.615885434416667e-05,
+      "loss": 0.0749,
+      "step": 5606
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.612523704220264e-05,
+      "loss": 0.0686,
+      "step": 5607
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.609162831642658e-05,
+      "loss": 0.0697,
+      "step": 5608
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.605802817218855e-05,
+      "loss": 0.073,
+      "step": 5609
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.602443661483729e-05,
+      "loss": 0.0718,
+      "step": 5610
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.59908536497202e-05,
+      "loss": 0.0713,
+      "step": 5611
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.595727928218318e-05,
+      "loss": 0.0721,
+      "step": 5612
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.592371351757092e-05,
+      "loss": 0.0702,
+      "step": 5613
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.589015636122669e-05,
+      "loss": 0.0714,
+      "step": 5614
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.5856607818492324e-05,
+      "loss": 0.072,
+      "step": 5615
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.582306789470826e-05,
+      "loss": 0.0725,
+      "step": 5616
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.578953659521379e-05,
+      "loss": 0.0717,
+      "step": 5617
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.575601392534659e-05,
+      "loss": 0.0719,
+      "step": 5618
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.572249989044302e-05,
+      "loss": 0.072,
+      "step": 5619
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.568899449583813e-05,
+      "loss": 0.0728,
+      "step": 5620
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.5655497746865604e-05,
+      "loss": 0.0726,
+      "step": 5621
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5622009648857625e-05,
+      "loss": 0.0721,
+      "step": 5622
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.558853020714511e-05,
+      "loss": 0.0718,
+      "step": 5623
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.555505942705761e-05,
+      "loss": 0.0681,
+      "step": 5624
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5521597313923224e-05,
+      "loss": 0.0715,
+      "step": 5625
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.548814387306865e-05,
+      "loss": 0.0715,
+      "step": 5626
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5454699109819307e-05,
+      "loss": 0.0703,
+      "step": 5627
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5421263029499214e-05,
+      "loss": 0.0707,
+      "step": 5628
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5387835637430906e-05,
+      "loss": 0.0744,
+      "step": 5629
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5354416938935654e-05,
+      "loss": 0.0718,
+      "step": 5630
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.53210069393333e-05,
+      "loss": 0.0708,
+      "step": 5631
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5287605643942246e-05,
+      "loss": 0.0726,
+      "step": 5632
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.525421305807963e-05,
+      "loss": 0.0717,
+      "step": 5633
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.5220829187061065e-05,
+      "loss": 0.0722,
+      "step": 5634
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.518745403620088e-05,
+      "loss": 0.0743,
+      "step": 5635
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.515408761081201e-05,
+      "loss": 0.0731,
+      "step": 5636
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.5120729916205916e-05,
+      "loss": 0.0726,
+      "step": 5637
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.5087380957692784e-05,
+      "loss": 0.071,
+      "step": 5638
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.505404074058127e-05,
+      "loss": 0.0702,
+      "step": 5639
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.502070927017878e-05,
+      "loss": 0.0741,
+      "step": 5640
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.498738655179129e-05,
+      "loss": 0.0714,
+      "step": 5641
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.4954072590723285e-05,
+      "loss": 0.0703,
+      "step": 5642
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.4920767392278015e-05,
+      "loss": 0.0732,
+      "step": 5643
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.488747096175716e-05,
+      "loss": 0.0691,
+      "step": 5644
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.4854183304461136e-05,
+      "loss": 0.0704,
+      "step": 5645
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.482090442568898e-05,
+      "loss": 0.0707,
+      "step": 5646
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.478763433073817e-05,
+      "loss": 0.0678,
+      "step": 5647
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.475437302490497e-05,
+      "loss": 0.0746,
+      "step": 5648
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.472112051348411e-05,
+      "loss": 0.0707,
+      "step": 5649
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.468787680176902e-05,
+      "loss": 0.0723,
+      "step": 5650
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.465464189505163e-05,
+      "loss": 0.0747,
+      "step": 5651
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.462141579862255e-05,
+      "loss": 0.0719,
+      "step": 5652
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.4588198517770975e-05,
+      "loss": 0.0733,
+      "step": 5653
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.455499005778464e-05,
+      "loss": 0.071,
+      "step": 5654
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.452179042394994e-05,
+      "loss": 0.073,
+      "step": 5655
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.448859962155187e-05,
+      "loss": 0.0721,
+      "step": 5656
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.445541765587394e-05,
+      "loss": 0.0723,
+      "step": 5657
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.442224453219835e-05,
+      "loss": 0.0697,
+      "step": 5658
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.438908025580578e-05,
+      "loss": 0.0727,
+      "step": 5659
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.435592483197566e-05,
+      "loss": 0.0717,
+      "step": 5660
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.432277826598582e-05,
+      "loss": 0.0743,
+      "step": 5661
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.428964056311282e-05,
+      "loss": 0.0702,
+      "step": 5662
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.425651172863181e-05,
+      "loss": 0.0727,
+      "step": 5663
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.422339176781642e-05,
+      "loss": 0.0708,
+      "step": 5664
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.419028068593896e-05,
+      "loss": 0.0712,
+      "step": 5665
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.415717848827033e-05,
+      "loss": 0.0688,
+      "step": 5666
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.4124085180079976e-05,
+      "loss": 0.0701,
+      "step": 5667
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.409100076663587e-05,
+      "loss": 0.0734,
+      "step": 5668
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.405792525320469e-05,
+      "loss": 0.071,
+      "step": 5669
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.402485864505167e-05,
+      "loss": 0.0705,
+      "step": 5670
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.399180094744053e-05,
+      "loss": 0.0721,
+      "step": 5671
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.39587521656337e-05,
+      "loss": 0.0739,
+      "step": 5672
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.392571230489214e-05,
+      "loss": 0.0698,
+      "step": 5673
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.3892681370475354e-05,
+      "loss": 0.0692,
+      "step": 5674
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.385965936764138e-05,
+      "loss": 0.0706,
+      "step": 5675
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3826646301647065e-05,
+      "loss": 0.0698,
+      "step": 5676
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3793642177747596e-05,
+      "loss": 0.0684,
+      "step": 5677
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3760647001196774e-05,
+      "loss": 0.0722,
+      "step": 5678
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.372766077724706e-05,
+      "loss": 0.0711,
+      "step": 5679
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.369468351114948e-05,
+      "loss": 0.0745,
+      "step": 5680
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.366171520815353e-05,
+      "loss": 0.0717,
+      "step": 5681
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.36287558735074e-05,
+      "loss": 0.0722,
+      "step": 5682
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.359580551245782e-05,
+      "loss": 0.0691,
+      "step": 5683
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3562864130250056e-05,
+      "loss": 0.0755,
+      "step": 5684
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.352993173212788e-05,
+      "loss": 0.0725,
+      "step": 5685
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.3497008323333864e-05,
+      "loss": 0.073,
+      "step": 5686
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.346409390910894e-05,
+      "loss": 0.0723,
+      "step": 5687
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.343118849469262e-05,
+      "loss": 0.0716,
+      "step": 5688
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.339829208532309e-05,
+      "loss": 0.0693,
+      "step": 5689
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.336540468623705e-05,
+      "loss": 0.0699,
+      "step": 5690
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.333252630266973e-05,
+      "loss": 0.0717,
+      "step": 5691
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.329965693985497e-05,
+      "loss": 0.0703,
+      "step": 5692
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.32667966030252e-05,
+      "loss": 0.0742,
+      "step": 5693
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3233945297411336e-05,
+      "loss": 0.0727,
+      "step": 5694
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3201103028242827e-05,
+      "loss": 0.0716,
+      "step": 5695
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3168269800747896e-05,
+      "loss": 0.0716,
+      "step": 5696
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.31354456201531e-05,
+      "loss": 0.0738,
+      "step": 5697
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.310263049168362e-05,
+      "loss": 0.0745,
+      "step": 5698
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3069824420563233e-05,
+      "loss": 0.0714,
+      "step": 5699
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.303702741201431e-05,
+      "loss": 0.0703,
+      "step": 5700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3004239471257626e-05,
+      "loss": 0.0733,
+      "step": 5701
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.297146060351266e-05,
+      "loss": 0.075,
+      "step": 5702
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.293869081399744e-05,
+      "loss": 0.07,
+      "step": 5703
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.290593010792846e-05,
+      "loss": 0.0777,
+      "step": 5704
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.287317849052075e-05,
+      "loss": 0.0725,
+      "step": 5705
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.28404359669881e-05,
+      "loss": 0.0735,
+      "step": 5706
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.280770254254264e-05,
+      "loss": 0.0744,
+      "step": 5707
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.277497822239507e-05,
+      "loss": 0.0702,
+      "step": 5708
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.274226301175475e-05,
+      "loss": 0.0724,
+      "step": 5709
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2709556915829554e-05,
+      "loss": 0.0731,
+      "step": 5710
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2676859939825806e-05,
+      "loss": 0.0727,
+      "step": 5711
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.2644172088948506e-05,
+      "loss": 0.0756,
+      "step": 5712
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.261149336840119e-05,
+      "loss": 0.0708,
+      "step": 5713
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.257882378338586e-05,
+      "loss": 0.074,
+      "step": 5714
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.254616333910305e-05,
+      "loss": 0.069,
+      "step": 5715
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.251351204075196e-05,
+      "loss": 0.0735,
+      "step": 5716
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2480869893530276e-05,
+      "loss": 0.073,
+      "step": 5717
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2448236902634165e-05,
+      "loss": 0.0721,
+      "step": 5718
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.241561307325842e-05,
+      "loss": 0.0699,
+      "step": 5719
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2382998410596394e-05,
+      "loss": 0.0737,
+      "step": 5720
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2350392919839846e-05,
+      "loss": 0.0765,
+      "step": 5721
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.23177966061792e-05,
+      "loss": 0.0677,
+      "step": 5722
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2285209474803435e-05,
+      "loss": 0.0699,
+      "step": 5723
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.225263153089996e-05,
+      "loss": 0.0708,
+      "step": 5724
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.222006277965474e-05,
+      "loss": 0.0732,
+      "step": 5725
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.218750322625234e-05,
+      "loss": 0.0741,
+      "step": 5726
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.21549528758759e-05,
+      "loss": 0.0727,
+      "step": 5727
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.212241173370694e-05,
+      "loss": 0.0734,
+      "step": 5728
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.2089879804925626e-05,
+      "loss": 0.0715,
+      "step": 5729
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2057357094710695e-05,
+      "loss": 0.0716,
+      "step": 5730
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.2024843608239264e-05,
+      "loss": 0.0738,
+      "step": 5731
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.199233935068714e-05,
+      "loss": 0.0754,
+      "step": 5732
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.195984432722855e-05,
+      "loss": 0.0717,
+      "step": 5733
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.192735854303633e-05,
+      "loss": 0.0739,
+      "step": 5734
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.189488200328178e-05,
+      "loss": 0.0705,
+      "step": 5735
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.186241471313476e-05,
+      "loss": 0.0717,
+      "step": 5736
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.18299566777637e-05,
+      "loss": 0.0711,
+      "step": 5737
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.179750790233545e-05,
+      "loss": 0.0716,
+      "step": 5738
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.1765068392015525e-05,
+      "loss": 0.0733,
+      "step": 5739
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.1732638151967805e-05,
+      "loss": 0.0724,
+      "step": 5740
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.1700217187354815e-05,
+      "loss": 0.0716,
+      "step": 5741
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.166780550333761e-05,
+      "loss": 0.0723,
+      "step": 5742
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.1635403105075656e-05,
+      "loss": 0.0713,
+      "step": 5743
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.160300999772706e-05,
+      "loss": 0.0723,
+      "step": 5744
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.1570626186448346e-05,
+      "loss": 0.0702,
+      "step": 5745
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.153825167639464e-05,
+      "loss": 0.0743,
+      "step": 5746
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.15058864727196e-05,
+      "loss": 0.0695,
+      "step": 5747
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.147353058057528e-05,
+      "loss": 0.0746,
+      "step": 5748
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.144118400511242e-05,
+      "loss": 0.0719,
+      "step": 5749
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.140884675148011e-05,
+      "loss": 0.0728,
+      "step": 5750
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.137651882482607e-05,
+      "loss": 0.0741,
+      "step": 5751
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.134420023029654e-05,
+      "loss": 0.0731,
+      "step": 5752
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.131189097303615e-05,
+      "loss": 0.0712,
+      "step": 5753
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.127959105818823e-05,
+      "loss": 0.0722,
+      "step": 5754
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.124730049089442e-05,
+      "loss": 0.0704,
+      "step": 5755
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.1215019276295076e-05,
+      "loss": 0.0738,
+      "step": 5756
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.118274741952887e-05,
+      "loss": 0.0723,
+      "step": 5757
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.115048492573311e-05,
+      "loss": 0.0711,
+      "step": 5758
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.111823180004364e-05,
+      "loss": 0.0691,
+      "step": 5759
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.108598804759466e-05,
+      "loss": 0.0728,
+      "step": 5760
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.105375367351902e-05,
+      "loss": 0.0725,
+      "step": 5761
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.1021528682948066e-05,
+      "loss": 0.0736,
+      "step": 5762
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.0989313081011525e-05,
+      "loss": 0.0725,
+      "step": 5763
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.095710687283781e-05,
+      "loss": 0.0736,
+      "step": 5764
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.092491006355367e-05,
+      "loss": 0.0709,
+      "step": 5765
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.0892722658284485e-05,
+      "loss": 0.072,
+      "step": 5766
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.086054466215404e-05,
+      "loss": 0.0708,
+      "step": 5767
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.08283760802847e-05,
+      "loss": 0.0687,
+      "step": 5768
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.079621691779734e-05,
+      "loss": 0.0752,
+      "step": 5769
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.0764067179811205e-05,
+      "loss": 0.0696,
+      "step": 5770
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.073192687144418e-05,
+      "loss": 0.0712,
+      "step": 5771
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.069979599781264e-05,
+      "loss": 0.0751,
+      "step": 5772
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.0667674564031365e-05,
+      "loss": 0.0703,
+      "step": 5773
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.063556257521368e-05,
+      "loss": 0.0727,
+      "step": 5774
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.060346003647142e-05,
+      "loss": 0.0727,
+      "step": 5775
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.057136695291495e-05,
+      "loss": 0.0751,
+      "step": 5776
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.053928332965302e-05,
+      "loss": 0.0744,
+      "step": 5777
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.050720917179297e-05,
+      "loss": 0.0686,
+      "step": 5778
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.0475144484440655e-05,
+      "loss": 0.0721,
+      "step": 5779
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.044308927270032e-05,
+      "loss": 0.0712,
+      "step": 5780
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.04110435416747e-05,
+      "loss": 0.0702,
+      "step": 5781
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.037900729646519e-05,
+      "loss": 0.0738,
+      "step": 5782
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.034698054217151e-05,
+      "loss": 0.0715,
+      "step": 5783
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.031496328389187e-05,
+      "loss": 0.0736,
+      "step": 5784
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.0282955526723065e-05,
+      "loss": 0.0715,
+      "step": 5785
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.025095727576036e-05,
+      "loss": 0.0721,
+      "step": 5786
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.02189685360974e-05,
+      "loss": 0.0714,
+      "step": 5787
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.018698931282644e-05,
+      "loss": 0.0686,
+      "step": 5788
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.015501961103818e-05,
+      "loss": 0.0734,
+      "step": 5789
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.01230594358218e-05,
+      "loss": 0.073,
+      "step": 5790
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.0091108792264856e-05,
+      "loss": 0.0715,
+      "step": 5791
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.005916768545365e-05,
+      "loss": 0.0724,
+      "step": 5792
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.0027236120472724e-05,
+      "loss": 0.0702,
+      "step": 5793
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.9995314102405156e-05,
+      "loss": 0.0701,
+      "step": 5794
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.9963401636332574e-05,
+      "loss": 0.0734,
+      "step": 5795
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.993149872733506e-05,
+      "loss": 0.0709,
+      "step": 5796
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.98996053804911e-05,
+      "loss": 0.0724,
+      "step": 5797
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.986772160087775e-05,
+      "loss": 0.0713,
+      "step": 5798
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.983584739357055e-05,
+      "loss": 0.0678,
+      "step": 5799
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.980398276364342e-05,
+      "loss": 0.0723,
+      "step": 5800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.977212771616875e-05,
+      "loss": 0.0727,
+      "step": 5801
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.974028225621762e-05,
+      "loss": 0.0716,
+      "step": 5802
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9708446388859335e-05,
+      "loss": 0.0743,
+      "step": 5803
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.967662011916175e-05,
+      "loss": 0.0732,
+      "step": 5804
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.964480345219123e-05,
+      "loss": 0.074,
+      "step": 5805
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9612996393012626e-05,
+      "loss": 0.0709,
+      "step": 5806
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9581198946689166e-05,
+      "loss": 0.0732,
+      "step": 5807
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.954941111828263e-05,
+      "loss": 0.0707,
+      "step": 5808
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9517632912853286e-05,
+      "loss": 0.0735,
+      "step": 5809
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.9485864335459785e-05,
+      "loss": 0.0688,
+      "step": 5810
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.945410539115921e-05,
+      "loss": 0.079,
+      "step": 5811
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.942235608500735e-05,
+      "loss": 0.0694,
+      "step": 5812
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.93906164220582e-05,
+      "loss": 0.0701,
+      "step": 5813
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.93588864073643e-05,
+      "loss": 0.0723,
+      "step": 5814
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.93271660459767e-05,
+      "loss": 0.0724,
+      "step": 5815
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.929545534294493e-05,
+      "loss": 0.0713,
+      "step": 5816
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.926375430331685e-05,
+      "loss": 0.0695,
+      "step": 5817
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.923206293213891e-05,
+      "loss": 0.0727,
+      "step": 5818
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.9200381234456016e-05,
+      "loss": 0.0729,
+      "step": 5819
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.9168709215311473e-05,
+      "loss": 0.0721,
+      "step": 5820
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.913704687974701e-05,
+      "loss": 0.0753,
+      "step": 5821
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.910539423280293e-05,
+      "loss": 0.071,
+      "step": 5822
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.907375127951797e-05,
+      "loss": 0.0734,
+      "step": 5823
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.904211802492922e-05,
+      "loss": 0.0721,
+      "step": 5824
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.901049447407233e-05,
+      "loss": 0.0703,
+      "step": 5825
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.897888063198142e-05,
+      "loss": 0.0718,
+      "step": 5826
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.894727650368892e-05,
+      "loss": 0.0719,
+      "step": 5827
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.891568209422588e-05,
+      "loss": 0.0722,
+      "step": 5828
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.888409740862175e-05,
+      "loss": 0.0718,
+      "step": 5829
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.885252245190439e-05,
+      "loss": 0.0718,
+      "step": 5830
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.882095722910011e-05,
+      "loss": 0.0706,
+      "step": 5831
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.878940174523371e-05,
+      "loss": 0.0754,
+      "step": 5832
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.8757856005328484e-05,
+      "loss": 0.0733,
+      "step": 5833
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.8726320014406035e-05,
+      "loss": 0.0718,
+      "step": 5834
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.869479377748655e-05,
+      "loss": 0.0705,
+      "step": 5835
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.866327729958863e-05,
+      "loss": 0.0697,
+      "step": 5836
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.863177058572924e-05,
+      "loss": 0.0727,
+      "step": 5837
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.860027364092392e-05,
+      "loss": 0.0748,
+      "step": 5838
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.8568786470186536e-05,
+      "loss": 0.073,
+      "step": 5839
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.8537309078529494e-05,
+      "loss": 0.0736,
+      "step": 5840
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.850584147096356e-05,
+      "loss": 0.0701,
+      "step": 5841
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.847438365249799e-05,
+      "loss": 0.0736,
+      "step": 5842
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.8442935628140545e-05,
+      "loss": 0.0743,
+      "step": 5843
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.841149740289725e-05,
+      "loss": 0.0705,
+      "step": 5844
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.838006898177276e-05,
+      "loss": 0.0681,
+      "step": 5845
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.834865036977002e-05,
+      "loss": 0.0738,
+      "step": 5846
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.831724157189053e-05,
+      "loss": 0.0708,
+      "step": 5847
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.828584259313418e-05,
+      "loss": 0.0752,
+      "step": 5848
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.825445343849925e-05,
+      "loss": 0.0706,
+      "step": 5849
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.822307411298256e-05,
+      "loss": 0.074,
+      "step": 5850
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.819170462157924e-05,
+      "loss": 0.0689,
+      "step": 5851
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.8160344969282944e-05,
+      "loss": 0.0711,
+      "step": 5852
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.812899516108579e-05,
+      "loss": 0.0725,
+      "step": 5853
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.80976552019782e-05,
+      "loss": 0.0719,
+      "step": 5854
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.806632509694915e-05,
+      "loss": 0.0698,
+      "step": 5855
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.8035004850985964e-05,
+      "loss": 0.0715,
+      "step": 5856
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.8003694469074446e-05,
+      "loss": 0.0704,
+      "step": 5857
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.797239395619887e-05,
+      "loss": 0.074,
+      "step": 5858
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.79411033173418e-05,
+      "loss": 0.072,
+      "step": 5859
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.790982255748438e-05,
+      "loss": 0.0716,
+      "step": 5860
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.787855168160606e-05,
+      "loss": 0.0709,
+      "step": 5861
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7847290694684834e-05,
+      "loss": 0.0721,
+      "step": 5862
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.7816039601696995e-05,
+      "loss": 0.0725,
+      "step": 5863
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.778479840761736e-05,
+      "loss": 0.0702,
+      "step": 5864
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.775356711741918e-05,
+      "loss": 0.07,
+      "step": 5865
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.772234573607398e-05,
+      "loss": 0.0709,
+      "step": 5866
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.76911342685519e-05,
+      "loss": 0.073,
+      "step": 5867
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.76599327198214e-05,
+      "loss": 0.0734,
+      "step": 5868
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.7628741094849374e-05,
+      "loss": 0.0743,
+      "step": 5869
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.75975593986011e-05,
+      "loss": 0.0717,
+      "step": 5870
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.756638763604033e-05,
+      "loss": 0.0714,
+      "step": 5871
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.753522581212927e-05,
+      "loss": 0.0673,
+      "step": 5872
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.750407393182842e-05,
+      "loss": 0.069,
+      "step": 5873
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.7472932000096805e-05,
+      "loss": 0.0738,
+      "step": 5874
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.744180002189186e-05,
+      "loss": 0.0724,
+      "step": 5875
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.7410678002169343e-05,
+      "loss": 0.0725,
+      "step": 5876
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.737956594588351e-05,
+      "loss": 0.0734,
+      "step": 5877
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.734846385798707e-05,
+      "loss": 0.0755,
+      "step": 5878
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.731737174343103e-05,
+      "loss": 0.0707,
+      "step": 5879
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.728628960716485e-05,
+      "loss": 0.0714,
+      "step": 5880
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.7255217454136425e-05,
+      "loss": 0.07,
+      "step": 5881
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.7224155289292115e-05,
+      "loss": 0.0704,
+      "step": 5882
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.719310311757656e-05,
+      "loss": 0.0694,
+      "step": 5883
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.716206094393288e-05,
+      "loss": 0.0707,
+      "step": 5884
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.713102877330266e-05,
+      "loss": 0.0691,
+      "step": 5885
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.7100006610625784e-05,
+      "loss": 0.0745,
+      "step": 5886
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.7068994460840544e-05,
+      "loss": 0.0727,
+      "step": 5887
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.703799232888381e-05,
+      "loss": 0.072,
+      "step": 5888
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.700700021969066e-05,
+      "loss": 0.0744,
+      "step": 5889
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.697601813819462e-05,
+      "loss": 0.0719,
+      "step": 5890
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.6945046089327693e-05,
+      "loss": 0.0712,
+      "step": 5891
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.691408407802026e-05,
+      "loss": 0.0707,
+      "step": 5892
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.688313210920103e-05,
+      "loss": 0.0726,
+      "step": 5893
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.685219018779721e-05,
+      "loss": 0.0737,
+      "step": 5894
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6821258318734376e-05,
+      "loss": 0.0684,
+      "step": 5895
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.679033650693647e-05,
+      "loss": 0.0727,
+      "step": 5896
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6759424757325814e-05,
+      "loss": 0.0697,
+      "step": 5897
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6728523074823274e-05,
+      "loss": 0.0722,
+      "step": 5898
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6697631464347945e-05,
+      "loss": 0.0705,
+      "step": 5899
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.666674993081738e-05,
+      "loss": 0.0729,
+      "step": 5900
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.6635878479147544e-05,
+      "loss": 0.0757,
+      "step": 5901
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.660501711425282e-05,
+      "loss": 0.0759,
+      "step": 5902
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.657416584104589e-05,
+      "loss": 0.073,
+      "step": 5903
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.654332466443792e-05,
+      "loss": 0.0742,
+      "step": 5904
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.651249358933848e-05,
+      "loss": 0.075,
+      "step": 5905
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.648167262065545e-05,
+      "loss": 0.0745,
+      "step": 5906
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.6450861763295074e-05,
+      "loss": 0.0708,
+      "step": 5907
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.642006102216219e-05,
+      "loss": 0.0709,
+      "step": 5908
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.638927040215984e-05,
+      "loss": 0.0699,
+      "step": 5909
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.635848990818944e-05,
+      "loss": 0.0729,
+      "step": 5910
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.632771954515092e-05,
+      "loss": 0.0724,
+      "step": 5911
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.629695931794257e-05,
+      "loss": 0.0724,
+      "step": 5912
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.626620923146095e-05,
+      "loss": 0.0759,
+      "step": 5913
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.623546929060112e-05,
+      "loss": 0.0705,
+      "step": 5914
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.620473950025654e-05,
+      "loss": 0.0734,
+      "step": 5915
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.617401986531899e-05,
+      "loss": 0.0736,
+      "step": 5916
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.614331039067854e-05,
+      "loss": 0.0738,
+      "step": 5917
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.6112611081223935e-05,
+      "loss": 0.0756,
+      "step": 5918
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.6081921941842024e-05,
+      "loss": 0.0706,
+      "step": 5919
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.605124297741811e-05,
+      "loss": 0.0715,
+      "step": 5920
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.6020574192835935e-05,
+      "loss": 0.0715,
+      "step": 5921
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5989915592977605e-05,
+      "loss": 0.0709,
+      "step": 5922
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.595926718272354e-05,
+      "loss": 0.0747,
+      "step": 5923
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.59286289669526e-05,
+      "loss": 0.0753,
+      "step": 5924
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.589800095054207e-05,
+      "loss": 0.0766,
+      "step": 5925
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.586738313836746e-05,
+      "loss": 0.0739,
+      "step": 5926
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.583677553530276e-05,
+      "loss": 0.0725,
+      "step": 5927
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.580617814622031e-05,
+      "loss": 0.0724,
+      "step": 5928
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5775590975990904e-05,
+      "loss": 0.07,
+      "step": 5929
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.574501402948354e-05,
+      "loss": 0.0712,
+      "step": 5930
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.571444731156572e-05,
+      "loss": 0.0734,
+      "step": 5931
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.568389082710334e-05,
+      "loss": 0.0714,
+      "step": 5932
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5653344580960525e-05,
+      "loss": 0.0703,
+      "step": 5933
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5622808577999924e-05,
+      "loss": 0.0708,
+      "step": 5934
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.559228282308241e-05,
+      "loss": 0.0723,
+      "step": 5935
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.556176732106738e-05,
+      "loss": 0.07,
+      "step": 5936
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5531262076812464e-05,
+      "loss": 0.0711,
+      "step": 5937
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5500767095173734e-05,
+      "loss": 0.0703,
+      "step": 5938
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.547028238100564e-05,
+      "loss": 0.0726,
+      "step": 5939
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5439807939160904e-05,
+      "loss": 0.0732,
+      "step": 5940
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.540934377449071e-05,
+      "loss": 0.0729,
+      "step": 5941
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.537888989184461e-05,
+      "loss": 0.0712,
+      "step": 5942
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5348446296070413e-05,
+      "loss": 0.0711,
+      "step": 5943
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5318012992014416e-05,
+      "loss": 0.0735,
+      "step": 5944
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.5287589984521154e-05,
+      "loss": 0.0735,
+      "step": 5945
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.525717727843367e-05,
+      "loss": 0.0727,
+      "step": 5946
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.522677487859321e-05,
+      "loss": 0.0724,
+      "step": 5947
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.519638278983948e-05,
+      "loss": 0.0749,
+      "step": 5948
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.516600101701056e-05,
+      "loss": 0.0698,
+      "step": 5949
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.5135629564942796e-05,
+      "loss": 0.072,
+      "step": 5950
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.510526843847099e-05,
+      "loss": 0.0715,
+      "step": 5951
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.507491764242821e-05,
+      "loss": 0.0719,
+      "step": 5952
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.504457718164592e-05,
+      "loss": 0.0695,
+      "step": 5953
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.501424706095401e-05,
+      "loss": 0.0733,
+      "step": 5954
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.498392728518056e-05,
+      "loss": 0.0726,
+      "step": 5955
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.495361785915219e-05,
+      "loss": 0.0718,
+      "step": 5956
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.49233187876937e-05,
+      "loss": 0.0711,
+      "step": 5957
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.489303007562836e-05,
+      "loss": 0.0697,
+      "step": 5958
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.4862751727777797e-05,
+      "loss": 0.071,
+      "step": 5959
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.483248374896186e-05,
+      "loss": 0.0751,
+      "step": 5960
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4802226143998916e-05,
+      "loss": 0.0724,
+      "step": 5961
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.477197891770552e-05,
+      "loss": 0.0716,
+      "step": 5962
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4741742074896674e-05,
+      "loss": 0.0715,
+      "step": 5963
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.471151562038577e-05,
+      "loss": 0.0696,
+      "step": 5964
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.468129955898439e-05,
+      "loss": 0.0704,
+      "step": 5965
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.46510938955026e-05,
+      "loss": 0.0735,
+      "step": 5966
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.462089863474878e-05,
+      "loss": 0.0682,
+      "step": 5967
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4590713781529616e-05,
+      "loss": 0.069,
+      "step": 5968
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4560539340650125e-05,
+      "loss": 0.072,
+      "step": 5969
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4530375316913734e-05,
+      "loss": 0.0738,
+      "step": 5970
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4500221715122205e-05,
+      "loss": 0.0737,
+      "step": 5971
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.447007854007556e-05,
+      "loss": 0.0703,
+      "step": 5972
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.4439945796572225e-05,
+      "loss": 0.0727,
+      "step": 5973
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.440982348940902e-05,
+      "loss": 0.0739,
+      "step": 5974
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.437971162338098e-05,
+      "loss": 0.0733,
+      "step": 5975
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.434961020328149e-05,
+      "loss": 0.0734,
+      "step": 5976
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4319519233902444e-05,
+      "loss": 0.0711,
+      "step": 5977
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.428943872003387e-05,
+      "loss": 0.0722,
+      "step": 5978
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.425936866646419e-05,
+      "loss": 0.0729,
+      "step": 5979
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.422930907798021e-05,
+      "loss": 0.0754,
+      "step": 5980
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.419925995936708e-05,
+      "loss": 0.0731,
+      "step": 5981
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.416922131540816e-05,
+      "loss": 0.07,
+      "step": 5982
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4139193150885287e-05,
+      "loss": 0.0696,
+      "step": 5983
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4109175470578566e-05,
+      "loss": 0.0714,
+      "step": 5984
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.407916827926644e-05,
+      "loss": 0.0729,
+      "step": 5985
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.404917158172558e-05,
+      "loss": 0.0733,
+      "step": 5986
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4019185382731234e-05,
+      "loss": 0.0681,
+      "step": 5987
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.398920968705677e-05,
+      "loss": 0.0721,
+      "step": 5988
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.395924449947389e-05,
+      "loss": 0.0748,
+      "step": 5989
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.392928982475272e-05,
+      "loss": 0.071,
+      "step": 5990
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3899345667661705e-05,
+      "loss": 0.0741,
+      "step": 5991
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.386941203296755e-05,
+      "loss": 0.0734,
+      "step": 5992
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3839488925435246e-05,
+      "loss": 0.0712,
+      "step": 5993
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.380957634982831e-05,
+      "loss": 0.0727,
+      "step": 5994
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.377967431090837e-05,
+      "loss": 0.0723,
+      "step": 5995
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.374978281343542e-05,
+      "loss": 0.0753,
+      "step": 5996
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.37199018621679e-05,
+      "loss": 0.0713,
+      "step": 5997
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.369003146186246e-05,
+      "loss": 0.0745,
+      "step": 5998
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.366017161727404e-05,
+      "loss": 0.0719,
+      "step": 5999
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.3630322333156e-05,
+      "loss": 0.0747,
+      "step": 6000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3600483614259984e-05,
+      "loss": 0.0723,
+      "step": 6001
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3570655465335924e-05,
+      "loss": 0.0741,
+      "step": 6002
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.354083789113203e-05,
+      "loss": 0.072,
+      "step": 6003
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.351103089639499e-05,
+      "loss": 0.074,
+      "step": 6004
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.348123448586967e-05,
+      "loss": 0.0735,
+      "step": 6005
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.34514486642992e-05,
+      "loss": 0.0735,
+      "step": 6006
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.342167343642526e-05,
+      "loss": 0.0752,
+      "step": 6007
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3391908806987604e-05,
+      "loss": 0.0679,
+      "step": 6008
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3362154780724374e-05,
+      "loss": 0.0737,
+      "step": 6009
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.333241136237206e-05,
+      "loss": 0.071,
+      "step": 6010
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.330267855666548e-05,
+      "loss": 0.0738,
+      "step": 6011
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.327295636833766e-05,
+      "loss": 0.0752,
+      "step": 6012
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3243244802120036e-05,
+      "loss": 0.0701,
+      "step": 6013
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.3213543862742346e-05,
+      "loss": 0.0728,
+      "step": 6014
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.3183853554932576e-05,
+      "loss": 0.0701,
+      "step": 6015
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.315417388341702e-05,
+      "loss": 0.0753,
+      "step": 6016
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.312450485292032e-05,
+      "loss": 0.0716,
+      "step": 6017
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.30948464681655e-05,
+      "loss": 0.0726,
+      "step": 6018
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.306519873387368e-05,
+      "loss": 0.0727,
+      "step": 6019
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.303556165476448e-05,
+      "loss": 0.0716,
+      "step": 6020
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.300593523555576e-05,
+      "loss": 0.0738,
+      "step": 6021
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.2976319480963634e-05,
+      "loss": 0.0747,
+      "step": 6022
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.294671439570258e-05,
+      "loss": 0.0716,
+      "step": 6023
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.291711998448539e-05,
+      "loss": 0.0719,
+      "step": 6024
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.2887536252023085e-05,
+      "loss": 0.0738,
+      "step": 6025
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.2857963203025e-05,
+      "loss": 0.0703,
+      "step": 6026
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.282840084219883e-05,
+      "loss": 0.0702,
+      "step": 6027
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.279884917425054e-05,
+      "loss": 0.0716,
+      "step": 6028
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.276930820388436e-05,
+      "loss": 0.0714,
+      "step": 6029
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.273977793580285e-05,
+      "loss": 0.066,
+      "step": 6030
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.27102583747069e-05,
+      "loss": 0.0721,
+      "step": 6031
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2680749525295594e-05,
+      "loss": 0.0701,
+      "step": 6032
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.265125139226642e-05,
+      "loss": 0.0721,
+      "step": 6033
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.262176398031506e-05,
+      "loss": 0.0713,
+      "step": 6034
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.259228729413561e-05,
+      "loss": 0.072,
+      "step": 6035
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.25628213384203e-05,
+      "loss": 0.0728,
+      "step": 6036
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.25333661178598e-05,
+      "loss": 0.0718,
+      "step": 6037
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.250392163714303e-05,
+      "loss": 0.0722,
+      "step": 6038
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.247448790095713e-05,
+      "loss": 0.071,
+      "step": 6039
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.244506491398764e-05,
+      "loss": 0.0752,
+      "step": 6040
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.2415652680918264e-05,
+      "loss": 0.0714,
+      "step": 6041
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.238625120643111e-05,
+      "loss": 0.0738,
+      "step": 6042
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.235686049520652e-05,
+      "loss": 0.0717,
+      "step": 6043
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.232748055192311e-05,
+      "loss": 0.0726,
+      "step": 6044
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.229811138125782e-05,
+      "loss": 0.0689,
+      "step": 6045
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.226875298788583e-05,
+      "loss": 0.0706,
+      "step": 6046
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.2239405376480634e-05,
+      "loss": 0.0748,
+      "step": 6047
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.221006855171405e-05,
+      "loss": 0.0724,
+      "step": 6048
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.218074251825605e-05,
+      "loss": 0.0716,
+      "step": 6049
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.215142728077505e-05,
+      "loss": 0.0723,
+      "step": 6050
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.2122122843937606e-05,
+      "loss": 0.0724,
+      "step": 6051
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.209282921240866e-05,
+      "loss": 0.0743,
+      "step": 6052
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.2063546390851395e-05,
+      "loss": 0.0691,
+      "step": 6053
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.203427438392723e-05,
+      "loss": 0.0704,
+      "step": 6054
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.2005013196295954e-05,
+      "loss": 0.0715,
+      "step": 6055
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.197576283261553e-05,
+      "loss": 0.0703,
+      "step": 6056
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.19465232975423e-05,
+      "loss": 0.0719,
+      "step": 6057
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1917294595730764e-05,
+      "loss": 0.0747,
+      "step": 6058
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1888076731833815e-05,
+      "loss": 0.0721,
+      "step": 6059
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.185886971050259e-05,
+      "loss": 0.071,
+      "step": 6060
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.182967353638643e-05,
+      "loss": 0.069,
+      "step": 6061
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.1800488214133014e-05,
+      "loss": 0.0714,
+      "step": 6062
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.177131374838833e-05,
+      "loss": 0.0735,
+      "step": 6063
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.174215014379652e-05,
+      "loss": 0.0737,
+      "step": 6064
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.171299740500012e-05,
+      "loss": 0.0708,
+      "step": 6065
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.168385553663983e-05,
+      "loss": 0.072,
+      "step": 6066
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.165472454335472e-05,
+      "loss": 0.0712,
+      "step": 6067
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.162560442978203e-05,
+      "loss": 0.0702,
+      "step": 6068
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.159649520055733e-05,
+      "loss": 0.0697,
+      "step": 6069
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.15673968603145e-05,
+      "loss": 0.0716,
+      "step": 6070
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.153830941368555e-05,
+      "loss": 0.0748,
+      "step": 6071
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.150923286530089e-05,
+      "loss": 0.0715,
+      "step": 6072
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.148016721978914e-05,
+      "loss": 0.0718,
+      "step": 6073
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.145111248177719e-05,
+      "loss": 0.0743,
+      "step": 6074
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1422068655890135e-05,
+      "loss": 0.07,
+      "step": 6075
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.139303574675144e-05,
+      "loss": 0.0721,
+      "step": 6076
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1364013758982804e-05,
+      "loss": 0.0718,
+      "step": 6077
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1335002697204084e-05,
+      "loss": 0.07,
+      "step": 6078
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.130600256603354e-05,
+      "loss": 0.0705,
+      "step": 6079
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1277013370087635e-05,
+      "loss": 0.0754,
+      "step": 6080
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.1248035113981075e-05,
+      "loss": 0.0721,
+      "step": 6081
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.121906780232676e-05,
+      "loss": 0.0762,
+      "step": 6082
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.119011143973606e-05,
+      "loss": 0.0746,
+      "step": 6083
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.116116603081839e-05,
+      "loss": 0.0725,
+      "step": 6084
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.113223158018148e-05,
+      "loss": 0.0716,
+      "step": 6085
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.110330809243134e-05,
+      "loss": 0.0712,
+      "step": 6086
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.107439557217229e-05,
+      "loss": 0.0718,
+      "step": 6087
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.1045494024006746e-05,
+      "loss": 0.0736,
+      "step": 6088
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.101660345253553e-05,
+      "loss": 0.0734,
+      "step": 6089
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0987723862357675e-05,
+      "loss": 0.0712,
+      "step": 6090
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.095885525807043e-05,
+      "loss": 0.0748,
+      "step": 6091
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.092999764426925e-05,
+      "loss": 0.0742,
+      "step": 6092
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0901151025548026e-05,
+      "loss": 0.0671,
+      "step": 6093
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.087231540649872e-05,
+      "loss": 0.0692,
+      "step": 6094
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.084349079171156e-05,
+      "loss": 0.0723,
+      "step": 6095
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.081467718577512e-05,
+      "loss": 0.0729,
+      "step": 6096
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.078587459327616e-05,
+      "loss": 0.0712,
+      "step": 6097
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.075708301879967e-05,
+      "loss": 0.0727,
+      "step": 6098
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.072830246692892e-05,
+      "loss": 0.0723,
+      "step": 6099
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.069953294224545e-05,
+      "loss": 0.0728,
+      "step": 6100
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0670774449328956e-05,
+      "loss": 0.0716,
+      "step": 6101
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.064202699275739e-05,
+      "loss": 0.0735,
+      "step": 6102
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0613290577107115e-05,
+      "loss": 0.0691,
+      "step": 6103
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0584565206952534e-05,
+      "loss": 0.0737,
+      "step": 6104
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.055585088686633e-05,
+      "loss": 0.0729,
+      "step": 6105
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0527147621419504e-05,
+      "loss": 0.0736,
+      "step": 6106
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0498455415181293e-05,
+      "loss": 0.0733,
+      "step": 6107
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.0469774272719076e-05,
+      "loss": 0.0695,
+      "step": 6108
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0441104198598547e-05,
+      "loss": 0.072,
+      "step": 6109
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0412445197383664e-05,
+      "loss": 0.0736,
+      "step": 6110
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.038379727363655e-05,
+      "loss": 0.0723,
+      "step": 6111
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.035516043191753e-05,
+      "loss": 0.0733,
+      "step": 6112
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0326534676785357e-05,
+      "loss": 0.0721,
+      "step": 6113
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0297920012796843e-05,
+      "loss": 0.0729,
+      "step": 6114
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0269316444507033e-05,
+      "loss": 0.071,
+      "step": 6115
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.02407239764693e-05,
+      "loss": 0.0745,
+      "step": 6116
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0212142613235238e-05,
+      "loss": 0.073,
+      "step": 6117
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0183572359354572e-05,
+      "loss": 0.0729,
+      "step": 6118
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.015501321937537e-05,
+      "loss": 0.0702,
+      "step": 6119
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0126465197843913e-05,
+      "loss": 0.0692,
+      "step": 6120
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0097928299304667e-05,
+      "loss": 0.0752,
+      "step": 6121
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.0069402528300307e-05,
+      "loss": 0.0743,
+      "step": 6122
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.0040887889371815e-05,
+      "loss": 0.0727,
+      "step": 6123
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.001238438705839e-05,
+      "loss": 0.073,
+      "step": 6124
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9983892025897386e-05,
+      "loss": 0.0719,
+      "step": 6125
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9955410810424446e-05,
+      "loss": 0.0693,
+      "step": 6126
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.992694074517346e-05,
+      "loss": 0.0724,
+      "step": 6127
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.989848183467645e-05,
+      "loss": 0.0706,
+      "step": 6128
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9870034083463737e-05,
+      "loss": 0.0729,
+      "step": 6129
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9841597496063877e-05,
+      "loss": 0.0727,
+      "step": 6130
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9813172077003604e-05,
+      "loss": 0.0718,
+      "step": 6131
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9784757830807854e-05,
+      "loss": 0.0719,
+      "step": 6132
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9756354761999838e-05,
+      "loss": 0.0703,
+      "step": 6133
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9727962875101e-05,
+      "loss": 0.0725,
+      "step": 6134
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.9699582174630923e-05,
+      "loss": 0.0728,
+      "step": 6135
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9671212665107496e-05,
+      "loss": 0.0746,
+      "step": 6136
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.96428543510468e-05,
+      "loss": 0.0732,
+      "step": 6137
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9614507236963073e-05,
+      "loss": 0.0693,
+      "step": 6138
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.958617132736887e-05,
+      "loss": 0.0756,
+      "step": 6139
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9557846626774876e-05,
+      "loss": 0.0744,
+      "step": 6140
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.952953313969008e-05,
+      "loss": 0.072,
+      "step": 6141
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9501230870621565e-05,
+      "loss": 0.0751,
+      "step": 6142
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.947293982407474e-05,
+      "loss": 0.0704,
+      "step": 6143
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9444660004553205e-05,
+      "loss": 0.0738,
+      "step": 6144
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9416391416558698e-05,
+      "loss": 0.0691,
+      "step": 6145
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9388134064591287e-05,
+      "loss": 0.071,
+      "step": 6146
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9359887953149124e-05,
+      "loss": 0.0697,
+      "step": 6147
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9331653086728673e-05,
+      "loss": 0.0748,
+      "step": 6148
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.9303429469824596e-05,
+      "loss": 0.0719,
+      "step": 6149
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9275217106929677e-05,
+      "loss": 0.071,
+      "step": 6150
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9247016002535044e-05,
+      "loss": 0.069,
+      "step": 6151
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9218826161129875e-05,
+      "loss": 0.072,
+      "step": 6152
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.91906475872017e-05,
+      "loss": 0.0716,
+      "step": 6153
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9162480285236203e-05,
+      "loss": 0.0754,
+      "step": 6154
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9134324259717217e-05,
+      "loss": 0.0722,
+      "step": 6155
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9106179515126887e-05,
+      "loss": 0.0741,
+      "step": 6156
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.907804605594544e-05,
+      "loss": 0.0742,
+      "step": 6157
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.9049923886651396e-05,
+      "loss": 0.0706,
+      "step": 6158
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.90218130117215e-05,
+      "loss": 0.0741,
+      "step": 6159
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.8993713435630575e-05,
+      "loss": 0.073,
+      "step": 6160
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.8965625162851796e-05,
+      "loss": 0.0718,
+      "step": 6161
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.893754819785639e-05,
+      "loss": 0.0701,
+      "step": 6162
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.890948254511393e-05,
+      "loss": 0.0695,
+      "step": 6163
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8881428209092052e-05,
+      "loss": 0.0722,
+      "step": 6164
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8853385194256676e-05,
+      "loss": 0.0717,
+      "step": 6165
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.882535350507195e-05,
+      "loss": 0.0698,
+      "step": 6166
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8797333146000084e-05,
+      "loss": 0.0717,
+      "step": 6167
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8769324121501616e-05,
+      "loss": 0.0717,
+      "step": 6168
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8741326436035255e-05,
+      "loss": 0.0701,
+      "step": 6169
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.871334009405785e-05,
+      "loss": 0.0744,
+      "step": 6170
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8685365100024453e-05,
+      "loss": 0.0701,
+      "step": 6171
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8657401458388367e-05,
+      "loss": 0.0697,
+      "step": 6172
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8629449173601064e-05,
+      "loss": 0.0717,
+      "step": 6173
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8601508250112162e-05,
+      "loss": 0.0712,
+      "step": 6174
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.857357869236952e-05,
+      "loss": 0.0746,
+      "step": 6175
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.8545660504819206e-05,
+      "loss": 0.0724,
+      "step": 6176
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.851775369190539e-05,
+      "loss": 0.0725,
+      "step": 6177
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8489858258070514e-05,
+      "loss": 0.0708,
+      "step": 6178
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8461974207755215e-05,
+      "loss": 0.0707,
+      "step": 6179
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8434101545398252e-05,
+      "loss": 0.0739,
+      "step": 6180
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.840624027543658e-05,
+      "loss": 0.0701,
+      "step": 6181
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8378390402305388e-05,
+      "loss": 0.0744,
+      "step": 6182
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8350551930438064e-05,
+      "loss": 0.0715,
+      "step": 6183
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.832272486426608e-05,
+      "loss": 0.0712,
+      "step": 6184
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8294909208219178e-05,
+      "loss": 0.0724,
+      "step": 6185
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.826710496672531e-05,
+      "loss": 0.0719,
+      "step": 6186
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8239312144210516e-05,
+      "loss": 0.0765,
+      "step": 6187
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.821153074509901e-05,
+      "loss": 0.073,
+      "step": 6188
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.8183760773813383e-05,
+      "loss": 0.0713,
+      "step": 6189
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.815600223477418e-05,
+      "loss": 0.07,
+      "step": 6190
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.8128255132400194e-05,
+      "loss": 0.0726,
+      "step": 6191
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.8100519471108445e-05,
+      "loss": 0.0714,
+      "step": 6192
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.8072795255314134e-05,
+      "loss": 0.0714,
+      "step": 6193
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.8045082489430552e-05,
+      "loss": 0.0706,
+      "step": 6194
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.801738117786925e-05,
+      "loss": 0.0725,
+      "step": 6195
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.798969132503997e-05,
+      "loss": 0.0715,
+      "step": 6196
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7962012935350535e-05,
+      "loss": 0.0688,
+      "step": 6197
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.793434601320697e-05,
+      "loss": 0.0736,
+      "step": 6198
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7906690563013592e-05,
+      "loss": 0.0684,
+      "step": 6199
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7879046589172775e-05,
+      "loss": 0.0707,
+      "step": 6200
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7851414096085036e-05,
+      "loss": 0.0689,
+      "step": 6201
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7823793088149164e-05,
+      "loss": 0.0721,
+      "step": 6202
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7796183569762103e-05,
+      "loss": 0.0724,
+      "step": 6203
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7768585545318893e-05,
+      "loss": 0.0732,
+      "step": 6204
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.774099901921282e-05,
+      "loss": 0.074,
+      "step": 6205
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7713423995835343e-05,
+      "loss": 0.0743,
+      "step": 6206
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.768586047957602e-05,
+      "loss": 0.0723,
+      "step": 6207
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7658308474822568e-05,
+      "loss": 0.0739,
+      "step": 6208
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7630767985961038e-05,
+      "loss": 0.0732,
+      "step": 6209
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7603239017375482e-05,
+      "loss": 0.0784,
+      "step": 6210
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.757572157344812e-05,
+      "loss": 0.0711,
+      "step": 6211
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.754821565855942e-05,
+      "loss": 0.0725,
+      "step": 6212
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7520721277088024e-05,
+      "loss": 0.0709,
+      "step": 6213
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7493238433410605e-05,
+      "loss": 0.0712,
+      "step": 6214
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7465767131902153e-05,
+      "loss": 0.0696,
+      "step": 6215
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7438307376935757e-05,
+      "loss": 0.0716,
+      "step": 6216
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.741085917288264e-05,
+      "loss": 0.0735,
+      "step": 6217
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.738342252411217e-05,
+      "loss": 0.0728,
+      "step": 6218
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.735599743499202e-05,
+      "loss": 0.0705,
+      "step": 6219
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7328583909887872e-05,
+      "loss": 0.069,
+      "step": 6220
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.730118195316358e-05,
+      "loss": 0.076,
+      "step": 6221
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.727379156918123e-05,
+      "loss": 0.0732,
+      "step": 6222
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7246412762301043e-05,
+      "loss": 0.0665,
+      "step": 6223
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.721904553688134e-05,
+      "loss": 0.067,
+      "step": 6224
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.719168989727866e-05,
+      "loss": 0.0735,
+      "step": 6225
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7164345847847705e-05,
+      "loss": 0.0669,
+      "step": 6226
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.713701339294129e-05,
+      "loss": 0.0719,
+      "step": 6227
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7109692536910357e-05,
+      "loss": 0.0719,
+      "step": 6228
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7082383284104083e-05,
+      "loss": 0.072,
+      "step": 6229
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.7055085638869783e-05,
+      "loss": 0.071,
+      "step": 6230
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.702779960555284e-05,
+      "loss": 0.0732,
+      "step": 6231
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.7000525188496884e-05,
+      "loss": 0.0708,
+      "step": 6232
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6973262392043685e-05,
+      "loss": 0.0725,
+      "step": 6233
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6946011220533084e-05,
+      "loss": 0.0702,
+      "step": 6234
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.691877167830319e-05,
+      "loss": 0.0699,
+      "step": 6235
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.689154376969012e-05,
+      "loss": 0.0746,
+      "step": 6236
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6864327499028295e-05,
+      "loss": 0.0723,
+      "step": 6237
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6837122870650134e-05,
+      "loss": 0.0711,
+      "step": 6238
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.680992988888631e-05,
+      "loss": 0.069,
+      "step": 6239
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.678274855806564e-05,
+      "loss": 0.0731,
+      "step": 6240
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6755578882514976e-05,
+      "loss": 0.0668,
+      "step": 6241
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6728420866559423e-05,
+      "loss": 0.0675,
+      "step": 6242
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6701274514522246e-05,
+      "loss": 0.072,
+      "step": 6243
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.667413983072472e-05,
+      "loss": 0.0706,
+      "step": 6244
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6647016819486424e-05,
+      "loss": 0.0728,
+      "step": 6245
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.661990548512493e-05,
+      "loss": 0.0703,
+      "step": 6246
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6592805831956104e-05,
+      "loss": 0.073,
+      "step": 6247
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6565717864293783e-05,
+      "loss": 0.0694,
+      "step": 6248
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6538641586450074e-05,
+      "loss": 0.0712,
+      "step": 6249
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.651157700273522e-05,
+      "loss": 0.0707,
+      "step": 6250
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.64845241174575e-05,
+      "loss": 0.0752,
+      "step": 6251
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.645748293492345e-05,
+      "loss": 0.0726,
+      "step": 6252
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6430453459437632e-05,
+      "loss": 0.0752,
+      "step": 6253
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6403435695302825e-05,
+      "loss": 0.0724,
+      "step": 6254
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6376429646819956e-05,
+      "loss": 0.0717,
+      "step": 6255
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6349435318287973e-05,
+      "loss": 0.0706,
+      "step": 6256
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.632245271400412e-05,
+      "loss": 0.0705,
+      "step": 6257
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6295481838263626e-05,
+      "loss": 0.0714,
+      "step": 6258
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6268522695359932e-05,
+      "loss": 0.074,
+      "step": 6259
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.624157528958463e-05,
+      "loss": 0.0708,
+      "step": 6260
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6214639625227356e-05,
+      "loss": 0.0688,
+      "step": 6261
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6187715706575987e-05,
+      "loss": 0.0691,
+      "step": 6262
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.616080353791641e-05,
+      "loss": 0.0697,
+      "step": 6263
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6133903123532744e-05,
+      "loss": 0.073,
+      "step": 6264
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6107014467707237e-05,
+      "loss": 0.071,
+      "step": 6265
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.608013757472013e-05,
+      "loss": 0.072,
+      "step": 6266
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6053272448849986e-05,
+      "loss": 0.0727,
+      "step": 6267
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.6026419094373322e-05,
+      "loss": 0.0751,
+      "step": 6268
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.599957751556492e-05,
+      "loss": 0.0706,
+      "step": 6269
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.5972747716697553e-05,
+      "loss": 0.0716,
+      "step": 6270
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.5945929702042226e-05,
+      "loss": 0.0722,
+      "step": 6271
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5919123475868067e-05,
+      "loss": 0.0746,
+      "step": 6272
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.589232904244222e-05,
+      "loss": 0.0717,
+      "step": 6273
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5865546406030062e-05,
+      "loss": 0.0738,
+      "step": 6274
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.583877557089508e-05,
+      "loss": 0.0701,
+      "step": 6275
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5812016541298834e-05,
+      "loss": 0.0699,
+      "step": 6276
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.578526932150098e-05,
+      "loss": 0.0675,
+      "step": 6277
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5758533915759386e-05,
+      "loss": 0.073,
+      "step": 6278
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.573181032833002e-05,
+      "loss": 0.0723,
+      "step": 6279
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5705098563466866e-05,
+      "loss": 0.0717,
+      "step": 6280
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.567839862542216e-05,
+      "loss": 0.0724,
+      "step": 6281
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5651710518446215e-05,
+      "loss": 0.0716,
+      "step": 6282
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5625034246787382e-05,
+      "loss": 0.0667,
+      "step": 6283
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.5598369814692224e-05,
+      "loss": 0.0715,
+      "step": 6284
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5571717226405413e-05,
+      "loss": 0.0712,
+      "step": 6285
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5545076486169673e-05,
+      "loss": 0.0698,
+      "step": 6286
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5518447598225847e-05,
+      "loss": 0.072,
+      "step": 6287
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5491830566812958e-05,
+      "loss": 0.0716,
+      "step": 6288
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5465225396168134e-05,
+      "loss": 0.0718,
+      "step": 6289
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5438632090526516e-05,
+      "loss": 0.0701,
+      "step": 6290
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5412050654121468e-05,
+      "loss": 0.0728,
+      "step": 6291
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.538548109118444e-05,
+      "loss": 0.0696,
+      "step": 6292
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5358923405944946e-05,
+      "loss": 0.0727,
+      "step": 6293
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5332377602630584e-05,
+      "loss": 0.0716,
+      "step": 6294
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.530584368546722e-05,
+      "loss": 0.0726,
+      "step": 6295
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.527932165867868e-05,
+      "loss": 0.0709,
+      "step": 6296
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5252811526486874e-05,
+      "loss": 0.0732,
+      "step": 6297
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5226313293111992e-05,
+      "loss": 0.0761,
+      "step": 6298
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.5199826962772166e-05,
+      "loss": 0.0746,
+      "step": 6299
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.5173352539683658e-05,
+      "loss": 0.0733,
+      "step": 6300
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.514689002806091e-05,
+      "loss": 0.0716,
+      "step": 6301
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.5120439432116427e-05,
+      "loss": 0.0744,
+      "step": 6302
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.50940007560608e-05,
+      "loss": 0.0689,
+      "step": 6303
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.5067574004102678e-05,
+      "loss": 0.0712,
+      "step": 6304
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.504115918044897e-05,
+      "loss": 0.0705,
+      "step": 6305
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.5014756289304543e-05,
+      "loss": 0.0703,
+      "step": 6306
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4988365334872345e-05,
+      "loss": 0.0731,
+      "step": 6307
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.496198632135359e-05,
+      "loss": 0.0716,
+      "step": 6308
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.493561925294745e-05,
+      "loss": 0.0706,
+      "step": 6309
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4909264133851186e-05,
+      "loss": 0.0715,
+      "step": 6310
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4882920968260238e-05,
+      "loss": 0.0745,
+      "step": 6311
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4856589760368143e-05,
+      "loss": 0.0695,
+      "step": 6312
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.483027051436646e-05,
+      "loss": 0.0741,
+      "step": 6313
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4803963234444826e-05,
+      "loss": 0.071,
+      "step": 6314
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.477766792479116e-05,
+      "loss": 0.0729,
+      "step": 6315
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4751384589591277e-05,
+      "loss": 0.0721,
+      "step": 6316
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.472511323302913e-05,
+      "loss": 0.0725,
+      "step": 6317
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4698853859286807e-05,
+      "loss": 0.0708,
+      "step": 6318
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4672606472544524e-05,
+      "loss": 0.0732,
+      "step": 6319
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4646371076980457e-05,
+      "loss": 0.0735,
+      "step": 6320
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4620147676770987e-05,
+      "loss": 0.071,
+      "step": 6321
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.459393627609059e-05,
+      "loss": 0.0709,
+      "step": 6322
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.456773687911176e-05,
+      "loss": 0.0692,
+      "step": 6323
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.4541549490005056e-05,
+      "loss": 0.0738,
+      "step": 6324
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.451537411293928e-05,
+      "loss": 0.0725,
+      "step": 6325
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4489210752081194e-05,
+      "loss": 0.07,
+      "step": 6326
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4463059411595634e-05,
+      "loss": 0.0704,
+      "step": 6327
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.443692009564559e-05,
+      "loss": 0.0765,
+      "step": 6328
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4410792808392156e-05,
+      "loss": 0.0744,
+      "step": 6329
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4384677553994406e-05,
+      "loss": 0.0763,
+      "step": 6330
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4358574336609585e-05,
+      "loss": 0.0703,
+      "step": 6331
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.433248316039304e-05,
+      "loss": 0.0713,
+      "step": 6332
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4306404029498108e-05,
+      "loss": 0.0698,
+      "step": 6333
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4280336948076267e-05,
+      "loss": 0.0712,
+      "step": 6334
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4254281920277056e-05,
+      "loss": 0.0713,
+      "step": 6335
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.422823895024817e-05,
+      "loss": 0.0711,
+      "step": 6336
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4202208042135266e-05,
+      "loss": 0.0732,
+      "step": 6337
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4176189200082146e-05,
+      "loss": 0.0685,
+      "step": 6338
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.415018242823073e-05,
+      "loss": 0.0717,
+      "step": 6339
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4124187730720917e-05,
+      "loss": 0.0716,
+      "step": 6340
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4098205111690774e-05,
+      "loss": 0.0714,
+      "step": 6341
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4072234575276364e-05,
+      "loss": 0.0677,
+      "step": 6342
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.40462761256119e-05,
+      "loss": 0.0726,
+      "step": 6343
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4020329766829675e-05,
+      "loss": 0.0706,
+      "step": 6344
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3994395503059953e-05,
+      "loss": 0.0742,
+      "step": 6345
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.396847333843122e-05,
+      "loss": 0.0718,
+      "step": 6346
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3942563277069897e-05,
+      "loss": 0.0755,
+      "step": 6347
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3916665323100564e-05,
+      "loss": 0.0713,
+      "step": 6348
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3890779480645875e-05,
+      "loss": 0.0714,
+      "step": 6349
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3864905753826485e-05,
+      "loss": 0.0708,
+      "step": 6350
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3839044146761225e-05,
+      "loss": 0.0726,
+      "step": 6351
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3813194663566886e-05,
+      "loss": 0.0711,
+      "step": 6352
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3787357308358395e-05,
+      "loss": 0.0709,
+      "step": 6353
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3761532085248784e-05,
+      "loss": 0.0708,
+      "step": 6354
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.373571899834902e-05,
+      "loss": 0.0739,
+      "step": 6355
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.370991805176832e-05,
+      "loss": 0.0741,
+      "step": 6356
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3684129249613774e-05,
+      "loss": 0.073,
+      "step": 6357
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3658352595990717e-05,
+      "loss": 0.0706,
+      "step": 6358
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3632588095002418e-05,
+      "loss": 0.0715,
+      "step": 6359
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3606835750750266e-05,
+      "loss": 0.0721,
+      "step": 6360
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3581095567333767e-05,
+      "loss": 0.0707,
+      "step": 6361
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.355536754885037e-05,
+      "loss": 0.0706,
+      "step": 6362
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3529651699395672e-05,
+      "loss": 0.0704,
+      "step": 6363
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.350394802306335e-05,
+      "loss": 0.0702,
+      "step": 6364
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3478256523945065e-05,
+      "loss": 0.0734,
+      "step": 6365
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.345257720613061e-05,
+      "loss": 0.0746,
+      "step": 6366
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3426910073707775e-05,
+      "loss": 0.0716,
+      "step": 6367
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.34012551307625e-05,
+      "loss": 0.0717,
+      "step": 6368
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3375612381378665e-05,
+      "loss": 0.0715,
+      "step": 6369
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3349981829638312e-05,
+      "loss": 0.07,
+      "step": 6370
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3324363479621536e-05,
+      "loss": 0.074,
+      "step": 6371
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3298757335406396e-05,
+      "loss": 0.0709,
+      "step": 6372
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3273163401069086e-05,
+      "loss": 0.0719,
+      "step": 6373
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3247581680683885e-05,
+      "loss": 0.0702,
+      "step": 6374
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.322201217832305e-05,
+      "loss": 0.0732,
+      "step": 6375
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3196454898056895e-05,
+      "loss": 0.0716,
+      "step": 6376
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3170909843953857e-05,
+      "loss": 0.0716,
+      "step": 6377
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3145377020080404e-05,
+      "loss": 0.0677,
+      "step": 6378
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3119856430500987e-05,
+      "loss": 0.0723,
+      "step": 6379
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3094348079278192e-05,
+      "loss": 0.0724,
+      "step": 6380
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3068851970472673e-05,
+      "loss": 0.0717,
+      "step": 6381
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.3043368108143047e-05,
+      "loss": 0.071,
+      "step": 6382
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.301789649634597e-05,
+      "loss": 0.0743,
+      "step": 6383
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2992437139136324e-05,
+      "loss": 0.0664,
+      "step": 6384
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.296699004056686e-05,
+      "loss": 0.0726,
+      "step": 6385
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2941555204688413e-05,
+      "loss": 0.0739,
+      "step": 6386
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2916132635549915e-05,
+      "loss": 0.08,
+      "step": 6387
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2890722337198346e-05,
+      "loss": 0.0691,
+      "step": 6388
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.286532431367866e-05,
+      "loss": 0.0726,
+      "step": 6389
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2839938569033924e-05,
+      "loss": 0.0723,
+      "step": 6390
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2814565107305262e-05,
+      "loss": 0.0704,
+      "step": 6391
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.278920393253179e-05,
+      "loss": 0.071,
+      "step": 6392
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2763855048750616e-05,
+      "loss": 0.0717,
+      "step": 6393
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2738518459997094e-05,
+      "loss": 0.0712,
+      "step": 6394
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.271319417030443e-05,
+      "loss": 0.0728,
+      "step": 6395
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2687882183703914e-05,
+      "loss": 0.0717,
+      "step": 6396
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2662582504224906e-05,
+      "loss": 0.0712,
+      "step": 6397
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2637295135894843e-05,
+      "loss": 0.0744,
+      "step": 6398
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2612020082739094e-05,
+      "loss": 0.0737,
+      "step": 6399
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2586757348781162e-05,
+      "loss": 0.0723,
+      "step": 6400
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2561506938042586e-05,
+      "loss": 0.0692,
+      "step": 6401
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2536268854542886e-05,
+      "loss": 0.0693,
+      "step": 6402
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.25110431022996e-05,
+      "loss": 0.0705,
+      "step": 6403
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2485829685328452e-05,
+      "loss": 0.0717,
+      "step": 6404
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.2460628607643052e-05,
+      "loss": 0.0706,
+      "step": 6405
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.243543987325508e-05,
+      "loss": 0.0717,
+      "step": 6406
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2410263486174276e-05,
+      "loss": 0.0712,
+      "step": 6407
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2385099450408442e-05,
+      "loss": 0.0726,
+      "step": 6408
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2359947769963318e-05,
+      "loss": 0.074,
+      "step": 6409
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.233480844884277e-05,
+      "loss": 0.0726,
+      "step": 6410
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2309681491048685e-05,
+      "loss": 0.0687,
+      "step": 6411
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2284566900580937e-05,
+      "loss": 0.0697,
+      "step": 6412
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2259464681437404e-05,
+      "loss": 0.0711,
+      "step": 6413
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2234374837614148e-05,
+      "loss": 0.0755,
+      "step": 6414
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.220929737310511e-05,
+      "loss": 0.0747,
+      "step": 6415
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2184232291902273e-05,
+      "loss": 0.0697,
+      "step": 6416
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2159179597995705e-05,
+      "loss": 0.0737,
+      "step": 6417
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2134139295373533e-05,
+      "loss": 0.0749,
+      "step": 6418
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.2109111388021787e-05,
+      "loss": 0.0723,
+      "step": 6419
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.2084095879924616e-05,
+      "loss": 0.073,
+      "step": 6420
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.205909277506423e-05,
+      "loss": 0.0726,
+      "step": 6421
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.2034102077420758e-05,
+      "loss": 0.0756,
+      "step": 6422
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.2009123790972385e-05,
+      "loss": 0.0739,
+      "step": 6423
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1984157919695382e-05,
+      "loss": 0.073,
+      "step": 6424
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1959204467564022e-05,
+      "loss": 0.0701,
+      "step": 6425
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.193426343855052e-05,
+      "loss": 0.0698,
+      "step": 6426
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1909334836625216e-05,
+      "loss": 0.0716,
+      "step": 6427
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1884418665756444e-05,
+      "loss": 0.0691,
+      "step": 6428
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1859514929910508e-05,
+      "loss": 0.0727,
+      "step": 6429
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1834623633051787e-05,
+      "loss": 0.0743,
+      "step": 6430
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1809744779142693e-05,
+      "loss": 0.0692,
+      "step": 6431
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1784878372143602e-05,
+      "loss": 0.0745,
+      "step": 6432
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.1760024416012914e-05,
+      "loss": 0.0725,
+      "step": 6433
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1735182914707087e-05,
+      "loss": 0.0714,
+      "step": 6434
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.17103538721806e-05,
+      "loss": 0.0724,
+      "step": 6435
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1685537292385882e-05,
+      "loss": 0.0727,
+      "step": 6436
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.166073317927345e-05,
+      "loss": 0.0711,
+      "step": 6437
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1635941536791826e-05,
+      "loss": 0.0727,
+      "step": 6438
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1611162368887484e-05,
+      "loss": 0.0713,
+      "step": 6439
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1586395679505e-05,
+      "loss": 0.0733,
+      "step": 6440
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.156164147258688e-05,
+      "loss": 0.0709,
+      "step": 6441
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1536899752073724e-05,
+      "loss": 0.0752,
+      "step": 6442
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1512170521904062e-05,
+      "loss": 0.07,
+      "step": 6443
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1487453786014512e-05,
+      "loss": 0.0708,
+      "step": 6444
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1462749548339668e-05,
+      "loss": 0.0701,
+      "step": 6445
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1438057812812108e-05,
+      "loss": 0.0694,
+      "step": 6446
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.1413378583362487e-05,
+      "loss": 0.0714,
+      "step": 6447
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1388711863919387e-05,
+      "loss": 0.0707,
+      "step": 6448
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1364057658409454e-05,
+      "loss": 0.0712,
+      "step": 6449
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.133941597075737e-05,
+      "loss": 0.0717,
+      "step": 6450
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1314786804885712e-05,
+      "loss": 0.067,
+      "step": 6451
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1290170164715205e-05,
+      "loss": 0.0723,
+      "step": 6452
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1265566054164456e-05,
+      "loss": 0.0713,
+      "step": 6453
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1240974477150143e-05,
+      "loss": 0.0721,
+      "step": 6454
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1216395437586977e-05,
+      "loss": 0.0727,
+      "step": 6455
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.119182893938758e-05,
+      "loss": 0.0715,
+      "step": 6456
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.116727498646268e-05,
+      "loss": 0.0678,
+      "step": 6457
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1142733582720907e-05,
+      "loss": 0.0739,
+      "step": 6458
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1118204732068982e-05,
+      "loss": 0.0703,
+      "step": 6459
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.1093688438411597e-05,
+      "loss": 0.0733,
+      "step": 6460
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.106918470565139e-05,
+      "loss": 0.0697,
+      "step": 6461
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.1044693537689118e-05,
+      "loss": 0.0681,
+      "step": 6462
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.10202149384234e-05,
+      "loss": 0.0733,
+      "step": 6463
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0995748911750978e-05,
+      "loss": 0.0696,
+      "step": 6464
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0971295461566498e-05,
+      "loss": 0.0761,
+      "step": 6465
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0946854591762653e-05,
+      "loss": 0.0738,
+      "step": 6466
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.092242630623016e-05,
+      "loss": 0.0681,
+      "step": 6467
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0898010608857633e-05,
+      "loss": 0.0705,
+      "step": 6468
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.087360750353178e-05,
+      "loss": 0.0748,
+      "step": 6469
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.08492169941373e-05,
+      "loss": 0.0711,
+      "step": 6470
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0824839084556823e-05,
+      "loss": 0.0699,
+      "step": 6471
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0800473778670983e-05,
+      "loss": 0.0723,
+      "step": 6472
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0776121080358446e-05,
+      "loss": 0.0727,
+      "step": 6473
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.0751780993495907e-05,
+      "loss": 0.0705,
+      "step": 6474
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.072745352195794e-05,
+      "loss": 0.0723,
+      "step": 6475
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0703138669617196e-05,
+      "loss": 0.0711,
+      "step": 6476
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.067883644034433e-05,
+      "loss": 0.073,
+      "step": 6477
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0654546838007882e-05,
+      "loss": 0.0685,
+      "step": 6478
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0630269866474495e-05,
+      "loss": 0.073,
+      "step": 6479
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0606005529608786e-05,
+      "loss": 0.0706,
+      "step": 6480
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0581753831273308e-05,
+      "loss": 0.0716,
+      "step": 6481
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.055751477532859e-05,
+      "loss": 0.0689,
+      "step": 6482
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.053328836563322e-05,
+      "loss": 0.0698,
+      "step": 6483
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0509074606043765e-05,
+      "loss": 0.0719,
+      "step": 6484
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0484873500414702e-05,
+      "loss": 0.0687,
+      "step": 6485
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.0460685052598562e-05,
+      "loss": 0.075,
+      "step": 6486
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.043650926644588e-05,
+      "loss": 0.0711,
+      "step": 6487
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0412346145805104e-05,
+      "loss": 0.0695,
+      "step": 6488
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.038819569452266e-05,
+      "loss": 0.0697,
+      "step": 6489
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0364057916443092e-05,
+      "loss": 0.069,
+      "step": 6490
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.033993281540878e-05,
+      "loss": 0.0707,
+      "step": 6491
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.031582039526011e-05,
+      "loss": 0.0763,
+      "step": 6492
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0291720659835524e-05,
+      "loss": 0.0706,
+      "step": 6493
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0267633612971394e-05,
+      "loss": 0.0719,
+      "step": 6494
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0243559258502033e-05,
+      "loss": 0.0733,
+      "step": 6495
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.021949760025982e-05,
+      "loss": 0.0739,
+      "step": 6496
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0195448642075075e-05,
+      "loss": 0.0702,
+      "step": 6497
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.017141238777608e-05,
+      "loss": 0.0728,
+      "step": 6498
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0147388841189054e-05,
+      "loss": 0.0717,
+      "step": 6499
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0123378006138327e-05,
+      "loss": 0.0728,
+      "step": 6500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.00993798864461e-05,
+      "loss": 0.0728,
+      "step": 6501
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.007539448593253e-05,
+      "loss": 0.0698,
+      "step": 6502
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.0051421808415827e-05,
+      "loss": 0.0682,
+      "step": 6503
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.002746185771216e-05,
+      "loss": 0.0756,
+      "step": 6504
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.0003514637635612e-05,
+      "loss": 0.0722,
+      "step": 6505
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.997958015199829e-05,
+      "loss": 0.0659,
+      "step": 6506
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.99556584046103e-05,
+      "loss": 0.0708,
+      "step": 6507
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9931749399279664e-05,
+      "loss": 0.0723,
+      "step": 6508
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9907853139812337e-05,
+      "loss": 0.072,
+      "step": 6509
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.988396963001241e-05,
+      "loss": 0.0719,
+      "step": 6510
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9860098873681787e-05,
+      "loss": 0.0707,
+      "step": 6511
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.983624087462037e-05,
+      "loss": 0.0735,
+      "step": 6512
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.981239563662608e-05,
+      "loss": 0.0693,
+      "step": 6513
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.97885631634948e-05,
+      "loss": 0.068,
+      "step": 6514
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.976474345902032e-05,
+      "loss": 0.0735,
+      "step": 6515
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9740936526994458e-05,
+      "loss": 0.0722,
+      "step": 6516
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9717142371206998e-05,
+      "loss": 0.0734,
+      "step": 6517
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9693360995445663e-05,
+      "loss": 0.073,
+      "step": 6518
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9669592403496085e-05,
+      "loss": 0.0713,
+      "step": 6519
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9645836599142032e-05,
+      "loss": 0.0688,
+      "step": 6520
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.962209358616508e-05,
+      "loss": 0.072,
+      "step": 6521
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9598363368344807e-05,
+      "loss": 0.0717,
+      "step": 6522
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.957464594945877e-05,
+      "loss": 0.0694,
+      "step": 6523
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.9550941333282523e-05,
+      "loss": 0.0705,
+      "step": 6524
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.95272495235895e-05,
+      "loss": 0.0719,
+      "step": 6525
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.950357052415115e-05,
+      "loss": 0.0693,
+      "step": 6526
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.947990433873691e-05,
+      "loss": 0.0745,
+      "step": 6527
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.945625097111411e-05,
+      "loss": 0.0709,
+      "step": 6528
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9432610425048046e-05,
+      "loss": 0.076,
+      "step": 6529
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9408982704302015e-05,
+      "loss": 0.0711,
+      "step": 6530
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.93853678126373e-05,
+      "loss": 0.0731,
+      "step": 6531
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9361765753813033e-05,
+      "loss": 0.0736,
+      "step": 6532
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9338176531586383e-05,
+      "loss": 0.0729,
+      "step": 6533
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.931460014971249e-05,
+      "loss": 0.0743,
+      "step": 6534
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.929103661194438e-05,
+      "loss": 0.0717,
+      "step": 6535
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9267485922033114e-05,
+      "loss": 0.0714,
+      "step": 6536
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9243948083727626e-05,
+      "loss": 0.0734,
+      "step": 6537
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9220423100774877e-05,
+      "loss": 0.0729,
+      "step": 6538
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9196910976919703e-05,
+      "loss": 0.071,
+      "step": 6539
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9173411715904988e-05,
+      "loss": 0.0739,
+      "step": 6540
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.9149925321471517e-05,
+      "loss": 0.0711,
+      "step": 6541
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.9126451797357992e-05,
+      "loss": 0.0729,
+      "step": 6542
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.9102991147301142e-05,
+      "loss": 0.073,
+      "step": 6543
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.90795433750356e-05,
+      "loss": 0.0723,
+      "step": 6544
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.9056108484293944e-05,
+      "loss": 0.0726,
+      "step": 6545
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.903268647880675e-05,
+      "loss": 0.0702,
+      "step": 6546
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.9009277362302458e-05,
+      "loss": 0.0693,
+      "step": 6547
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8985881138507555e-05,
+      "loss": 0.0716,
+      "step": 6548
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8962497811146386e-05,
+      "loss": 0.0691,
+      "step": 6549
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8939127383941292e-05,
+      "loss": 0.074,
+      "step": 6550
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8915769860612588e-05,
+      "loss": 0.0727,
+      "step": 6551
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8892425244878452e-05,
+      "loss": 0.0696,
+      "step": 6552
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8869093540455108e-05,
+      "loss": 0.0722,
+      "step": 6553
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.88457747510566e-05,
+      "loss": 0.0718,
+      "step": 6554
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.8822468880395038e-05,
+      "loss": 0.0719,
+      "step": 6555
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8799175932180436e-05,
+      "loss": 0.0726,
+      "step": 6556
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.877589591012069e-05,
+      "loss": 0.0729,
+      "step": 6557
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8752628817921747e-05,
+      "loss": 0.0713,
+      "step": 6558
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.872937465928738e-05,
+      "loss": 0.0688,
+      "step": 6559
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8706133437919382e-05,
+      "loss": 0.0752,
+      "step": 6560
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8682905157517493e-05,
+      "loss": 0.0706,
+      "step": 6561
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8659689821779324e-05,
+      "loss": 0.0718,
+      "step": 6562
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8636487434400496e-05,
+      "loss": 0.0747,
+      "step": 6563
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.861329799907451e-05,
+      "loss": 0.0711,
+      "step": 6564
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8590121519492843e-05,
+      "loss": 0.0703,
+      "step": 6565
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8566957999344924e-05,
+      "loss": 0.0712,
+      "step": 6566
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8543807442318063e-05,
+      "loss": 0.075,
+      "step": 6567
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.8520669852097573e-05,
+      "loss": 0.0721,
+      "step": 6568
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.849754523236663e-05,
+      "loss": 0.0727,
+      "step": 6569
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8474433586806415e-05,
+      "loss": 0.0715,
+      "step": 6570
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8451334919095987e-05,
+      "loss": 0.0695,
+      "step": 6571
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8428249232912375e-05,
+      "loss": 0.0738,
+      "step": 6572
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8405176531930567e-05,
+      "loss": 0.0733,
+      "step": 6573
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.838211681982338e-05,
+      "loss": 0.0708,
+      "step": 6574
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8359070100261677e-05,
+      "loss": 0.0713,
+      "step": 6575
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8336036376914212e-05,
+      "loss": 0.0722,
+      "step": 6576
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8313015653447663e-05,
+      "loss": 0.0717,
+      "step": 6577
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8290007933526597e-05,
+      "loss": 0.072,
+      "step": 6578
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.826701322081359e-05,
+      "loss": 0.0724,
+      "step": 6579
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.824403151896914e-05,
+      "loss": 0.0761,
+      "step": 6580
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.822106283165158e-05,
+      "loss": 0.07,
+      "step": 6581
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.8198107162517285e-05,
+      "loss": 0.0691,
+      "step": 6582
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8175164515220523e-05,
+      "loss": 0.0731,
+      "step": 6583
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8152234893413424e-05,
+      "loss": 0.0737,
+      "step": 6584
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8129318300746124e-05,
+      "loss": 0.071,
+      "step": 6585
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8106414740866683e-05,
+      "loss": 0.0691,
+      "step": 6586
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8083524217421045e-05,
+      "loss": 0.0748,
+      "step": 6587
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8060646734053054e-05,
+      "loss": 0.0692,
+      "step": 6588
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.803778229440455e-05,
+      "loss": 0.0715,
+      "step": 6589
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.80149309021153e-05,
+      "loss": 0.0698,
+      "step": 6590
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7992092560822894e-05,
+      "loss": 0.0723,
+      "step": 6591
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.796926727416295e-05,
+      "loss": 0.0713,
+      "step": 6592
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7946455045768983e-05,
+      "loss": 0.0704,
+      "step": 6593
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7923655879272393e-05,
+      "loss": 0.0738,
+      "step": 6594
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7900869778302477e-05,
+      "loss": 0.0685,
+      "step": 6595
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7878096746486582e-05,
+      "loss": 0.0674,
+      "step": 6596
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.785533678744985e-05,
+      "loss": 0.0708,
+      "step": 6597
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7832589904815366e-05,
+      "loss": 0.0674,
+      "step": 6598
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7809856102204147e-05,
+      "loss": 0.0715,
+      "step": 6599
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.778713538323519e-05,
+      "loss": 0.0724,
+      "step": 6600
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.776442775152527e-05,
+      "loss": 0.075,
+      "step": 6601
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7741733210689193e-05,
+      "loss": 0.0714,
+      "step": 6602
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.771905176433968e-05,
+      "loss": 0.0749,
+      "step": 6603
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7696383416087292e-05,
+      "loss": 0.0706,
+      "step": 6604
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7673728169540505e-05,
+      "loss": 0.0702,
+      "step": 6605
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.765108602830584e-05,
+      "loss": 0.0737,
+      "step": 6606
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.762845699598762e-05,
+      "loss": 0.0698,
+      "step": 6607
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.7605841076188058e-05,
+      "loss": 0.0758,
+      "step": 6608
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.758323827250734e-05,
+      "loss": 0.0714,
+      "step": 6609
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7560648588543595e-05,
+      "loss": 0.0724,
+      "step": 6610
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7538072027892762e-05,
+      "loss": 0.0678,
+      "step": 6611
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7515508594148768e-05,
+      "loss": 0.0699,
+      "step": 6612
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.749295829090346e-05,
+      "loss": 0.0742,
+      "step": 6613
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7470421121746538e-05,
+      "loss": 0.0727,
+      "step": 6614
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7447897090265585e-05,
+      "loss": 0.0733,
+      "step": 6615
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7425386200046234e-05,
+      "loss": 0.0733,
+      "step": 6616
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7402888454671907e-05,
+      "loss": 0.0722,
+      "step": 6617
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7380403857723925e-05,
+      "loss": 0.0739,
+      "step": 6618
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7357932412781598e-05,
+      "loss": 0.0713,
+      "step": 6619
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.733547412342209e-05,
+      "loss": 0.0703,
+      "step": 6620
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7313028993220458e-05,
+      "loss": 0.072,
+      "step": 6621
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.72905970257497e-05,
+      "loss": 0.0721,
+      "step": 6622
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7268178224580735e-05,
+      "loss": 0.0738,
+      "step": 6623
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7245772593282327e-05,
+      "loss": 0.0713,
+      "step": 6624
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7223380135421117e-05,
+      "loss": 0.0714,
+      "step": 6625
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.720100085456182e-05,
+      "loss": 0.0723,
+      "step": 6626
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7178634754266866e-05,
+      "loss": 0.0678,
+      "step": 6627
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7156281838096643e-05,
+      "loss": 0.0746,
+      "step": 6628
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7133942109609492e-05,
+      "loss": 0.069,
+      "step": 6629
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7111615572361628e-05,
+      "loss": 0.0742,
+      "step": 6630
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.70893022299071e-05,
+      "loss": 0.0707,
+      "step": 6631
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7067002085797966e-05,
+      "loss": 0.074,
+      "step": 6632
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7044715143584123e-05,
+      "loss": 0.0735,
+      "step": 6633
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.702244140681337e-05,
+      "loss": 0.07,
+      "step": 6634
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.7000180879031367e-05,
+      "loss": 0.0709,
+      "step": 6635
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6977933563781758e-05,
+      "loss": 0.0691,
+      "step": 6636
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6955699464606047e-05,
+      "loss": 0.0698,
+      "step": 6637
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.693347858504356e-05,
+      "loss": 0.0678,
+      "step": 6638
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.691127092863164e-05,
+      "loss": 0.0727,
+      "step": 6639
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6889076498905464e-05,
+      "loss": 0.072,
+      "step": 6640
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.686689529939808e-05,
+      "loss": 0.073,
+      "step": 6641
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6844727333640486e-05,
+      "loss": 0.0722,
+      "step": 6642
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6822572605161514e-05,
+      "loss": 0.0708,
+      "step": 6643
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6800431117487956e-05,
+      "loss": 0.0725,
+      "step": 6644
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6778302874144426e-05,
+      "loss": 0.0737,
+      "step": 6645
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.675618787865347e-05,
+      "loss": 0.0715,
+      "step": 6646
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.6734086134535552e-05,
+      "loss": 0.0699,
+      "step": 6647
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.671199764530894e-05,
+      "loss": 0.0702,
+      "step": 6648
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.668992241448988e-05,
+      "loss": 0.0728,
+      "step": 6649
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6667860445592477e-05,
+      "loss": 0.0738,
+      "step": 6650
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6645811742128692e-05,
+      "loss": 0.0703,
+      "step": 6651
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.662377630760844e-05,
+      "loss": 0.0721,
+      "step": 6652
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.660175414553944e-05,
+      "loss": 0.0721,
+      "step": 6653
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6579745259427394e-05,
+      "loss": 0.0698,
+      "step": 6654
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.65577496527758e-05,
+      "loss": 0.0724,
+      "step": 6655
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6535767329086095e-05,
+      "loss": 0.0729,
+      "step": 6656
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6513798291857608e-05,
+      "loss": 0.072,
+      "step": 6657
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.649184254458751e-05,
+      "loss": 0.0726,
+      "step": 6658
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6469900090770917e-05,
+      "loss": 0.0742,
+      "step": 6659
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6447970933900737e-05,
+      "loss": 0.0735,
+      "step": 6660
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.642605507746786e-05,
+      "loss": 0.0721,
+      "step": 6661
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6404152524961015e-05,
+      "loss": 0.0736,
+      "step": 6662
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.6382263279866794e-05,
+      "loss": 0.0705,
+      "step": 6663
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6360387345669715e-05,
+      "loss": 0.0711,
+      "step": 6664
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6338524725852123e-05,
+      "loss": 0.0759,
+      "step": 6665
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6316675423894278e-05,
+      "loss": 0.0719,
+      "step": 6666
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.629483944327437e-05,
+      "loss": 0.0746,
+      "step": 6667
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6273016787468332e-05,
+      "loss": 0.0736,
+      "step": 6668
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6251207459950126e-05,
+      "loss": 0.0729,
+      "step": 6669
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6229411464191457e-05,
+      "loss": 0.0682,
+      "step": 6670
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.620762880366202e-05,
+      "loss": 0.0701,
+      "step": 6671
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6185859481829367e-05,
+      "loss": 0.0755,
+      "step": 6672
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6164103502158846e-05,
+      "loss": 0.0724,
+      "step": 6673
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6142360868113748e-05,
+      "loss": 0.0685,
+      "step": 6674
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6120631583155266e-05,
+      "loss": 0.0716,
+      "step": 6675
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6098915650742407e-05,
+      "loss": 0.0739,
+      "step": 6676
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6077213074332042e-05,
+      "loss": 0.072,
+      "step": 6677
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6055523857378986e-05,
+      "loss": 0.0708,
+      "step": 6678
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.6033848003335904e-05,
+      "loss": 0.0701,
+      "step": 6679
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.601218551565329e-05,
+      "loss": 0.0693,
+      "step": 6680
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5990536397779544e-05,
+      "loss": 0.0731,
+      "step": 6681
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5968900653160955e-05,
+      "loss": 0.0718,
+      "step": 6682
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5947278285241664e-05,
+      "loss": 0.0735,
+      "step": 6683
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.592566929746362e-05,
+      "loss": 0.0704,
+      "step": 6684
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5904073693266796e-05,
+      "loss": 0.0696,
+      "step": 6685
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.58824914760889e-05,
+      "loss": 0.0725,
+      "step": 6686
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5860922649365516e-05,
+      "loss": 0.0716,
+      "step": 6687
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5839367216530176e-05,
+      "loss": 0.0713,
+      "step": 6688
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5817825181014246e-05,
+      "loss": 0.0712,
+      "step": 6689
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.579629654624689e-05,
+      "loss": 0.0721,
+      "step": 6690
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5774781315655242e-05,
+      "loss": 0.0718,
+      "step": 6691
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5753279492664262e-05,
+      "loss": 0.0706,
+      "step": 6692
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5731791080696767e-05,
+      "loss": 0.0727,
+      "step": 6693
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5710316083173382e-05,
+      "loss": 0.0689,
+      "step": 6694
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.568885450351275e-05,
+      "loss": 0.0752,
+      "step": 6695
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5667406345131253e-05,
+      "loss": 0.0675,
+      "step": 6696
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5645971611443133e-05,
+      "loss": 0.0684,
+      "step": 6697
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5624550305860554e-05,
+      "loss": 0.0686,
+      "step": 6698
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.560314243179356e-05,
+      "loss": 0.0724,
+      "step": 6699
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5581747992649975e-05,
+      "loss": 0.0721,
+      "step": 6700
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5560366991835484e-05,
+      "loss": 0.0729,
+      "step": 6701
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5538999432753777e-05,
+      "loss": 0.0692,
+      "step": 6702
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.551764531880624e-05,
+      "loss": 0.0694,
+      "step": 6703
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5496304653392146e-05,
+      "loss": 0.0705,
+      "step": 6704
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5474977439908757e-05,
+      "loss": 0.0704,
+      "step": 6705
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5453663681751042e-05,
+      "loss": 0.0723,
+      "step": 6706
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5432363382311865e-05,
+      "loss": 0.0735,
+      "step": 6707
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5411076544981994e-05,
+      "loss": 0.0701,
+      "step": 6708
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5389803173150043e-05,
+      "loss": 0.0737,
+      "step": 6709
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.536854327020245e-05,
+      "loss": 0.0711,
+      "step": 6710
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5347296839523472e-05,
+      "loss": 0.0713,
+      "step": 6711
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.532606388449538e-05,
+      "loss": 0.0699,
+      "step": 6712
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5304844408498132e-05,
+      "loss": 0.0731,
+      "step": 6713
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5283638414909573e-05,
+      "loss": 0.0743,
+      "step": 6714
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5262445907105514e-05,
+      "loss": 0.0717,
+      "step": 6715
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5241266888459494e-05,
+      "loss": 0.0725,
+      "step": 6716
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.522010136234292e-05,
+      "loss": 0.0722,
+      "step": 6717
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5198949332125112e-05,
+      "loss": 0.0689,
+      "step": 6718
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.517781080117322e-05,
+      "loss": 0.0728,
+      "step": 6719
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5156685772852198e-05,
+      "loss": 0.0701,
+      "step": 6720
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5135574250524897e-05,
+      "loss": 0.0709,
+      "step": 6721
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5114476237552044e-05,
+      "loss": 0.0703,
+      "step": 6722
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.509339173729214e-05,
+      "loss": 0.0717,
+      "step": 6723
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5072320753101565e-05,
+      "loss": 0.0725,
+      "step": 6724
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5051263288334572e-05,
+      "loss": 0.0699,
+      "step": 6725
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5030219346343265e-05,
+      "loss": 0.0733,
+      "step": 6726
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5009188930477547e-05,
+      "loss": 0.0665,
+      "step": 6727
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4988172044085202e-05,
+      "loss": 0.0759,
+      "step": 6728
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4967168690511891e-05,
+      "loss": 0.072,
+      "step": 6729
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4946178873101024e-05,
+      "loss": 0.0722,
+      "step": 6730
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4925202595193943e-05,
+      "loss": 0.0738,
+      "step": 6731
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4904239860129854e-05,
+      "loss": 0.0691,
+      "step": 6732
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4883290671245709e-05,
+      "loss": 0.0716,
+      "step": 6733
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4862355031876352e-05,
+      "loss": 0.0739,
+      "step": 6734
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4841432945354495e-05,
+      "loss": 0.0726,
+      "step": 6735
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4820524415010695e-05,
+      "loss": 0.0722,
+      "step": 6736
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.479962944417328e-05,
+      "loss": 0.0715,
+      "step": 6737
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4778748036168489e-05,
+      "loss": 0.0738,
+      "step": 6738
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4757880194320405e-05,
+      "loss": 0.0686,
+      "step": 6739
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4737025921950875e-05,
+      "loss": 0.071,
+      "step": 6740
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4716185222379709e-05,
+      "loss": 0.0676,
+      "step": 6741
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4695358098924417e-05,
+      "loss": 0.0687,
+      "step": 6742
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.467454455490046e-05,
+      "loss": 0.0698,
+      "step": 6743
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4653744593621066e-05,
+      "loss": 0.0711,
+      "step": 6744
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.463295821839733e-05,
+      "loss": 0.0707,
+      "step": 6745
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4612185432538228e-05,
+      "loss": 0.0726,
+      "step": 6746
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4591426239350459e-05,
+      "loss": 0.073,
+      "step": 6747
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.457068064213869e-05,
+      "loss": 0.0705,
+      "step": 6748
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4549948644205303e-05,
+      "loss": 0.0713,
+      "step": 6749
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4529230248850611e-05,
+      "loss": 0.0698,
+      "step": 6750
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4508525459372724e-05,
+      "loss": 0.0699,
+      "step": 6751
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4487834279067558e-05,
+      "loss": 0.0714,
+      "step": 6752
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4467156711228935e-05,
+      "loss": 0.0701,
+      "step": 6753
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4446492759148411e-05,
+      "loss": 0.0716,
+      "step": 6754
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4425842426115443e-05,
+      "loss": 0.0707,
+      "step": 6755
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4405205715417359e-05,
+      "loss": 0.0694,
+      "step": 6756
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4384582630339184e-05,
+      "loss": 0.0715,
+      "step": 6757
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4363973174163914e-05,
+      "loss": 0.0708,
+      "step": 6758
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4343377350172283e-05,
+      "loss": 0.0687,
+      "step": 6759
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.43227951616429e-05,
+      "loss": 0.072,
+      "step": 6760
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4302226611852221e-05,
+      "loss": 0.0744,
+      "step": 6761
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.428167170407444e-05,
+      "loss": 0.0707,
+      "step": 6762
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4261130441581704e-05,
+      "loss": 0.0702,
+      "step": 6763
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4240602827643868e-05,
+      "loss": 0.0709,
+      "step": 6764
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4220088865528725e-05,
+      "loss": 0.0727,
+      "step": 6765
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4199588558501798e-05,
+      "loss": 0.0677,
+      "step": 6766
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4179101909826487e-05,
+      "loss": 0.0673,
+      "step": 6767
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.415862892276405e-05,
+      "loss": 0.0665,
+      "step": 6768
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.413816960057347e-05,
+      "loss": 0.0656,
+      "step": 6769
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4117723946511653e-05,
+      "loss": 0.0657,
+      "step": 6770
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.4097291963833304e-05,
+      "loss": 0.0643,
+      "step": 6771
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4076873655790912e-05,
+      "loss": 0.0636,
+      "step": 6772
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.405646902563481e-05,
+      "loss": 0.0709,
+      "step": 6773
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4036078076613157e-05,
+      "loss": 0.0644,
+      "step": 6774
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.4015700811971987e-05,
+      "loss": 0.0634,
+      "step": 6775
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3995337234955041e-05,
+      "loss": 0.0677,
+      "step": 6776
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.397498734880398e-05,
+      "loss": 0.0667,
+      "step": 6777
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.395465115675827e-05,
+      "loss": 0.0695,
+      "step": 6778
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3934328662055119e-05,
+      "loss": 0.0619,
+      "step": 6779
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3914019867929661e-05,
+      "loss": 0.0681,
+      "step": 6780
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3893724777614814e-05,
+      "loss": 0.0671,
+      "step": 6781
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3873443394341268e-05,
+      "loss": 0.0641,
+      "step": 6782
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3853175721337552e-05,
+      "loss": 0.0661,
+      "step": 6783
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.3832921761830064e-05,
+      "loss": 0.0653,
+      "step": 6784
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.381268151904298e-05,
+      "loss": 0.0662,
+      "step": 6785
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3792454996198256e-05,
+      "loss": 0.063,
+      "step": 6786
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3772242196515728e-05,
+      "loss": 0.0643,
+      "step": 6787
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3752043123213032e-05,
+      "loss": 0.0628,
+      "step": 6788
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3731857779505608e-05,
+      "loss": 0.0665,
+      "step": 6789
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3711686168606641e-05,
+      "loss": 0.0662,
+      "step": 6790
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3691528293727297e-05,
+      "loss": 0.0701,
+      "step": 6791
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3671384158076416e-05,
+      "loss": 0.0635,
+      "step": 6792
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3651253764860672e-05,
+      "loss": 0.0656,
+      "step": 6793
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3631137117284576e-05,
+      "loss": 0.062,
+      "step": 6794
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.36110342185505e-05,
+      "loss": 0.0645,
+      "step": 6795
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3590945071858496e-05,
+      "loss": 0.064,
+      "step": 6796
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3570869680406551e-05,
+      "loss": 0.0648,
+      "step": 6797
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.3550808047390429e-05,
+      "loss": 0.0668,
+      "step": 6798
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3530760176003677e-05,
+      "loss": 0.0639,
+      "step": 6799
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3510726069437607e-05,
+      "loss": 0.0653,
+      "step": 6800
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3490705730881492e-05,
+      "loss": 0.0659,
+      "step": 6801
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3470699163522282e-05,
+      "loss": 0.0626,
+      "step": 6802
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3450706370544742e-05,
+      "loss": 0.0658,
+      "step": 6803
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.343072735513149e-05,
+      "loss": 0.0655,
+      "step": 6804
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3410762120462972e-05,
+      "loss": 0.0682,
+      "step": 6805
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3390810669717357e-05,
+      "loss": 0.0631,
+      "step": 6806
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3370873006070684e-05,
+      "loss": 0.0691,
+      "step": 6807
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.33509491326968e-05,
+      "loss": 0.0653,
+      "step": 6808
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3331039052767313e-05,
+      "loss": 0.0663,
+      "step": 6809
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3311142769451623e-05,
+      "loss": 0.066,
+      "step": 6810
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.329126028591705e-05,
+      "loss": 0.0667,
+      "step": 6811
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.3271391605328598e-05,
+      "loss": 0.0651,
+      "step": 6812
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3251536730849079e-05,
+      "loss": 0.0675,
+      "step": 6813
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3231695665639189e-05,
+      "loss": 0.0677,
+      "step": 6814
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3211868412857364e-05,
+      "loss": 0.0698,
+      "step": 6815
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.319205497565983e-05,
+      "loss": 0.0646,
+      "step": 6816
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3172255357200658e-05,
+      "loss": 0.0687,
+      "step": 6817
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3152469560631731e-05,
+      "loss": 0.0683,
+      "step": 6818
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3132697589102649e-05,
+      "loss": 0.0667,
+      "step": 6819
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3112939445760853e-05,
+      "loss": 0.0655,
+      "step": 6820
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3093195133751657e-05,
+      "loss": 0.0672,
+      "step": 6821
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3073464656218082e-05,
+      "loss": 0.0656,
+      "step": 6822
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3053748016300937e-05,
+      "loss": 0.0664,
+      "step": 6823
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.3034045217138879e-05,
+      "loss": 0.0696,
+      "step": 6824
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.30143562618684e-05,
+      "loss": 0.064,
+      "step": 6825
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2994681153623656e-05,
+      "loss": 0.0672,
+      "step": 6826
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2975019895536723e-05,
+      "loss": 0.0704,
+      "step": 6827
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2955372490737438e-05,
+      "loss": 0.0642,
+      "step": 6828
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2935738942353403e-05,
+      "loss": 0.065,
+      "step": 6829
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2916119253510017e-05,
+      "loss": 0.0658,
+      "step": 6830
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2896513427330504e-05,
+      "loss": 0.0663,
+      "step": 6831
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2876921466935899e-05,
+      "loss": 0.0635,
+      "step": 6832
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.285734337544493e-05,
+      "loss": 0.0662,
+      "step": 6833
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2837779155974228e-05,
+      "loss": 0.0685,
+      "step": 6834
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2818228811638178e-05,
+      "loss": 0.0685,
+      "step": 6835
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2798692345548924e-05,
+      "loss": 0.0671,
+      "step": 6836
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2779169760816457e-05,
+      "loss": 0.065,
+      "step": 6837
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.275966106054849e-05,
+      "loss": 0.0659,
+      "step": 6838
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.2740166247850605e-05,
+      "loss": 0.0658,
+      "step": 6839
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2720685325826088e-05,
+      "loss": 0.0668,
+      "step": 6840
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2701218297576078e-05,
+      "loss": 0.0648,
+      "step": 6841
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2681765166199511e-05,
+      "loss": 0.0658,
+      "step": 6842
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2662325934793029e-05,
+      "loss": 0.0643,
+      "step": 6843
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2642900606451147e-05,
+      "loss": 0.0642,
+      "step": 6844
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2623489184266158e-05,
+      "loss": 0.0665,
+      "step": 6845
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2604091671328055e-05,
+      "loss": 0.0662,
+      "step": 6846
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2584708070724737e-05,
+      "loss": 0.0678,
+      "step": 6847
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2565338385541792e-05,
+      "loss": 0.0662,
+      "step": 6848
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2545982618862662e-05,
+      "loss": 0.0656,
+      "step": 6849
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2526640773768506e-05,
+      "loss": 0.0601,
+      "step": 6850
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2507312853338316e-05,
+      "loss": 0.0655,
+      "step": 6851
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.2487998860648898e-05,
+      "loss": 0.0682,
+      "step": 6852
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2468698798774737e-05,
+      "loss": 0.0657,
+      "step": 6853
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2449412670788208e-05,
+      "loss": 0.0694,
+      "step": 6854
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2430140479759378e-05,
+      "loss": 0.066,
+      "step": 6855
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.241088222875616e-05,
+      "loss": 0.0661,
+      "step": 6856
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2391637920844257e-05,
+      "loss": 0.0681,
+      "step": 6857
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2372407559087063e-05,
+      "loss": 0.0646,
+      "step": 6858
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2353191146545862e-05,
+      "loss": 0.0679,
+      "step": 6859
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2333988686279629e-05,
+      "loss": 0.068,
+      "step": 6860
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2314800181345166e-05,
+      "loss": 0.0679,
+      "step": 6861
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.229562563479707e-05,
+      "loss": 0.0648,
+      "step": 6862
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2276465049687647e-05,
+      "loss": 0.0651,
+      "step": 6863
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2257318429067067e-05,
+      "loss": 0.0699,
+      "step": 6864
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2238185775983192e-05,
+      "loss": 0.0683,
+      "step": 6865
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.2219067093481706e-05,
+      "loss": 0.0652,
+      "step": 6866
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.219996238460609e-05,
+      "loss": 0.0659,
+      "step": 6867
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2180871652397552e-05,
+      "loss": 0.0644,
+      "step": 6868
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2161794899895118e-05,
+      "loss": 0.065,
+      "step": 6869
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2142732130135537e-05,
+      "loss": 0.0634,
+      "step": 6870
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2123683346153403e-05,
+      "loss": 0.0698,
+      "step": 6871
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2104648550981e-05,
+      "loss": 0.0696,
+      "step": 6872
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2085627747648454e-05,
+      "loss": 0.0657,
+      "step": 6873
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2066620939183647e-05,
+      "loss": 0.0701,
+      "step": 6874
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2047628128612198e-05,
+      "loss": 0.0685,
+      "step": 6875
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2028649318957552e-05,
+      "loss": 0.066,
+      "step": 6876
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.2009684513240892e-05,
+      "loss": 0.0639,
+      "step": 6877
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.1990733714481184e-05,
+      "loss": 0.0663,
+      "step": 6878
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.1971796925695122e-05,
+      "loss": 0.0683,
+      "step": 6879
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.1952874149897231e-05,
+      "loss": 0.0661,
+      "step": 6880
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.19339653900998e-05,
+      "loss": 0.0686,
+      "step": 6881
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1915070649312832e-05,
+      "loss": 0.0634,
+      "step": 6882
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1896189930544133e-05,
+      "loss": 0.0661,
+      "step": 6883
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1877323236799321e-05,
+      "loss": 0.0636,
+      "step": 6884
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1858470571081693e-05,
+      "loss": 0.0658,
+      "step": 6885
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1839631936392359e-05,
+      "loss": 0.0641,
+      "step": 6886
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1820807335730244e-05,
+      "loss": 0.0658,
+      "step": 6887
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1801996772091938e-05,
+      "loss": 0.0644,
+      "step": 6888
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.178320024847185e-05,
+      "loss": 0.0673,
+      "step": 6889
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1764417767862169e-05,
+      "loss": 0.0675,
+      "step": 6890
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1745649333252828e-05,
+      "loss": 0.0675,
+      "step": 6891
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1726894947631516e-05,
+      "loss": 0.0669,
+      "step": 6892
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.1708154613983702e-05,
+      "loss": 0.0648,
+      "step": 6893
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1689428335292629e-05,
+      "loss": 0.0651,
+      "step": 6894
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1670716114539282e-05,
+      "loss": 0.0649,
+      "step": 6895
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1652017954702354e-05,
+      "loss": 0.0687,
+      "step": 6896
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1633333858758444e-05,
+      "loss": 0.0634,
+      "step": 6897
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1614663829681794e-05,
+      "loss": 0.068,
+      "step": 6898
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1596007870444414e-05,
+      "loss": 0.0658,
+      "step": 6899
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.157736598401612e-05,
+      "loss": 0.0671,
+      "step": 6900
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.155873817336448e-05,
+      "loss": 0.0636,
+      "step": 6901
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.154012444145478e-05,
+      "loss": 0.0629,
+      "step": 6902
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1521524791250104e-05,
+      "loss": 0.0685,
+      "step": 6903
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1502939225711307e-05,
+      "loss": 0.0649,
+      "step": 6904
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1484367747796954e-05,
+      "loss": 0.067,
+      "step": 6905
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1465810360463358e-05,
+      "loss": 0.0671,
+      "step": 6906
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.1447267066664713e-05,
+      "loss": 0.067,
+      "step": 6907
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1428737869352812e-05,
+      "loss": 0.0683,
+      "step": 6908
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1410222771477274e-05,
+      "loss": 0.0627,
+      "step": 6909
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1391721775985476e-05,
+      "loss": 0.0675,
+      "step": 6910
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1373234885822582e-05,
+      "loss": 0.0658,
+      "step": 6911
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1354762103931415e-05,
+      "loss": 0.0675,
+      "step": 6912
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1336303433252638e-05,
+      "loss": 0.0705,
+      "step": 6913
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1317858876724663e-05,
+      "loss": 0.0668,
+      "step": 6914
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.129942843728362e-05,
+      "loss": 0.0652,
+      "step": 6915
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1281012117863342e-05,
+      "loss": 0.0639,
+      "step": 6916
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1262609921395583e-05,
+      "loss": 0.0641,
+      "step": 6917
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1244221850809678e-05,
+      "loss": 0.064,
+      "step": 6918
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1225847909032783e-05,
+      "loss": 0.0649,
+      "step": 6919
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.1207488098989794e-05,
+      "loss": 0.0676,
+      "step": 6920
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1189142423603394e-05,
+      "loss": 0.0665,
+      "step": 6921
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1170810885793948e-05,
+      "loss": 0.0679,
+      "step": 6922
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1152493488479631e-05,
+      "loss": 0.0632,
+      "step": 6923
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1134190234576347e-05,
+      "loss": 0.0667,
+      "step": 6924
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1115901126997741e-05,
+      "loss": 0.0652,
+      "step": 6925
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1097626168655173e-05,
+      "loss": 0.0681,
+      "step": 6926
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.107936536245785e-05,
+      "loss": 0.0669,
+      "step": 6927
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1061118711312623e-05,
+      "loss": 0.0687,
+      "step": 6928
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.1042886218124137e-05,
+      "loss": 0.0687,
+      "step": 6929
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.102466788579477e-05,
+      "loss": 0.0684,
+      "step": 6930
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.100646371722468e-05,
+      "loss": 0.0669,
+      "step": 6931
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.0988273715311715e-05,
+      "loss": 0.0651,
+      "step": 6932
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.0970097882951502e-05,
+      "loss": 0.0697,
+      "step": 6933
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.0951936223037429e-05,
+      "loss": 0.0655,
+      "step": 6934
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0933788738460582e-05,
+      "loss": 0.0652,
+      "step": 6935
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0915655432109806e-05,
+      "loss": 0.0678,
+      "step": 6936
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0897536306871692e-05,
+      "loss": 0.0656,
+      "step": 6937
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0879431365630633e-05,
+      "loss": 0.0646,
+      "step": 6938
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0861340611268633e-05,
+      "loss": 0.0645,
+      "step": 6939
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0843264046665557e-05,
+      "loss": 0.0646,
+      "step": 6940
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0825201674698982e-05,
+      "loss": 0.0649,
+      "step": 6941
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0807153498244172e-05,
+      "loss": 0.0647,
+      "step": 6942
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0789119520174207e-05,
+      "loss": 0.0657,
+      "step": 6943
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0771099743359836e-05,
+      "loss": 0.0645,
+      "step": 6944
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.075309417066962e-05,
+      "loss": 0.0658,
+      "step": 6945
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0735102804969766e-05,
+      "loss": 0.0633,
+      "step": 6946
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0717125649124327e-05,
+      "loss": 0.0668,
+      "step": 6947
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0699162705995025e-05,
+      "loss": 0.0686,
+      "step": 6948
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.068121397844133e-05,
+      "loss": 0.0681,
+      "step": 6949
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0663279469320442e-05,
+      "loss": 0.0646,
+      "step": 6950
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0645359181487357e-05,
+      "loss": 0.0637,
+      "step": 6951
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0627453117794706e-05,
+      "loss": 0.0659,
+      "step": 6952
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.060956128109295e-05,
+      "loss": 0.0675,
+      "step": 6953
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0591683674230212e-05,
+      "loss": 0.0636,
+      "step": 6954
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0573820300052428e-05,
+      "loss": 0.0689,
+      "step": 6955
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0555971161403166e-05,
+      "loss": 0.0663,
+      "step": 6956
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.053813626112382e-05,
+      "loss": 0.0669,
+      "step": 6957
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0520315602053498e-05,
+      "loss": 0.0678,
+      "step": 6958
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0502509187028986e-05,
+      "loss": 0.0681,
+      "step": 6959
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0484717018884882e-05,
+      "loss": 0.0652,
+      "step": 6960
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.0466939100453443e-05,
+      "loss": 0.0652,
+      "step": 6961
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0449175434564707e-05,
+      "loss": 0.0621,
+      "step": 6962
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0431426024046431e-05,
+      "loss": 0.0675,
+      "step": 6963
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.041369087172409e-05,
+      "loss": 0.0634,
+      "step": 6964
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0395969980420916e-05,
+      "loss": 0.0689,
+      "step": 6965
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0378263352957818e-05,
+      "loss": 0.0679,
+      "step": 6966
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0360570992153506e-05,
+      "loss": 0.0655,
+      "step": 6967
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0342892900824353e-05,
+      "loss": 0.0665,
+      "step": 6968
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0325229081784504e-05,
+      "loss": 0.0648,
+      "step": 6969
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0307579537845835e-05,
+      "loss": 0.0633,
+      "step": 6970
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0289944271817897e-05,
+      "loss": 0.0661,
+      "step": 6971
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0272323286508023e-05,
+      "loss": 0.0672,
+      "step": 6972
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.025471658472127e-05,
+      "loss": 0.0686,
+      "step": 6973
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.023712416926037e-05,
+      "loss": 0.0673,
+      "step": 6974
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0219546042925843e-05,
+      "loss": 0.0673,
+      "step": 6975
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0201982208515882e-05,
+      "loss": 0.0631,
+      "step": 6976
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0184432668826471e-05,
+      "loss": 0.0663,
+      "step": 6977
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0166897426651223e-05,
+      "loss": 0.0652,
+      "step": 6978
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0149376484781548e-05,
+      "loss": 0.0649,
+      "step": 6979
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0131869846006603e-05,
+      "loss": 0.066,
+      "step": 6980
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.011437751311316e-05,
+      "loss": 0.0673,
+      "step": 6981
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0096899488885813e-05,
+      "loss": 0.0626,
+      "step": 6982
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.007943577610686e-05,
+      "loss": 0.0681,
+      "step": 6983
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0061986377556298e-05,
+      "loss": 0.0654,
+      "step": 6984
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0044551296011818e-05,
+      "loss": 0.0638,
+      "step": 6985
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0027130534248885e-05,
+      "loss": 0.069,
+      "step": 6986
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.0009724095040696e-05,
+      "loss": 0.0679,
+      "step": 6987
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.992331981158098e-06,
+      "loss": 0.0662,
+      "step": 6988
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.974954195369712e-06,
+      "loss": 0.0702,
+      "step": 6989
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.95759074044188e-06,
+      "loss": 0.0677,
+      "step": 6990
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.940241619138623e-06,
+      "loss": 0.0605,
+      "step": 6991
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.922906834221712e-06,
+      "loss": 0.0669,
+      "step": 6992
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.90558638845065e-06,
+      "loss": 0.0691,
+      "step": 6993
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.888280284582608e-06,
+      "loss": 0.0671,
+      "step": 6994
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.870988525372494e-06,
+      "loss": 0.0634,
+      "step": 6995
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.853711113572961e-06,
+      "loss": 0.0643,
+      "step": 6996
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.836448051934366e-06,
+      "loss": 0.0655,
+      "step": 6997
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.819199343204733e-06,
+      "loss": 0.066,
+      "step": 6998
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.801964990129863e-06,
+      "loss": 0.067,
+      "step": 6999
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.784744995453277e-06,
+      "loss": 0.0632,
+      "step": 7000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.767539361916144e-06,
+      "loss": 0.067,
+      "step": 7001
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.750348092257367e-06,
+      "loss": 0.0623,
+      "step": 7002
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.733171189213641e-06,
+      "loss": 0.0692,
+      "step": 7003
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.716008655519293e-06,
+      "loss": 0.0649,
+      "step": 7004
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.698860493906336e-06,
+      "loss": 0.0675,
+      "step": 7005
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.68172670710462e-06,
+      "loss": 0.0659,
+      "step": 7006
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.664607297841589e-06,
+      "loss": 0.0618,
+      "step": 7007
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.647502268842412e-06,
+      "loss": 0.0648,
+      "step": 7008
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.630411622830038e-06,
+      "loss": 0.0668,
+      "step": 7009
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.61333536252509e-06,
+      "loss": 0.0667,
+      "step": 7010
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.596273490645857e-06,
+      "loss": 0.066,
+      "step": 7011
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.57922600990837e-06,
+      "loss": 0.0658,
+      "step": 7012
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.562192923026435e-06,
+      "loss": 0.0624,
+      "step": 7013
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.545174232711462e-06,
+      "loss": 0.0656,
+      "step": 7014
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.528169941672594e-06,
+      "loss": 0.0655,
+      "step": 7015
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.511180052616753e-06,
+      "loss": 0.068,
+      "step": 7016
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.4942045682485e-06,
+      "loss": 0.0657,
+      "step": 7017
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.477243491270094e-06,
+      "loss": 0.0675,
+      "step": 7018
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.460296824381542e-06,
+      "loss": 0.064,
+      "step": 7019
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.443364570280554e-06,
+      "loss": 0.0656,
+      "step": 7020
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.42644673166253e-06,
+      "loss": 0.0644,
+      "step": 7021
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.409543311220526e-06,
+      "loss": 0.066,
+      "step": 7022
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.392654311645444e-06,
+      "loss": 0.0667,
+      "step": 7023
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.375779735625745e-06,
+      "loss": 0.0654,
+      "step": 7024
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.358919585847648e-06,
+      "loss": 0.0654,
+      "step": 7025
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.342073864995093e-06,
+      "loss": 0.0654,
+      "step": 7026
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.325242575749726e-06,
+      "loss": 0.0665,
+      "step": 7027
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.308425720790837e-06,
+      "loss": 0.0682,
+      "step": 7028
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.291623302795482e-06,
+      "loss": 0.0666,
+      "step": 7029
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.274835324438414e-06,
+      "loss": 0.0635,
+      "step": 7030
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.258061788392037e-06,
+      "loss": 0.0674,
+      "step": 7031
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.24130269732647e-06,
+      "loss": 0.0616,
+      "step": 7032
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.224558053909615e-06,
+      "loss": 0.0636,
+      "step": 7033
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.207827860806973e-06,
+      "loss": 0.065,
+      "step": 7034
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.191112120681767e-06,
+      "loss": 0.0659,
+      "step": 7035
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.174410836194946e-06,
+      "loss": 0.0662,
+      "step": 7036
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.157724010005175e-06,
+      "loss": 0.0656,
+      "step": 7037
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.141051644768739e-06,
+      "loss": 0.0629,
+      "step": 7038
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.124393743139692e-06,
+      "loss": 0.064,
+      "step": 7039
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.10775030776978e-06,
+      "loss": 0.0662,
+      "step": 7040
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.091121341308417e-06,
+      "loss": 0.0681,
+      "step": 7041
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.074506846402708e-06,
+      "loss": 0.0654,
+      "step": 7042
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.057906825697481e-06,
+      "loss": 0.0655,
+      "step": 7043
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.04132128183528e-06,
+      "loss": 0.0653,
+      "step": 7044
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.024750217456279e-06,
+      "loss": 0.0656,
+      "step": 7045
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.008193635198392e-06,
+      "loss": 0.0675,
+      "step": 7046
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.991651537697255e-06,
+      "loss": 0.0659,
+      "step": 7047
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.975123927586104e-06,
+      "loss": 0.0669,
+      "step": 7048
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.958610807495982e-06,
+      "loss": 0.0707,
+      "step": 7049
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.942112180055528e-06,
+      "loss": 0.0647,
+      "step": 7050
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.925628047891144e-06,
+      "loss": 0.0659,
+      "step": 7051
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.909158413626906e-06,
+      "loss": 0.0647,
+      "step": 7052
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.89270327988453e-06,
+      "loss": 0.0641,
+      "step": 7053
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.876262649283518e-06,
+      "loss": 0.0691,
+      "step": 7054
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 8.859836524440979e-06,
+      "loss": 0.0658,
+      "step": 7055
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.843424907971743e-06,
+      "loss": 0.0653,
+      "step": 7056
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.827027802488375e-06,
+      "loss": 0.0653,
+      "step": 7057
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.810645210601042e-06,
+      "loss": 0.0663,
+      "step": 7058
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.794277134917695e-06,
+      "loss": 0.0663,
+      "step": 7059
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.777923578043868e-06,
+      "loss": 0.0676,
+      "step": 7060
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.761584542582879e-06,
+      "loss": 0.0677,
+      "step": 7061
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.745260031135704e-06,
+      "loss": 0.0653,
+      "step": 7062
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.728950046300988e-06,
+      "loss": 0.0656,
+      "step": 7063
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.712654590675085e-06,
+      "loss": 0.0648,
+      "step": 7064
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.696373666851999e-06,
+      "loss": 0.0657,
+      "step": 7065
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.6801072774235e-06,
+      "loss": 0.0667,
+      "step": 7066
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.663855424978939e-06,
+      "loss": 0.0653,
+      "step": 7067
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.647618112105438e-06,
+      "loss": 0.0671,
+      "step": 7068
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 8.631395341387783e-06,
+      "loss": 0.0635,
+      "step": 7069
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.615187115408408e-06,
+      "loss": 0.0679,
+      "step": 7070
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.598993436747472e-06,
+      "loss": 0.068,
+      "step": 7071
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.582814307982823e-06,
+      "loss": 0.0676,
+      "step": 7072
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.566649731689968e-06,
+      "loss": 0.0681,
+      "step": 7073
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.550499710442072e-06,
+      "loss": 0.0659,
+      "step": 7074
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.534364246810045e-06,
+      "loss": 0.0685,
+      "step": 7075
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.518243343362476e-06,
+      "loss": 0.0664,
+      "step": 7076
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.502137002665556e-06,
+      "loss": 0.0666,
+      "step": 7077
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.486045227283236e-06,
+      "loss": 0.0643,
+      "step": 7078
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.469968019777153e-06,
+      "loss": 0.0671,
+      "step": 7079
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.45390538270655e-06,
+      "loss": 0.0693,
+      "step": 7080
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.437857318628428e-06,
+      "loss": 0.0659,
+      "step": 7081
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 8.421823830097442e-06,
+      "loss": 0.0629,
+      "step": 7082
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.405804919665917e-06,
+      "loss": 0.0671,
+      "step": 7083
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.389800589883845e-06,
+      "loss": 0.0657,
+      "step": 7084
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.373810843298925e-06,
+      "loss": 0.0674,
+      "step": 7085
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.357835682456538e-06,
+      "loss": 0.0632,
+      "step": 7086
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.3418751098997e-06,
+      "loss": 0.0663,
+      "step": 7087
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.325929128169163e-06,
+      "loss": 0.0664,
+      "step": 7088
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.309997739803332e-06,
+      "loss": 0.0644,
+      "step": 7089
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.294080947338268e-06,
+      "loss": 0.0673,
+      "step": 7090
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.278178753307698e-06,
+      "loss": 0.0687,
+      "step": 7091
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.262291160243119e-06,
+      "loss": 0.0685,
+      "step": 7092
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.246418170673597e-06,
+      "loss": 0.0652,
+      "step": 7093
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.230559787125891e-06,
+      "loss": 0.0643,
+      "step": 7094
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.21471601212449e-06,
+      "loss": 0.0662,
+      "step": 7095
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 8.198886848191545e-06,
+      "loss": 0.0656,
+      "step": 7096
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.183072297846806e-06,
+      "loss": 0.0662,
+      "step": 7097
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.16727236360777e-06,
+      "loss": 0.0684,
+      "step": 7098
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.151487047989625e-06,
+      "loss": 0.0642,
+      "step": 7099
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.135716353505163e-06,
+      "loss": 0.0634,
+      "step": 7100
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.119960282664863e-06,
+      "loss": 0.0684,
+      "step": 7101
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.10421883797694e-06,
+      "loss": 0.0631,
+      "step": 7102
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.088492021947214e-06,
+      "loss": 0.0677,
+      "step": 7103
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.072779837079181e-06,
+      "loss": 0.0651,
+      "step": 7104
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.05708228587404e-06,
+      "loss": 0.0648,
+      "step": 7105
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.04139937083066e-06,
+      "loss": 0.0638,
+      "step": 7106
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.025731094445532e-06,
+      "loss": 0.0677,
+      "step": 7107
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 8.010077459212861e-06,
+      "loss": 0.0685,
+      "step": 7108
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.994438467624532e-06,
+      "loss": 0.0648,
+      "step": 7109
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 7.978814122170065e-06,
+      "loss": 0.065,
+      "step": 7110
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.963204425336623e-06,
+      "loss": 0.0671,
+      "step": 7111
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.947609379609133e-06,
+      "loss": 0.0658,
+      "step": 7112
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.932028987470108e-06,
+      "loss": 0.0676,
+      "step": 7113
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.916463251399731e-06,
+      "loss": 0.0665,
+      "step": 7114
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.900912173875885e-06,
+      "loss": 0.0646,
+      "step": 7115
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.885375757374136e-06,
+      "loss": 0.0631,
+      "step": 7116
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.869854004367639e-06,
+      "loss": 0.066,
+      "step": 7117
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.854346917327283e-06,
+      "loss": 0.0656,
+      "step": 7118
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.838854498721627e-06,
+      "loss": 0.0627,
+      "step": 7119
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.823376751016843e-06,
+      "loss": 0.0652,
+      "step": 7120
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.80791367667677e-06,
+      "loss": 0.0659,
+      "step": 7121
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.792465278162997e-06,
+      "loss": 0.0645,
+      "step": 7122
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 7.777031557934678e-06,
+      "loss": 0.0636,
+      "step": 7123
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.761612518448669e-06,
+      "loss": 0.0687,
+      "step": 7124
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.746208162159485e-06,
+      "loss": 0.0633,
+      "step": 7125
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.730818491519343e-06,
+      "loss": 0.0654,
+      "step": 7126
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.715443508978027e-06,
+      "loss": 0.0671,
+      "step": 7127
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.70008321698309e-06,
+      "loss": 0.065,
+      "step": 7128
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.684737617979687e-06,
+      "loss": 0.063,
+      "step": 7129
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.669406714410632e-06,
+      "loss": 0.0654,
+      "step": 7130
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.654090508716415e-06,
+      "loss": 0.0667,
+      "step": 7131
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.63878900333519e-06,
+      "loss": 0.0639,
+      "step": 7132
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.623502200702781e-06,
+      "loss": 0.0668,
+      "step": 7133
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.608230103252622e-06,
+      "loss": 0.0651,
+      "step": 7134
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.592972713415858e-06,
+      "loss": 0.0663,
+      "step": 7135
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.577730033621289e-06,
+      "loss": 0.0637,
+      "step": 7136
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 7.5625020662953186e-06,
+      "loss": 0.0654,
+      "step": 7137
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.547288813862097e-06,
+      "loss": 0.0685,
+      "step": 7138
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.5320902787433425e-06,
+      "loss": 0.0684,
+      "step": 7139
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.516906463358509e-06,
+      "loss": 0.0627,
+      "step": 7140
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.5017373701246396e-06,
+      "loss": 0.0647,
+      "step": 7141
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.486583001456471e-06,
+      "loss": 0.068,
+      "step": 7142
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.471443359766406e-06,
+      "loss": 0.064,
+      "step": 7143
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.45631844746445e-06,
+      "loss": 0.0651,
+      "step": 7144
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.441208266958332e-06,
+      "loss": 0.0677,
+      "step": 7145
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.426112820653408e-06,
+      "loss": 0.0639,
+      "step": 7146
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.411032110952653e-06,
+      "loss": 0.0628,
+      "step": 7147
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.395966140256761e-06,
+      "loss": 0.0638,
+      "step": 7148
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.38091491096401e-06,
+      "loss": 0.067,
+      "step": 7149
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 7.365878425470407e-06,
+      "loss": 0.0656,
+      "step": 7150
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.350856686169538e-06,
+      "loss": 0.0653,
+      "step": 7151
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.33584969545269e-06,
+      "loss": 0.065,
+      "step": 7152
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.320857455708796e-06,
+      "loss": 0.068,
+      "step": 7153
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.305879969324414e-06,
+      "loss": 0.0665,
+      "step": 7154
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.290917238683803e-06,
+      "loss": 0.0643,
+      "step": 7155
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.275969266168803e-06,
+      "loss": 0.0649,
+      "step": 7156
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.261036054158965e-06,
+      "loss": 0.0651,
+      "step": 7157
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.246117605031489e-06,
+      "loss": 0.0666,
+      "step": 7158
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.231213921161162e-06,
+      "loss": 0.0631,
+      "step": 7159
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.21632500492051e-06,
+      "loss": 0.0652,
+      "step": 7160
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.201450858679615e-06,
+      "loss": 0.0643,
+      "step": 7161
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.186591484806282e-06,
+      "loss": 0.0672,
+      "step": 7162
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.171746885665942e-06,
+      "loss": 0.0657,
+      "step": 7163
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 7.15691706362166e-06,
+      "loss": 0.068,
+      "step": 7164
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.1421020210341695e-06,
+      "loss": 0.0666,
+      "step": 7165
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.127301760261806e-06,
+      "loss": 0.0624,
+      "step": 7166
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.11251628366062e-06,
+      "loss": 0.0672,
+      "step": 7167
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.097745593584282e-06,
+      "loss": 0.0669,
+      "step": 7168
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.082989692384057e-06,
+      "loss": 0.0654,
+      "step": 7169
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.068248582408943e-06,
+      "loss": 0.066,
+      "step": 7170
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.053522266005519e-06,
+      "loss": 0.0666,
+      "step": 7171
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.0388107455180426e-06,
+      "loss": 0.0696,
+      "step": 7172
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.024114023288375e-06,
+      "loss": 0.0655,
+      "step": 7173
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 7.0094321016560775e-06,
+      "loss": 0.0677,
+      "step": 7174
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.994764982958335e-06,
+      "loss": 0.0662,
+      "step": 7175
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.980112669529936e-06,
+      "loss": 0.0663,
+      "step": 7176
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 6.965475163703372e-06,
+      "loss": 0.065,
+      "step": 7177
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.950852467808755e-06,
+      "loss": 0.0655,
+      "step": 7178
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.9362445841738236e-06,
+      "loss": 0.0642,
+      "step": 7179
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.92165151512395e-06,
+      "loss": 0.0685,
+      "step": 7180
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.9070732629822e-06,
+      "loss": 0.0632,
+      "step": 7181
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.892509830069249e-06,
+      "loss": 0.0677,
+      "step": 7182
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.877961218703388e-06,
+      "loss": 0.067,
+      "step": 7183
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.863427431200586e-06,
+      "loss": 0.0661,
+      "step": 7184
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.848908469874449e-06,
+      "loss": 0.0657,
+      "step": 7185
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.834404337036204e-06,
+      "loss": 0.0683,
+      "step": 7186
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.819915034994728e-06,
+      "loss": 0.0667,
+      "step": 7187
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.805440566056553e-06,
+      "loss": 0.07,
+      "step": 7188
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.790980932525814e-06,
+      "loss": 0.0638,
+      "step": 7189
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.776536136704293e-06,
+      "loss": 0.0652,
+      "step": 7190
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 6.762106180891437e-06,
+      "loss": 0.0651,
+      "step": 7191
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.7476910673843316e-06,
+      "loss": 0.0694,
+      "step": 7192
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.733290798477654e-06,
+      "loss": 0.0679,
+      "step": 7193
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.7189053764637465e-06,
+      "loss": 0.0615,
+      "step": 7194
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.704534803632623e-06,
+      "loss": 0.0656,
+      "step": 7195
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.690179082271875e-06,
+      "loss": 0.0695,
+      "step": 7196
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.675838214666719e-06,
+      "loss": 0.0645,
+      "step": 7197
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.661512203100107e-06,
+      "loss": 0.0662,
+      "step": 7198
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.647201049852525e-06,
+      "loss": 0.0683,
+      "step": 7199
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.63290475720213e-06,
+      "loss": 0.0693,
+      "step": 7200
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.618623327424711e-06,
+      "loss": 0.0649,
+      "step": 7201
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.604356762793718e-06,
+      "loss": 0.0627,
+      "step": 7202
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.590105065580166e-06,
+      "loss": 0.0659,
+      "step": 7203
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 6.575868238052785e-06,
+      "loss": 0.0659,
+      "step": 7204
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.561646282477896e-06,
+      "loss": 0.066,
+      "step": 7205
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.547439201119443e-06,
+      "loss": 0.0657,
+      "step": 7206
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.533246996238984e-06,
+      "loss": 0.0635,
+      "step": 7207
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.519069670095812e-06,
+      "loss": 0.0641,
+      "step": 7208
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.504907224946743e-06,
+      "loss": 0.0676,
+      "step": 7209
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.490759663046253e-06,
+      "loss": 0.0639,
+      "step": 7210
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.476626986646461e-06,
+      "loss": 0.0648,
+      "step": 7211
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.462509197997124e-06,
+      "loss": 0.0676,
+      "step": 7212
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.4484062993456e-06,
+      "loss": 0.0642,
+      "step": 7213
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.4343182929369025e-06,
+      "loss": 0.0663,
+      "step": 7214
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.420245181013684e-06,
+      "loss": 0.0653,
+      "step": 7215
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.4061869658161855e-06,
+      "loss": 0.0632,
+      "step": 7216
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.392143649582283e-06,
+      "loss": 0.0672,
+      "step": 7217
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 6.378115234547544e-06,
+      "loss": 0.0637,
+      "step": 7218
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.364101722945082e-06,
+      "loss": 0.0663,
+      "step": 7219
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.350103117005679e-06,
+      "loss": 0.0666,
+      "step": 7220
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.336119418957731e-06,
+      "loss": 0.0654,
+      "step": 7221
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.32215063102729e-06,
+      "loss": 0.0659,
+      "step": 7222
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.308196755437989e-06,
+      "loss": 0.0651,
+      "step": 7223
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.2942577944111185e-06,
+      "loss": 0.0643,
+      "step": 7224
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.280333750165601e-06,
+      "loss": 0.069,
+      "step": 7225
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.266424624917966e-06,
+      "loss": 0.0625,
+      "step": 7226
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.25253042088233e-06,
+      "loss": 0.0654,
+      "step": 7227
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.238651140270535e-06,
+      "loss": 0.0682,
+      "step": 7228
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.22478678529197e-06,
+      "loss": 0.0696,
+      "step": 7229
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.210937358153634e-06,
+      "loss": 0.0674,
+      "step": 7230
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 6.197102861060211e-06,
+      "loss": 0.0662,
+      "step": 7231
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.183283296214004e-06,
+      "loss": 0.0676,
+      "step": 7232
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.169478665814865e-06,
+      "loss": 0.0655,
+      "step": 7233
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.155688972060347e-06,
+      "loss": 0.0641,
+      "step": 7234
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.141914217145595e-06,
+      "loss": 0.0675,
+      "step": 7235
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.1281544032633865e-06,
+      "loss": 0.0658,
+      "step": 7236
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.114409532604082e-06,
+      "loss": 0.0691,
+      "step": 7237
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.100679607355708e-06,
+      "loss": 0.0653,
+      "step": 7238
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.086964629703906e-06,
+      "loss": 0.0647,
+      "step": 7239
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.073264601831918e-06,
+      "loss": 0.0655,
+      "step": 7240
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.059579525920611e-06,
+      "loss": 0.0632,
+      "step": 7241
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.045909404148508e-06,
+      "loss": 0.0635,
+      "step": 7242
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.032254238691692e-06,
+      "loss": 0.0653,
+      "step": 7243
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.0186140317239125e-06,
+      "loss": 0.0666,
+      "step": 7244
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 6.004988785416487e-06,
+      "loss": 0.0646,
+      "step": 7245
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.9913785019384276e-06,
+      "loss": 0.0655,
+      "step": 7246
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.977783183456287e-06,
+      "loss": 0.0648,
+      "step": 7247
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.964202832134292e-06,
+      "loss": 0.0645,
+      "step": 7248
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.950637450134255e-06,
+      "loss": 0.068,
+      "step": 7249
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.937087039615619e-06,
+      "loss": 0.0648,
+      "step": 7250
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.923551602735422e-06,
+      "loss": 0.0628,
+      "step": 7251
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.910031141648375e-06,
+      "loss": 0.0692,
+      "step": 7252
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.896525658506724e-06,
+      "loss": 0.0649,
+      "step": 7253
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.883035155460404e-06,
+      "loss": 0.0629,
+      "step": 7254
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.869559634656907e-06,
+      "loss": 0.0675,
+      "step": 7255
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.856099098241396e-06,
+      "loss": 0.0705,
+      "step": 7256
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.842653548356602e-06,
+      "loss": 0.066,
+      "step": 7257
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 5.82922298714288e-06,
+      "loss": 0.0642,
+      "step": 7258
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.815807416738239e-06,
+      "loss": 0.0691,
+      "step": 7259
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.80240683927823e-06,
+      "loss": 0.0633,
+      "step": 7260
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.789021256896099e-06,
+      "loss": 0.067,
+      "step": 7261
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.775650671722621e-06,
+      "loss": 0.0647,
+      "step": 7262
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.762295085886249e-06,
+      "loss": 0.0627,
+      "step": 7263
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.748954501513037e-06,
+      "loss": 0.064,
+      "step": 7264
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.73562892072661e-06,
+      "loss": 0.067,
+      "step": 7265
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.722318345648259e-06,
+      "loss": 0.0664,
+      "step": 7266
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.709022778396844e-06,
+      "loss": 0.0644,
+      "step": 7267
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.695742221088862e-06,
+      "loss": 0.0695,
+      "step": 7268
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.6824766758383996e-06,
+      "loss": 0.0663,
+      "step": 7269
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.669226144757178e-06,
+      "loss": 0.0628,
+      "step": 7270
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.655990629954522e-06,
+      "loss": 0.0644,
+      "step": 7271
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 5.642770133537334e-06,
+      "loss": 0.0665,
+      "step": 7272
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.629564657610176e-06,
+      "loss": 0.0632,
+      "step": 7273
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.616374204275199e-06,
+      "loss": 0.065,
+      "step": 7274
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.603198775632123e-06,
+      "loss": 0.0661,
+      "step": 7275
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.5900383737783505e-06,
+      "loss": 0.0652,
+      "step": 7276
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.576893000808825e-06,
+      "loss": 0.0651,
+      "step": 7277
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.563762658816152e-06,
+      "loss": 0.0638,
+      "step": 7278
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.550647349890481e-06,
+      "loss": 0.0653,
+      "step": 7279
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.5375470761196316e-06,
+      "loss": 0.0657,
+      "step": 7280
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.524461839589013e-06,
+      "loss": 0.0675,
+      "step": 7281
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.511391642381602e-06,
+      "loss": 0.068,
+      "step": 7282
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.498336486578015e-06,
+      "loss": 0.0674,
+      "step": 7283
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.485296374256499e-06,
+      "loss": 0.0633,
+      "step": 7284
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 5.47227130749286e-06,
+      "loss": 0.0673,
+      "step": 7285
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.459261288360506e-06,
+      "loss": 0.0654,
+      "step": 7286
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.446266318930482e-06,
+      "loss": 0.0655,
+      "step": 7287
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.433286401271442e-06,
+      "loss": 0.0693,
+      "step": 7288
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.420321537449602e-06,
+      "loss": 0.0655,
+      "step": 7289
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.40737172952881e-06,
+      "loss": 0.0665,
+      "step": 7290
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.3944369795705385e-06,
+      "loss": 0.066,
+      "step": 7291
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.381517289633797e-06,
+      "loss": 0.0659,
+      "step": 7292
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.368612661775274e-06,
+      "loss": 0.0656,
+      "step": 7293
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.355723098049214e-06,
+      "loss": 0.0662,
+      "step": 7294
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.342848600507478e-06,
+      "loss": 0.0671,
+      "step": 7295
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.329989171199501e-06,
+      "loss": 0.0662,
+      "step": 7296
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.317144812172359e-06,
+      "loss": 0.0656,
+      "step": 7297
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.304315525470726e-06,
+      "loss": 0.0674,
+      "step": 7298
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 5.291501313136848e-06,
+      "loss": 0.0669,
+      "step": 7299
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.278702177210581e-06,
+      "loss": 0.063,
+      "step": 7300
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.265918119729418e-06,
+      "loss": 0.0667,
+      "step": 7301
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.253149142728398e-06,
+      "loss": 0.0656,
+      "step": 7302
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.24039524824016e-06,
+      "loss": 0.0673,
+      "step": 7303
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.2276564382950145e-06,
+      "loss": 0.067,
+      "step": 7304
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.214932714920795e-06,
+      "loss": 0.0671,
+      "step": 7305
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.202224080142948e-06,
+      "loss": 0.0636,
+      "step": 7306
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.189530535984532e-06,
+      "loss": 0.0661,
+      "step": 7307
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.176852084466233e-06,
+      "loss": 0.0676,
+      "step": 7308
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.164188727606267e-06,
+      "loss": 0.0638,
+      "step": 7309
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.151540467420479e-06,
+      "loss": 0.0655,
+      "step": 7310
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.138907305922358e-06,
+      "loss": 0.0649,
+      "step": 7311
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 5.126289245122906e-06,
+      "loss": 0.0658,
+      "step": 7312
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.113686287030739e-06,
+      "loss": 0.0665,
+      "step": 7313
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.101098433652152e-06,
+      "loss": 0.0653,
+      "step": 7314
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.08852568699093e-06,
+      "loss": 0.0657,
+      "step": 7315
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.075968049048508e-06,
+      "loss": 0.0658,
+      "step": 7316
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.063425521823906e-06,
+      "loss": 0.0667,
+      "step": 7317
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.0508981073137395e-06,
+      "loss": 0.0631,
+      "step": 7318
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.038385807512213e-06,
+      "loss": 0.0643,
+      "step": 7319
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.025888624411123e-06,
+      "loss": 0.0658,
+      "step": 7320
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.013406559999878e-06,
+      "loss": 0.0656,
+      "step": 7321
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 5.000939616265466e-06,
+      "loss": 0.0655,
+      "step": 7322
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.988487795192431e-06,
+      "loss": 0.0662,
+      "step": 7323
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.976051098763012e-06,
+      "loss": 0.0681,
+      "step": 7324
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.963629528956937e-06,
+      "loss": 0.0646,
+      "step": 7325
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.951223087751544e-06,
+      "loss": 0.0692,
+      "step": 7326
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.9388317771218106e-06,
+      "loss": 0.0642,
+      "step": 7327
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.926455599040302e-06,
+      "loss": 0.0654,
+      "step": 7328
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.9140945554771e-06,
+      "loss": 0.0655,
+      "step": 7329
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.901748648399951e-06,
+      "loss": 0.0637,
+      "step": 7330
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.889417879774183e-06,
+      "loss": 0.0621,
+      "step": 7331
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.8771022515626815e-06,
+      "loss": 0.0634,
+      "step": 7332
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.864801765725924e-06,
+      "loss": 0.0644,
+      "step": 7333
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.852516424222053e-06,
+      "loss": 0.065,
+      "step": 7334
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.8402462290066845e-06,
+      "loss": 0.0676,
+      "step": 7335
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.827991182033098e-06,
+      "loss": 0.0688,
+      "step": 7336
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.8157512852521455e-06,
+      "loss": 0.0648,
+      "step": 7337
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.803526540612268e-06,
+      "loss": 0.0672,
+      "step": 7338
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.791316950059488e-06,
+      "loss": 0.0633,
+      "step": 7339
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 4.7791225155374044e-06,
+      "loss": 0.0664,
+      "step": 7340
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.766943238987265e-06,
+      "loss": 0.0671,
+      "step": 7341
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.754779122347808e-06,
+      "loss": 0.0647,
+      "step": 7342
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.742630167555428e-06,
+      "loss": 0.064,
+      "step": 7343
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.73049637654408e-06,
+      "loss": 0.0686,
+      "step": 7344
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.7183777512453396e-06,
+      "loss": 0.064,
+      "step": 7345
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.706274293588297e-06,
+      "loss": 0.0678,
+      "step": 7346
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.69418600549969e-06,
+      "loss": 0.0656,
+      "step": 7347
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.682112888903845e-06,
+      "loss": 0.0651,
+      "step": 7348
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.670054945722624e-06,
+      "loss": 0.0662,
+      "step": 7349
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.658012177875515e-06,
+      "loss": 0.0632,
+      "step": 7350
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.645984587279551e-06,
+      "loss": 0.0618,
+      "step": 7351
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.63397217584941e-06,
+      "loss": 0.0659,
+      "step": 7352
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.621974945497287e-06,
+      "loss": 0.0677,
+      "step": 7353
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.609992898133009e-06,
+      "loss": 0.0666,
+      "step": 7354
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.598026035663961e-06,
+      "loss": 0.0677,
+      "step": 7355
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.586074359995119e-06,
+      "loss": 0.0657,
+      "step": 7356
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.5741378730290384e-06,
+      "loss": 0.0648,
+      "step": 7357
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.562216576665868e-06,
+      "loss": 0.0652,
+      "step": 7358
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.550310472803309e-06,
+      "loss": 0.0685,
+      "step": 7359
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.5384195633366934e-06,
+      "loss": 0.0654,
+      "step": 7360
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.526543850158871e-06,
+      "loss": 0.065,
+      "step": 7361
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.514683335160341e-06,
+      "loss": 0.0655,
+      "step": 7362
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.502838020229116e-06,
+      "loss": 0.0626,
+      "step": 7363
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.4910079072508436e-06,
+      "loss": 0.0652,
+      "step": 7364
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.479192998108727e-06,
+      "loss": 0.0641,
+      "step": 7365
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.4673932946835524e-06,
+      "loss": 0.0669,
+      "step": 7366
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.455608798853683e-06,
+      "loss": 0.064,
+      "step": 7367
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.443839512495041e-06,
+      "loss": 0.0653,
+      "step": 7368
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.4320854374811815e-06,
+      "loss": 0.069,
+      "step": 7369
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.42034657568321e-06,
+      "loss": 0.0654,
+      "step": 7370
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.4086229289697635e-06,
+      "loss": 0.0671,
+      "step": 7371
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.39691449920715e-06,
+      "loss": 0.0642,
+      "step": 7372
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.385221288259167e-06,
+      "loss": 0.0678,
+      "step": 7373
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.37354329798726e-06,
+      "loss": 0.0629,
+      "step": 7374
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.361880530250384e-06,
+      "loss": 0.0651,
+      "step": 7375
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.350232986905123e-06,
+      "loss": 0.0672,
+      "step": 7376
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.338600669805637e-06,
+      "loss": 0.0658,
+      "step": 7377
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.326983580803612e-06,
+      "loss": 0.0661,
+      "step": 7378
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.315381721748357e-06,
+      "loss": 0.0658,
+      "step": 7379
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 4.303795094486762e-06,
+      "loss": 0.0669,
+      "step": 7380
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.292223700863251e-06,
+      "loss": 0.0642,
+      "step": 7381
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.2806675427198515e-06,
+      "loss": 0.0663,
+      "step": 7382
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.269126621896147e-06,
+      "loss": 0.0633,
+      "step": 7383
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.2576009402293335e-06,
+      "loss": 0.0645,
+      "step": 7384
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.246090499554134e-06,
+      "loss": 0.0622,
+      "step": 7385
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.234595301702859e-06,
+      "loss": 0.0671,
+      "step": 7386
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.223115348505446e-06,
+      "loss": 0.0665,
+      "step": 7387
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.2116506417893e-06,
+      "loss": 0.0653,
+      "step": 7388
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.2002011833794926e-06,
+      "loss": 0.0668,
+      "step": 7389
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.188766975098646e-06,
+      "loss": 0.0631,
+      "step": 7390
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.177348018766913e-06,
+      "loss": 0.0649,
+      "step": 7391
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.1659443162020415e-06,
+      "loss": 0.0628,
+      "step": 7392
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.154555869219412e-06,
+      "loss": 0.0638,
+      "step": 7393
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.143182679631885e-06,
+      "loss": 0.0649,
+      "step": 7394
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.131824749249913e-06,
+      "loss": 0.0671,
+      "step": 7395
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.12048207988156e-06,
+      "loss": 0.0639,
+      "step": 7396
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.109154673332449e-06,
+      "loss": 0.0705,
+      "step": 7397
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.0978425314057374e-06,
+      "loss": 0.0663,
+      "step": 7398
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.086545655902186e-06,
+      "loss": 0.064,
+      "step": 7399
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.07526404862012e-06,
+      "loss": 0.067,
+      "step": 7400
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.063997711355427e-06,
+      "loss": 0.0656,
+      "step": 7401
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.05274664590155e-06,
+      "loss": 0.0624,
+      "step": 7402
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.041510854049546e-06,
+      "loss": 0.0646,
+      "step": 7403
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.030290337588005e-06,
+      "loss": 0.0688,
+      "step": 7404
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.019085098303077e-06,
+      "loss": 0.0664,
+      "step": 7405
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 4.007895137978502e-06,
+      "loss": 0.0657,
+      "step": 7406
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.996720458395609e-06,
+      "loss": 0.0656,
+      "step": 7407
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.985561061333232e-06,
+      "loss": 0.0642,
+      "step": 7408
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.974416948567794e-06,
+      "loss": 0.0624,
+      "step": 7409
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.963288121873354e-06,
+      "loss": 0.0651,
+      "step": 7410
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.952174583021451e-06,
+      "loss": 0.0647,
+      "step": 7411
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.941076333781201e-06,
+      "loss": 0.0686,
+      "step": 7412
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.92999337591935e-06,
+      "loss": 0.066,
+      "step": 7413
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.918925711200139e-06,
+      "loss": 0.0636,
+      "step": 7414
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.907873341385404e-06,
+      "loss": 0.0657,
+      "step": 7415
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.896836268234538e-06,
+      "loss": 0.0651,
+      "step": 7416
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.885814493504525e-06,
+      "loss": 0.0637,
+      "step": 7417
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.874808018949883e-06,
+      "loss": 0.0674,
+      "step": 7418
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.863816846322677e-06,
+      "loss": 0.0639,
+      "step": 7419
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.852840977372618e-06,
+      "loss": 0.0631,
+      "step": 7420
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.841880413846899e-06,
+      "loss": 0.0667,
+      "step": 7421
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.830935157490278e-06,
+      "loss": 0.0679,
+      "step": 7422
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.820005210045164e-06,
+      "loss": 0.0673,
+      "step": 7423
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.8090905732514283e-06,
+      "loss": 0.0665,
+      "step": 7424
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.7981912488465386e-06,
+      "loss": 0.0644,
+      "step": 7425
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.7873072385655405e-06,
+      "loss": 0.0662,
+      "step": 7426
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.7764385441410475e-06,
+      "loss": 0.0654,
+      "step": 7427
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.7655851673031985e-06,
+      "loss": 0.0673,
+      "step": 7428
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.7547471097797103e-06,
+      "loss": 0.0639,
+      "step": 7429
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.743924373295904e-06,
+      "loss": 0.0649,
+      "step": 7430
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.7331169595745896e-06,
+      "loss": 0.0657,
+      "step": 7431
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.722324870336169e-06,
+      "loss": 0.0622,
+      "step": 7432
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.7115481072986237e-06,
+      "loss": 0.0669,
+      "step": 7433
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.700786672177492e-06,
+      "loss": 0.0654,
+      "step": 7434
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.6900405666858263e-06,
+      "loss": 0.0658,
+      "step": 7435
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.679309792534291e-06,
+      "loss": 0.0655,
+      "step": 7436
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.668594351431098e-06,
+      "loss": 0.0693,
+      "step": 7437
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.657894245081994e-06,
+      "loss": 0.0641,
+      "step": 7438
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.647209475190327e-06,
+      "loss": 0.066,
+      "step": 7439
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.6365400434569374e-06,
+      "loss": 0.0625,
+      "step": 7440
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.62588595158031e-06,
+      "loss": 0.0689,
+      "step": 7441
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.615247201256411e-06,
+      "loss": 0.0665,
+      "step": 7442
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.6046237941788073e-06,
+      "loss": 0.0657,
+      "step": 7443
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.594015732038625e-06,
+      "loss": 0.0624,
+      "step": 7444
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.583423016524512e-06,
+      "loss": 0.0648,
+      "step": 7445
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.572845649322709e-06,
+      "loss": 0.0633,
+      "step": 7446
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.562283632117003e-06,
+      "loss": 0.0655,
+      "step": 7447
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.551736966588726e-06,
+      "loss": 0.0686,
+      "step": 7448
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.5412056544167902e-06,
+      "loss": 0.0683,
+      "step": 7449
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.530689697277623e-06,
+      "loss": 0.0615,
+      "step": 7450
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.520189096845261e-06,
+      "loss": 0.0641,
+      "step": 7451
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.5097038547912463e-06,
+      "loss": 0.0619,
+      "step": 7452
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.4992339727847103e-06,
+      "loss": 0.0665,
+      "step": 7453
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.4887794524923302e-06,
+      "loss": 0.0666,
+      "step": 7454
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.4783402955783193e-06,
+      "loss": 0.066,
+      "step": 7455
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.4679165037044825e-06,
+      "loss": 0.0671,
+      "step": 7456
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.457508078530136e-06,
+      "loss": 0.0661,
+      "step": 7457
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.4471150217121774e-06,
+      "loss": 0.0658,
+      "step": 7458
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.4367373349050604e-06,
+      "loss": 0.0696,
+      "step": 7459
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.4263750197607635e-06,
+      "loss": 0.0645,
+      "step": 7460
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.416028077928868e-06,
+      "loss": 0.0637,
+      "step": 7461
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.405696511056455e-06,
+      "loss": 0.0654,
+      "step": 7462
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.395380320788177e-06,
+      "loss": 0.0643,
+      "step": 7463
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.3850795087662533e-06,
+      "loss": 0.0647,
+      "step": 7464
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.37479407663045e-06,
+      "loss": 0.0666,
+      "step": 7465
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.364524026018068e-06,
+      "loss": 0.0662,
+      "step": 7466
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.3542693585639662e-06,
+      "loss": 0.0699,
+      "step": 7467
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.3440300759005616e-06,
+      "loss": 0.0693,
+      "step": 7468
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.33380617965785e-06,
+      "loss": 0.0685,
+      "step": 7469
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.3235976714632965e-06,
+      "loss": 0.0685,
+      "step": 7470
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.313404552942012e-06,
+      "loss": 0.0645,
+      "step": 7471
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.3032268257165875e-06,
+      "loss": 0.065,
+      "step": 7472
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.293064491407205e-06,
+      "loss": 0.0648,
+      "step": 7473
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.282917551631559e-06,
+      "loss": 0.0656,
+      "step": 7474
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.2727860080049355e-06,
+      "loss": 0.0654,
+      "step": 7475
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.2626698621401554e-06,
+      "loss": 0.0675,
+      "step": 7476
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.2525691156475523e-06,
+      "loss": 0.068,
+      "step": 7477
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.242483770135063e-06,
+      "loss": 0.0646,
+      "step": 7478
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.2324138272081584e-06,
+      "loss": 0.0636,
+      "step": 7479
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.222359288469823e-06,
+      "loss": 0.0691,
+      "step": 7480
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.2123201555206094e-06,
+      "loss": 0.0655,
+      "step": 7481
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.20229642995864e-06,
+      "loss": 0.0675,
+      "step": 7482
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.1922881133795825e-06,
+      "loss": 0.0633,
+      "step": 7483
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.1822952073765846e-06,
+      "loss": 0.0627,
+      "step": 7484
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.1723177135404293e-06,
+      "loss": 0.0652,
+      "step": 7485
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.1623556334594128e-06,
+      "loss": 0.0611,
+      "step": 7486
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.152408968719356e-06,
+      "loss": 0.0654,
+      "step": 7487
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.142477720903636e-06,
+      "loss": 0.0633,
+      "step": 7488
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.132561891593222e-06,
+      "loss": 0.0642,
+      "step": 7489
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1226614823665514e-06,
+      "loss": 0.0684,
+      "step": 7490
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1127764947996518e-06,
+      "loss": 0.0626,
+      "step": 7491
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1029069304660984e-06,
+      "loss": 0.0657,
+      "step": 7492
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.093052790937012e-06,
+      "loss": 0.0669,
+      "step": 7493
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0832140777810157e-06,
+      "loss": 0.0639,
+      "step": 7494
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0733907925643456e-06,
+      "loss": 0.0642,
+      "step": 7495
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0635829368507287e-06,
+      "loss": 0.0673,
+      "step": 7496
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0537905122014598e-06,
+      "loss": 0.0672,
+      "step": 7497
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0440135201753374e-06,
+      "loss": 0.0639,
+      "step": 7498
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0342519623287935e-06,
+      "loss": 0.0634,
+      "step": 7499
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.024505840215708e-06,
+      "loss": 0.0647,
+      "step": 7500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0147751553875393e-06,
+      "loss": 0.0655,
+      "step": 7501
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.005059909393304e-06,
+      "loss": 0.0657,
+      "step": 7502
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.995360103779554e-06,
+      "loss": 0.0639,
+      "step": 7503
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.985675740090366e-06,
+      "loss": 0.0628,
+      "step": 7504
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.9760068198673607e-06,
+      "loss": 0.0667,
+      "step": 7505
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.9663533446497304e-06,
+      "loss": 0.0696,
+      "step": 7506
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.956715315974179e-06,
+      "loss": 0.0629,
+      "step": 7507
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.9470927353749346e-06,
+      "loss": 0.0662,
+      "step": 7508
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.937485604383838e-06,
+      "loss": 0.0667,
+      "step": 7509
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.9278939245302006e-06,
+      "loss": 0.0674,
+      "step": 7510
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.9183176973408886e-06,
+      "loss": 0.0677,
+      "step": 7511
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.9087569243403277e-06,
+      "loss": 0.0681,
+      "step": 7512
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.8992116070504895e-06,
+      "loss": 0.0653,
+      "step": 7513
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.8896817469908355e-06,
+      "loss": 0.0635,
+      "step": 7514
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.8801673456784086e-06,
+      "loss": 0.0671,
+      "step": 7515
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.8706684046278077e-06,
+      "loss": 0.0688,
+      "step": 7516
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.8611849253511346e-06,
+      "loss": 0.0668,
+      "step": 7517
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.8517169093580044e-06,
+      "loss": 0.0654,
+      "step": 7518
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.842264358155655e-06,
+      "loss": 0.0652,
+      "step": 7519
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.832827273248806e-06,
+      "loss": 0.0613,
+      "step": 7520
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.823405656139699e-06,
+      "loss": 0.0639,
+      "step": 7521
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.813999508328147e-06,
+      "loss": 0.0642,
+      "step": 7522
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.804608831311506e-06,
+      "loss": 0.0647,
+      "step": 7523
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.7952336265846367e-06,
+      "loss": 0.0677,
+      "step": 7524
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.785873895639968e-06,
+      "loss": 0.0654,
+      "step": 7525
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.776529639967451e-06,
+      "loss": 0.0664,
+      "step": 7526
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.767200861054564e-06,
+      "loss": 0.0642,
+      "step": 7527
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.7578875603863187e-06,
+      "loss": 0.0644,
+      "step": 7528
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.7485897394453064e-06,
+      "loss": 0.0636,
+      "step": 7529
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.7393073997116216e-06,
+      "loss": 0.0659,
+      "step": 7530
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.7300405426628706e-06,
+      "loss": 0.067,
+      "step": 7531
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.7207891697742405e-06,
+      "loss": 0.0649,
+      "step": 7532
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.711553282518442e-06,
+      "loss": 0.0647,
+      "step": 7533
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.702332882365688e-06,
+      "loss": 0.0658,
+      "step": 7534
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.6931279707837597e-06,
+      "loss": 0.0646,
+      "step": 7535
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.6839385492379856e-06,
+      "loss": 0.0647,
+      "step": 7536
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.6747646191911837e-06,
+      "loss": 0.0646,
+      "step": 7537
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.6656061821037193e-06,
+      "loss": 0.0655,
+      "step": 7538
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.6564632394335263e-06,
+      "loss": 0.0649,
+      "step": 7539
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.6473357926360407e-06,
+      "loss": 0.0667,
+      "step": 7540
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.6382238431642225e-06,
+      "loss": 0.0663,
+      "step": 7541
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.629127392468589e-06,
+      "loss": 0.0631,
+      "step": 7542
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.6200464419972035e-06,
+      "loss": 0.0652,
+      "step": 7543
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.6109809931956e-06,
+      "loss": 0.0678,
+      "step": 7544
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.601931047506911e-06,
+      "loss": 0.0693,
+      "step": 7545
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.592896606371764e-06,
+      "loss": 0.0682,
+      "step": 7546
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.583877671228341e-06,
+      "loss": 0.0656,
+      "step": 7547
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.574874243512326e-06,
+      "loss": 0.0654,
+      "step": 7548
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.565886324656952e-06,
+      "loss": 0.0634,
+      "step": 7549
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.556913916093007e-06,
+      "loss": 0.0636,
+      "step": 7550
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.5479570192487702e-06,
+      "loss": 0.0665,
+      "step": 7551
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.539015635550057e-06,
+      "loss": 0.0642,
+      "step": 7552
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.5300897664202517e-06,
+      "loss": 0.0645,
+      "step": 7553
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.521179413280206e-06,
+      "loss": 0.0653,
+      "step": 7554
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.5122845775483737e-06,
+      "loss": 0.0656,
+      "step": 7555
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.503405260640668e-06,
+      "loss": 0.0643,
+      "step": 7556
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.49454146397059e-06,
+      "loss": 0.0671,
+      "step": 7557
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.485693188949123e-06,
+      "loss": 0.065,
+      "step": 7558
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.476860436984818e-06,
+      "loss": 0.0649,
+      "step": 7559
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.468043209483739e-06,
+      "loss": 0.0655,
+      "step": 7560
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.4592415078494635e-06,
+      "loss": 0.0659,
+      "step": 7561
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.450455333483137e-06,
+      "loss": 0.0658,
+      "step": 7562
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.4416846877833855e-06,
+      "loss": 0.0666,
+      "step": 7563
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.432929572146392e-06,
+      "loss": 0.0683,
+      "step": 7564
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.4241899879658634e-06,
+      "loss": 0.0637,
+      "step": 7565
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.4154659366330323e-06,
+      "loss": 0.0655,
+      "step": 7566
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.406757419536665e-06,
+      "loss": 0.0637,
+      "step": 7567
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.398064438063041e-06,
+      "loss": 0.0638,
+      "step": 7568
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.3893869935959767e-06,
+      "loss": 0.0651,
+      "step": 7569
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.3807250875167998e-06,
+      "loss": 0.0634,
+      "step": 7570
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.3720787212043847e-06,
+      "loss": 0.0679,
+      "step": 7571
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.3634478960351423e-06,
+      "loss": 0.0683,
+      "step": 7572
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.3548326133829736e-06,
+      "loss": 0.0653,
+      "step": 7573
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.346232874619325e-06,
+      "loss": 0.0686,
+      "step": 7574
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.3376486811131693e-06,
+      "loss": 0.0668,
+      "step": 7575
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.3290800342310126e-06,
+      "loss": 0.0654,
+      "step": 7576
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.320526935336864e-06,
+      "loss": 0.0658,
+      "step": 7577
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.311989385792268e-06,
+      "loss": 0.0649,
+      "step": 7578
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.303467386956304e-06,
+      "loss": 0.0673,
+      "step": 7579
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2949609401855644e-06,
+      "loss": 0.0651,
+      "step": 7580
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.286470046834166e-06,
+      "loss": 0.0644,
+      "step": 7581
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.2779947082537612e-06,
+      "loss": 0.0668,
+      "step": 7582
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.269534925793515e-06,
+      "loss": 0.0662,
+      "step": 7583
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.261090700800106e-06,
+      "loss": 0.0651,
+      "step": 7584
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2526620346177697e-06,
+      "loss": 0.0638,
+      "step": 7585
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2442489285882327e-06,
+      "loss": 0.0633,
+      "step": 7586
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2358513840507465e-06,
+      "loss": 0.064,
+      "step": 7587
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2274694023421196e-06,
+      "loss": 0.0683,
+      "step": 7588
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2191029847966394e-06,
+      "loss": 0.0659,
+      "step": 7589
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2107521327461412e-06,
+      "loss": 0.0657,
+      "step": 7590
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.2024168475199614e-06,
+      "loss": 0.0684,
+      "step": 7591
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1940971304450166e-06,
+      "loss": 0.0648,
+      "step": 7592
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.185792982845647e-06,
+      "loss": 0.0619,
+      "step": 7593
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1775044060437954e-06,
+      "loss": 0.0641,
+      "step": 7594
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1692314013589177e-06,
+      "loss": 0.0645,
+      "step": 7595
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.160973970107949e-06,
+      "loss": 0.0659,
+      "step": 7596
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.15273211360536e-06,
+      "loss": 0.0647,
+      "step": 7597
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1445058331631685e-06,
+      "loss": 0.0668,
+      "step": 7598
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1362951300908927e-06,
+      "loss": 0.0649,
+      "step": 7599
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1281000056955656e-06,
+      "loss": 0.0686,
+      "step": 7600
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1199204612817657e-06,
+      "loss": 0.0643,
+      "step": 7601
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1117564981515624e-06,
+      "loss": 0.0655,
+      "step": 7602
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1036081176045606e-06,
+      "loss": 0.0667,
+      "step": 7603
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.0954753209378564e-06,
+      "loss": 0.065,
+      "step": 7604
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.0873581094461357e-06,
+      "loss": 0.0646,
+      "step": 7605
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.079256484421521e-06,
+      "loss": 0.064,
+      "step": 7606
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.071170447153703e-06,
+      "loss": 0.0661,
+      "step": 7607
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.0630999989298626e-06,
+      "loss": 0.0628,
+      "step": 7608
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.0550451410347503e-06,
+      "loss": 0.0657,
+      "step": 7609
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.0470058747505516e-06,
+      "loss": 0.067,
+      "step": 7610
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.0389822013570426e-06,
+      "loss": 0.0659,
+      "step": 7611
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.0309741221315015e-06,
+      "loss": 0.0662,
+      "step": 7612
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.0229816383487087e-06,
+      "loss": 0.0626,
+      "step": 7613
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.0150047512809356e-06,
+      "loss": 0.0626,
+      "step": 7614
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.007043462198044e-06,
+      "loss": 0.067,
+      "step": 7615
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.999097772367353e-06,
+      "loss": 0.069,
+      "step": 7616
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.9911676830537187e-06,
+      "loss": 0.0634,
+      "step": 7617
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.9832531955195076e-06,
+      "loss": 0.0651,
+      "step": 7618
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.9753543110246242e-06,
+      "loss": 0.0636,
+      "step": 7619
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.967471030826451e-06,
+      "loss": 0.0684,
+      "step": 7620
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.9596033561799176e-06,
+      "loss": 0.0656,
+      "step": 7621
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.951751288337467e-06,
+      "loss": 0.0649,
+      "step": 7622
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.9439148285490427e-06,
+      "loss": 0.0658,
+      "step": 7623
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.9360939780620925e-06,
+      "loss": 0.0632,
+      "step": 7624
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.928288738121631e-06,
+      "loss": 0.0663,
+      "step": 7625
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.920499109970142e-06,
+      "loss": 0.0623,
+      "step": 7626
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.9127250948476226e-06,
+      "loss": 0.0629,
+      "step": 7627
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.9049666939916055e-06,
+      "loss": 0.0659,
+      "step": 7628
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8972239086371468e-06,
+      "loss": 0.0683,
+      "step": 7629
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8894967400167828e-06,
+      "loss": 0.0657,
+      "step": 7630
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8817851893605964e-06,
+      "loss": 0.0653,
+      "step": 7631
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8740892578961611e-06,
+      "loss": 0.0633,
+      "step": 7632
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8664089468485746e-06,
+      "loss": 0.065,
+      "step": 7633
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8587442574404368e-06,
+      "loss": 0.0657,
+      "step": 7634
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8510951908918938e-06,
+      "loss": 0.0659,
+      "step": 7635
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8434617484205607e-06,
+      "loss": 0.0653,
+      "step": 7636
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.8358439312415876e-06,
+      "loss": 0.0687,
+      "step": 7637
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.8282417405676378e-06,
+      "loss": 0.0672,
+      "step": 7638
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.8206551776088987e-06,
+      "loss": 0.0631,
+      "step": 7639
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.8130842435730377e-06,
+      "loss": 0.0684,
+      "step": 7640
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.8055289396652575e-06,
+      "loss": 0.0666,
+      "step": 7641
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7979892670882737e-06,
+      "loss": 0.0653,
+      "step": 7642
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.790465227042315e-06,
+      "loss": 0.0658,
+      "step": 7643
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7829568207250903e-06,
+      "loss": 0.0633,
+      "step": 7644
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.775464049331854e-06,
+      "loss": 0.0639,
+      "step": 7645
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.767986914055375e-06,
+      "loss": 0.068,
+      "step": 7646
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.760525416085912e-06,
+      "loss": 0.0664,
+      "step": 7647
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7530795566112257e-06,
+      "loss": 0.0627,
+      "step": 7648
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.745649336816635e-06,
+      "loss": 0.0654,
+      "step": 7649
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7382347578849046e-06,
+      "loss": 0.0658,
+      "step": 7650
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.7308358209963682e-06,
+      "loss": 0.0665,
+      "step": 7651
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7234525273288283e-06,
+      "loss": 0.0652,
+      "step": 7652
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7160848780576334e-06,
+      "loss": 0.0686,
+      "step": 7653
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7087328743555898e-06,
+      "loss": 0.0666,
+      "step": 7654
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7013965173930723e-06,
+      "loss": 0.0685,
+      "step": 7655
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.694075808337925e-06,
+      "loss": 0.0663,
+      "step": 7656
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.686770748355504e-06,
+      "loss": 0.0655,
+      "step": 7657
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6794813386087016e-06,
+      "loss": 0.0632,
+      "step": 7658
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6722075802578896e-06,
+      "loss": 0.0657,
+      "step": 7659
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6649494744609638e-06,
+      "loss": 0.0614,
+      "step": 7660
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6577070223733337e-06,
+      "loss": 0.067,
+      "step": 7661
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6504802251478768e-06,
+      "loss": 0.0668,
+      "step": 7662
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.6432690839350394e-06,
+      "loss": 0.0653,
+      "step": 7663
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.636073599882726e-06,
+      "loss": 0.0658,
+      "step": 7664
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6288937741363753e-06,
+      "loss": 0.0659,
+      "step": 7665
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.62172960783894e-06,
+      "loss": 0.0645,
+      "step": 7666
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.61458110213083e-06,
+      "loss": 0.0689,
+      "step": 7667
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.607448258150035e-06,
+      "loss": 0.0639,
+      "step": 7668
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.6003310770319913e-06,
+      "loss": 0.0648,
+      "step": 7669
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5932295599096702e-06,
+      "loss": 0.0628,
+      "step": 7670
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5861437079135566e-06,
+      "loss": 0.0718,
+      "step": 7671
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5790735221716035e-06,
+      "loss": 0.065,
+      "step": 7672
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.572019003809322e-06,
+      "loss": 0.0707,
+      "step": 7673
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5649801539496912e-06,
+      "loss": 0.0656,
+      "step": 7674
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.557956973713215e-06,
+      "loss": 0.0644,
+      "step": 7675
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5509494642178767e-06,
+      "loss": 0.0683,
+      "step": 7676
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.5439576265792066e-06,
+      "loss": 0.0664,
+      "step": 7677
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.536981461910214e-06,
+      "loss": 0.0662,
+      "step": 7678
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.5300209713213998e-06,
+      "loss": 0.0647,
+      "step": 7679
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.5230761559207996e-06,
+      "loss": 0.0652,
+      "step": 7680
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.5161470168139513e-06,
+      "loss": 0.0621,
+      "step": 7681
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.5092335551038616e-06,
+      "loss": 0.0655,
+      "step": 7682
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.5023357718910946e-06,
+      "loss": 0.0649,
+      "step": 7683
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4954536682736719e-06,
+      "loss": 0.0658,
+      "step": 7684
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4885872453471506e-06,
+      "loss": 0.0656,
+      "step": 7685
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4817365042045671e-06,
+      "loss": 0.0674,
+      "step": 7686
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.474901445936472e-06,
+      "loss": 0.068,
+      "step": 7687
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4680820716309496e-06,
+      "loss": 0.0647,
+      "step": 7688
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.461278382373521e-06,
+      "loss": 0.0637,
+      "step": 7689
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.454490379247264e-06,
+      "loss": 0.0647,
+      "step": 7690
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4477180633327591e-06,
+      "loss": 0.0617,
+      "step": 7691
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4409614357080548e-06,
+      "loss": 0.067,
+      "step": 7692
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4342204974487238e-06,
+      "loss": 0.065,
+      "step": 7693
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4274952496278415e-06,
+      "loss": 0.0625,
+      "step": 7694
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4207856933159957e-06,
+      "loss": 0.0628,
+      "step": 7695
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4140918295812434e-06,
+      "loss": 0.066,
+      "step": 7696
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.407413659489165e-06,
+      "loss": 0.0629,
+      "step": 7697
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4007511841028664e-06,
+      "loss": 0.0648,
+      "step": 7698
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3941044044828988e-06,
+      "loss": 0.0664,
+      "step": 7699
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3874733216873714e-06,
+      "loss": 0.066,
+      "step": 7700
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.380857936771851e-06,
+      "loss": 0.0673,
+      "step": 7701
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3742582507894397e-06,
+      "loss": 0.0642,
+      "step": 7702
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3676742647907193e-06,
+      "loss": 0.0649,
+      "step": 7703
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3611059798237624e-06,
+      "loss": 0.0666,
+      "step": 7704
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.3545533969341885e-06,
+      "loss": 0.0642,
+      "step": 7705
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.3480165171650517e-06,
+      "loss": 0.0652,
+      "step": 7706
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.3414953415569641e-06,
+      "loss": 0.0663,
+      "step": 7707
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.334989871148018e-06,
+      "loss": 0.0654,
+      "step": 7708
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.3285001069737957e-06,
+      "loss": 0.0619,
+      "step": 7709
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.322026050067382e-06,
+      "loss": 0.0669,
+      "step": 7710
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.3155677014593748e-06,
+      "loss": 0.0638,
+      "step": 7711
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.3091250621778517e-06,
+      "loss": 0.0626,
+      "step": 7712
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.3026981332484034e-06,
+      "loss": 0.0648,
+      "step": 7713
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2962869156941226e-06,
+      "loss": 0.067,
+      "step": 7714
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.289891410535593e-06,
+      "loss": 0.0668,
+      "step": 7715
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2835116187908891e-06,
+      "loss": 0.0643,
+      "step": 7716
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.2771475414756096e-06,
+      "loss": 0.0683,
+      "step": 7717
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.270799179602833e-06,
+      "loss": 0.0666,
+      "step": 7718
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2644665341831286e-06,
+      "loss": 0.0663,
+      "step": 7719
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.258149606224579e-06,
+      "loss": 0.0648,
+      "step": 7720
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2518483967327689e-06,
+      "loss": 0.0662,
+      "step": 7721
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2455629067107622e-06,
+      "loss": 0.0653,
+      "step": 7722
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2392931371591144e-06,
+      "loss": 0.0663,
+      "step": 7723
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.233039089075938e-06,
+      "loss": 0.0649,
+      "step": 7724
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.226800763456759e-06,
+      "loss": 0.0646,
+      "step": 7725
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.220578161294661e-06,
+      "loss": 0.0667,
+      "step": 7726
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2143712835801956e-06,
+      "loss": 0.0676,
+      "step": 7727
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2081801313014285e-06,
+      "loss": 0.0688,
+      "step": 7728
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.2020047054439042e-06,
+      "loss": 0.0667,
+      "step": 7729
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.19584500699067e-06,
+      "loss": 0.0642,
+      "step": 7730
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.189701036922297e-06,
+      "loss": 0.0657,
+      "step": 7731
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.1835727962168031e-06,
+      "loss": 0.065,
+      "step": 7732
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1774602858497297e-06,
+      "loss": 0.0647,
+      "step": 7733
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1713635067941208e-06,
+      "loss": 0.0656,
+      "step": 7734
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1652824600205115e-06,
+      "loss": 0.0656,
+      "step": 7735
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.159217146496916e-06,
+      "loss": 0.0653,
+      "step": 7736
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1531675671888619e-06,
+      "loss": 0.0657,
+      "step": 7737
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1471337230593682e-06,
+      "loss": 0.0679,
+      "step": 7738
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.141115615068944e-06,
+      "loss": 0.064,
+      "step": 7739
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1351132441756006e-06,
+      "loss": 0.064,
+      "step": 7740
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1291266113348298e-06,
+      "loss": 0.0659,
+      "step": 7741
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1231557174996466e-06,
+      "loss": 0.0644,
+      "step": 7742
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1172005636205352e-06,
+      "loss": 0.0651,
+      "step": 7743
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1112611506454706e-06,
+      "loss": 0.0673,
+      "step": 7744
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.1053374795199412e-06,
+      "loss": 0.0651,
+      "step": 7745
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0994295511869257e-06,
+      "loss": 0.066,
+      "step": 7746
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.093537366586883e-06,
+      "loss": 0.0648,
+      "step": 7747
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0876609266577964e-06,
+      "loss": 0.0661,
+      "step": 7748
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.081800232335095e-06,
+      "loss": 0.0652,
+      "step": 7749
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0759552845517329e-06,
+      "loss": 0.0638,
+      "step": 7750
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0701260842381654e-06,
+      "loss": 0.0635,
+      "step": 7751
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0643126323223284e-06,
+      "loss": 0.066,
+      "step": 7752
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0585149297296259e-06,
+      "loss": 0.0688,
+      "step": 7753
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0527329773830086e-06,
+      "loss": 0.0682,
+      "step": 7754
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0469667762028846e-06,
+      "loss": 0.0688,
+      "step": 7755
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0412163271071417e-06,
+      "loss": 0.0633,
+      "step": 7756
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0354816310112038e-06,
+      "loss": 0.0601,
+      "step": 7757
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.029762688827951e-06,
+      "loss": 0.0682,
+      "step": 7758
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0240595014677668e-06,
+      "loss": 0.0648,
+      "step": 7759
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.0183720698385246e-06,
+      "loss": 0.0649,
+      "step": 7760
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.0127003948456004e-06,
+      "loss": 0.0657,
+      "step": 7761
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.0070444773918609e-06,
+      "loss": 0.0648,
+      "step": 7762
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.00140431837763e-06,
+      "loss": 0.0648,
+      "step": 7763
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.957799187007788e-07,
+      "loss": 0.0641,
+      "step": 7764
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.901712792566243e-07,
+      "loss": 0.0628,
+      "step": 7765
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.845784009379965e-07,
+      "loss": 0.0648,
+      "step": 7766
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.790012846352282e-07,
+      "loss": 0.0697,
+      "step": 7767
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.734399312360975e-07,
+      "loss": 0.0657,
+      "step": 7768
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.678943416259078e-07,
+      "loss": 0.065,
+      "step": 7769
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.623645166874751e-07,
+      "loss": 0.0636,
+      "step": 7770
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.568504573010396e-07,
+      "loss": 0.0606,
+      "step": 7771
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.51352164344399e-07,
+      "loss": 0.0649,
+      "step": 7772
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.458696386927979e-07,
+      "loss": 0.0644,
+      "step": 7773
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.404028812190047e-07,
+      "loss": 0.0654,
+      "step": 7774
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.349518927932344e-07,
+      "loss": 0.0638,
+      "step": 7775
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.295166742832262e-07,
+      "loss": 0.0637,
+      "step": 7776
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.240972265541991e-07,
+      "loss": 0.0661,
+      "step": 7777
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.186935504688632e-07,
+      "loss": 0.0678,
+      "step": 7778
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.13305646887419e-07,
+      "loss": 0.0666,
+      "step": 7779
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.079335166675474e-07,
+      "loss": 0.064,
+      "step": 7780
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.025771606644306e-07,
+      "loss": 0.0676,
+      "step": 7781
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.972365797307203e-07,
+      "loss": 0.0633,
+      "step": 7782
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.919117747165695e-07,
+      "loss": 0.0658,
+      "step": 7783
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.866027464696447e-07,
+      "loss": 0.0665,
+      "step": 7784
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.813094958350587e-07,
+      "loss": 0.0659,
+      "step": 7785
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.760320236554264e-07,
+      "loss": 0.0656,
+      "step": 7786
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.707703307708759e-07,
+      "loss": 0.0645,
+      "step": 7787
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.655244180189814e-07,
+      "loss": 0.0654,
+      "step": 7788
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.602942862348306e-07,
+      "loss": 0.0651,
+      "step": 7789
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.550799362510132e-07,
+      "loss": 0.0644,
+      "step": 7790
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.498813688975649e-07,
+      "loss": 0.068,
+      "step": 7791
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.44698585002035e-07,
+      "loss": 0.0641,
+      "step": 7792
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.395315853894747e-07,
+      "loss": 0.067,
+      "step": 7793
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.343803708823927e-07,
+      "loss": 0.0651,
+      "step": 7794
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.292449423007997e-07,
+      "loss": 0.0654,
+      "step": 7795
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.241253004621974e-07,
+      "loss": 0.0626,
+      "step": 7796
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.190214461815671e-07,
+      "loss": 0.0661,
+      "step": 7797
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.1393338027137e-07,
+      "loss": 0.0668,
+      "step": 7798
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.088611035415583e-07,
+      "loss": 0.0652,
+      "step": 7799
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.038046167996083e-07,
+      "loss": 0.0647,
+      "step": 7800
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.987639208504094e-07,
+      "loss": 0.0637,
+      "step": 7801
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.937390164963976e-07,
+      "loss": 0.0661,
+      "step": 7802
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.887299045374664e-07,
+      "loss": 0.066,
+      "step": 7803
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.837365857710222e-07,
+      "loss": 0.0666,
+      "step": 7804
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.787590609919182e-07,
+      "loss": 0.0609,
+      "step": 7805
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.737973309925095e-07,
+      "loss": 0.0627,
+      "step": 7806
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.688513965626642e-07,
+      "loss": 0.0647,
+      "step": 7807
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.639212584897082e-07,
+      "loss": 0.0657,
+      "step": 7808
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.590069175584246e-07,
+      "loss": 0.0665,
+      "step": 7809
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.541083745511657e-07,
+      "loss": 0.0689,
+      "step": 7810
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.492256302476963e-07,
+      "loss": 0.0672,
+      "step": 7811
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.443586854252727e-07,
+      "loss": 0.0681,
+      "step": 7812
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.395075408586749e-07,
+      "loss": 0.0635,
+      "step": 7813
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.346721973201298e-07,
+      "loss": 0.0667,
+      "step": 7814
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.298526555793661e-07,
+      "loss": 0.0645,
+      "step": 7815
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.250489164035923e-07,
+      "loss": 0.065,
+      "step": 7816
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.202609805575189e-07,
+      "loss": 0.0685,
+      "step": 7817
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.15488848803314e-07,
+      "loss": 0.0607,
+      "step": 7818
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.107325219006366e-07,
+      "loss": 0.0658,
+      "step": 7819
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.059920006066367e-07,
+      "loss": 0.0674,
+      "step": 7820
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.01267285675955e-07,
+      "loss": 0.065,
+      "step": 7821
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.96558377860701e-07,
+      "loss": 0.0655,
+      "step": 7822
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.918652779104751e-07,
+      "loss": 0.0619,
+      "step": 7823
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.871879865723574e-07,
+      "loss": 0.0653,
+      "step": 7824
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.825265045909079e-07,
+      "loss": 0.0662,
+      "step": 7825
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.778808327081887e-07,
+      "loss": 0.0658,
+      "step": 7826
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 6.732509716637303e-07,
+      "loss": 0.0651,
+      "step": 7827
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.686369221945432e-07,
+      "loss": 0.0648,
+      "step": 7828
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.640386850351177e-07,
+      "loss": 0.0655,
+      "step": 7829
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.594562609174681e-07,
+      "loss": 0.0663,
+      "step": 7830
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.548896505710333e-07,
+      "loss": 0.067,
+      "step": 7831
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.50338854722754e-07,
+      "loss": 0.0663,
+      "step": 7832
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.458038740970728e-07,
+      "loss": 0.0636,
+      "step": 7833
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.412847094159013e-07,
+      "loss": 0.0682,
+      "step": 7834
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.367813613986418e-07,
+      "loss": 0.0695,
+      "step": 7835
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.32293830762154e-07,
+      "loss": 0.0663,
+      "step": 7836
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.278221182207999e-07,
+      "loss": 0.0642,
+      "step": 7837
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.233662244864324e-07,
+      "loss": 0.0652,
+      "step": 7838
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.189261502683619e-07,
+      "loss": 0.0646,
+      "step": 7839
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 6.145018962734006e-07,
+      "loss": 0.0651,
+      "step": 7840
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.100934632058297e-07,
+      "loss": 0.0645,
+      "step": 7841
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.057008517674212e-07,
+      "loss": 0.0659,
+      "step": 7842
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.013240626574157e-07,
+      "loss": 0.0656,
+      "step": 7843
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.969630965725447e-07,
+      "loss": 0.0631,
+      "step": 7844
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.926179542070309e-07,
+      "loss": 0.0634,
+      "step": 7845
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.882886362525541e-07,
+      "loss": 0.0677,
+      "step": 7846
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.839751433982855e-07,
+      "loss": 0.065,
+      "step": 7847
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.796774763308976e-07,
+      "loss": 0.064,
+      "step": 7848
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.753956357344992e-07,
+      "loss": 0.0662,
+      "step": 7849
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.711296222907336e-07,
+      "loss": 0.0654,
+      "step": 7850
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.668794366786689e-07,
+      "loss": 0.0654,
+      "step": 7851
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.626450795749083e-07,
+      "loss": 0.0648,
+      "step": 7852
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.584265516534904e-07,
+      "loss": 0.0631,
+      "step": 7853
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 5.542238535859556e-07,
+      "loss": 0.0639,
+      "step": 7854
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.500369860413134e-07,
+      "loss": 0.0653,
+      "step": 7855
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.458659496860863e-07,
+      "loss": 0.066,
+      "step": 7856
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.417107451842207e-07,
+      "loss": 0.0662,
+      "step": 7857
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.375713731971988e-07,
+      "loss": 0.0669,
+      "step": 7858
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.334478343839377e-07,
+      "loss": 0.0696,
+      "step": 7859
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.293401294008571e-07,
+      "loss": 0.0655,
+      "step": 7860
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.25248258901867e-07,
+      "loss": 0.0626,
+      "step": 7861
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.211722235383243e-07,
+      "loss": 0.0651,
+      "step": 7862
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.171120239590877e-07,
+      "loss": 0.0666,
+      "step": 7863
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.130676608104845e-07,
+      "loss": 0.066,
+      "step": 7864
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.090391347363444e-07,
+      "loss": 0.0647,
+      "step": 7865
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.050264463779431e-07,
+      "loss": 0.0653,
+      "step": 7866
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 5.010295963740475e-07,
+      "loss": 0.0668,
+      "step": 7867
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.970485853609264e-07,
+      "loss": 0.0628,
+      "step": 7868
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.93083413972284e-07,
+      "loss": 0.0676,
+      "step": 7869
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.891340828393487e-07,
+      "loss": 0.0692,
+      "step": 7870
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.852005925907732e-07,
+      "loss": 0.0683,
+      "step": 7871
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.812829438527567e-07,
+      "loss": 0.0683,
+      "step": 7872
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.773811372489112e-07,
+      "loss": 0.0651,
+      "step": 7873
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.734951734003623e-07,
+      "loss": 0.0647,
+      "step": 7874
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.696250529257151e-07,
+      "loss": 0.064,
+      "step": 7875
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.657707764410435e-07,
+      "loss": 0.0624,
+      "step": 7876
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.6193234455989e-07,
+      "loss": 0.0662,
+      "step": 7877
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.5810975789329914e-07,
+      "loss": 0.0632,
+      "step": 7878
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.54303017049762e-07,
+      "loss": 0.0648,
+      "step": 7879
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.5051212263528265e-07,
+      "loss": 0.0671,
+      "step": 7880
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 4.4673707525330067e-07,
+      "loss": 0.0659,
+      "step": 7881
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.4297787550477974e-07,
+      "loss": 0.0652,
+      "step": 7882
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.3923452398811904e-07,
+      "loss": 0.0658,
+      "step": 7883
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.355070212992307e-07,
+      "loss": 0.0621,
+      "step": 7884
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.317953680314735e-07,
+      "loss": 0.068,
+      "step": 7885
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.28099564775708e-07,
+      "loss": 0.0673,
+      "step": 7886
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.244196121202415e-07,
+      "loss": 0.0638,
+      "step": 7887
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.2075551065089424e-07,
+      "loss": 0.066,
+      "step": 7888
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.17107260950933e-07,
+      "loss": 0.0663,
+      "step": 7889
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.1347486360112655e-07,
+      "loss": 0.0663,
+      "step": 7890
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.098583191796901e-07,
+      "loss": 0.0639,
+      "step": 7891
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.0625762826234094e-07,
+      "loss": 0.0655,
+      "step": 7892
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 4.02672791422265e-07,
+      "loss": 0.063,
+      "step": 7893
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.99103809230128e-07,
+      "loss": 0.0654,
+      "step": 7894
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.9555068225405335e-07,
+      "loss": 0.0641,
+      "step": 7895
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.9201341105967735e-07,
+      "loss": 0.0637,
+      "step": 7896
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.8849199621007195e-07,
+      "loss": 0.0664,
+      "step": 7897
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.8498643826579974e-07,
+      "loss": 0.0627,
+      "step": 7898
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.8149673778490323e-07,
+      "loss": 0.066,
+      "step": 7899
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.780228953229048e-07,
+      "loss": 0.062,
+      "step": 7900
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.745649114328065e-07,
+      "loss": 0.0675,
+      "step": 7901
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.7112278666504575e-07,
+      "loss": 0.0696,
+      "step": 7902
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.676965215676065e-07,
+      "loss": 0.0646,
+      "step": 7903
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.6428611668587464e-07,
+      "loss": 0.0634,
+      "step": 7904
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.608915725627493e-07,
+      "loss": 0.0671,
+      "step": 7905
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.575128897386093e-07,
+      "loss": 0.0672,
+      "step": 7906
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.541500687512911e-07,
+      "loss": 0.0659,
+      "step": 7907
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.50803110136122e-07,
+      "loss": 0.0648,
+      "step": 7908
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.47472014425887e-07,
+      "loss": 0.0642,
+      "step": 7909
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.4415678215086176e-07,
+      "loss": 0.064,
+      "step": 7910
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.40857413838791e-07,
+      "loss": 0.0654,
+      "step": 7911
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.375739100148767e-07,
+      "loss": 0.0655,
+      "step": 7912
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.343062712018341e-07,
+      "loss": 0.0667,
+      "step": 7913
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.310544979198138e-07,
+      "loss": 0.0667,
+      "step": 7914
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.2781859068645726e-07,
+      "loss": 0.0673,
+      "step": 7915
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.245985500168969e-07,
+      "loss": 0.065,
+      "step": 7916
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.213943764237115e-07,
+      "loss": 0.0629,
+      "step": 7917
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.1820607041695983e-07,
+      "loss": 0.0669,
+      "step": 7918
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.1503363250419136e-07,
+      "loss": 0.0652,
+      "step": 7919
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.118770631904133e-07,
+      "loss": 0.0645,
+      "step": 7920
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.087363629781126e-07,
+      "loss": 0.0663,
+      "step": 7921
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0561153236724484e-07,
+      "loss": 0.0686,
+      "step": 7922
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0250257185525655e-07,
+      "loss": 0.0668,
+      "step": 7923
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.9940948193704076e-07,
+      "loss": 0.066,
+      "step": 7924
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.9633226310498143e-07,
+      "loss": 0.0666,
+      "step": 7925
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.932709158489422e-07,
+      "loss": 0.0644,
+      "step": 7926
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.9022544065625545e-07,
+      "loss": 0.0623,
+      "step": 7927
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.8719583801170014e-07,
+      "loss": 0.0655,
+      "step": 7928
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.841821083975682e-07,
+      "loss": 0.0639,
+      "step": 7929
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.8118425229359814e-07,
+      "loss": 0.0666,
+      "step": 7930
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.782022701770304e-07,
+      "loss": 0.0625,
+      "step": 7931
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.752361625225297e-07,
+      "loss": 0.0668,
+      "step": 7932
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.7228592980228506e-07,
+      "loss": 0.0665,
+      "step": 7933
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.69351572485943e-07,
+      "loss": 0.0682,
+      "step": 7934
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.6643309104058545e-07,
+      "loss": 0.0647,
+      "step": 7935
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.6353048593082964e-07,
+      "loss": 0.0655,
+      "step": 7936
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.606437576187171e-07,
+      "loss": 0.0673,
+      "step": 7937
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.5777290656378017e-07,
+      "loss": 0.0649,
+      "step": 7938
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.5491793322302004e-07,
+      "loss": 0.0641,
+      "step": 7939
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.520788380509287e-07,
+      "loss": 0.0659,
+      "step": 7940
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.492556214994446e-07,
+      "loss": 0.0639,
+      "step": 7941
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.4644828401797495e-07,
+      "loss": 0.0639,
+      "step": 7942
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.4365682605343997e-07,
+      "loss": 0.0674,
+      "step": 7943
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.408812480501843e-07,
+      "loss": 0.0655,
+      "step": 7944
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.3812155045004335e-07,
+      "loss": 0.0677,
+      "step": 7945
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.3537773369234352e-07,
+      "loss": 0.0655,
+      "step": 7946
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.3264979821385757e-07,
+      "loss": 0.0662,
+      "step": 7947
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.299377444488271e-07,
+      "loss": 0.0642,
+      "step": 7948
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.2724157282899562e-07,
+      "loss": 0.0638,
+      "step": 7949
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.245612837835531e-07,
+      "loss": 0.0622,
+      "step": 7950
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.2189687773915835e-07,
+      "loss": 0.0657,
+      "step": 7951
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.19248355119972e-07,
+      "loss": 0.0608,
+      "step": 7952
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.1661571634759015e-07,
+      "loss": 0.0652,
+      "step": 7953
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.139989618410998e-07,
+      "loss": 0.067,
+      "step": 7954
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.1139809201705664e-07,
+      "loss": 0.0646,
+      "step": 7955
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.088131072894739e-07,
+      "loss": 0.0654,
+      "step": 7956
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0624400806987797e-07,
+      "loss": 0.0661,
+      "step": 7957
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0369079476720842e-07,
+      "loss": 0.0634,
+      "step": 7958
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0115346778791788e-07,
+      "loss": 0.0636,
+      "step": 7959
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.9863202753591657e-07,
+      "loss": 0.0658,
+      "step": 7960
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.9612647441258347e-07,
+      "loss": 0.0661,
+      "step": 7961
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.9363680881677727e-07,
+      "loss": 0.0655,
+      "step": 7962
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.9116303114480315e-07,
+      "loss": 0.0679,
+      "step": 7963
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.887051417904795e-07,
+      "loss": 0.0636,
+      "step": 7964
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.8626314114505994e-07,
+      "loss": 0.0643,
+      "step": 7965
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.838370295972891e-07,
+      "loss": 0.0645,
+      "step": 7966
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.8142680753335805e-07,
+      "loss": 0.0669,
+      "step": 7967
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.7903247533695988e-07,
+      "loss": 0.067,
+      "step": 7968
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.7665403338922305e-07,
+      "loss": 0.0656,
+      "step": 7969
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.742914820687891e-07,
+      "loss": 0.0643,
+      "step": 7970
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.7194482175172388e-07,
+      "loss": 0.0655,
+      "step": 7971
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.6961405281160635e-07,
+      "loss": 0.0656,
+      "step": 7972
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.6729917561945086e-07,
+      "loss": 0.0668,
+      "step": 7973
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.6500019054376258e-07,
+      "loss": 0.0645,
+      "step": 7974
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.627170979505155e-07,
+      "loss": 0.0622,
+      "step": 7975
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.6044989820314104e-07,
+      "loss": 0.0658,
+      "step": 7976
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.581985916625617e-07,
+      "loss": 0.0652,
+      "step": 7977
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.5596317868713517e-07,
+      "loss": 0.0677,
+      "step": 7978
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.537436596327435e-07,
+      "loss": 0.0658,
+      "step": 7979
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.5154003485267076e-07,
+      "loss": 0.0665,
+      "step": 7980
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.493523046977252e-07,
+      "loss": 0.0654,
+      "step": 7981
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.471804695161727e-07,
+      "loss": 0.0632,
+      "step": 7982
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.450245296537256e-07,
+      "loss": 0.068,
+      "step": 7983
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.4288448545359822e-07,
+      "loss": 0.0675,
+      "step": 7984
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.4076033725645143e-07,
+      "loss": 0.0674,
+      "step": 7985
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3865208540041474e-07,
+      "loss": 0.0619,
+      "step": 7986
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3655973022110858e-07,
+      "loss": 0.063,
+      "step": 7987
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3448327205161094e-07,
+      "loss": 0.0643,
+      "step": 7988
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.324227112224574e-07,
+      "loss": 0.0708,
+      "step": 7989
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3037804806166344e-07,
+      "loss": 0.0699,
+      "step": 7990
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2834928289472416e-07,
+      "loss": 0.066,
+      "step": 7991
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2633641604458125e-07,
+      "loss": 0.066,
+      "step": 7992
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2433944783166729e-07,
+      "loss": 0.0638,
+      "step": 7993
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2235837857387246e-07,
+      "loss": 0.0652,
+      "step": 7994
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.2039320858655557e-07,
+      "loss": 0.0662,
+      "step": 7995
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.1844393818255528e-07,
+      "loss": 0.0652,
+      "step": 7996
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.1651056767215673e-07,
+      "loss": 0.0658,
+      "step": 7997
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.1459309736313595e-07,
+      "loss": 0.0666,
+      "step": 7998
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.1269152756073764e-07,
+      "loss": 0.0648,
+      "step": 7999
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.1080585856766412e-07,
+      "loss": 0.0639,
+      "step": 8000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.0893609068407528e-07,
+      "loss": 0.0668,
+      "step": 8001
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.0708222420763303e-07,
+      "loss": 0.0679,
+      "step": 8002
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.0524425943344574e-07,
+      "loss": 0.0679,
+      "step": 8003
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.0342219665407938e-07,
+      "loss": 0.0619,
+      "step": 8004
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.0161603615961301e-07,
+      "loss": 0.0628,
+      "step": 8005
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.982577823753891e-08,
+      "loss": 0.0667,
+      "step": 8006
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.805142317285132e-08,
+      "loss": 0.067,
+      "step": 8007
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.629297124800208e-08,
+      "loss": 0.0623,
+      "step": 8008
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.455042274293391e-08,
+      "loss": 0.0658,
+      "step": 8009
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.282377793501385e-08,
+      "loss": 0.0634,
+      "step": 8010
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.11130370991109e-08,
+      "loss": 0.0647,
+      "step": 8011
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.941820050756277e-08,
+      "loss": 0.0626,
+      "step": 8012
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.773926843015367e-08,
+      "loss": 0.0642,
+      "step": 8013
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.607624113415869e-08,
+      "loss": 0.068,
+      "step": 8014
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.442911888429939e-08,
+      "loss": 0.0655,
+      "step": 8015
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.279790194278824e-08,
+      "loss": 0.0674,
+      "step": 8016
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 8.118259056928423e-08,
+      "loss": 0.0656,
+      "step": 8017
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.95831850209261e-08,
+      "loss": 0.0648,
+      "step": 8018
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.799968555233239e-08,
+      "loss": 0.0686,
+      "step": 8019
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.643209241556815e-08,
+      "loss": 0.0653,
+      "step": 8020
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.488040586016709e-08,
+      "loss": 0.0639,
+      "step": 8021
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.334462613315385e-08,
+      "loss": 0.0637,
+      "step": 8022
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.182475347898842e-08,
+      "loss": 0.0669,
+      "step": 8023
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.032078813962173e-08,
+      "loss": 0.0644,
+      "step": 8024
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.883273035447335e-08,
+      "loss": 0.0639,
+      "step": 8025
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.736058036042048e-08,
+      "loss": 0.0638,
+      "step": 8026
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.590433839179788e-08,
+      "loss": 0.0678,
+      "step": 8027
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.446400468044233e-08,
+      "loss": 0.0652,
+      "step": 8028
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.303957945562599e-08,
+      "loss": 0.064,
+      "step": 8029
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 6.163106294408972e-08,
+      "loss": 0.0674,
+      "step": 8030
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 6.023845537006522e-08,
+      "loss": 0.0671,
+      "step": 8031
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.886175695524188e-08,
+      "loss": 0.0658,
+      "step": 8032
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.750096791876658e-08,
+      "loss": 0.0647,
+      "step": 8033
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.6156088477254955e-08,
+      "loss": 0.0648,
+      "step": 8034
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.482711884480241e-08,
+      "loss": 0.064,
+      "step": 8035
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.351405923295083e-08,
+      "loss": 0.0656,
+      "step": 8036
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.221690985074412e-08,
+      "loss": 0.0661,
+      "step": 8037
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 5.093567090465046e-08,
+      "loss": 0.0656,
+      "step": 8038
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.967034259865111e-08,
+      "loss": 0.0645,
+      "step": 8039
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.842092513415164e-08,
+      "loss": 0.0653,
+      "step": 8040
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.71874187100485e-08,
+      "loss": 0.0676,
+      "step": 8041
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.596982352269574e-08,
+      "loss": 0.0648,
+      "step": 8042
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 4.4768139765927196e-08,
+      "loss": 0.0646,
+      "step": 8043
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.358236763103429e-08,
+      "loss": 0.0648,
+      "step": 8044
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.241250730677715e-08,
+      "loss": 0.0671,
+      "step": 8045
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.1258558979384575e-08,
+      "loss": 0.0658,
+      "step": 8046
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 4.012052283254297e-08,
+      "loss": 0.0681,
+      "step": 8047
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.899839904742964e-08,
+      "loss": 0.065,
+      "step": 8048
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.789218780264614e-08,
+      "loss": 0.0668,
+      "step": 8049
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.6801889274318266e-08,
+      "loss": 0.0617,
+      "step": 8050
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.5727503635984985e-08,
+      "loss": 0.068,
+      "step": 8051
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.466903105867614e-08,
+      "loss": 0.0662,
+      "step": 8052
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.3626471710901386e-08,
+      "loss": 0.0651,
+      "step": 8053
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.2599825758616863e-08,
+      "loss": 0.0659,
+      "step": 8054
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.15890933652474e-08,
+      "loss": 0.0633,
+      "step": 8055
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.059427469168652e-08,
+      "loss": 0.0668,
+      "step": 8056
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.9615369896307533e-08,
+      "loss": 0.064,
+      "step": 8057
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.8652379134941344e-08,
+      "loss": 0.0634,
+      "step": 8058
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.7705302560865344e-08,
+      "loss": 0.0655,
+      "step": 8059
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.677414032485892e-08,
+      "loss": 0.0633,
+      "step": 8060
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.5858892575147953e-08,
+      "loss": 0.0636,
+      "step": 8061
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.4959559457427006e-08,
+      "loss": 0.0655,
+      "step": 8062
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.407614111485934e-08,
+      "loss": 0.0668,
+      "step": 8063
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.3208637688065804e-08,
+      "loss": 0.0628,
+      "step": 8064
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.235704931515814e-08,
+      "loss": 0.0634,
+      "step": 8065
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.152137613169458e-08,
+      "loss": 0.0635,
+      "step": 8066
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.0701618270690948e-08,
+      "loss": 0.0655,
+      "step": 8067
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.9897775862653957e-08,
+      "loss": 0.0674,
+      "step": 8068
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.910984903554791e-08,
+      "loss": 0.0673,
+      "step": 8069
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.8337837914783605e-08,
+      "loss": 0.0662,
+      "step": 8070
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.7581742623273834e-08,
+      "loss": 0.0674,
+      "step": 8071
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.684156328137787e-08,
+      "loss": 0.0653,
+      "step": 8072
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.6117300006912584e-08,
+      "loss": 0.0662,
+      "step": 8073
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.5408952915174634e-08,
+      "loss": 0.0654,
+      "step": 8074
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.4716522118929377e-08,
+      "loss": 0.0663,
+      "step": 8075
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.4040007728399752e-08,
+      "loss": 0.0664,
+      "step": 8076
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3379409851277391e-08,
+      "loss": 0.0643,
+      "step": 8077
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.2734728592722623e-08,
+      "loss": 0.0668,
+      "step": 8078
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.210596405536446e-08,
+      "loss": 0.0647,
+      "step": 8079
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.1493116339289512e-08,
+      "loss": 0.0652,
+      "step": 8080
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.0896185542053072e-08,
+      "loss": 0.0625,
+      "step": 8081
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.0315171758690235e-08,
+      "loss": 0.0683,
+      "step": 8082
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 9.750075081682575e-09,
+      "loss": 0.0619,
+      "step": 8083
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 9.200895600980363e-09,
+      "loss": 0.0659,
+      "step": 8084
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 8.667633404013665e-09,
+      "loss": 0.0651,
+      "step": 8085
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 8.150288575670128e-09,
+      "loss": 0.0658,
+      "step": 8086
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.648861198306101e-09,
+      "loss": 0.0657,
+      "step": 8087
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.163351351746616e-09,
+      "loss": 0.0638,
+      "step": 8088
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.6937591132743e-09,
+      "loss": 0.0676,
+      "step": 8089
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 6.240084557629367e-09,
+      "loss": 0.0663,
+      "step": 8090
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 5.8023277570540266e-09,
+      "loss": 0.065,
+      "step": 8091
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 5.380488781225879e-09,
+      "loss": 0.0664,
+      "step": 8092
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.974567697280108e-09,
+      "loss": 0.0671,
+      "step": 8093
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.5845645698539e-09,
+      "loss": 0.0643,
+      "step": 8094
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 4.210479461030925e-09,
+      "loss": 0.0624,
+      "step": 8095
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.852312430352445e-09,
+      "loss": 0.0654,
+      "step": 8096
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.5100635348284116e-09,
+      "loss": 0.0629,
+      "step": 8097
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.1837328289596737e-09,
+      "loss": 0.066,
+      "step": 8098
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.873320364671361e-09,
+      "loss": 0.0623,
+      "step": 8099
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.5788261913906044e-09,
+      "loss": 0.0641,
+      "step": 8100
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.3002503560021204e-09,
+      "loss": 0.0627,
+      "step": 8101
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0375929028371156e-09,
+      "loss": 0.0686,
+      "step": 8102
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.7908538737287928e-09,
+      "loss": 0.0627,
+      "step": 8103
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.5600333079235363e-09,
+      "loss": 0.0638,
+      "step": 8104
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3451312421919327e-09,
+      "loss": 0.0659,
+      "step": 8105
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.1461477107399531e-09,
+      "loss": 0.0657,
+      "step": 8106
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 9.630827452311586e-10,
+      "loss": 0.0675,
+      "step": 8107
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.959363748089032e-10,
+      "loss": 0.0634,
+      "step": 8108
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 6.447086260963353e-10,
+      "loss": 0.0633,
+      "step": 8109
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 5.093995231519877e-10,
+      "loss": 0.0672,
+      "step": 8110
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.900090875141871e-10,
+      "loss": 0.066,
+      "step": 8111
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.865373382010539e-10,
+      "loss": 0.064,
+      "step": 8112
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.9898429167719557e-10,
+      "loss": 0.0656,
+      "step": 8113
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.273499618759111e-10,
+      "loss": 0.0636,
+      "step": 8114
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.163436021029313e-11,
+      "loss": 0.0648,
+      "step": 8115
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.183749553992144e-11,
+      "loss": 0.0657,
+      "step": 8116
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.959374204169478e-12,
+      "loss": 0.0624,
+      "step": 8117
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0,
+      "loss": 0.0638,
+      "step": 8118
+    },
+    {
+      "epoch": 6.0,
+      "step": 8118,
+      "total_flos": 1.7294893396918272e+16,
+      "train_loss": 0.07726468138229703,
+      "train_runtime": 93632.2893,
+      "train_samples_per_second": 8.323,
+      "train_steps_per_second": 0.087
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 8118,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 5000,
+  "total_flos": 1.7294893396918272e+16,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}