diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,115981 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 19326,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.448275862068966e-08,
+      "loss": 0.7661,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.896551724137931e-08,
+      "loss": 1.4468,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0344827586206898e-07,
+      "loss": 1.4534,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3793103448275863e-07,
+      "loss": 1.3823,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7241379310344828e-07,
+      "loss": 1.5125,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0689655172413796e-07,
+      "loss": 1.4889,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.413793103448276e-07,
+      "loss": 1.2908,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7586206896551726e-07,
+      "loss": 1.4012,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1034482758620694e-07,
+      "loss": 1.6395,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4482758620689656e-07,
+      "loss": 1.4805,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7931034482758624e-07,
+      "loss": 1.4205,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.137931034482759e-07,
+      "loss": 1.4153,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4827586206896554e-07,
+      "loss": 1.5244,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.827586206896552e-07,
+      "loss": 1.4868,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.172413793103449e-07,
+      "loss": 1.4583,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.517241379310345e-07,
+      "loss": 1.3198,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.862068965517241e-07,
+      "loss": 1.4207,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.206896551724139e-07,
+      "loss": 1.4197,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.551724137931036e-07,
+      "loss": 1.2316,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.896551724137931e-07,
+      "loss": 1.2242,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.241379310344829e-07,
+      "loss": 1.4196,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.586206896551725e-07,
+      "loss": 1.3197,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.931034482758622e-07,
+      "loss": 1.3804,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.275862068965518e-07,
+      "loss": 1.1549,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.620689655172415e-07,
+      "loss": 1.1625,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.965517241379311e-07,
+      "loss": 1.2245,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.310344827586208e-07,
+      "loss": 1.1765,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.655172413793103e-07,
+      "loss": 1.2505,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.2489,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0344827586206898e-06,
+      "loss": 1.2489,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0689655172413794e-06,
+      "loss": 1.2981,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.103448275862069e-06,
+      "loss": 1.1525,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1379310344827587e-06,
+      "loss": 1.1891,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1724137931034483e-06,
+      "loss": 1.2227,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2068965517241381e-06,
+      "loss": 1.0931,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2413793103448277e-06,
+      "loss": 1.2038,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2758620689655174e-06,
+      "loss": 1.2126,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3103448275862072e-06,
+      "loss": 1.1193,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3448275862068966e-06,
+      "loss": 1.1189,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3793103448275862e-06,
+      "loss": 1.149,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.413793103448276e-06,
+      "loss": 1.1338,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4482758620689657e-06,
+      "loss": 1.0924,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4827586206896551e-06,
+      "loss": 1.1437,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.517241379310345e-06,
+      "loss": 1.1361,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5517241379310346e-06,
+      "loss": 1.1329,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5862068965517244e-06,
+      "loss": 1.0889,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6206896551724138e-06,
+      "loss": 1.1228,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6551724137931037e-06,
+      "loss": 1.1711,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6896551724137933e-06,
+      "loss": 1.1266,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.724137931034483e-06,
+      "loss": 0.9776,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7586206896551725e-06,
+      "loss": 1.0834,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7931034482758622e-06,
+      "loss": 1.0392,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.827586206896552e-06,
+      "loss": 1.0128,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8620689655172416e-06,
+      "loss": 0.9954,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.896551724137931e-06,
+      "loss": 1.0568,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9310344827586207e-06,
+      "loss": 1.0168,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9655172413793105e-06,
+      "loss": 1.0812,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.9991,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0344827586206897e-06,
+      "loss": 0.9781,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0689655172413796e-06,
+      "loss": 1.1157,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.103448275862069e-06,
+      "loss": 0.9789,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.137931034482759e-06,
+      "loss": 0.9445,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1724137931034482e-06,
+      "loss": 1.0741,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.206896551724138e-06,
+      "loss": 1.0257,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.241379310344828e-06,
+      "loss": 1.1248,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2758620689655173e-06,
+      "loss": 1.0347,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.310344827586207e-06,
+      "loss": 1.0985,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3448275862068966e-06,
+      "loss": 1.0784,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3793103448275864e-06,
+      "loss": 1.0295,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4137931034482762e-06,
+      "loss": 1.0654,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4482758620689657e-06,
+      "loss": 0.2081,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4827586206896555e-06,
+      "loss": 1.0142,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.517241379310345e-06,
+      "loss": 1.0655,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5517241379310347e-06,
+      "loss": 1.121,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5862068965517246e-06,
+      "loss": 0.9749,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6206896551724144e-06,
+      "loss": 0.9956,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6551724137931034e-06,
+      "loss": 0.9807,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6896551724137932e-06,
+      "loss": 1.067,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.724137931034483e-06,
+      "loss": 0.8956,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7586206896551725e-06,
+      "loss": 0.9143,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7931034482758623e-06,
+      "loss": 1.0723,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.827586206896552e-06,
+      "loss": 0.9982,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8620689655172416e-06,
+      "loss": 0.9525,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8965517241379314e-06,
+      "loss": 1.1261,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.931034482758621e-06,
+      "loss": 0.9525,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9655172413793102e-06,
+      "loss": 0.9985,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.976,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.03448275862069e-06,
+      "loss": 1.0782,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0689655172413797e-06,
+      "loss": 0.9833,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.103448275862069e-06,
+      "loss": 0.9842,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.137931034482759e-06,
+      "loss": 0.8678,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.172413793103449e-06,
+      "loss": 1.0472,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.206896551724138e-06,
+      "loss": 0.9529,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2413793103448277e-06,
+      "loss": 1.0049,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2758620689655175e-06,
+      "loss": 0.9554,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3103448275862073e-06,
+      "loss": 1.0129,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3448275862068967e-06,
+      "loss": 0.9656,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3793103448275866e-06,
+      "loss": 0.9625,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4137931034482764e-06,
+      "loss": 0.9667,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.448275862068966e-06,
+      "loss": 1.0007,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4827586206896552e-06,
+      "loss": 0.9371,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.517241379310345e-06,
+      "loss": 0.9363,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5517241379310345e-06,
+      "loss": 0.9696,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5862068965517243e-06,
+      "loss": 0.9869,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.620689655172414e-06,
+      "loss": 0.9125,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.655172413793104e-06,
+      "loss": 1.1107,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6896551724137934e-06,
+      "loss": 0.907,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7241379310344832e-06,
+      "loss": 1.0371,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7586206896551727e-06,
+      "loss": 1.0004,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.793103448275862e-06,
+      "loss": 1.0234,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.827586206896552e-06,
+      "loss": 0.959,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.862068965517241e-06,
+      "loss": 0.9911,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.896551724137932e-06,
+      "loss": 1.0712,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.931034482758621e-06,
+      "loss": 0.9388,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96551724137931e-06,
+      "loss": 0.9955,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.96,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.03448275862069e-06,
+      "loss": 0.9607,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0689655172413795e-06,
+      "loss": 0.9907,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.103448275862069e-06,
+      "loss": 0.9505,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.137931034482759e-06,
+      "loss": 0.9041,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1724137931034486e-06,
+      "loss": 1.0004,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.206896551724138e-06,
+      "loss": 1.0608,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.241379310344828e-06,
+      "loss": 1.0328,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.275862068965518e-06,
+      "loss": 0.9387,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.310344827586207e-06,
+      "loss": 1.0109,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3448275862068965e-06,
+      "loss": 0.9294,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.379310344827587e-06,
+      "loss": 0.9803,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.413793103448276e-06,
+      "loss": 0.9731,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4482758620689656e-06,
+      "loss": 0.9433,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.482758620689656e-06,
+      "loss": 0.9783,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.517241379310345e-06,
+      "loss": 0.9999,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.551724137931035e-06,
+      "loss": 0.9289,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.586206896551724e-06,
+      "loss": 0.9642,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.620689655172414e-06,
+      "loss": 0.95,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.655172413793104e-06,
+      "loss": 0.9408,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.689655172413793e-06,
+      "loss": 0.9799,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.724137931034483e-06,
+      "loss": 0.9636,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.758620689655173e-06,
+      "loss": 0.9216,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.793103448275862e-06,
+      "loss": 0.9573,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8275862068965525e-06,
+      "loss": 1.0178,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.862068965517242e-06,
+      "loss": 0.9156,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.896551724137931e-06,
+      "loss": 0.9965,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931034482758621e-06,
+      "loss": 0.9743,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965517241379311e-06,
+      "loss": 0.9325,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.9291,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.03448275862069e-06,
+      "loss": 0.9585,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.06896551724138e-06,
+      "loss": 0.8847,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1034482758620695e-06,
+      "loss": 0.9603,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.137931034482759e-06,
+      "loss": 1.0121,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.172413793103449e-06,
+      "loss": 0.8952,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.206896551724139e-06,
+      "loss": 0.9236,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.241379310344829e-06,
+      "loss": 0.9778,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.275862068965518e-06,
+      "loss": 0.9311,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.310344827586207e-06,
+      "loss": 0.9696,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.344827586206896e-06,
+      "loss": 0.9322,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3793103448275865e-06,
+      "loss": 0.9547,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.413793103448276e-06,
+      "loss": 0.9259,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.448275862068966e-06,
+      "loss": 0.9672,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4827586206896556e-06,
+      "loss": 0.9185,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.517241379310345e-06,
+      "loss": 0.9391,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.551724137931035e-06,
+      "loss": 0.9728,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.586206896551725e-06,
+      "loss": 0.9919,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.620689655172414e-06,
+      "loss": 0.9056,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.655172413793104e-06,
+      "loss": 0.9389,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.689655172413794e-06,
+      "loss": 0.1928,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.724137931034483e-06,
+      "loss": 0.9342,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.758620689655173e-06,
+      "loss": 0.951,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.793103448275863e-06,
+      "loss": 0.9896,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.827586206896553e-06,
+      "loss": 0.9545,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.862068965517242e-06,
+      "loss": 0.9583,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.896551724137931e-06,
+      "loss": 0.984,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9310344827586205e-06,
+      "loss": 0.9392,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.965517241379311e-06,
+      "loss": 0.9097,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 0.9496,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.03448275862069e-06,
+      "loss": 0.8673,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.06896551724138e-06,
+      "loss": 0.9645,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.103448275862069e-06,
+      "loss": 0.8877,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1379310344827595e-06,
+      "loss": 0.9724,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.172413793103449e-06,
+      "loss": 0.9003,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.206896551724138e-06,
+      "loss": 0.942,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.241379310344829e-06,
+      "loss": 1.0144,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.275862068965518e-06,
+      "loss": 1.0101,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.310344827586207e-06,
+      "loss": 0.9259,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.344827586206898e-06,
+      "loss": 0.9347,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.379310344827587e-06,
+      "loss": 0.9123,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.413793103448276e-06,
+      "loss": 0.9353,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.448275862068966e-06,
+      "loss": 0.8659,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.482758620689655e-06,
+      "loss": 0.947,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.517241379310345e-06,
+      "loss": 0.9771,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.551724137931035e-06,
+      "loss": 0.9377,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.586206896551724e-06,
+      "loss": 0.972,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.620689655172415e-06,
+      "loss": 0.9448,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.655172413793104e-06,
+      "loss": 0.9142,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6896551724137935e-06,
+      "loss": 0.9394,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.724137931034484e-06,
+      "loss": 0.9539,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.758620689655173e-06,
+      "loss": 0.9197,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7931034482758626e-06,
+      "loss": 0.9929,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.827586206896553e-06,
+      "loss": 0.9135,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.862068965517242e-06,
+      "loss": 0.9034,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.896551724137932e-06,
+      "loss": 1.0114,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.931034482758622e-06,
+      "loss": 0.8987,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9655172413793105e-06,
+      "loss": 0.9276,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-06,
+      "loss": 0.9005,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.03448275862069e-06,
+      "loss": 0.9297,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0689655172413796e-06,
+      "loss": 0.2271,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.103448275862069e-06,
+      "loss": 0.932,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.137931034482759e-06,
+      "loss": 0.9832,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.172413793103449e-06,
+      "loss": 0.8943,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.206896551724139e-06,
+      "loss": 0.9578,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.241379310344828e-06,
+      "loss": 0.9462,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.275862068965518e-06,
+      "loss": 0.9294,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.310344827586208e-06,
+      "loss": 0.9919,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.344827586206897e-06,
+      "loss": 0.9448,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.379310344827587e-06,
+      "loss": 0.9719,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.413793103448277e-06,
+      "loss": 0.9214,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4482758620689665e-06,
+      "loss": 0.9542,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.482758620689656e-06,
+      "loss": 0.8927,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.517241379310345e-06,
+      "loss": 0.9551,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.551724137931035e-06,
+      "loss": 0.9949,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.586206896551724e-06,
+      "loss": 0.9608,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.620689655172414e-06,
+      "loss": 0.9818,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.655172413793104e-06,
+      "loss": 0.8763,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.689655172413794e-06,
+      "loss": 0.9424,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.724137931034483e-06,
+      "loss": 0.9865,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.758620689655173e-06,
+      "loss": 0.8132,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.793103448275863e-06,
+      "loss": 0.8521,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.827586206896552e-06,
+      "loss": 0.9802,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.862068965517242e-06,
+      "loss": 0.8809,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.896551724137932e-06,
+      "loss": 0.8787,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.93103448275862e-06,
+      "loss": 0.876,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.965517241379311e-06,
+      "loss": 0.9167,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.9652,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.034482758620692e-06,
+      "loss": 0.9176,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.06896551724138e-06,
+      "loss": 0.8914,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.103448275862069e-06,
+      "loss": 0.9396,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.137931034482759e-06,
+      "loss": 0.9259,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.17241379310345e-06,
+      "loss": 0.8922,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.206896551724138e-06,
+      "loss": 0.2221,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.241379310344828e-06,
+      "loss": 1.0323,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.275862068965518e-06,
+      "loss": 0.912,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.310344827586207e-06,
+      "loss": 0.9443,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.344827586206897e-06,
+      "loss": 0.8907,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.379310344827587e-06,
+      "loss": 0.7676,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.413793103448276e-06,
+      "loss": 0.8277,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.448275862068966e-06,
+      "loss": 1.0103,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.482758620689656e-06,
+      "loss": 0.8851,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.517241379310345e-06,
+      "loss": 0.8927,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.551724137931035e-06,
+      "loss": 0.2389,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.586206896551726e-06,
+      "loss": 0.9438,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.620689655172414e-06,
+      "loss": 0.9441,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.655172413793104e-06,
+      "loss": 0.8297,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.689655172413793e-06,
+      "loss": 0.9279,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.724137931034483e-06,
+      "loss": 0.9268,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.758620689655173e-06,
+      "loss": 0.9518,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.793103448275862e-06,
+      "loss": 0.9355,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.827586206896552e-06,
+      "loss": 0.9119,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.862068965517243e-06,
+      "loss": 0.968,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.896551724137931e-06,
+      "loss": 0.8895,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.931034482758621e-06,
+      "loss": 0.9421,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.965517241379312e-06,
+      "loss": 0.9203,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-06,
+      "loss": 0.9395,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.03448275862069e-06,
+      "loss": 0.8252,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.06896551724138e-06,
+      "loss": 0.9193,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.10344827586207e-06,
+      "loss": 0.9283,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.13793103448276e-06,
+      "loss": 0.8454,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.172413793103448e-06,
+      "loss": 0.9739,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.206896551724138e-06,
+      "loss": 0.8889,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.241379310344829e-06,
+      "loss": 0.9871,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.275862068965517e-06,
+      "loss": 0.841,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.310344827586207e-06,
+      "loss": 0.853,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.344827586206898e-06,
+      "loss": 0.9629,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.379310344827586e-06,
+      "loss": 0.963,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.413793103448277e-06,
+      "loss": 0.8615,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.448275862068967e-06,
+      "loss": 0.9183,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.482758620689655e-06,
+      "loss": 0.9161,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.517241379310346e-06,
+      "loss": 0.8921,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.551724137931036e-06,
+      "loss": 0.8796,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.586206896551724e-06,
+      "loss": 0.8935,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.620689655172415e-06,
+      "loss": 0.8568,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.655172413793105e-06,
+      "loss": 0.9678,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.689655172413794e-06,
+      "loss": 0.9287,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.724137931034484e-06,
+      "loss": 0.9476,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.758620689655172e-06,
+      "loss": 0.8284,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.793103448275863e-06,
+      "loss": 0.9161,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.827586206896553e-06,
+      "loss": 0.827,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.862068965517241e-06,
+      "loss": 0.8203,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.896551724137932e-06,
+      "loss": 0.8569,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.931034482758622e-06,
+      "loss": 0.9117,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.96551724137931e-06,
+      "loss": 0.9683,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.8414,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.003448275862069e-05,
+      "loss": 0.834,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.006896551724138e-05,
+      "loss": 0.9755,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0103448275862068e-05,
+      "loss": 0.9302,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.013793103448276e-05,
+      "loss": 0.9875,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0172413793103449e-05,
+      "loss": 0.9006,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0206896551724139e-05,
+      "loss": 0.967,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0241379310344828e-05,
+      "loss": 0.8594,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0275862068965518e-05,
+      "loss": 0.8845,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0310344827586208e-05,
+      "loss": 0.8471,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0344827586206898e-05,
+      "loss": 0.9202,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0379310344827587e-05,
+      "loss": 0.9077,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0413793103448277e-05,
+      "loss": 0.8975,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0448275862068966e-05,
+      "loss": 0.898,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0482758620689658e-05,
+      "loss": 0.9112,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0517241379310346e-05,
+      "loss": 0.9006,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0551724137931037e-05,
+      "loss": 0.9347,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0586206896551725e-05,
+      "loss": 0.8846,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0620689655172414e-05,
+      "loss": 0.8803,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0655172413793104e-05,
+      "loss": 0.2245,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0689655172413792e-05,
+      "loss": 0.204,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0724137931034484e-05,
+      "loss": 1.0225,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0758620689655173e-05,
+      "loss": 0.8541,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0793103448275863e-05,
+      "loss": 0.8768,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0827586206896552e-05,
+      "loss": 0.8969,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0862068965517242e-05,
+      "loss": 0.872,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0896551724137932e-05,
+      "loss": 0.9498,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0931034482758623e-05,
+      "loss": 0.8475,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0965517241379311e-05,
+      "loss": 0.8705,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.7968,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.103448275862069e-05,
+      "loss": 0.8827,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1068965517241382e-05,
+      "loss": 0.9146,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.110344827586207e-05,
+      "loss": 0.8599,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1137931034482759e-05,
+      "loss": 0.8562,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.117241379310345e-05,
+      "loss": 0.9085,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1206896551724138e-05,
+      "loss": 0.8734,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1241379310344828e-05,
+      "loss": 0.8741,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1275862068965517e-05,
+      "loss": 0.899,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1310344827586209e-05,
+      "loss": 0.8983,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1344827586206897e-05,
+      "loss": 0.8434,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1379310344827587e-05,
+      "loss": 0.8502,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1413793103448276e-05,
+      "loss": 0.921,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1448275862068966e-05,
+      "loss": 0.9232,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1482758620689657e-05,
+      "loss": 0.9161,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1517241379310347e-05,
+      "loss": 0.9063,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1551724137931035e-05,
+      "loss": 0.2117,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1586206896551726e-05,
+      "loss": 0.8961,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1620689655172414e-05,
+      "loss": 0.903,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1655172413793106e-05,
+      "loss": 0.9056,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1689655172413795e-05,
+      "loss": 0.8917,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1724137931034483e-05,
+      "loss": 0.8835,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1758620689655174e-05,
+      "loss": 0.8894,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1793103448275862e-05,
+      "loss": 0.878,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1827586206896552e-05,
+      "loss": 0.9104,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1862068965517241e-05,
+      "loss": 0.9556,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1896551724137933e-05,
+      "loss": 0.8744,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1931034482758621e-05,
+      "loss": 0.8369,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1965517241379312e-05,
+      "loss": 0.9094,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.8411,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.203448275862069e-05,
+      "loss": 0.9856,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.206896551724138e-05,
+      "loss": 0.863,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2103448275862071e-05,
+      "loss": 0.8555,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.213793103448276e-05,
+      "loss": 0.9689,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.217241379310345e-05,
+      "loss": 0.86,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2206896551724138e-05,
+      "loss": 0.9261,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2241379310344827e-05,
+      "loss": 0.8467,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2275862068965519e-05,
+      "loss": 0.8926,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2310344827586208e-05,
+      "loss": 0.9702,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2344827586206898e-05,
+      "loss": 0.8448,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2379310344827586e-05,
+      "loss": 0.238,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2413793103448277e-05,
+      "loss": 0.8962,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2448275862068965e-05,
+      "loss": 0.8866,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2482758620689657e-05,
+      "loss": 0.8894,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2517241379310346e-05,
+      "loss": 0.9111,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2551724137931036e-05,
+      "loss": 0.8894,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2586206896551725e-05,
+      "loss": 0.8758,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2620689655172415e-05,
+      "loss": 0.8702,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2655172413793105e-05,
+      "loss": 0.8926,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2689655172413795e-05,
+      "loss": 0.8297,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2724137931034484e-05,
+      "loss": 0.8751,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2758620689655174e-05,
+      "loss": 0.802,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2793103448275863e-05,
+      "loss": 0.8839,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2827586206896551e-05,
+      "loss": 0.8028,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2862068965517243e-05,
+      "loss": 0.8352,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2896551724137932e-05,
+      "loss": 0.8866,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2931034482758622e-05,
+      "loss": 0.9747,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.296551724137931e-05,
+      "loss": 0.943,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.9017,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.303448275862069e-05,
+      "loss": 0.8544,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3068965517241381e-05,
+      "loss": 0.8945,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.310344827586207e-05,
+      "loss": 0.8539,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.313793103448276e-05,
+      "loss": 0.891,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3172413793103449e-05,
+      "loss": 0.944,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3206896551724139e-05,
+      "loss": 0.8249,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.324137931034483e-05,
+      "loss": 0.912,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.327586206896552e-05,
+      "loss": 0.9079,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3310344827586208e-05,
+      "loss": 0.8612,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3344827586206897e-05,
+      "loss": 0.8493,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3379310344827587e-05,
+      "loss": 0.8518,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3413793103448276e-05,
+      "loss": 0.887,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3448275862068967e-05,
+      "loss": 0.9358,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3482758620689656e-05,
+      "loss": 0.8964,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3517241379310346e-05,
+      "loss": 0.8948,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3551724137931035e-05,
+      "loss": 0.9226,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3586206896551725e-05,
+      "loss": 0.936,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3620689655172414e-05,
+      "loss": 0.9113,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3655172413793106e-05,
+      "loss": 0.787,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3689655172413794e-05,
+      "loss": 0.8984,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3724137931034484e-05,
+      "loss": 0.8736,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3758620689655173e-05,
+      "loss": 0.9919,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3793103448275863e-05,
+      "loss": 0.9407,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3827586206896554e-05,
+      "loss": 0.9617,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3862068965517244e-05,
+      "loss": 0.927,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3896551724137932e-05,
+      "loss": 0.9143,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3931034482758621e-05,
+      "loss": 0.8091,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3965517241379311e-05,
+      "loss": 0.9573,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 0.9197,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4034482758620692e-05,
+      "loss": 0.8351,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.406896551724138e-05,
+      "loss": 0.9188,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.410344827586207e-05,
+      "loss": 0.9504,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4137931034482759e-05,
+      "loss": 0.908,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.417241379310345e-05,
+      "loss": 0.9122,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4206896551724138e-05,
+      "loss": 0.8767,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.424137931034483e-05,
+      "loss": 0.8837,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4275862068965518e-05,
+      "loss": 0.8742,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4310344827586209e-05,
+      "loss": 0.2242,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4344827586206897e-05,
+      "loss": 0.8823,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4379310344827588e-05,
+      "loss": 0.9119,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4413793103448278e-05,
+      "loss": 0.948,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4448275862068966e-05,
+      "loss": 0.8313,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4482758620689657e-05,
+      "loss": 0.9419,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4517241379310345e-05,
+      "loss": 0.8406,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4551724137931035e-05,
+      "loss": 0.7904,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4586206896551724e-05,
+      "loss": 0.8774,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4620689655172416e-05,
+      "loss": 0.9131,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4655172413793105e-05,
+      "loss": 0.8932,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4689655172413795e-05,
+      "loss": 0.9293,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4724137931034483e-05,
+      "loss": 0.7985,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4758620689655174e-05,
+      "loss": 0.7732,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4793103448275862e-05,
+      "loss": 0.8644,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4827586206896554e-05,
+      "loss": 0.9237,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4862068965517243e-05,
+      "loss": 0.8719,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4896551724137933e-05,
+      "loss": 0.8925,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4931034482758622e-05,
+      "loss": 0.8849,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4965517241379312e-05,
+      "loss": 0.8538,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9023,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.503448275862069e-05,
+      "loss": 0.8306,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5068965517241381e-05,
+      "loss": 0.9075,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.510344827586207e-05,
+      "loss": 0.8954,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.513793103448276e-05,
+      "loss": 0.8965,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5172413793103448e-05,
+      "loss": 0.8748,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.520689655172414e-05,
+      "loss": 0.9299,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5241379310344829e-05,
+      "loss": 0.9186,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5275862068965517e-05,
+      "loss": 0.7601,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5310344827586208e-05,
+      "loss": 0.8611,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5344827586206898e-05,
+      "loss": 0.874,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5379310344827588e-05,
+      "loss": 0.9011,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.541379310344828e-05,
+      "loss": 0.825,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5448275862068965e-05,
+      "loss": 0.8676,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.548275862068966e-05,
+      "loss": 0.8911,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5517241379310346e-05,
+      "loss": 0.8799,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5551724137931036e-05,
+      "loss": 0.8845,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5586206896551726e-05,
+      "loss": 0.8979,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5620689655172413e-05,
+      "loss": 0.8667,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5655172413793103e-05,
+      "loss": 0.9,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5689655172413794e-05,
+      "loss": 0.869,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5724137931034484e-05,
+      "loss": 0.8741,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5758620689655174e-05,
+      "loss": 0.8544,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5793103448275864e-05,
+      "loss": 0.8599,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.582758620689655e-05,
+      "loss": 0.9285,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.586206896551724e-05,
+      "loss": 0.9287,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5896551724137932e-05,
+      "loss": 0.9145,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5931034482758622e-05,
+      "loss": 0.9257,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5965517241379312e-05,
+      "loss": 0.892,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.8288,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.603448275862069e-05,
+      "loss": 0.9175,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6068965517241383e-05,
+      "loss": 0.7983,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.610344827586207e-05,
+      "loss": 0.8075,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.613793103448276e-05,
+      "loss": 0.8411,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.617241379310345e-05,
+      "loss": 0.8439,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6206896551724137e-05,
+      "loss": 0.8905,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6241379310344828e-05,
+      "loss": 0.8542,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6275862068965518e-05,
+      "loss": 0.8696,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6310344827586208e-05,
+      "loss": 0.7691,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.63448275862069e-05,
+      "loss": 0.8199,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.637931034482759e-05,
+      "loss": 0.8572,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6413793103448276e-05,
+      "loss": 0.7998,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6448275862068966e-05,
+      "loss": 0.861,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6482758620689656e-05,
+      "loss": 0.2314,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6517241379310346e-05,
+      "loss": 0.8129,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6551724137931037e-05,
+      "loss": 0.8634,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6586206896551727e-05,
+      "loss": 0.8633,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6620689655172414e-05,
+      "loss": 0.8884,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6655172413793107e-05,
+      "loss": 0.862,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6689655172413794e-05,
+      "loss": 0.8924,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6724137931034485e-05,
+      "loss": 0.7723,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6758620689655175e-05,
+      "loss": 0.851,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6793103448275862e-05,
+      "loss": 0.8333,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6827586206896552e-05,
+      "loss": 0.8658,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6862068965517242e-05,
+      "loss": 0.8394,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6896551724137932e-05,
+      "loss": 0.8962,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6931034482758623e-05,
+      "loss": 0.861,
+      "step": 491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6965517241379313e-05,
+      "loss": 0.8752,
+      "step": 492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7e-05,
+      "loss": 0.8405,
+      "step": 493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.703448275862069e-05,
+      "loss": 0.8268,
+      "step": 494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.706896551724138e-05,
+      "loss": 0.7806,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.710344827586207e-05,
+      "loss": 0.86,
+      "step": 496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.713793103448276e-05,
+      "loss": 0.8766,
+      "step": 497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.717241379310345e-05,
+      "loss": 0.8674,
+      "step": 498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7206896551724138e-05,
+      "loss": 0.8713,
+      "step": 499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7241379310344828e-05,
+      "loss": 0.8607,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.727586206896552e-05,
+      "loss": 0.8246,
+      "step": 501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.731034482758621e-05,
+      "loss": 0.9045,
+      "step": 502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.73448275862069e-05,
+      "loss": 0.7871,
+      "step": 503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7379310344827586e-05,
+      "loss": 0.8648,
+      "step": 504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7413793103448276e-05,
+      "loss": 0.891,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7448275862068966e-05,
+      "loss": 0.8325,
+      "step": 506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7482758620689657e-05,
+      "loss": 0.8588,
+      "step": 507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7517241379310347e-05,
+      "loss": 0.8306,
+      "step": 508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7551724137931037e-05,
+      "loss": 0.8649,
+      "step": 509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7586206896551724e-05,
+      "loss": 0.884,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7620689655172414e-05,
+      "loss": 0.9029,
+      "step": 511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7655172413793105e-05,
+      "loss": 0.8381,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7689655172413795e-05,
+      "loss": 0.876,
+      "step": 513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7724137931034485e-05,
+      "loss": 0.9444,
+      "step": 514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7758620689655175e-05,
+      "loss": 0.9087,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7793103448275862e-05,
+      "loss": 0.9033,
+      "step": 516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7827586206896553e-05,
+      "loss": 0.9435,
+      "step": 517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7862068965517243e-05,
+      "loss": 0.8557,
+      "step": 518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7896551724137933e-05,
+      "loss": 0.8755,
+      "step": 519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7931034482758623e-05,
+      "loss": 0.9907,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.796551724137931e-05,
+      "loss": 0.8678,
+      "step": 521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 0.198,
+      "step": 522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.803448275862069e-05,
+      "loss": 0.886,
+      "step": 523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.806896551724138e-05,
+      "loss": 0.8282,
+      "step": 524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.810344827586207e-05,
+      "loss": 0.7745,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.813793103448276e-05,
+      "loss": 0.8904,
+      "step": 526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.817241379310345e-05,
+      "loss": 0.9006,
+      "step": 527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.820689655172414e-05,
+      "loss": 0.8379,
+      "step": 528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.824137931034483e-05,
+      "loss": 0.8073,
+      "step": 529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.827586206896552e-05,
+      "loss": 0.8813,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.831034482758621e-05,
+      "loss": 0.8782,
+      "step": 531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8344827586206896e-05,
+      "loss": 0.7704,
+      "step": 532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8379310344827587e-05,
+      "loss": 0.9133,
+      "step": 533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8413793103448277e-05,
+      "loss": 0.7954,
+      "step": 534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8448275862068967e-05,
+      "loss": 0.8169,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8482758620689657e-05,
+      "loss": 0.8923,
+      "step": 536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8517241379310348e-05,
+      "loss": 0.8849,
+      "step": 537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8551724137931034e-05,
+      "loss": 0.8336,
+      "step": 538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8586206896551725e-05,
+      "loss": 0.8587,
+      "step": 539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8620689655172415e-05,
+      "loss": 0.8616,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8655172413793105e-05,
+      "loss": 0.8986,
+      "step": 541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8689655172413795e-05,
+      "loss": 0.9197,
+      "step": 542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8724137931034486e-05,
+      "loss": 0.861,
+      "step": 543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8758620689655173e-05,
+      "loss": 0.8474,
+      "step": 544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8793103448275863e-05,
+      "loss": 0.857,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8827586206896553e-05,
+      "loss": 0.9313,
+      "step": 546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8862068965517243e-05,
+      "loss": 0.9248,
+      "step": 547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8896551724137934e-05,
+      "loss": 0.8719,
+      "step": 548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.893103448275862e-05,
+      "loss": 0.8412,
+      "step": 549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.896551724137931e-05,
+      "loss": 0.8186,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9e-05,
+      "loss": 0.8425,
+      "step": 551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.903448275862069e-05,
+      "loss": 0.9104,
+      "step": 552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.906896551724138e-05,
+      "loss": 1.0118,
+      "step": 553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9103448275862072e-05,
+      "loss": 0.8661,
+      "step": 554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.913793103448276e-05,
+      "loss": 0.8695,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.917241379310345e-05,
+      "loss": 0.8101,
+      "step": 556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.920689655172414e-05,
+      "loss": 0.9098,
+      "step": 557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.924137931034483e-05,
+      "loss": 0.8514,
+      "step": 558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.927586206896552e-05,
+      "loss": 0.8687,
+      "step": 559
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.931034482758621e-05,
+      "loss": 0.8844,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9344827586206897e-05,
+      "loss": 0.8682,
+      "step": 561
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9379310344827587e-05,
+      "loss": 0.8511,
+      "step": 562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9413793103448277e-05,
+      "loss": 0.9209,
+      "step": 563
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9448275862068968e-05,
+      "loss": 0.8778,
+      "step": 564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9482758620689658e-05,
+      "loss": 0.8745,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9517241379310345e-05,
+      "loss": 0.8555,
+      "step": 566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9551724137931035e-05,
+      "loss": 0.9094,
+      "step": 567
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9586206896551725e-05,
+      "loss": 0.8312,
+      "step": 568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9620689655172416e-05,
+      "loss": 0.9972,
+      "step": 569
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9655172413793106e-05,
+      "loss": 0.8637,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9689655172413796e-05,
+      "loss": 0.8564,
+      "step": 571
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9724137931034483e-05,
+      "loss": 0.8463,
+      "step": 572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9758620689655173e-05,
+      "loss": 0.8007,
+      "step": 573
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9793103448275863e-05,
+      "loss": 0.8481,
+      "step": 574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9827586206896554e-05,
+      "loss": 0.9219,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9862068965517244e-05,
+      "loss": 0.922,
+      "step": 576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9896551724137934e-05,
+      "loss": 0.8981,
+      "step": 577
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.993103448275862e-05,
+      "loss": 0.8264,
+      "step": 578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996551724137931e-05,
+      "loss": 0.9277,
+      "step": 579
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.8726,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999859572385e-05,
+      "loss": 0.8601,
+      "step": 581
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999943828954e-05,
+      "loss": 0.9338,
+      "step": 582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999873615148e-05,
+      "loss": 0.8527,
+      "step": 583
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999997753158224e-05,
+      "loss": 0.2119,
+      "step": 584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996489309796e-05,
+      "loss": 0.8968,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994944606235e-05,
+      "loss": 0.9244,
+      "step": 586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993119047586e-05,
+      "loss": 0.8502,
+      "step": 587
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999991012633897e-05,
+      "loss": 0.7938,
+      "step": 588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999988625365227e-05,
+      "loss": 0.8777,
+      "step": 589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999985957241645e-05,
+      "loss": 0.8892,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999983008263225e-05,
+      "loss": 0.8784,
+      "step": 591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997977843005e-05,
+      "loss": 0.8846,
+      "step": 592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997626774221e-05,
+      "loss": 0.9625,
+      "step": 593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999972476199807e-05,
+      "loss": 0.8521,
+      "step": 594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996840380294e-05,
+      "loss": 0.8279,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999964050551736e-05,
+      "loss": 0.8574,
+      "step": 596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999959416446303e-05,
+      "loss": 0.8876,
+      "step": 597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995450148678e-05,
+      "loss": 0.8572,
+      "step": 598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999493056733e-05,
+      "loss": 0.8427,
+      "step": 599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999943829006015e-05,
+      "loss": 0.7954,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999938071485077e-05,
+      "loss": 0.9589,
+      "step": 601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999932033110642e-05,
+      "loss": 0.8395,
+      "step": 602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999925713882885e-05,
+      "loss": 0.8469,
+      "step": 603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999919113801983e-05,
+      "loss": 0.9484,
+      "step": 604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999912232868118e-05,
+      "loss": 0.884,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999905071081485e-05,
+      "loss": 0.821,
+      "step": 606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999897628442292e-05,
+      "loss": 0.8597,
+      "step": 607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999889904950737e-05,
+      "loss": 0.8766,
+      "step": 608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999881900607042e-05,
+      "loss": 0.892,
+      "step": 609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999873615411433e-05,
+      "loss": 0.8608,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999986504936414e-05,
+      "loss": 0.9668,
+      "step": 611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999856202465406e-05,
+      "loss": 0.8077,
+      "step": 612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999847074715482e-05,
+      "loss": 0.8419,
+      "step": 613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999837666114613e-05,
+      "loss": 0.8304,
+      "step": 614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999827976663078e-05,
+      "loss": 0.8309,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999981800636114e-05,
+      "loss": 0.8925,
+      "step": 616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999980775520908e-05,
+      "loss": 0.8431,
+      "step": 617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999797223207188e-05,
+      "loss": 0.7758,
+      "step": 618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999978641035576e-05,
+      "loss": 0.9313,
+      "step": 619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999775316655095e-05,
+      "loss": 0.8376,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999976394210551e-05,
+      "loss": 0.8313,
+      "step": 621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999752286707324e-05,
+      "loss": 0.8885,
+      "step": 622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999974035046086e-05,
+      "loss": 0.8939,
+      "step": 623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972813336646e-05,
+      "loss": 0.9073,
+      "step": 624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999971563542446e-05,
+      "loss": 0.9055,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999702856635217e-05,
+      "loss": 0.835,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999689796999088e-05,
+      "loss": 0.8678,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999676456516435e-05,
+      "loss": 0.87,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999966283518764e-05,
+      "loss": 0.85,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999648933013082e-05,
+      "loss": 0.8558,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999963474999315e-05,
+      "loss": 0.863,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999620286128243e-05,
+      "loss": 0.7934,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999605541418772e-05,
+      "loss": 0.8264,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999590515865145e-05,
+      "loss": 0.8413,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999575209467784e-05,
+      "loss": 0.873,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999955962222713e-05,
+      "loss": 0.8663,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999543754143602e-05,
+      "loss": 0.8782,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999952760521766e-05,
+      "loss": 0.8436,
+      "step": 638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999511175449753e-05,
+      "loss": 0.7482,
+      "step": 639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999494464840342e-05,
+      "loss": 0.8723,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999477473389896e-05,
+      "loss": 0.8617,
+      "step": 641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999460201098898e-05,
+      "loss": 0.913,
+      "step": 642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999442647967823e-05,
+      "loss": 0.7538,
+      "step": 643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999942481399717e-05,
+      "loss": 0.8733,
+      "step": 644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999406699187443e-05,
+      "loss": 0.8759,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999388303539143e-05,
+      "loss": 0.8236,
+      "step": 646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999369627052794e-05,
+      "loss": 0.8885,
+      "step": 647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999350669728918e-05,
+      "loss": 0.829,
+      "step": 648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999331431568042e-05,
+      "loss": 0.8245,
+      "step": 649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999311912570712e-05,
+      "loss": 0.8883,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999292112737478e-05,
+      "loss": 0.8904,
+      "step": 651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999927203206889e-05,
+      "loss": 0.8973,
+      "step": 652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999251670565518e-05,
+      "loss": 0.8511,
+      "step": 653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999231028227927e-05,
+      "loss": 0.8593,
+      "step": 654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999210105056704e-05,
+      "loss": 0.7918,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999918890105243e-05,
+      "loss": 0.252,
+      "step": 656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999167416215707e-05,
+      "loss": 0.8566,
+      "step": 657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999145650547137e-05,
+      "loss": 0.8942,
+      "step": 658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999123604047326e-05,
+      "loss": 0.8663,
+      "step": 659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99991012767169e-05,
+      "loss": 0.8665,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999907866855648e-05,
+      "loss": 0.8581,
+      "step": 661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999905577956671e-05,
+      "loss": 0.8114,
+      "step": 662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999032609748218e-05,
+      "loss": 0.8104,
+      "step": 663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999900915910167e-05,
+      "loss": 0.8076,
+      "step": 664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998985427627712e-05,
+      "loss": 0.8327,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999896141532702e-05,
+      "loss": 0.8073,
+      "step": 666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998937122200265e-05,
+      "loss": 0.9706,
+      "step": 667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998912548248132e-05,
+      "loss": 0.8246,
+      "step": 668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998887693471305e-05,
+      "loss": 0.9159,
+      "step": 669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998862557870487e-05,
+      "loss": 0.903,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998837141446378e-05,
+      "loss": 0.8829,
+      "step": 671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998811444199703e-05,
+      "loss": 0.8032,
+      "step": 672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998785466131173e-05,
+      "loss": 0.8764,
+      "step": 673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998759207241522e-05,
+      "loss": 0.8937,
+      "step": 674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998732667531487e-05,
+      "loss": 0.8417,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998705847001813e-05,
+      "loss": 0.8773,
+      "step": 676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998678745653256e-05,
+      "loss": 0.913,
+      "step": 677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998651363486576e-05,
+      "loss": 0.894,
+      "step": 678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998623700502537e-05,
+      "loss": 0.7801,
+      "step": 679
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998595756701923e-05,
+      "loss": 0.8977,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998567532085516e-05,
+      "loss": 0.9324,
+      "step": 681
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998539026654106e-05,
+      "loss": 0.8485,
+      "step": 682
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998510240408495e-05,
+      "loss": 0.825,
+      "step": 683
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998481173349497e-05,
+      "loss": 0.7867,
+      "step": 684
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998451825477918e-05,
+      "loss": 0.8466,
+      "step": 685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998422196794595e-05,
+      "loss": 0.7931,
+      "step": 686
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998392287300348e-05,
+      "loss": 0.8555,
+      "step": 687
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998362096996024e-05,
+      "loss": 0.9427,
+      "step": 688
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998331625882466e-05,
+      "loss": 0.886,
+      "step": 689
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999830087396054e-05,
+      "loss": 0.8471,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998269841231097e-05,
+      "loss": 0.7301,
+      "step": 691
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998238527695016e-05,
+      "loss": 0.8405,
+      "step": 692
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999820693335317e-05,
+      "loss": 0.8332,
+      "step": 693
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998175058206457e-05,
+      "loss": 0.8156,
+      "step": 694
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998142902255768e-05,
+      "loss": 0.9154,
+      "step": 695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998110465501998e-05,
+      "loss": 0.85,
+      "step": 696
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999807774794607e-05,
+      "loss": 0.8712,
+      "step": 697
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998044749588896e-05,
+      "loss": 0.8878,
+      "step": 698
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99980114704314e-05,
+      "loss": 0.8659,
+      "step": 699
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997977910474523e-05,
+      "loss": 0.9115,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997944069719205e-05,
+      "loss": 0.9052,
+      "step": 701
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99979099481664e-05,
+      "loss": 0.8796,
+      "step": 702
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999787554581706e-05,
+      "loss": 0.8903,
+      "step": 703
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997840862672153e-05,
+      "loss": 0.7581,
+      "step": 704
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997805898732654e-05,
+      "loss": 0.8472,
+      "step": 705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997770653999548e-05,
+      "loss": 0.7802,
+      "step": 706
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999773512847382e-05,
+      "loss": 0.9588,
+      "step": 707
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997699322156473e-05,
+      "loss": 0.8696,
+      "step": 708
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997663235048508e-05,
+      "loss": 0.9844,
+      "step": 709
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999762686715094e-05,
+      "loss": 0.8485,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999759021846479e-05,
+      "loss": 0.8771,
+      "step": 711
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997553288991085e-05,
+      "loss": 0.761,
+      "step": 712
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999751607873087e-05,
+      "loss": 0.865,
+      "step": 713
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997478587685183e-05,
+      "loss": 0.8811,
+      "step": 714
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999744081585508e-05,
+      "loss": 0.7997,
+      "step": 715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997402763241617e-05,
+      "loss": 0.8726,
+      "step": 716
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999736442984587e-05,
+      "loss": 0.855,
+      "step": 717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999732581566891e-05,
+      "loss": 0.8399,
+      "step": 718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997286920711828e-05,
+      "loss": 0.8404,
+      "step": 719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997247744975707e-05,
+      "loss": 0.859,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997208288461655e-05,
+      "loss": 0.8515,
+      "step": 721
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997168551170774e-05,
+      "loss": 0.8146,
+      "step": 722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997128533104192e-05,
+      "loss": 0.8784,
+      "step": 723
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997088234263017e-05,
+      "loss": 0.837,
+      "step": 724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999704765464839e-05,
+      "loss": 0.8434,
+      "step": 725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997006794261448e-05,
+      "loss": 0.7979,
+      "step": 726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999696565310334e-05,
+      "loss": 0.864,
+      "step": 727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996924231175224e-05,
+      "loss": 0.8287,
+      "step": 728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996882528478258e-05,
+      "loss": 0.8811,
+      "step": 729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996840545013617e-05,
+      "loss": 0.8296,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996798280782473e-05,
+      "loss": 0.8804,
+      "step": 731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996755735786023e-05,
+      "loss": 0.8787,
+      "step": 732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999671291002546e-05,
+      "loss": 0.8348,
+      "step": 733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996669803501982e-05,
+      "loss": 0.8209,
+      "step": 734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996626416216803e-05,
+      "loss": 0.8821,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999658274817114e-05,
+      "loss": 0.8984,
+      "step": 736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999653879936622e-05,
+      "loss": 0.8073,
+      "step": 737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996494569803276e-05,
+      "loss": 0.8525,
+      "step": 738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996450059483553e-05,
+      "loss": 0.828,
+      "step": 739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996405268408303e-05,
+      "loss": 0.8153,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996360196578777e-05,
+      "loss": 0.8693,
+      "step": 741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996314843996245e-05,
+      "loss": 0.8573,
+      "step": 742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996269210661982e-05,
+      "loss": 0.9077,
+      "step": 743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996223296577266e-05,
+      "loss": 0.8623,
+      "step": 744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999617710174339e-05,
+      "loss": 0.8456,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996130626161648e-05,
+      "loss": 0.8496,
+      "step": 746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996083869833346e-05,
+      "loss": 0.8204,
+      "step": 747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996036832759803e-05,
+      "loss": 0.8929,
+      "step": 748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999598951494233e-05,
+      "loss": 0.8591,
+      "step": 749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995941916382267e-05,
+      "loss": 0.8959,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999589403708094e-05,
+      "loss": 0.7747,
+      "step": 751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995845877039705e-05,
+      "loss": 0.8638,
+      "step": 752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995797436259902e-05,
+      "loss": 0.8613,
+      "step": 753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99957487147429e-05,
+      "loss": 0.8347,
+      "step": 754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995699712490067e-05,
+      "loss": 0.8793,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995650429502776e-05,
+      "loss": 0.8324,
+      "step": 756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995600865782413e-05,
+      "loss": 0.8531,
+      "step": 757
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999555102133037e-05,
+      "loss": 0.7748,
+      "step": 758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999550089614805e-05,
+      "loss": 0.8323,
+      "step": 759
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995450490236858e-05,
+      "loss": 0.8726,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99953998035982e-05,
+      "loss": 0.8113,
+      "step": 761
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995348836233517e-05,
+      "loss": 0.795,
+      "step": 762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999529758814423e-05,
+      "loss": 0.9078,
+      "step": 763
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995246059331782e-05,
+      "loss": 0.8041,
+      "step": 764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995194249797617e-05,
+      "loss": 0.7751,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999514215954319e-05,
+      "loss": 0.8162,
+      "step": 766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995089788569965e-05,
+      "loss": 0.8058,
+      "step": 767
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999503713687942e-05,
+      "loss": 0.8867,
+      "step": 768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999498420447302e-05,
+      "loss": 0.8217,
+      "step": 769
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994930991352263e-05,
+      "loss": 0.8221,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999487749751864e-05,
+      "loss": 0.8177,
+      "step": 771
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994823722973652e-05,
+      "loss": 0.8787,
+      "step": 772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999476966771881e-05,
+      "loss": 0.8951,
+      "step": 773
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999471533175563e-05,
+      "loss": 0.8702,
+      "step": 774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994660715085645e-05,
+      "loss": 0.8295,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994605817710385e-05,
+      "loss": 0.8931,
+      "step": 776
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994550639631387e-05,
+      "loss": 0.8376,
+      "step": 777
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994495180850207e-05,
+      "loss": 0.856,
+      "step": 778
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994439441368404e-05,
+      "loss": 0.7915,
+      "step": 779
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994383421187533e-05,
+      "loss": 0.8239,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999432712030918e-05,
+      "loss": 0.8198,
+      "step": 781
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994270538734917e-05,
+      "loss": 0.8121,
+      "step": 782
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994213676466337e-05,
+      "loss": 0.8744,
+      "step": 783
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999415653350504e-05,
+      "loss": 0.9104,
+      "step": 784
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994099109852624e-05,
+      "loss": 0.9229,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994041405510705e-05,
+      "loss": 0.8131,
+      "step": 786
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993983420480906e-05,
+      "loss": 0.9437,
+      "step": 787
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993925154764856e-05,
+      "loss": 0.8743,
+      "step": 788
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993866608364184e-05,
+      "loss": 0.8415,
+      "step": 789
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999380778128054e-05,
+      "loss": 0.8967,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993748673515577e-05,
+      "loss": 0.859,
+      "step": 791
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993689285070956e-05,
+      "loss": 0.8682,
+      "step": 792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999362961594834e-05,
+      "loss": 0.7971,
+      "step": 793
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993569666149407e-05,
+      "loss": 0.8117,
+      "step": 794
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993509435675844e-05,
+      "loss": 0.7568,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999344892452934e-05,
+      "loss": 0.803,
+      "step": 796
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999338813271159e-05,
+      "loss": 0.8745,
+      "step": 797
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993327060224304e-05,
+      "loss": 0.8035,
+      "step": 798
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993265707069204e-05,
+      "loss": 0.8518,
+      "step": 799
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993204073248005e-05,
+      "loss": 0.8281,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993142158762443e-05,
+      "loss": 0.7887,
+      "step": 801
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999307996361425e-05,
+      "loss": 0.8718,
+      "step": 802
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993017487805183e-05,
+      "loss": 0.8474,
+      "step": 803
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992954731336992e-05,
+      "loss": 0.7589,
+      "step": 804
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992891694211437e-05,
+      "loss": 0.8654,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992828376430287e-05,
+      "loss": 0.8349,
+      "step": 806
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992764777995325e-05,
+      "loss": 0.7918,
+      "step": 807
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992700898908337e-05,
+      "loss": 0.8192,
+      "step": 808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992636739171115e-05,
+      "loss": 0.8134,
+      "step": 809
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992572298785467e-05,
+      "loss": 0.8286,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992507577753192e-05,
+      "loss": 0.8479,
+      "step": 811
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992442576076113e-05,
+      "loss": 0.9363,
+      "step": 812
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999237729375606e-05,
+      "loss": 0.8303,
+      "step": 813
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999231173079486e-05,
+      "loss": 0.2577,
+      "step": 814
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999224588719436e-05,
+      "loss": 0.9517,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99921797629564e-05,
+      "loss": 0.8592,
+      "step": 816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992113358082847e-05,
+      "loss": 0.9126,
+      "step": 817
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999204667257556e-05,
+      "loss": 0.8183,
+      "step": 818
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999197970643642e-05,
+      "loss": 0.8176,
+      "step": 819
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991912459667295e-05,
+      "loss": 0.8223,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991844932270084e-05,
+      "loss": 0.84,
+      "step": 821
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999177712424668e-05,
+      "loss": 0.8289,
+      "step": 822
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999170903559899e-05,
+      "loss": 0.808,
+      "step": 823
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999164066632892e-05,
+      "loss": 0.8252,
+      "step": 824
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991572016438396e-05,
+      "loss": 0.7923,
+      "step": 825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991503085929343e-05,
+      "loss": 0.8093,
+      "step": 826
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99914338748037e-05,
+      "loss": 0.8064,
+      "step": 827
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991364383063404e-05,
+      "loss": 0.9495,
+      "step": 828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991294610710416e-05,
+      "loss": 0.8829,
+      "step": 829
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999122455774669e-05,
+      "loss": 0.8122,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991154224174195e-05,
+      "loss": 0.8162,
+      "step": 831
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991083609994906e-05,
+      "loss": 0.8658,
+      "step": 832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991012715210804e-05,
+      "loss": 0.8105,
+      "step": 833
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990941539823886e-05,
+      "loss": 0.9136,
+      "step": 834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990870083836146e-05,
+      "loss": 0.8016,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990798347249593e-05,
+      "loss": 0.9506,
+      "step": 836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999072633006624e-05,
+      "loss": 0.8633,
+      "step": 837
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999065403228811e-05,
+      "loss": 0.8231,
+      "step": 838
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990581453917234e-05,
+      "loss": 0.8154,
+      "step": 839
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990508594955654e-05,
+      "loss": 0.8682,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999043545540541e-05,
+      "loss": 0.7299,
+      "step": 841
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999036203526856e-05,
+      "loss": 0.8588,
+      "step": 842
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990288334547166e-05,
+      "loss": 0.8756,
+      "step": 843
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990214353243295e-05,
+      "loss": 0.8119,
+      "step": 844
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999014009135903e-05,
+      "loss": 0.7622,
+      "step": 845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990065548896447e-05,
+      "loss": 0.8017,
+      "step": 846
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998999072585765e-05,
+      "loss": 0.8518,
+      "step": 847
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998991562224474e-05,
+      "loss": 0.8019,
+      "step": 848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989840238059815e-05,
+      "loss": 0.916,
+      "step": 849
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989764573305005e-05,
+      "loss": 0.7344,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998968862798243e-05,
+      "loss": 0.8219,
+      "step": 851
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989612402094223e-05,
+      "loss": 0.79,
+      "step": 852
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989535895642523e-05,
+      "loss": 0.8635,
+      "step": 853
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989459108629482e-05,
+      "loss": 0.8424,
+      "step": 854
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998938204105726e-05,
+      "loss": 0.86,
+      "step": 855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989304692928007e-05,
+      "loss": 0.8924,
+      "step": 856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998922706424391e-05,
+      "loss": 0.753,
+      "step": 857
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989149155007147e-05,
+      "loss": 0.816,
+      "step": 858
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99890709652199e-05,
+      "loss": 0.8083,
+      "step": 859
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998899249488437e-05,
+      "loss": 0.8158,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998891374400276e-05,
+      "loss": 0.8517,
+      "step": 861
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998883471257728e-05,
+      "loss": 0.8033,
+      "step": 862
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988755400610153e-05,
+      "loss": 0.7971,
+      "step": 863
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99886758081036e-05,
+      "loss": 0.8494,
+      "step": 864
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988595935059864e-05,
+      "loss": 0.8667,
+      "step": 865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998851578148118e-05,
+      "loss": 0.8649,
+      "step": 866
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998843534736981e-05,
+      "loss": 0.8703,
+      "step": 867
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988354632728005e-05,
+      "loss": 0.8569,
+      "step": 868
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988273637558033e-05,
+      "loss": 0.813,
+      "step": 869
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998819236186217e-05,
+      "loss": 0.8135,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99881108056427e-05,
+      "loss": 0.8963,
+      "step": 871
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988028968901912e-05,
+      "loss": 0.8601,
+      "step": 872
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987946851642102e-05,
+      "loss": 0.8522,
+      "step": 873
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998786445386558e-05,
+      "loss": 0.8083,
+      "step": 874
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998778177557466e-05,
+      "loss": 0.9171,
+      "step": 875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998769881677166e-05,
+      "loss": 0.8643,
+      "step": 876
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987615577458917e-05,
+      "loss": 0.8986,
+      "step": 877
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998753205763876e-05,
+      "loss": 0.8707,
+      "step": 878
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987448257313545e-05,
+      "loss": 0.9067,
+      "step": 879
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987364176485616e-05,
+      "loss": 0.8263,
+      "step": 880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987279815157342e-05,
+      "loss": 0.8328,
+      "step": 881
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987195173331087e-05,
+      "loss": 0.8088,
+      "step": 882
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998711025100923e-05,
+      "loss": 0.8342,
+      "step": 883
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987025048194154e-05,
+      "loss": 0.8329,
+      "step": 884
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986939564888258e-05,
+      "loss": 0.7975,
+      "step": 885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986853801093937e-05,
+      "loss": 0.8179,
+      "step": 886
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986767756813604e-05,
+      "loss": 0.8343,
+      "step": 887
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998668143204967e-05,
+      "loss": 0.7994,
+      "step": 888
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986594826804563e-05,
+      "loss": 0.8555,
+      "step": 889
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986507941080715e-05,
+      "loss": 0.8737,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986420774880567e-05,
+      "loss": 0.8845,
+      "step": 891
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986333328206567e-05,
+      "loss": 0.7921,
+      "step": 892
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998624560106117e-05,
+      "loss": 0.8701,
+      "step": 893
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986157593446842e-05,
+      "loss": 0.8287,
+      "step": 894
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998606930536605e-05,
+      "loss": 0.8813,
+      "step": 895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998598073682128e-05,
+      "loss": 0.8058,
+      "step": 896
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985891887815015e-05,
+      "loss": 0.8766,
+      "step": 897
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998580275834975e-05,
+      "loss": 0.8055,
+      "step": 898
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985713348427992e-05,
+      "loss": 0.8494,
+      "step": 899
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985623658052246e-05,
+      "loss": 0.9196,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985533687225037e-05,
+      "loss": 0.7694,
+      "step": 901
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998544343594889e-05,
+      "loss": 0.8721,
+      "step": 902
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985352904226342e-05,
+      "loss": 0.8023,
+      "step": 903
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998526209205993e-05,
+      "loss": 0.8645,
+      "step": 904
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985170999452207e-05,
+      "loss": 0.8366,
+      "step": 905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985079626405735e-05,
+      "loss": 0.7803,
+      "step": 906
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984987972923077e-05,
+      "loss": 0.8274,
+      "step": 907
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984896039006804e-05,
+      "loss": 0.8651,
+      "step": 908
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984803824659504e-05,
+      "loss": 0.8506,
+      "step": 909
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984711329883763e-05,
+      "loss": 0.8255,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984618554682183e-05,
+      "loss": 0.8837,
+      "step": 911
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984525499057365e-05,
+      "loss": 0.8501,
+      "step": 912
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984432163011923e-05,
+      "loss": 0.8211,
+      "step": 913
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984338546548482e-05,
+      "loss": 0.8135,
+      "step": 914
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984244649669667e-05,
+      "loss": 0.8302,
+      "step": 915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984150472378115e-05,
+      "loss": 0.7404,
+      "step": 916
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998405601467648e-05,
+      "loss": 0.8491,
+      "step": 917
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99839612765674e-05,
+      "loss": 0.8136,
+      "step": 918
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983866258053548e-05,
+      "loss": 0.8749,
+      "step": 919
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998377095913759e-05,
+      "loss": 0.8157,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99836753798222e-05,
+      "loss": 0.8531,
+      "step": 921
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998357952011006e-05,
+      "loss": 0.785,
+      "step": 922
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998348338000387e-05,
+      "loss": 0.8299,
+      "step": 923
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998338695950632e-05,
+      "loss": 0.9236,
+      "step": 924
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983290258620126e-05,
+      "loss": 0.8463,
+      "step": 925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983193277348002e-05,
+      "loss": 0.8405,
+      "step": 926
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983096015692672e-05,
+      "loss": 0.8263,
+      "step": 927
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998299847365687e-05,
+      "loss": 0.8281,
+      "step": 928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998290065124333e-05,
+      "loss": 0.8376,
+      "step": 929
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99828025484548e-05,
+      "loss": 0.8286,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998270416529404e-05,
+      "loss": 0.8398,
+      "step": 931
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982605501763806e-05,
+      "loss": 0.2531,
+      "step": 932
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998250655786688e-05,
+      "loss": 0.8586,
+      "step": 933
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998240733360603e-05,
+      "loss": 0.7075,
+      "step": 934
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998230782898405e-05,
+      "loss": 0.8257,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998220804400373e-05,
+      "loss": 0.8636,
+      "step": 936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982107978667873e-05,
+      "loss": 0.8898,
+      "step": 937
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998200763297929e-05,
+      "loss": 0.8487,
+      "step": 938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981907006940803e-05,
+      "loss": 0.7224,
+      "step": 939
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981806100555234e-05,
+      "loss": 0.8271,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981704913825415e-05,
+      "loss": 0.8572,
+      "step": 941
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981603446754192e-05,
+      "loss": 0.7836,
+      "step": 942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981501699344416e-05,
+      "loss": 0.9127,
+      "step": 943
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998139967159894e-05,
+      "loss": 0.8981,
+      "step": 944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998129736352063e-05,
+      "loss": 0.8159,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981194775112367e-05,
+      "loss": 0.925,
+      "step": 946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998109190637702e-05,
+      "loss": 0.8732,
+      "step": 947
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998098875731749e-05,
+      "loss": 0.858,
+      "step": 948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998088532793667e-05,
+      "loss": 0.7817,
+      "step": 949
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980781618237457e-05,
+      "loss": 0.7961,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980677628222772e-05,
+      "loss": 0.7517,
+      "step": 951
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980573357895533e-05,
+      "loss": 0.8424,
+      "step": 952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998046880725867e-05,
+      "loss": 0.8077,
+      "step": 953
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998036397631512e-05,
+      "loss": 0.8639,
+      "step": 954
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980258865067825e-05,
+      "loss": 0.8605,
+      "step": 955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980153473519736e-05,
+      "loss": 0.8194,
+      "step": 956
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980047801673818e-05,
+      "loss": 0.8383,
+      "step": 957
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979941849533037e-05,
+      "loss": 0.8328,
+      "step": 958
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979835617100362e-05,
+      "loss": 0.8521,
+      "step": 959
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979729104378783e-05,
+      "loss": 0.8167,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979622311371295e-05,
+      "loss": 0.8485,
+      "step": 961
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979515238080892e-05,
+      "loss": 0.7528,
+      "step": 962
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979407884510577e-05,
+      "loss": 0.786,
+      "step": 963
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979300250663375e-05,
+      "loss": 0.8027,
+      "step": 964
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979192336542302e-05,
+      "loss": 0.8843,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997908414215039e-05,
+      "loss": 0.8241,
+      "step": 966
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997897566749068e-05,
+      "loss": 0.3004,
+      "step": 967
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997886691256622e-05,
+      "loss": 0.7951,
+      "step": 968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978757877380056e-05,
+      "loss": 0.8119,
+      "step": 969
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997864856193526e-05,
+      "loss": 0.8259,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99785389662349e-05,
+      "loss": 0.8836,
+      "step": 971
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978429090282048e-05,
+      "loss": 0.7876,
+      "step": 972
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978318934079793e-05,
+      "loss": 0.8623,
+      "step": 973
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978208497631233e-05,
+      "loss": 0.8871,
+      "step": 974
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978097780939464e-05,
+      "loss": 0.8115,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99779867840076e-05,
+      "loss": 0.9442,
+      "step": 976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977875506838758e-05,
+      "loss": 0.8154,
+      "step": 977
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997776394943606e-05,
+      "loss": 0.2246,
+      "step": 978
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997765211180264e-05,
+      "loss": 0.8227,
+      "step": 979
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997753999394164e-05,
+      "loss": 0.7653,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997742759585621e-05,
+      "loss": 0.8652,
+      "step": 981
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977314917549505e-05,
+      "loss": 0.7995,
+      "step": 982
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997720195902469e-05,
+      "loss": 0.8167,
+      "step": 983
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997708872028494e-05,
+      "loss": 0.8831,
+      "step": 984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997697520133343e-05,
+      "loss": 0.828,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976861402173353e-05,
+      "loss": 0.8328,
+      "step": 986
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976747322807902e-05,
+      "loss": 0.9244,
+      "step": 987
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976632963240283e-05,
+      "loss": 0.8284,
+      "step": 988
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976518323473707e-05,
+      "loss": 0.8994,
+      "step": 989
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976403403511394e-05,
+      "loss": 0.9126,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976288203356572e-05,
+      "loss": 0.9218,
+      "step": 991
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997617272301248e-05,
+      "loss": 0.8043,
+      "step": 992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976056962482352e-05,
+      "loss": 0.8766,
+      "step": 993
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975940921769445e-05,
+      "loss": 0.7861,
+      "step": 994
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997582460087702e-05,
+      "loss": 0.8783,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975707999808338e-05,
+      "loss": 0.8606,
+      "step": 996
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997559111856668e-05,
+      "loss": 0.8268,
+      "step": 997
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975473957155326e-05,
+      "loss": 0.7837,
+      "step": 998
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975356515577566e-05,
+      "loss": 0.7588,
+      "step": 999
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99752387938367e-05,
+      "loss": 0.802,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975120791936028e-05,
+      "loss": 0.8945,
+      "step": 1001
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975002509878875e-05,
+      "loss": 0.8113,
+      "step": 1002
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974883947668557e-05,
+      "loss": 0.7661,
+      "step": 1003
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974765105308403e-05,
+      "loss": 0.9005,
+      "step": 1004
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997464598280175e-05,
+      "loss": 0.7825,
+      "step": 1005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997452658015195e-05,
+      "loss": 0.7748,
+      "step": 1006
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997440689736235e-05,
+      "loss": 0.8508,
+      "step": 1007
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974286934436313e-05,
+      "loss": 0.7629,
+      "step": 1008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997416669137721e-05,
+      "loss": 0.8407,
+      "step": 1009
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974046168188415e-05,
+      "loss": 0.7706,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973925364873317e-05,
+      "loss": 0.8358,
+      "step": 1011
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973804281435302e-05,
+      "loss": 0.8407,
+      "step": 1012
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997368291787778e-05,
+      "loss": 0.8304,
+      "step": 1013
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973561274204153e-05,
+      "loss": 0.8376,
+      "step": 1014
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997343935041784e-05,
+      "loss": 0.8165,
+      "step": 1015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973317146522264e-05,
+      "loss": 0.924,
+      "step": 1016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973194662520857e-05,
+      "loss": 0.8114,
+      "step": 1017
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973071898417064e-05,
+      "loss": 0.7606,
+      "step": 1018
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997294885421432e-05,
+      "loss": 0.8154,
+      "step": 1019
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99728255299161e-05,
+      "loss": 0.8762,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997270192552585e-05,
+      "loss": 0.838,
+      "step": 1021
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997257804104705e-05,
+      "loss": 0.7687,
+      "step": 1022
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972453876483175e-05,
+      "loss": 0.8647,
+      "step": 1023
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997232943183772e-05,
+      "loss": 0.9102,
+      "step": 1024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972204707114173e-05,
+      "loss": 0.8924,
+      "step": 1025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972079702316037e-05,
+      "loss": 0.8166,
+      "step": 1026
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971954417446827e-05,
+      "loss": 0.8336,
+      "step": 1027
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997182885251006e-05,
+      "loss": 0.7914,
+      "step": 1028
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997170300750926e-05,
+      "loss": 0.7858,
+      "step": 1029
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997157688244797e-05,
+      "loss": 0.8159,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997145047732972e-05,
+      "loss": 0.8455,
+      "step": 1031
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997132379215807e-05,
+      "loss": 0.923,
+      "step": 1032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997119682693657e-05,
+      "loss": 0.826,
+      "step": 1033
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971069581668795e-05,
+      "loss": 0.7783,
+      "step": 1034
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997094205635831e-05,
+      "loss": 0.7652,
+      "step": 1035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99708142510087e-05,
+      "loss": 0.8736,
+      "step": 1036
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970686165623554e-05,
+      "loss": 0.7927,
+      "step": 1037
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970557800206473e-05,
+      "loss": 0.7823,
+      "step": 1038
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970429154761058e-05,
+      "loss": 0.7914,
+      "step": 1039
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970300229290922e-05,
+      "loss": 0.7971,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970171023799693e-05,
+      "loss": 0.8208,
+      "step": 1041
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997004153829099e-05,
+      "loss": 0.7916,
+      "step": 1042
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969911772768453e-05,
+      "loss": 0.86,
+      "step": 1043
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996978172723573e-05,
+      "loss": 0.8699,
+      "step": 1044
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969651401696467e-05,
+      "loss": 0.829,
+      "step": 1045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969520796154328e-05,
+      "loss": 0.7791,
+      "step": 1046
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969389910612987e-05,
+      "loss": 0.8065,
+      "step": 1047
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996925874507611e-05,
+      "loss": 0.8127,
+      "step": 1048
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969127299547387e-05,
+      "loss": 0.8096,
+      "step": 1049
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968995574030506e-05,
+      "loss": 0.834,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968863568529168e-05,
+      "loss": 0.8336,
+      "step": 1051
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968731283047083e-05,
+      "loss": 0.8264,
+      "step": 1052
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996859871758796e-05,
+      "loss": 0.8265,
+      "step": 1053
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968465872155532e-05,
+      "loss": 0.8783,
+      "step": 1054
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996833274675352e-05,
+      "loss": 0.8282,
+      "step": 1055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996819934138567e-05,
+      "loss": 0.873,
+      "step": 1056
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968065656055722e-05,
+      "loss": 0.8792,
+      "step": 1057
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967931690767438e-05,
+      "loss": 0.8848,
+      "step": 1058
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967797445524576e-05,
+      "loss": 0.7923,
+      "step": 1059
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967662920330907e-05,
+      "loss": 0.7871,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996752811519021e-05,
+      "loss": 0.8035,
+      "step": 1061
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967393030106274e-05,
+      "loss": 0.752,
+      "step": 1062
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996725766508288e-05,
+      "loss": 0.7808,
+      "step": 1063
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967122020123847e-05,
+      "loss": 0.7595,
+      "step": 1064
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966986095232977e-05,
+      "loss": 0.7776,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966849890414086e-05,
+      "loss": 0.8679,
+      "step": 1066
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966713405671003e-05,
+      "loss": 0.8192,
+      "step": 1067
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996657664100756e-05,
+      "loss": 0.8075,
+      "step": 1068
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966439596427593e-05,
+      "loss": 0.8069,
+      "step": 1069
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966302271934955e-05,
+      "loss": 0.8086,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966164667533506e-05,
+      "loss": 0.7864,
+      "step": 1071
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966026783227107e-05,
+      "loss": 0.886,
+      "step": 1072
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965888619019632e-05,
+      "loss": 0.8517,
+      "step": 1073
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996575017491496e-05,
+      "loss": 0.8606,
+      "step": 1074
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996561145091698e-05,
+      "loss": 0.8347,
+      "step": 1075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996547244702959e-05,
+      "loss": 0.8275,
+      "step": 1076
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965333163256687e-05,
+      "loss": 0.7914,
+      "step": 1077
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965193599602193e-05,
+      "loss": 0.7835,
+      "step": 1078
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965053756070023e-05,
+      "loss": 0.8719,
+      "step": 1079
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964913632664102e-05,
+      "loss": 0.7817,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996477322938837e-05,
+      "loss": 0.8039,
+      "step": 1081
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964632546246766e-05,
+      "loss": 0.7797,
+      "step": 1082
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964491583243246e-05,
+      "loss": 0.8137,
+      "step": 1083
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964350340381763e-05,
+      "loss": 0.7571,
+      "step": 1084
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964208817666287e-05,
+      "loss": 0.8479,
+      "step": 1085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964067015100796e-05,
+      "loss": 0.8087,
+      "step": 1086
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963924932689265e-05,
+      "loss": 0.7764,
+      "step": 1087
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996378257043569e-05,
+      "loss": 0.8103,
+      "step": 1088
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963639928344075e-05,
+      "loss": 0.8158,
+      "step": 1089
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996349700641841e-05,
+      "loss": 0.8828,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963353804662727e-05,
+      "loss": 0.8462,
+      "step": 1091
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963210323081032e-05,
+      "loss": 0.8218,
+      "step": 1092
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963066561677365e-05,
+      "loss": 0.8665,
+      "step": 1093
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996292252045576e-05,
+      "loss": 0.8134,
+      "step": 1094
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962778199420265e-05,
+      "loss": 0.835,
+      "step": 1095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962633598574932e-05,
+      "loss": 0.823,
+      "step": 1096
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996248871792382e-05,
+      "loss": 0.8004,
+      "step": 1097
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962343557471002e-05,
+      "loss": 0.8896,
+      "step": 1098
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996219811722055e-05,
+      "loss": 0.8633,
+      "step": 1099
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962052397176553e-05,
+      "loss": 0.8763,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99619063973431e-05,
+      "loss": 0.7992,
+      "step": 1101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961760117724295e-05,
+      "loss": 0.8744,
+      "step": 1102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961613558324245e-05,
+      "loss": 0.7845,
+      "step": 1103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961466719147068e-05,
+      "loss": 0.842,
+      "step": 1104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961319600196884e-05,
+      "loss": 0.826,
+      "step": 1105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961172201477826e-05,
+      "loss": 0.8121,
+      "step": 1106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961024522994035e-05,
+      "loss": 0.8343,
+      "step": 1107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996087656474966e-05,
+      "loss": 0.8119,
+      "step": 1108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960728326748855e-05,
+      "loss": 0.8232,
+      "step": 1109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960579808995783e-05,
+      "loss": 0.7753,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960431011494618e-05,
+      "loss": 0.8075,
+      "step": 1111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960281934249532e-05,
+      "loss": 0.7433,
+      "step": 1112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996013257726472e-05,
+      "loss": 0.7747,
+      "step": 1113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959982940544368e-05,
+      "loss": 0.8694,
+      "step": 1114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995983302409269e-05,
+      "loss": 0.7231,
+      "step": 1115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959682827913888e-05,
+      "loss": 0.8273,
+      "step": 1116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995953235201218e-05,
+      "loss": 0.882,
+      "step": 1117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959381596391797e-05,
+      "loss": 0.7889,
+      "step": 1118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995923056105697e-05,
+      "loss": 0.7981,
+      "step": 1119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959079246011947e-05,
+      "loss": 0.7121,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958927651260967e-05,
+      "loss": 0.9014,
+      "step": 1121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958775776808296e-05,
+      "loss": 0.8767,
+      "step": 1122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958623622658194e-05,
+      "loss": 0.6978,
+      "step": 1123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995847118881494e-05,
+      "loss": 0.8634,
+      "step": 1124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995831847528281e-05,
+      "loss": 0.8588,
+      "step": 1125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958165482066094e-05,
+      "loss": 0.8604,
+      "step": 1126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958012209169095e-05,
+      "loss": 0.8024,
+      "step": 1127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957858656596108e-05,
+      "loss": 0.8056,
+      "step": 1128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957704824351453e-05,
+      "loss": 0.8012,
+      "step": 1129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957550712439446e-05,
+      "loss": 0.8391,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995739632086442e-05,
+      "loss": 0.8859,
+      "step": 1131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957241649630704e-05,
+      "loss": 0.8617,
+      "step": 1132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957086698742653e-05,
+      "loss": 0.9247,
+      "step": 1133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956931468204607e-05,
+      "loss": 0.8333,
+      "step": 1134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995677595802093e-05,
+      "loss": 0.7579,
+      "step": 1135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956620168195996e-05,
+      "loss": 0.8783,
+      "step": 1136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995646409873417e-05,
+      "loss": 0.8642,
+      "step": 1137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956307749639845e-05,
+      "loss": 0.7625,
+      "step": 1138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956151120917405e-05,
+      "loss": 0.89,
+      "step": 1139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955994212571253e-05,
+      "loss": 0.7658,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995583702460579e-05,
+      "loss": 0.8332,
+      "step": 1141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995567955702544e-05,
+      "loss": 0.8386,
+      "step": 1142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995552180983462e-05,
+      "loss": 0.7537,
+      "step": 1143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995536378303776e-05,
+      "loss": 0.8537,
+      "step": 1144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99552054766393e-05,
+      "loss": 0.7776,
+      "step": 1145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955046890643682e-05,
+      "loss": 0.8297,
+      "step": 1146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954888025055368e-05,
+      "loss": 0.886,
+      "step": 1147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954728879878812e-05,
+      "loss": 0.8184,
+      "step": 1148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995456945511849e-05,
+      "loss": 0.2798,
+      "step": 1149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954409750778872e-05,
+      "loss": 0.6885,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995424976686445e-05,
+      "loss": 0.7808,
+      "step": 1151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954089503379716e-05,
+      "loss": 0.7921,
+      "step": 1152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953928960329173e-05,
+      "loss": 0.7904,
+      "step": 1153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953768137717323e-05,
+      "loss": 0.8199,
+      "step": 1154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953607035548686e-05,
+      "loss": 0.8501,
+      "step": 1155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995344565382779e-05,
+      "loss": 0.7504,
+      "step": 1156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995328399255916e-05,
+      "loss": 0.775,
+      "step": 1157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953122051747347e-05,
+      "loss": 0.8155,
+      "step": 1158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952959831396893e-05,
+      "loss": 0.8517,
+      "step": 1159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952797331512353e-05,
+      "loss": 0.8285,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995263455209829e-05,
+      "loss": 0.8481,
+      "step": 1161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952471493159282e-05,
+      "loss": 0.8663,
+      "step": 1162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952308154699903e-05,
+      "loss": 0.8618,
+      "step": 1163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952144536724743e-05,
+      "loss": 0.8158,
+      "step": 1164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951980639238398e-05,
+      "loss": 0.822,
+      "step": 1165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951816462245468e-05,
+      "loss": 0.7799,
+      "step": 1166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951652005750567e-05,
+      "loss": 0.8218,
+      "step": 1167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995148726975831e-05,
+      "loss": 0.7668,
+      "step": 1168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951322254273328e-05,
+      "loss": 0.8112,
+      "step": 1169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951156959300252e-05,
+      "loss": 0.8386,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995099138484373e-05,
+      "loss": 0.8264,
+      "step": 1171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950825530908404e-05,
+      "loss": 0.8527,
+      "step": 1172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950659397498938e-05,
+      "loss": 0.7855,
+      "step": 1173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950492984619995e-05,
+      "loss": 0.8715,
+      "step": 1174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950326292276252e-05,
+      "loss": 0.8397,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995015932047239e-05,
+      "loss": 0.8765,
+      "step": 1176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949992069213097e-05,
+      "loss": 0.8771,
+      "step": 1177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994982453850307e-05,
+      "loss": 0.8238,
+      "step": 1178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949656728347012e-05,
+      "loss": 0.8084,
+      "step": 1179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949488638749643e-05,
+      "loss": 0.7839,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949320269715677e-05,
+      "loss": 0.7874,
+      "step": 1181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994915162124985e-05,
+      "loss": 0.8168,
+      "step": 1182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994898269335689e-05,
+      "loss": 0.7909,
+      "step": 1183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948813486041546e-05,
+      "loss": 0.7678,
+      "step": 1184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994864399930857e-05,
+      "loss": 0.8511,
+      "step": 1185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948474233162725e-05,
+      "loss": 0.7974,
+      "step": 1186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948304187608773e-05,
+      "loss": 0.8535,
+      "step": 1187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994813386265149e-05,
+      "loss": 0.6899,
+      "step": 1188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947963258295665e-05,
+      "loss": 0.8536,
+      "step": 1189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947792374546088e-05,
+      "loss": 0.7855,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994762121140756e-05,
+      "loss": 0.7372,
+      "step": 1191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994744976888488e-05,
+      "loss": 0.7501,
+      "step": 1192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994727804698287e-05,
+      "loss": 0.8279,
+      "step": 1193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947106045706356e-05,
+      "loss": 0.7991,
+      "step": 1194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946933765060155e-05,
+      "loss": 0.8149,
+      "step": 1195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994676120504912e-05,
+      "loss": 0.7233,
+      "step": 1196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946588365678093e-05,
+      "loss": 0.2971,
+      "step": 1197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946415246951928e-05,
+      "loss": 0.849,
+      "step": 1198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946241848875486e-05,
+      "loss": 0.8884,
+      "step": 1199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946068171453635e-05,
+      "loss": 0.8514,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994589421469126e-05,
+      "loss": 0.8417,
+      "step": 1201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945719978593237e-05,
+      "loss": 0.8034,
+      "step": 1202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945545463164467e-05,
+      "loss": 0.8041,
+      "step": 1203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994537066840985e-05,
+      "loss": 0.8114,
+      "step": 1204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945195594334294e-05,
+      "loss": 0.8299,
+      "step": 1205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945020240942717e-05,
+      "loss": 0.8764,
+      "step": 1206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994484460824004e-05,
+      "loss": 0.7384,
+      "step": 1207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99446686962312e-05,
+      "loss": 0.7954,
+      "step": 1208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944492504921137e-05,
+      "loss": 0.7839,
+      "step": 1209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944316034314797e-05,
+      "loss": 0.8275,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994413928441714e-05,
+      "loss": 0.7804,
+      "step": 1211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994396225523313e-05,
+      "loss": 0.7805,
+      "step": 1212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943784946767736e-05,
+      "loss": 0.8447,
+      "step": 1213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943607359025935e-05,
+      "loss": 0.902,
+      "step": 1214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943429492012723e-05,
+      "loss": 0.8004,
+      "step": 1215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994325134573309e-05,
+      "loss": 0.807,
+      "step": 1216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943072920192042e-05,
+      "loss": 0.8248,
+      "step": 1217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942894215394585e-05,
+      "loss": 0.7809,
+      "step": 1218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942715231345744e-05,
+      "loss": 0.7946,
+      "step": 1219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942535968050543e-05,
+      "loss": 0.8329,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942356425514018e-05,
+      "loss": 0.7983,
+      "step": 1221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942176603741207e-05,
+      "loss": 0.8524,
+      "step": 1222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994199650273717e-05,
+      "loss": 0.813,
+      "step": 1223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941816122506958e-05,
+      "loss": 0.7376,
+      "step": 1224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941635463055634e-05,
+      "loss": 0.7662,
+      "step": 1225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994145452438828e-05,
+      "loss": 0.8693,
+      "step": 1226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941273306509973e-05,
+      "loss": 0.8034,
+      "step": 1227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941091809425806e-05,
+      "loss": 0.7984,
+      "step": 1228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994091003314087e-05,
+      "loss": 0.78,
+      "step": 1229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940727977660275e-05,
+      "loss": 0.6966,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940545642989137e-05,
+      "loss": 0.7982,
+      "step": 1231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940363029132572e-05,
+      "loss": 0.9012,
+      "step": 1232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994018013609571e-05,
+      "loss": 0.7674,
+      "step": 1233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993999696388369e-05,
+      "loss": 0.8115,
+      "step": 1234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939813512501653e-05,
+      "loss": 0.7992,
+      "step": 1235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939629781954753e-05,
+      "loss": 0.8007,
+      "step": 1236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939445772248148e-05,
+      "loss": 0.7659,
+      "step": 1237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993926148338701e-05,
+      "loss": 0.8208,
+      "step": 1238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993907691537651e-05,
+      "loss": 0.8396,
+      "step": 1239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938892068221837e-05,
+      "loss": 0.8148,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993870694192818e-05,
+      "loss": 0.8139,
+      "step": 1241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938521536500738e-05,
+      "loss": 0.8378,
+      "step": 1242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993833585194472e-05,
+      "loss": 0.7882,
+      "step": 1243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938149888265338e-05,
+      "loss": 0.7747,
+      "step": 1244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937963645467813e-05,
+      "loss": 0.7705,
+      "step": 1245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993777712355739e-05,
+      "loss": 0.8324,
+      "step": 1246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937590322539284e-05,
+      "loss": 0.8154,
+      "step": 1247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993740324241876e-05,
+      "loss": 0.8515,
+      "step": 1248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937215883201068e-05,
+      "loss": 0.7655,
+      "step": 1249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937028244891468e-05,
+      "loss": 0.7619,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936840327495228e-05,
+      "loss": 0.8831,
+      "step": 1251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993665213101763e-05,
+      "loss": 0.9129,
+      "step": 1252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993646365546396e-05,
+      "loss": 0.7672,
+      "step": 1253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936274900839505e-05,
+      "loss": 0.8417,
+      "step": 1254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936085867149573e-05,
+      "loss": 0.8699,
+      "step": 1255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9935896554399472e-05,
+      "loss": 0.8211,
+      "step": 1256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935706962594516e-05,
+      "loss": 0.7857,
+      "step": 1257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993551709174003e-05,
+      "loss": 0.8357,
+      "step": 1258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935326941841348e-05,
+      "loss": 0.7658,
+      "step": 1259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935136512903813e-05,
+      "loss": 0.8058,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993494580493277e-05,
+      "loss": 0.8344,
+      "step": 1261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934754817933576e-05,
+      "loss": 0.7173,
+      "step": 1262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934563551911594e-05,
+      "loss": 0.7914,
+      "step": 1263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934372006872195e-05,
+      "loss": 0.8904,
+      "step": 1264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934180182820767e-05,
+      "loss": 0.7415,
+      "step": 1265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933988079762683e-05,
+      "loss": 0.7782,
+      "step": 1266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993379569770335e-05,
+      "loss": 0.8102,
+      "step": 1267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933603036648164e-05,
+      "loss": 0.8116,
+      "step": 1268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933410096602544e-05,
+      "loss": 0.8895,
+      "step": 1269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933216877571905e-05,
+      "loss": 0.7677,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933023379561667e-05,
+      "loss": 0.7814,
+      "step": 1271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932829602577273e-05,
+      "loss": 0.8029,
+      "step": 1272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932635546624163e-05,
+      "loss": 0.8084,
+      "step": 1273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932441211707785e-05,
+      "loss": 0.7962,
+      "step": 1274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99322465978336e-05,
+      "loss": 0.8745,
+      "step": 1275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993205170500707e-05,
+      "loss": 0.761,
+      "step": 1276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931856533233676e-05,
+      "loss": 0.785,
+      "step": 1277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931661082518893e-05,
+      "loss": 0.7203,
+      "step": 1278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993146535286821e-05,
+      "loss": 0.741,
+      "step": 1279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993126934428713e-05,
+      "loss": 0.7636,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993107305678115e-05,
+      "loss": 0.8719,
+      "step": 1281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993087649035579e-05,
+      "loss": 0.7526,
+      "step": 1282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930679645016563e-05,
+      "loss": 0.8163,
+      "step": 1283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930482520769008e-05,
+      "loss": 0.6999,
+      "step": 1284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930285117618653e-05,
+      "loss": 0.8066,
+      "step": 1285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930087435571045e-05,
+      "loss": 0.7613,
+      "step": 1286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929889474631738e-05,
+      "loss": 0.795,
+      "step": 1287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929691234806287e-05,
+      "loss": 0.8312,
+      "step": 1288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929492716100262e-05,
+      "loss": 0.7852,
+      "step": 1289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992929391851924e-05,
+      "loss": 0.7692,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99290948420688e-05,
+      "loss": 0.8196,
+      "step": 1291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928895486754542e-05,
+      "loss": 0.8239,
+      "step": 1292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928695852582054e-05,
+      "loss": 0.8654,
+      "step": 1293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928495939556952e-05,
+      "loss": 0.8302,
+      "step": 1294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928295747684846e-05,
+      "loss": 0.887,
+      "step": 1295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928095276971357e-05,
+      "loss": 0.8167,
+      "step": 1296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927894527422117e-05,
+      "loss": 0.7507,
+      "step": 1297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992769349904277e-05,
+      "loss": 0.8216,
+      "step": 1298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992749219183895e-05,
+      "loss": 0.742,
+      "step": 1299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992729060581632e-05,
+      "loss": 0.8459,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992708874098054e-05,
+      "loss": 0.7505,
+      "step": 1301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992688659733728e-05,
+      "loss": 0.8539,
+      "step": 1302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926684174892215e-05,
+      "loss": 0.8168,
+      "step": 1303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992648147365103e-05,
+      "loss": 0.8298,
+      "step": 1304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992627849361942e-05,
+      "loss": 0.8331,
+      "step": 1305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926075234803087e-05,
+      "loss": 0.8203,
+      "step": 1306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925871697207733e-05,
+      "loss": 0.851,
+      "step": 1307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992566788083908e-05,
+      "loss": 0.7525,
+      "step": 1308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925463785702856e-05,
+      "loss": 0.7563,
+      "step": 1309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992525941180478e-05,
+      "loss": 0.7588,
+      "step": 1310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925054759150607e-05,
+      "loss": 0.7356,
+      "step": 1311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924849827746075e-05,
+      "loss": 0.7916,
+      "step": 1312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924644617596946e-05,
+      "loss": 0.8267,
+      "step": 1313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924439128708976e-05,
+      "loss": 0.7975,
+      "step": 1314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924233361087945e-05,
+      "loss": 0.8605,
+      "step": 1315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924027314739623e-05,
+      "loss": 0.813,
+      "step": 1316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99238209896698e-05,
+      "loss": 0.8268,
+      "step": 1317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992361438588428e-05,
+      "loss": 0.7724,
+      "step": 1318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992340750338885e-05,
+      "loss": 0.8323,
+      "step": 1319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923200342189333e-05,
+      "loss": 0.7765,
+      "step": 1320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992299290229154e-05,
+      "loss": 0.7906,
+      "step": 1321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99227851837013e-05,
+      "loss": 0.7919,
+      "step": 1322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992257718642445e-05,
+      "loss": 0.8054,
+      "step": 1323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992236891046682e-05,
+      "loss": 0.8284,
+      "step": 1324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922160355834273e-05,
+      "loss": 0.7875,
+      "step": 1325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992195152253266e-05,
+      "loss": 0.8057,
+      "step": 1326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921742410567848e-05,
+      "loss": 0.722,
+      "step": 1327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921533019945707e-05,
+      "loss": 0.816,
+      "step": 1328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921323350672123e-05,
+      "loss": 0.6976,
+      "step": 1329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992111340275298e-05,
+      "loss": 0.8458,
+      "step": 1330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920903176194172e-05,
+      "loss": 0.8836,
+      "step": 1331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920692671001614e-05,
+      "loss": 0.8932,
+      "step": 1332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992048188718121e-05,
+      "loss": 0.7727,
+      "step": 1333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920270824738882e-05,
+      "loss": 0.8564,
+      "step": 1334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920059483680553e-05,
+      "loss": 0.7711,
+      "step": 1335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919847864012167e-05,
+      "loss": 0.7812,
+      "step": 1336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919635965739662e-05,
+      "loss": 0.8458,
+      "step": 1337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991942378886899e-05,
+      "loss": 0.8488,
+      "step": 1338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919211333406114e-05,
+      "loss": 0.8531,
+      "step": 1339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918998599356998e-05,
+      "loss": 0.826,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918785586727612e-05,
+      "loss": 0.7408,
+      "step": 1341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918572295523947e-05,
+      "loss": 0.7977,
+      "step": 1342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991835872575199e-05,
+      "loss": 0.8427,
+      "step": 1343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918144877417735e-05,
+      "loss": 0.8754,
+      "step": 1344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917930750527194e-05,
+      "loss": 0.795,
+      "step": 1345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917716345086377e-05,
+      "loss": 0.8513,
+      "step": 1346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917501661101307e-05,
+      "loss": 0.8074,
+      "step": 1347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917286698578014e-05,
+      "loss": 0.7959,
+      "step": 1348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991707145752254e-05,
+      "loss": 0.8276,
+      "step": 1349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916855937940924e-05,
+      "loss": 0.8673,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916640139839215e-05,
+      "loss": 0.8523,
+      "step": 1351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991642406322348e-05,
+      "loss": 0.7623,
+      "step": 1352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991620770809979e-05,
+      "loss": 0.8895,
+      "step": 1353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915991074474216e-05,
+      "loss": 0.8448,
+      "step": 1354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915774162352847e-05,
+      "loss": 0.8562,
+      "step": 1355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915556971741773e-05,
+      "loss": 0.8406,
+      "step": 1356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991533950264709e-05,
+      "loss": 0.7914,
+      "step": 1357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991512175507491e-05,
+      "loss": 0.2669,
+      "step": 1358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914903729031347e-05,
+      "loss": 0.7304,
+      "step": 1359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914685424522525e-05,
+      "loss": 0.8495,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991446684155458e-05,
+      "loss": 0.8398,
+      "step": 1361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991424798013364e-05,
+      "loss": 0.7826,
+      "step": 1362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991402884026586e-05,
+      "loss": 0.8093,
+      "step": 1363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913809421957395e-05,
+      "loss": 0.8691,
+      "step": 1364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913589725214405e-05,
+      "loss": 0.7726,
+      "step": 1365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991336975004306e-05,
+      "loss": 0.8606,
+      "step": 1366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991314949644954e-05,
+      "loss": 0.7646,
+      "step": 1367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912928964440025e-05,
+      "loss": 0.8406,
+      "step": 1368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912708154020716e-05,
+      "loss": 0.7885,
+      "step": 1369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991248706519781e-05,
+      "loss": 0.8107,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912265697977524e-05,
+      "loss": 0.2668,
+      "step": 1371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912044052366066e-05,
+      "loss": 0.8557,
+      "step": 1372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911822128369668e-05,
+      "loss": 0.8011,
+      "step": 1373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911599925994553e-05,
+      "loss": 0.8877,
+      "step": 1374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911377445246973e-05,
+      "loss": 0.8525,
+      "step": 1375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911154686133168e-05,
+      "loss": 0.803,
+      "step": 1376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99109316486594e-05,
+      "loss": 0.7856,
+      "step": 1377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991070833283193e-05,
+      "loss": 0.7964,
+      "step": 1378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910484738657035e-05,
+      "loss": 0.7835,
+      "step": 1379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910260866140986e-05,
+      "loss": 0.8486,
+      "step": 1380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910036715290077e-05,
+      "loss": 0.7523,
+      "step": 1381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909812286110602e-05,
+      "loss": 0.7571,
+      "step": 1382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909587578608865e-05,
+      "loss": 0.8174,
+      "step": 1383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909362592791176e-05,
+      "loss": 0.8578,
+      "step": 1384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990913732866385e-05,
+      "loss": 0.7922,
+      "step": 1385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990891178623322e-05,
+      "loss": 0.861,
+      "step": 1386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990868596550562e-05,
+      "loss": 0.8415,
+      "step": 1387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908459866487385e-05,
+      "loss": 0.8061,
+      "step": 1388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908233489184875e-05,
+      "loss": 0.8488,
+      "step": 1389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908006833604442e-05,
+      "loss": 0.6816,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907779899752453e-05,
+      "loss": 0.7818,
+      "step": 1391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990755268763528e-05,
+      "loss": 0.8457,
+      "step": 1392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990732519725931e-05,
+      "loss": 0.7976,
+      "step": 1393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907097428630923e-05,
+      "loss": 0.801,
+      "step": 1394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906869381756526e-05,
+      "loss": 0.8132,
+      "step": 1395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906641056642515e-05,
+      "loss": 0.7976,
+      "step": 1396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990641245329531e-05,
+      "loss": 0.8386,
+      "step": 1397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906183571721326e-05,
+      "loss": 0.8804,
+      "step": 1398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905954411926992e-05,
+      "loss": 0.8094,
+      "step": 1399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905724973918748e-05,
+      "loss": 0.7486,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905495257703033e-05,
+      "loss": 0.8274,
+      "step": 1401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905265263286302e-05,
+      "loss": 0.8336,
+      "step": 1402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905034990675013e-05,
+      "loss": 0.7789,
+      "step": 1403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904804439875635e-05,
+      "loss": 0.8635,
+      "step": 1404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990457361089464e-05,
+      "loss": 0.8479,
+      "step": 1405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904342503738513e-05,
+      "loss": 0.8264,
+      "step": 1406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904111118413744e-05,
+      "loss": 0.8221,
+      "step": 1407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903879454926834e-05,
+      "loss": 0.8307,
+      "step": 1408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990364751328429e-05,
+      "loss": 0.791,
+      "step": 1409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903415293492618e-05,
+      "loss": 0.8204,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990318279555835e-05,
+      "loss": 0.7255,
+      "step": 1411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990295001948801e-05,
+      "loss": 0.7878,
+      "step": 1412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902716965288137e-05,
+      "loss": 0.8475,
+      "step": 1413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990248363296528e-05,
+      "loss": 0.8551,
+      "step": 1414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902250022525985e-05,
+      "loss": 0.7836,
+      "step": 1415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990201613397682e-05,
+      "loss": 0.9221,
+      "step": 1416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990178196732435e-05,
+      "loss": 0.8597,
+      "step": 1417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990154752257515e-05,
+      "loss": 0.7818,
+      "step": 1418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990131279973581e-05,
+      "loss": 0.8854,
+      "step": 1419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990107779881292e-05,
+      "loss": 0.7704,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900842519813077e-05,
+      "loss": 0.8605,
+      "step": 1421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900606962742893e-05,
+      "loss": 0.8344,
+      "step": 1422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990037112760898e-05,
+      "loss": 0.7913,
+      "step": 1423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990013501441797e-05,
+      "loss": 0.8219,
+      "step": 1424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989989862317648e-05,
+      "loss": 0.7947,
+      "step": 1425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899661953891167e-05,
+      "loss": 0.8923,
+      "step": 1426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989942500656866e-05,
+      "loss": 0.8276,
+      "step": 1427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899187781215623e-05,
+      "loss": 0.8437,
+      "step": 1428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989895027783872e-05,
+      "loss": 0.8084,
+      "step": 1429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989871249644462e-05,
+      "loss": 0.792,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989847443704e-05,
+      "loss": 0.8677,
+      "step": 1431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898236099631543e-05,
+      "loss": 0.788,
+      "step": 1432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897997484225947e-05,
+      "loss": 0.8152,
+      "step": 1433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897758590829915e-05,
+      "loss": 0.8243,
+      "step": 1434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897519419450156e-05,
+      "loss": 0.712,
+      "step": 1435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897279970093383e-05,
+      "loss": 0.8159,
+      "step": 1436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897040242766323e-05,
+      "loss": 0.7593,
+      "step": 1437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989680023747571e-05,
+      "loss": 0.8175,
+      "step": 1438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896559954228283e-05,
+      "loss": 0.8114,
+      "step": 1439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896319393030796e-05,
+      "loss": 0.773,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896078553889997e-05,
+      "loss": 0.7537,
+      "step": 1441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895837436812653e-05,
+      "loss": 0.7361,
+      "step": 1442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989559604180554e-05,
+      "loss": 0.7714,
+      "step": 1443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895354368875432e-05,
+      "loss": 0.868,
+      "step": 1444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895112418029125e-05,
+      "loss": 0.7642,
+      "step": 1445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98948701892734e-05,
+      "loss": 0.7922,
+      "step": 1446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989462768261507e-05,
+      "loss": 0.7778,
+      "step": 1447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989438489806095e-05,
+      "loss": 0.7496,
+      "step": 1448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894141835617852e-05,
+      "loss": 0.7679,
+      "step": 1449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98938984952926e-05,
+      "loss": 0.8112,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893654877092032e-05,
+      "loss": 0.8208,
+      "step": 1451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893410981022994e-05,
+      "loss": 0.773,
+      "step": 1452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893166807092327e-05,
+      "loss": 0.8077,
+      "step": 1453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98929223553069e-05,
+      "loss": 0.7922,
+      "step": 1454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989267762567357e-05,
+      "loss": 0.8827,
+      "step": 1455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989243261819921e-05,
+      "loss": 0.7533,
+      "step": 1456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892187332890708e-05,
+      "loss": 0.8118,
+      "step": 1457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891941769754947e-05,
+      "loss": 0.7398,
+      "step": 1458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891695928798828e-05,
+      "loss": 0.7977,
+      "step": 1459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891449810029247e-05,
+      "loss": 0.8169,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891203413453128e-05,
+      "loss": 0.7429,
+      "step": 1461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890956739077385e-05,
+      "loss": 0.7881,
+      "step": 1462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989070978690895e-05,
+      "loss": 0.8376,
+      "step": 1463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989046255695475e-05,
+      "loss": 0.809,
+      "step": 1464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890215049221738e-05,
+      "loss": 0.7852,
+      "step": 1465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889967263716858e-05,
+      "loss": 0.8656,
+      "step": 1466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889719200447076e-05,
+      "loss": 0.7706,
+      "step": 1467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889470859419355e-05,
+      "loss": 0.8157,
+      "step": 1468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988922224064067e-05,
+      "loss": 0.807,
+      "step": 1469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888973344118002e-05,
+      "loss": 0.8657,
+      "step": 1470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988872416985835e-05,
+      "loss": 0.7593,
+      "step": 1471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98884747178687e-05,
+      "loss": 0.7961,
+      "step": 1472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888224988156066e-05,
+      "loss": 0.7354,
+      "step": 1473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887974980727458e-05,
+      "loss": 0.8138,
+      "step": 1474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887724695589902e-05,
+      "loss": 0.8217,
+      "step": 1475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988747413275042e-05,
+      "loss": 0.8612,
+      "step": 1476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887223292216058e-05,
+      "loss": 0.7462,
+      "step": 1477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886972173993853e-05,
+      "loss": 0.7978,
+      "step": 1478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988672077809086e-05,
+      "loss": 0.7865,
+      "step": 1479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886469104514147e-05,
+      "loss": 0.8121,
+      "step": 1480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886217153270774e-05,
+      "loss": 0.7947,
+      "step": 1481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885964924367818e-05,
+      "loss": 0.8977,
+      "step": 1482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885712417812365e-05,
+      "loss": 0.8846,
+      "step": 1483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988545963361151e-05,
+      "loss": 0.8372,
+      "step": 1484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885206571772346e-05,
+      "loss": 0.783,
+      "step": 1485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884953232301982e-05,
+      "loss": 0.8086,
+      "step": 1486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884699615207534e-05,
+      "loss": 0.7867,
+      "step": 1487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884445720496127e-05,
+      "loss": 0.8137,
+      "step": 1488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884191548174892e-05,
+      "loss": 0.8081,
+      "step": 1489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883937098250962e-05,
+      "loss": 0.8137,
+      "step": 1490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988368237073149e-05,
+      "loss": 0.7791,
+      "step": 1491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883427365623626e-05,
+      "loss": 0.7383,
+      "step": 1492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883172082934534e-05,
+      "loss": 0.808,
+      "step": 1493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882916522671383e-05,
+      "loss": 0.7998,
+      "step": 1494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988266068484135e-05,
+      "loss": 0.7999,
+      "step": 1495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988240456945162e-05,
+      "loss": 0.8781,
+      "step": 1496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988214817650939e-05,
+      "loss": 0.7643,
+      "step": 1497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881891506021852e-05,
+      "loss": 0.8004,
+      "step": 1498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881634557996228e-05,
+      "loss": 0.7946,
+      "step": 1499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881377332439723e-05,
+      "loss": 0.8368,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881119829359568e-05,
+      "loss": 0.7854,
+      "step": 1501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880862048762986e-05,
+      "loss": 0.7149,
+      "step": 1502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988060399065723e-05,
+      "loss": 0.8288,
+      "step": 1503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988034565504954e-05,
+      "loss": 0.7798,
+      "step": 1504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880087041947173e-05,
+      "loss": 0.8868,
+      "step": 1505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879828151357394e-05,
+      "loss": 0.8149,
+      "step": 1506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879568983287468e-05,
+      "loss": 0.8321,
+      "step": 1507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879309537744676e-05,
+      "loss": 0.8275,
+      "step": 1508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987904981473631e-05,
+      "loss": 0.8016,
+      "step": 1509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878789814269662e-05,
+      "loss": 0.8453,
+      "step": 1510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878529536352032e-05,
+      "loss": 0.8405,
+      "step": 1511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878268980990732e-05,
+      "loss": 0.8867,
+      "step": 1512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878008148193077e-05,
+      "loss": 0.836,
+      "step": 1513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877747037966398e-05,
+      "loss": 0.8035,
+      "step": 1514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877485650318022e-05,
+      "loss": 0.7696,
+      "step": 1515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877223985255293e-05,
+      "loss": 0.7716,
+      "step": 1516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876962042785564e-05,
+      "loss": 0.8378,
+      "step": 1517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876699822916187e-05,
+      "loss": 0.2821,
+      "step": 1518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987643732565453e-05,
+      "loss": 0.8217,
+      "step": 1519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876174551007956e-05,
+      "loss": 0.9543,
+      "step": 1520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875911498983857e-05,
+      "loss": 0.7706,
+      "step": 1521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875648169589618e-05,
+      "loss": 0.8228,
+      "step": 1522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987538456283263e-05,
+      "loss": 0.8091,
+      "step": 1523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98751206787203e-05,
+      "loss": 0.7869,
+      "step": 1524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987485651726004e-05,
+      "loss": 0.8082,
+      "step": 1525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987459207845927e-05,
+      "loss": 0.7922,
+      "step": 1526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987432736232541e-05,
+      "loss": 0.8071,
+      "step": 1527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98740623688659e-05,
+      "loss": 0.7916,
+      "step": 1528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873797098088185e-05,
+      "loss": 0.847,
+      "step": 1529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873531549999713e-05,
+      "loss": 0.7388,
+      "step": 1530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987326572460794e-05,
+      "loss": 0.8115,
+      "step": 1531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987299962192033e-05,
+      "loss": 0.7491,
+      "step": 1532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872733241944364e-05,
+      "loss": 0.8358,
+      "step": 1533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987246658468752e-05,
+      "loss": 0.8458,
+      "step": 1534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872199650157286e-05,
+      "loss": 0.7743,
+      "step": 1535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871932438361156e-05,
+      "loss": 0.7968,
+      "step": 1536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987166494930664e-05,
+      "loss": 0.823,
+      "step": 1537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871397183001247e-05,
+      "loss": 0.741,
+      "step": 1538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871129139452502e-05,
+      "loss": 0.7794,
+      "step": 1539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870860818667932e-05,
+      "loss": 0.7815,
+      "step": 1540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870592220655068e-05,
+      "loss": 0.7674,
+      "step": 1541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870323345421457e-05,
+      "loss": 0.791,
+      "step": 1542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870054192974653e-05,
+      "loss": 0.7849,
+      "step": 1543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986978476332221e-05,
+      "loss": 0.7215,
+      "step": 1544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98695150564717e-05,
+      "loss": 0.8241,
+      "step": 1545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869245072430695e-05,
+      "loss": 0.8061,
+      "step": 1546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986897481120678e-05,
+      "loss": 0.8455,
+      "step": 1547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986870427280754e-05,
+      "loss": 0.9009,
+      "step": 1548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986843345724058e-05,
+      "loss": 0.7477,
+      "step": 1549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98681623645135e-05,
+      "loss": 0.8025,
+      "step": 1550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986789099463392e-05,
+      "loss": 0.8489,
+      "step": 1551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867619347609458e-05,
+      "loss": 0.7739,
+      "step": 1552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986734742344774e-05,
+      "loss": 0.7682,
+      "step": 1553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986707522215641e-05,
+      "loss": 0.8191,
+      "step": 1554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986680274374311e-05,
+      "loss": 0.7835,
+      "step": 1555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986652998821549e-05,
+      "loss": 0.7848,
+      "step": 1556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866256955581215e-05,
+      "loss": 0.8036,
+      "step": 1557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986598364584795e-05,
+      "loss": 0.8441,
+      "step": 1558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865710059023376e-05,
+      "loss": 0.8171,
+      "step": 1559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865436195115167e-05,
+      "loss": 0.8039,
+      "step": 1560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865162054131026e-05,
+      "loss": 0.8023,
+      "step": 1561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864887636078644e-05,
+      "loss": 0.7921,
+      "step": 1562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864612940965732e-05,
+      "loss": 0.7723,
+      "step": 1563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864337968800005e-05,
+      "loss": 0.7755,
+      "step": 1564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864062719589184e-05,
+      "loss": 0.8331,
+      "step": 1565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863787193341002e-05,
+      "loss": 0.8087,
+      "step": 1566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863511390063193e-05,
+      "loss": 0.8187,
+      "step": 1567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863235309763505e-05,
+      "loss": 0.7593,
+      "step": 1568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862958952449695e-05,
+      "loss": 0.8171,
+      "step": 1569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862682318129525e-05,
+      "loss": 0.8536,
+      "step": 1570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862405406810756e-05,
+      "loss": 0.7758,
+      "step": 1571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862128218501175e-05,
+      "loss": 0.8649,
+      "step": 1572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986185075320856e-05,
+      "loss": 0.7817,
+      "step": 1573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986157301094071e-05,
+      "loss": 0.8361,
+      "step": 1574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986129499170542e-05,
+      "loss": 0.7657,
+      "step": 1575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861016695510502e-05,
+      "loss": 0.6836,
+      "step": 1576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986073812236377e-05,
+      "loss": 0.7779,
+      "step": 1577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860459272273047e-05,
+      "loss": 0.8025,
+      "step": 1578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986018014524617e-05,
+      "loss": 0.8182,
+      "step": 1579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859900741290973e-05,
+      "loss": 0.8562,
+      "step": 1580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859621060415302e-05,
+      "loss": 0.8011,
+      "step": 1581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859341102627017e-05,
+      "loss": 0.8545,
+      "step": 1582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859060867933978e-05,
+      "loss": 0.7711,
+      "step": 1583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858780356344056e-05,
+      "loss": 0.7442,
+      "step": 1584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985849956786513e-05,
+      "loss": 0.8779,
+      "step": 1585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858218502505086e-05,
+      "loss": 0.8439,
+      "step": 1586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857937160271813e-05,
+      "loss": 0.8369,
+      "step": 1587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985765554117322e-05,
+      "loss": 0.9021,
+      "step": 1588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857373645217217e-05,
+      "loss": 0.8105,
+      "step": 1589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857091472411713e-05,
+      "loss": 0.8023,
+      "step": 1590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856809022764638e-05,
+      "loss": 0.8175,
+      "step": 1591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856526296283926e-05,
+      "loss": 0.8077,
+      "step": 1592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856243292977515e-05,
+      "loss": 0.7536,
+      "step": 1593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855960012853354e-05,
+      "loss": 0.7281,
+      "step": 1594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98556764559194e-05,
+      "loss": 0.8156,
+      "step": 1595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855392622183614e-05,
+      "loss": 0.7867,
+      "step": 1596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985510851165397e-05,
+      "loss": 0.7676,
+      "step": 1597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985482412433845e-05,
+      "loss": 0.7602,
+      "step": 1598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854539460245036e-05,
+      "loss": 0.8095,
+      "step": 1599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854254519381726e-05,
+      "loss": 0.773,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853969301756523e-05,
+      "loss": 0.786,
+      "step": 1601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853683807377432e-05,
+      "loss": 0.8789,
+      "step": 1602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985339803625248e-05,
+      "loss": 0.6484,
+      "step": 1603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985311198838969e-05,
+      "loss": 0.8553,
+      "step": 1604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852825663797087e-05,
+      "loss": 0.7817,
+      "step": 1605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852539062482724e-05,
+      "loss": 0.8395,
+      "step": 1606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852252184454646e-05,
+      "loss": 0.8014,
+      "step": 1607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851965029720916e-05,
+      "loss": 0.7759,
+      "step": 1608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851677598289588e-05,
+      "loss": 0.8502,
+      "step": 1609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851389890168738e-05,
+      "loss": 0.9247,
+      "step": 1610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851101905366456e-05,
+      "loss": 0.8171,
+      "step": 1611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850813643890813e-05,
+      "loss": 0.8074,
+      "step": 1612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850525105749923e-05,
+      "loss": 0.7349,
+      "step": 1613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850236290951877e-05,
+      "loss": 0.8305,
+      "step": 1614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849947199504794e-05,
+      "loss": 0.7783,
+      "step": 1615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849657831416787e-05,
+      "loss": 0.7465,
+      "step": 1616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849368186695988e-05,
+      "loss": 0.7799,
+      "step": 1617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984907826535053e-05,
+      "loss": 0.7992,
+      "step": 1618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848788067388555e-05,
+      "loss": 0.8734,
+      "step": 1619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848497592818214e-05,
+      "loss": 0.7592,
+      "step": 1620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848206841647665e-05,
+      "loss": 0.2946,
+      "step": 1621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847915813885076e-05,
+      "loss": 0.8154,
+      "step": 1622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847624509538614e-05,
+      "loss": 0.8312,
+      "step": 1623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984733292861647e-05,
+      "loss": 0.8007,
+      "step": 1624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847041071126827e-05,
+      "loss": 0.906,
+      "step": 1625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846748937077885e-05,
+      "loss": 0.8145,
+      "step": 1626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846456526477845e-05,
+      "loss": 0.6937,
+      "step": 1627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984616383933492e-05,
+      "loss": 0.7371,
+      "step": 1628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845870875657335e-05,
+      "loss": 0.8974,
+      "step": 1629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984557763545331e-05,
+      "loss": 0.8013,
+      "step": 1630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845284118731093e-05,
+      "loss": 0.7996,
+      "step": 1631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844990325498915e-05,
+      "loss": 0.8851,
+      "step": 1632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844696255765036e-05,
+      "loss": 0.8353,
+      "step": 1633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984440190953771e-05,
+      "loss": 0.8021,
+      "step": 1634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844107286825205e-05,
+      "loss": 0.9233,
+      "step": 1635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843812387635796e-05,
+      "loss": 0.8521,
+      "step": 1636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843517211977765e-05,
+      "loss": 0.881,
+      "step": 1637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843221759859407e-05,
+      "loss": 0.7525,
+      "step": 1638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984292603128901e-05,
+      "loss": 0.8358,
+      "step": 1639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842630026274888e-05,
+      "loss": 0.8214,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842333744825356e-05,
+      "loss": 0.8502,
+      "step": 1641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842037186948725e-05,
+      "loss": 0.8219,
+      "step": 1642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841740352653334e-05,
+      "loss": 0.7817,
+      "step": 1643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841443241947515e-05,
+      "loss": 0.8337,
+      "step": 1644
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841145854839608e-05,
+      "loss": 0.751,
+      "step": 1645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840848191337976e-05,
+      "loss": 0.809,
+      "step": 1646
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840550251450974e-05,
+      "loss": 0.7659,
+      "step": 1647
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840252035186967e-05,
+      "loss": 0.7872,
+      "step": 1648
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839953542554335e-05,
+      "loss": 0.8042,
+      "step": 1649
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839654773561455e-05,
+      "loss": 0.8054,
+      "step": 1650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839355728216727e-05,
+      "loss": 0.8016,
+      "step": 1651
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839056406528544e-05,
+      "loss": 0.651,
+      "step": 1652
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838756808505315e-05,
+      "loss": 0.7541,
+      "step": 1653
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838456934155452e-05,
+      "loss": 0.7714,
+      "step": 1654
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838156783487376e-05,
+      "loss": 0.8073,
+      "step": 1655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983785635650952e-05,
+      "loss": 0.747,
+      "step": 1656
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837555653230326e-05,
+      "loss": 0.7459,
+      "step": 1657
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837254673658227e-05,
+      "loss": 0.7697,
+      "step": 1658
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836953417801686e-05,
+      "loss": 0.8095,
+      "step": 1659
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836651885669165e-05,
+      "loss": 0.8217,
+      "step": 1660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836350077269126e-05,
+      "loss": 0.7395,
+      "step": 1661
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836047992610047e-05,
+      "loss": 0.8651,
+      "step": 1662
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835745631700418e-05,
+      "loss": 0.8152,
+      "step": 1663
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835442994548724e-05,
+      "loss": 0.775,
+      "step": 1664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835140081163468e-05,
+      "loss": 0.8099,
+      "step": 1665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983483689155316e-05,
+      "loss": 0.8491,
+      "step": 1666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834533425726303e-05,
+      "loss": 0.7343,
+      "step": 1667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834229683691436e-05,
+      "loss": 0.7892,
+      "step": 1668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983392566545708e-05,
+      "loss": 0.7788,
+      "step": 1669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833621371031778e-05,
+      "loss": 0.8286,
+      "step": 1670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833316800424078e-05,
+      "loss": 0.8254,
+      "step": 1671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833011953642525e-05,
+      "loss": 0.8282,
+      "step": 1672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832706830695688e-05,
+      "loss": 0.8466,
+      "step": 1673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983240143159214e-05,
+      "loss": 0.7414,
+      "step": 1674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832095756340444e-05,
+      "loss": 0.8209,
+      "step": 1675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98317898049492e-05,
+      "loss": 0.9048,
+      "step": 1676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831483577426997e-05,
+      "loss": 0.8046,
+      "step": 1677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983117707378243e-05,
+      "loss": 0.8615,
+      "step": 1678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983087029402411e-05,
+      "loss": 0.8268,
+      "step": 1679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983056323816066e-05,
+      "loss": 0.7843,
+      "step": 1680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983025590620069e-05,
+      "loss": 0.6957,
+      "step": 1681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982994829815285e-05,
+      "loss": 0.7873,
+      "step": 1682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829640414025758e-05,
+      "loss": 0.8386,
+      "step": 1683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829332253828077e-05,
+      "loss": 0.8478,
+      "step": 1684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829023817568455e-05,
+      "loss": 0.7586,
+      "step": 1685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982871510525556e-05,
+      "loss": 0.7853,
+      "step": 1686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828406116898056e-05,
+      "loss": 0.8448,
+      "step": 1687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828096852504624e-05,
+      "loss": 0.801,
+      "step": 1688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982778731208395e-05,
+      "loss": 0.7748,
+      "step": 1689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827477495644727e-05,
+      "loss": 0.7826,
+      "step": 1690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827167403195654e-05,
+      "loss": 0.7894,
+      "step": 1691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826857034745445e-05,
+      "loss": 0.8325,
+      "step": 1692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826546390302814e-05,
+      "loss": 0.7802,
+      "step": 1693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982623546987649e-05,
+      "loss": 0.7605,
+      "step": 1694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825924273475197e-05,
+      "loss": 0.8025,
+      "step": 1695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982561280110768e-05,
+      "loss": 0.8664,
+      "step": 1696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982530105278269e-05,
+      "loss": 0.8464,
+      "step": 1697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824989028508975e-05,
+      "loss": 0.8034,
+      "step": 1698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824676728295304e-05,
+      "loss": 0.7095,
+      "step": 1699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982436415215045e-05,
+      "loss": 0.8145,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824051300083185e-05,
+      "loss": 0.8572,
+      "step": 1701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823738172102297e-05,
+      "loss": 0.8102,
+      "step": 1702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823424768216585e-05,
+      "loss": 0.8282,
+      "step": 1703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823111088434848e-05,
+      "loss": 0.7551,
+      "step": 1704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822797132765897e-05,
+      "loss": 0.336,
+      "step": 1705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982248290121855e-05,
+      "loss": 0.8511,
+      "step": 1706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982216839380163e-05,
+      "loss": 0.8271,
+      "step": 1707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821853610523973e-05,
+      "loss": 0.7756,
+      "step": 1708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821538551394414e-05,
+      "loss": 0.7776,
+      "step": 1709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982122321642181e-05,
+      "loss": 0.3025,
+      "step": 1710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820907605615013e-05,
+      "loss": 0.7985,
+      "step": 1711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820591718982887e-05,
+      "loss": 0.8168,
+      "step": 1712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820275556534306e-05,
+      "loss": 0.8147,
+      "step": 1713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819959118278144e-05,
+      "loss": 0.7968,
+      "step": 1714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819642404223293e-05,
+      "loss": 0.8146,
+      "step": 1715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981932541437865e-05,
+      "loss": 0.7418,
+      "step": 1716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819008148753114e-05,
+      "loss": 0.7694,
+      "step": 1717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818690607355596e-05,
+      "loss": 0.8547,
+      "step": 1718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818372790195014e-05,
+      "loss": 0.8027,
+      "step": 1719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818054697280297e-05,
+      "loss": 0.7657,
+      "step": 1720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817736328620378e-05,
+      "loss": 0.8427,
+      "step": 1721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817417684224193e-05,
+      "loss": 0.783,
+      "step": 1722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98170987641007e-05,
+      "loss": 0.7719,
+      "step": 1723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981677956825885e-05,
+      "loss": 0.8513,
+      "step": 1724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816460096707605e-05,
+      "loss": 0.7021,
+      "step": 1725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816140349455946e-05,
+      "loss": 0.7556,
+      "step": 1726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981582032651285e-05,
+      "loss": 0.7148,
+      "step": 1727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815500027887303e-05,
+      "loss": 0.7864,
+      "step": 1728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815179453588304e-05,
+      "loss": 0.7938,
+      "step": 1729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814858603624854e-05,
+      "loss": 0.7798,
+      "step": 1730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814537478005964e-05,
+      "loss": 0.8431,
+      "step": 1731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814216076740654e-05,
+      "loss": 0.7854,
+      "step": 1732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981389439983795e-05,
+      "loss": 0.3441,
+      "step": 1733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813572447306888e-05,
+      "loss": 0.7607,
+      "step": 1734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981325021915651e-05,
+      "loss": 0.765,
+      "step": 1735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981292771539586e-05,
+      "loss": 0.7908,
+      "step": 1736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812604936034004e-05,
+      "loss": 0.762,
+      "step": 1737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812281881080007e-05,
+      "loss": 0.821,
+      "step": 1738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811958550542933e-05,
+      "loss": 0.7978,
+      "step": 1739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811634944431875e-05,
+      "loss": 0.8428,
+      "step": 1740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811311062755917e-05,
+      "loss": 0.7784,
+      "step": 1741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981098690552415e-05,
+      "loss": 0.9074,
+      "step": 1742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810662472745682e-05,
+      "loss": 0.7666,
+      "step": 1743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810337764429627e-05,
+      "loss": 0.7837,
+      "step": 1744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810012780585104e-05,
+      "loss": 0.7814,
+      "step": 1745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980968752122124e-05,
+      "loss": 0.7726,
+      "step": 1746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809361986347167e-05,
+      "loss": 0.8034,
+      "step": 1747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809036175972028e-05,
+      "loss": 0.8087,
+      "step": 1748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808710090104983e-05,
+      "loss": 0.7625,
+      "step": 1749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808383728755176e-05,
+      "loss": 0.7791,
+      "step": 1750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808057091931782e-05,
+      "loss": 0.7902,
+      "step": 1751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807730179643974e-05,
+      "loss": 0.8545,
+      "step": 1752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807402991900933e-05,
+      "loss": 0.8504,
+      "step": 1753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807075528711847e-05,
+      "loss": 0.7599,
+      "step": 1754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806747790085912e-05,
+      "loss": 0.8286,
+      "step": 1755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806419776032334e-05,
+      "loss": 0.8024,
+      "step": 1756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806091486560328e-05,
+      "loss": 0.8185,
+      "step": 1757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980576292167911e-05,
+      "loss": 0.7392,
+      "step": 1758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980543408139791e-05,
+      "loss": 0.7953,
+      "step": 1759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805104965725962e-05,
+      "loss": 0.799,
+      "step": 1760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804775574672514e-05,
+      "loss": 0.7638,
+      "step": 1761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980444590824681e-05,
+      "loss": 0.8352,
+      "step": 1762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804115966458118e-05,
+      "loss": 0.843,
+      "step": 1763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803785749315697e-05,
+      "loss": 0.8226,
+      "step": 1764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803455256828824e-05,
+      "loss": 0.8129,
+      "step": 1765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803124489006777e-05,
+      "loss": 0.8267,
+      "step": 1766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802793445858855e-05,
+      "loss": 0.7237,
+      "step": 1767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980246212739435e-05,
+      "loss": 0.8525,
+      "step": 1768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980213053362256e-05,
+      "loss": 0.783,
+      "step": 1769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801798664552813e-05,
+      "loss": 0.8298,
+      "step": 1770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980146652019442e-05,
+      "loss": 0.7797,
+      "step": 1771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801134100556708e-05,
+      "loss": 0.8057,
+      "step": 1772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800801405649023e-05,
+      "loss": 0.7595,
+      "step": 1773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800468435480696e-05,
+      "loss": 0.8167,
+      "step": 1774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980013519006109e-05,
+      "loss": 0.8019,
+      "step": 1775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799801669399556e-05,
+      "loss": 0.8537,
+      "step": 1776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799467873505467e-05,
+      "loss": 0.825,
+      "step": 1777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799133802388194e-05,
+      "loss": 0.7171,
+      "step": 1778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798799456057123e-05,
+      "loss": 0.8655,
+      "step": 1779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979846483452164e-05,
+      "loss": 0.7797,
+      "step": 1780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798129937791147e-05,
+      "loss": 0.8078,
+      "step": 1781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979779476587505e-05,
+      "loss": 0.8152,
+      "step": 1782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797459318782756e-05,
+      "loss": 0.7529,
+      "step": 1783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797123596523692e-05,
+      "loss": 0.8093,
+      "step": 1784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979678759910729e-05,
+      "loss": 0.8176,
+      "step": 1785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796451326542977e-05,
+      "loss": 0.8118,
+      "step": 1786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796114778840205e-05,
+      "loss": 0.7826,
+      "step": 1787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795777956008423e-05,
+      "loss": 0.8458,
+      "step": 1788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795440858057094e-05,
+      "loss": 0.8378,
+      "step": 1789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795103484995685e-05,
+      "loss": 0.7923,
+      "step": 1790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794765836833664e-05,
+      "loss": 0.803,
+      "step": 1791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794427913580523e-05,
+      "loss": 0.8554,
+      "step": 1792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794089715245748e-05,
+      "loss": 0.826,
+      "step": 1793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979375124183884e-05,
+      "loss": 0.8442,
+      "step": 1794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97934124933693e-05,
+      "loss": 0.7737,
+      "step": 1795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793073469846653e-05,
+      "loss": 0.83,
+      "step": 1796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979273417128041e-05,
+      "loss": 0.8248,
+      "step": 1797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792394597680106e-05,
+      "loss": 0.8231,
+      "step": 1798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792054749055275e-05,
+      "loss": 0.8224,
+      "step": 1799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979171462541546e-05,
+      "loss": 0.8564,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979137422677022e-05,
+      "loss": 0.7903,
+      "step": 1801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979103355312911e-05,
+      "loss": 0.7562,
+      "step": 1802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97906926045017e-05,
+      "loss": 0.7746,
+      "step": 1803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979035138089757e-05,
+      "loss": 0.8088,
+      "step": 1804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790009882326293e-05,
+      "loss": 0.7886,
+      "step": 1805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978966810879747e-05,
+      "loss": 0.7432,
+      "step": 1806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978932606032069e-05,
+      "loss": 0.7684,
+      "step": 1807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788983736905572e-05,
+      "loss": 0.8099,
+      "step": 1808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788641138561723e-05,
+      "loss": 0.8341,
+      "step": 1809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788298265298765e-05,
+      "loss": 0.8472,
+      "step": 1810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787955117126328e-05,
+      "loss": 0.7752,
+      "step": 1811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978761169405405e-05,
+      "loss": 0.8428,
+      "step": 1812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978726799609158e-05,
+      "loss": 0.8146,
+      "step": 1813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978692402324856e-05,
+      "loss": 0.8025,
+      "step": 1814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786579775534663e-05,
+      "loss": 0.8113,
+      "step": 1815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786235252959555e-05,
+      "loss": 0.8089,
+      "step": 1816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785890455532905e-05,
+      "loss": 0.7665,
+      "step": 1817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785545383264403e-05,
+      "loss": 0.6982,
+      "step": 1818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978520003616374e-05,
+      "loss": 0.8318,
+      "step": 1819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784854414240614e-05,
+      "loss": 0.8379,
+      "step": 1820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978450851750473e-05,
+      "loss": 0.7899,
+      "step": 1821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978416234596581e-05,
+      "loss": 0.8027,
+      "step": 1822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783815899633567e-05,
+      "loss": 0.7974,
+      "step": 1823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783469178517736e-05,
+      "loss": 0.7834,
+      "step": 1824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783122182628053e-05,
+      "loss": 0.7852,
+      "step": 1825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782774911974264e-05,
+      "loss": 0.8706,
+      "step": 1826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782427366566125e-05,
+      "loss": 0.8652,
+      "step": 1827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782079546413396e-05,
+      "loss": 0.8139,
+      "step": 1828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781731451525846e-05,
+      "loss": 0.8314,
+      "step": 1829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781383081913246e-05,
+      "loss": 0.7661,
+      "step": 1830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781034437585388e-05,
+      "loss": 0.7818,
+      "step": 1831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780685518552057e-05,
+      "loss": 0.742,
+      "step": 1832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780336324823057e-05,
+      "loss": 0.7469,
+      "step": 1833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779986856408195e-05,
+      "loss": 0.7527,
+      "step": 1834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779637113317285e-05,
+      "loss": 0.8116,
+      "step": 1835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977928709556015e-05,
+      "loss": 0.8403,
+      "step": 1836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778936803146618e-05,
+      "loss": 0.8683,
+      "step": 1837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778586236086533e-05,
+      "loss": 0.7538,
+      "step": 1838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778235394389733e-05,
+      "loss": 0.7892,
+      "step": 1839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977788427806608e-05,
+      "loss": 0.7667,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777532887125425e-05,
+      "loss": 0.8203,
+      "step": 1841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777181221577648e-05,
+      "loss": 0.8043,
+      "step": 1842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977682928143262e-05,
+      "loss": 0.8478,
+      "step": 1843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776477066700223e-05,
+      "loss": 0.8788,
+      "step": 1844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977612457739036e-05,
+      "loss": 0.8212,
+      "step": 1845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775771813512915e-05,
+      "loss": 0.7508,
+      "step": 1846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977541877507781e-05,
+      "loss": 0.3365,
+      "step": 1847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977506546209495e-05,
+      "loss": 0.7653,
+      "step": 1848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774711874574263e-05,
+      "loss": 0.7075,
+      "step": 1849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977435801252568e-05,
+      "loss": 0.8494,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774003875959135e-05,
+      "loss": 0.7682,
+      "step": 1851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773649464884577e-05,
+      "loss": 0.7858,
+      "step": 1852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977329477931196e-05,
+      "loss": 0.7634,
+      "step": 1853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772939819251247e-05,
+      "loss": 0.779,
+      "step": 1854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977258458471241e-05,
+      "loss": 0.8391,
+      "step": 1855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977222907570542e-05,
+      "loss": 0.7678,
+      "step": 1856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771873292240257e-05,
+      "loss": 0.8018,
+      "step": 1857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771517234326927e-05,
+      "loss": 0.8138,
+      "step": 1858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977116090197542e-05,
+      "loss": 0.7852,
+      "step": 1859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770804295195747e-05,
+      "loss": 0.8118,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770447413997922e-05,
+      "loss": 0.8548,
+      "step": 1861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977009025839197e-05,
+      "loss": 0.7831,
+      "step": 1862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976973282838792e-05,
+      "loss": 0.8183,
+      "step": 1863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769375123995814e-05,
+      "loss": 0.7689,
+      "step": 1864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769017145225696e-05,
+      "loss": 0.894,
+      "step": 1865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768658892087617e-05,
+      "loss": 0.7891,
+      "step": 1866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768300364591643e-05,
+      "loss": 0.8955,
+      "step": 1867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767941562747843e-05,
+      "loss": 0.2998,
+      "step": 1868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767582486566294e-05,
+      "loss": 0.8125,
+      "step": 1869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767223136057082e-05,
+      "loss": 0.8359,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766863511230294e-05,
+      "loss": 0.8347,
+      "step": 1871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766503612096036e-05,
+      "loss": 0.8494,
+      "step": 1872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766143438664412e-05,
+      "loss": 0.7594,
+      "step": 1873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765782990945546e-05,
+      "loss": 0.8615,
+      "step": 1874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765422268949548e-05,
+      "loss": 0.7346,
+      "step": 1875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976506127268656e-05,
+      "loss": 0.7639,
+      "step": 1876
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764700002166718e-05,
+      "loss": 0.8051,
+      "step": 1877
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976433845740016e-05,
+      "loss": 0.705,
+      "step": 1878
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763976638397057e-05,
+      "loss": 0.7027,
+      "step": 1879
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763614545167552e-05,
+      "loss": 0.7359,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763252177721834e-05,
+      "loss": 0.2907,
+      "step": 1881
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762889536070063e-05,
+      "loss": 0.7366,
+      "step": 1882
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762526620222434e-05,
+      "loss": 0.7363,
+      "step": 1883
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762163430189136e-05,
+      "loss": 0.8606,
+      "step": 1884
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976179996598037e-05,
+      "loss": 0.7589,
+      "step": 1885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761436227606344e-05,
+      "loss": 0.8301,
+      "step": 1886
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761072215077277e-05,
+      "loss": 0.8378,
+      "step": 1887
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760707928403385e-05,
+      "loss": 0.7531,
+      "step": 1888
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976034336759491e-05,
+      "loss": 0.7715,
+      "step": 1889
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975997853266208e-05,
+      "loss": 0.7891,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759613423615148e-05,
+      "loss": 0.7797,
+      "step": 1891
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759248040464367e-05,
+      "loss": 0.7599,
+      "step": 1892
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758882383219995e-05,
+      "loss": 0.7263,
+      "step": 1893
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975851645189231e-05,
+      "loss": 0.7891,
+      "step": 1894
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975815024649158e-05,
+      "loss": 0.7728,
+      "step": 1895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97577837670281e-05,
+      "loss": 0.874,
+      "step": 1896
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757417013512152e-05,
+      "loss": 0.8087,
+      "step": 1897
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757049985954047e-05,
+      "loss": 0.8035,
+      "step": 1898
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756682684364086e-05,
+      "loss": 0.788,
+      "step": 1899
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756315108752586e-05,
+      "loss": 0.83,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975594725912987e-05,
+      "loss": 0.7747,
+      "step": 1901
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755579135506275e-05,
+      "loss": 0.7751,
+      "step": 1902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755210737892133e-05,
+      "loss": 0.7651,
+      "step": 1903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975484206629779e-05,
+      "loss": 0.7946,
+      "step": 1904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975447312073361e-05,
+      "loss": 0.7941,
+      "step": 1905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754103901209943e-05,
+      "loss": 0.7938,
+      "step": 1906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975373440773717e-05,
+      "loss": 0.8588,
+      "step": 1907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975336464032566e-05,
+      "loss": 0.8127,
+      "step": 1908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752994598985802e-05,
+      "loss": 0.8311,
+      "step": 1909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752624283727987e-05,
+      "loss": 0.8192,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752253694562614e-05,
+      "loss": 0.8167,
+      "step": 1911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751882831500095e-05,
+      "loss": 0.8432,
+      "step": 1912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751511694550848e-05,
+      "loss": 0.7913,
+      "step": 1913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975114028372529e-05,
+      "loss": 0.312,
+      "step": 1914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975076859903385e-05,
+      "loss": 0.8351,
+      "step": 1915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975039664048698e-05,
+      "loss": 0.8217,
+      "step": 1916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750024408095115e-05,
+      "loss": 0.7666,
+      "step": 1917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749651901868713e-05,
+      "loss": 0.2793,
+      "step": 1918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.8121,
+      "step": 1919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748906067954156e-05,
+      "loss": 0.7491,
+      "step": 1920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748532740286948e-05,
+      "loss": 0.7643,
+      "step": 1921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748159138827097e-05,
+      "loss": 0.2987,
+      "step": 1922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747785263585096e-05,
+      "loss": 0.7406,
+      "step": 1923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747411114571445e-05,
+      "loss": 0.7961,
+      "step": 1924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974703669179665e-05,
+      "loss": 0.782,
+      "step": 1925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746661995271233e-05,
+      "loss": 0.7819,
+      "step": 1926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746287025005712e-05,
+      "loss": 0.7935,
+      "step": 1927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745911781010622e-05,
+      "loss": 0.7144,
+      "step": 1928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745536263296503e-05,
+      "loss": 0.8492,
+      "step": 1929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745160471873895e-05,
+      "loss": 0.8863,
+      "step": 1930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974478440675336e-05,
+      "loss": 0.738,
+      "step": 1931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744408067945455e-05,
+      "loss": 0.7384,
+      "step": 1932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744031455460747e-05,
+      "loss": 0.7454,
+      "step": 1933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743654569309824e-05,
+      "loss": 0.8269,
+      "step": 1934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743277409503257e-05,
+      "loss": 0.7732,
+      "step": 1935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974289997605165e-05,
+      "loss": 0.7912,
+      "step": 1936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97425222689656e-05,
+      "loss": 0.8107,
+      "step": 1937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742144288255712e-05,
+      "loss": 0.817,
+      "step": 1938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974176603393261e-05,
+      "loss": 0.7109,
+      "step": 1939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741387506006906e-05,
+      "loss": 0.8054,
+      "step": 1940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974100870448924e-05,
+      "loss": 0.8125,
+      "step": 1941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740629629390245e-05,
+      "loss": 0.8237,
+      "step": 1942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740250280720574e-05,
+      "loss": 0.7622,
+      "step": 1943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739870658490872e-05,
+      "loss": 0.7886,
+      "step": 1944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739490762711812e-05,
+      "loss": 0.7856,
+      "step": 1945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739110593394057e-05,
+      "loss": 0.7852,
+      "step": 1946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738730150548285e-05,
+      "loss": 0.7097,
+      "step": 1947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738349434185183e-05,
+      "loss": 0.7928,
+      "step": 1948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737968444315438e-05,
+      "loss": 0.688,
+      "step": 1949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737587180949755e-05,
+      "loss": 0.8664,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737205644098844e-05,
+      "loss": 0.8016,
+      "step": 1951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736823833773412e-05,
+      "loss": 0.7674,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736441749984193e-05,
+      "loss": 0.7288,
+      "step": 1953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736059392741914e-05,
+      "loss": 0.8016,
+      "step": 1954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973567676205731e-05,
+      "loss": 0.8358,
+      "step": 1955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973529385794113e-05,
+      "loss": 0.7799,
+      "step": 1956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973491068040413e-05,
+      "loss": 0.8185,
+      "step": 1957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973452722945707e-05,
+      "loss": 0.7535,
+      "step": 1958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973414350511072e-05,
+      "loss": 0.8404,
+      "step": 1959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733759507375854e-05,
+      "loss": 0.828,
+      "step": 1960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973337523626326e-05,
+      "loss": 0.7569,
+      "step": 1961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732990691783732e-05,
+      "loss": 0.7752,
+      "step": 1962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732605873948065e-05,
+      "loss": 0.7911,
+      "step": 1963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732220782767072e-05,
+      "loss": 0.8029,
+      "step": 1964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731835418251566e-05,
+      "loss": 0.8253,
+      "step": 1965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973144978041237e-05,
+      "loss": 0.8125,
+      "step": 1966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731063869260315e-05,
+      "loss": 0.8094,
+      "step": 1967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730677684806237e-05,
+      "loss": 0.8064,
+      "step": 1968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973029122706099e-05,
+      "loss": 0.7679,
+      "step": 1969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972990449603542e-05,
+      "loss": 0.7924,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972951749174039e-05,
+      "loss": 0.8213,
+      "step": 1971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729130214186774e-05,
+      "loss": 0.7852,
+      "step": 1972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972874266338544e-05,
+      "loss": 0.7782,
+      "step": 1973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972835483934729e-05,
+      "loss": 0.7915,
+      "step": 1974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727966742083192e-05,
+      "loss": 0.8563,
+      "step": 1975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972757837160406e-05,
+      "loss": 0.7623,
+      "step": 1976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727189727920802e-05,
+      "loss": 0.8799,
+      "step": 1977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726800811044336e-05,
+      "loss": 0.7879,
+      "step": 1978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972641162098557e-05,
+      "loss": 0.849,
+      "step": 1979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726022157755452e-05,
+      "loss": 0.7786,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725632421364912e-05,
+      "loss": 0.8107,
+      "step": 1981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725242411824895e-05,
+      "loss": 0.8466,
+      "step": 1982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724852129146358e-05,
+      "loss": 0.7503,
+      "step": 1983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724461573340258e-05,
+      "loss": 0.7836,
+      "step": 1984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972407074441757e-05,
+      "loss": 0.8581,
+      "step": 1985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723679642389264e-05,
+      "loss": 0.7722,
+      "step": 1986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723288267266327e-05,
+      "loss": 0.8105,
+      "step": 1987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722896619059755e-05,
+      "loss": 0.8644,
+      "step": 1988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722504697780542e-05,
+      "loss": 0.7092,
+      "step": 1989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97221125034397e-05,
+      "loss": 0.7431,
+      "step": 1990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972172003604824e-05,
+      "loss": 0.813,
+      "step": 1991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721327295617186e-05,
+      "loss": 0.8004,
+      "step": 1992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972093428215757e-05,
+      "loss": 0.8481,
+      "step": 1993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720540995680428e-05,
+      "loss": 0.7879,
+      "step": 1994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720147436196804e-05,
+      "loss": 0.7542,
+      "step": 1995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971975360371775e-05,
+      "loss": 0.8758,
+      "step": 1996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971935949825434e-05,
+      "loss": 0.9283,
+      "step": 1997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718965119817628e-05,
+      "loss": 0.8557,
+      "step": 1998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718570468418696e-05,
+      "loss": 0.7916,
+      "step": 1999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718175544068624e-05,
+      "loss": 0.7705,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717780346778516e-05,
+      "loss": 0.7873,
+      "step": 2001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717384876559453e-05,
+      "loss": 0.8391,
+      "step": 2002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716989133422557e-05,
+      "loss": 0.8904,
+      "step": 2003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716593117378933e-05,
+      "loss": 0.7841,
+      "step": 2004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716196828439712e-05,
+      "loss": 0.8206,
+      "step": 2005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715800266616016e-05,
+      "loss": 0.8074,
+      "step": 2006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715403431918985e-05,
+      "loss": 0.8189,
+      "step": 2007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715006324359768e-05,
+      "loss": 0.7987,
+      "step": 2008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714608943949516e-05,
+      "loss": 0.8151,
+      "step": 2009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971421129069939e-05,
+      "loss": 0.725,
+      "step": 2010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713813364620555e-05,
+      "loss": 0.7803,
+      "step": 2011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713415165724186e-05,
+      "loss": 0.8848,
+      "step": 2012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713016694021473e-05,
+      "loss": 0.7911,
+      "step": 2013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712617949523603e-05,
+      "loss": 0.7895,
+      "step": 2014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971221893224178e-05,
+      "loss": 0.8484,
+      "step": 2015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711819642187203e-05,
+      "loss": 0.8444,
+      "step": 2016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711420079371087e-05,
+      "loss": 0.8014,
+      "step": 2017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711020243804664e-05,
+      "loss": 0.7153,
+      "step": 2018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971062013549915e-05,
+      "loss": 0.7244,
+      "step": 2019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710219754465794e-05,
+      "loss": 0.7986,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709819100715833e-05,
+      "loss": 0.8429,
+      "step": 2021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 0.7773,
+      "step": 2022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709016975111123e-05,
+      "loss": 0.7968,
+      "step": 2023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708615503278897e-05,
+      "loss": 0.7703,
+      "step": 2024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970821375877513e-05,
+      "loss": 0.7778,
+      "step": 2025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97078117416111e-05,
+      "loss": 0.8291,
+      "step": 2026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97074094517981e-05,
+      "loss": 0.8728,
+      "step": 2027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707006889347423e-05,
+      "loss": 0.8661,
+      "step": 2028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9706604054270376e-05,
+      "loss": 0.7552,
+      "step": 2029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706200946578276e-05,
+      "loss": 0.7895,
+      "step": 2030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970579756628245e-05,
+      "loss": 0.7743,
+      "step": 2031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705393913394218e-05,
+      "loss": 0.7891,
+      "step": 2032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970498998792492e-05,
+      "loss": 0.7281,
+      "step": 2033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97045857898859e-05,
+      "loss": 0.7834,
+      "step": 2034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704181319288505e-05,
+      "loss": 0.763,
+      "step": 2035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703776576144106e-05,
+      "loss": 0.7919,
+      "step": 2036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703371560464065e-05,
+      "loss": 0.7385,
+      "step": 2037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702966272259755e-05,
+      "loss": 0.828,
+      "step": 2038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702560711542564e-05,
+      "loss": 0.7908,
+      "step": 2039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702154878323876e-05,
+      "loss": 0.7628,
+      "step": 2040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970174877261509e-05,
+      "loss": 0.843,
+      "step": 2041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701342394427616e-05,
+      "loss": 0.7494,
+      "step": 2042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700935743772864e-05,
+      "loss": 0.8677,
+      "step": 2043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700528820662256e-05,
+      "loss": 0.8121,
+      "step": 2044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970012162510722e-05,
+      "loss": 0.8165,
+      "step": 2045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699714157119193e-05,
+      "loss": 0.8467,
+      "step": 2046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969930641670962e-05,
+      "loss": 0.7399,
+      "step": 2047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969889840388995e-05,
+      "loss": 0.7534,
+      "step": 2048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698490118671647e-05,
+      "loss": 0.7899,
+      "step": 2049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698081561066173e-05,
+      "loss": 0.7974,
+      "step": 2050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697672731085002e-05,
+      "loss": 0.8613,
+      "step": 2051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969726362873962e-05,
+      "loss": 0.7838,
+      "step": 2052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696854254041514e-05,
+      "loss": 0.853,
+      "step": 2053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696444607002184e-05,
+      "loss": 0.8271,
+      "step": 2054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696034687633133e-05,
+      "loss": 0.8168,
+      "step": 2055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695624495945875e-05,
+      "loss": 0.834,
+      "step": 2056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695214031951932e-05,
+      "loss": 0.7526,
+      "step": 2057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969480329566283e-05,
+      "loss": 0.7807,
+      "step": 2058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694392287090102e-05,
+      "loss": 0.8307,
+      "step": 2059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693981006245292e-05,
+      "loss": 0.7563,
+      "step": 2060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969356945313996e-05,
+      "loss": 0.7645,
+      "step": 2061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693157627785654e-05,
+      "loss": 0.8704,
+      "step": 2062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692745530193943e-05,
+      "loss": 0.7565,
+      "step": 2063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692333160376407e-05,
+      "loss": 0.8191,
+      "step": 2064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969192051834462e-05,
+      "loss": 0.8123,
+      "step": 2065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691507604110175e-05,
+      "loss": 0.2729,
+      "step": 2066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691094417684668e-05,
+      "loss": 0.783,
+      "step": 2067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690680959079702e-05,
+      "loss": 0.7817,
+      "step": 2068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690267228306892e-05,
+      "loss": 0.8046,
+      "step": 2069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689853225377857e-05,
+      "loss": 0.7652,
+      "step": 2070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689438950304226e-05,
+      "loss": 0.8136,
+      "step": 2071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968902440309763e-05,
+      "loss": 0.8167,
+      "step": 2072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688609583769714e-05,
+      "loss": 0.8293,
+      "step": 2073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968819449233213e-05,
+      "loss": 0.8069,
+      "step": 2074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687779128796537e-05,
+      "loss": 0.8015,
+      "step": 2075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687363493174592e-05,
+      "loss": 0.7565,
+      "step": 2076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686947585477977e-05,
+      "loss": 0.8266,
+      "step": 2077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686531405718374e-05,
+      "loss": 0.7357,
+      "step": 2078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686114953907464e-05,
+      "loss": 0.7467,
+      "step": 2079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685698230056952e-05,
+      "loss": 0.797,
+      "step": 2080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685281234178536e-05,
+      "loss": 0.7916,
+      "step": 2081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684863966283928e-05,
+      "loss": 0.8287,
+      "step": 2082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684446426384846e-05,
+      "loss": 0.7967,
+      "step": 2083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968402861449302e-05,
+      "loss": 0.7924,
+      "step": 2084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683610530620188e-05,
+      "loss": 0.8193,
+      "step": 2085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683192174778086e-05,
+      "loss": 0.7784,
+      "step": 2086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968277354697846e-05,
+      "loss": 0.77,
+      "step": 2087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682354647233078e-05,
+      "loss": 0.7334,
+      "step": 2088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96819354755537e-05,
+      "loss": 0.7448,
+      "step": 2089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681516031952094e-05,
+      "loss": 0.7553,
+      "step": 2090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968109631644005e-05,
+      "loss": 0.8696,
+      "step": 2091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680676329029346e-05,
+      "loss": 0.7601,
+      "step": 2092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680256069731786e-05,
+      "loss": 0.8132,
+      "step": 2093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679835538559167e-05,
+      "loss": 0.8335,
+      "step": 2094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96794147355233e-05,
+      "loss": 0.8105,
+      "step": 2095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967899366063601e-05,
+      "loss": 0.6972,
+      "step": 2096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678572313909114e-05,
+      "loss": 0.7405,
+      "step": 2097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678150695354453e-05,
+      "loss": 0.9375,
+      "step": 2098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967772880498387e-05,
+      "loss": 0.7958,
+      "step": 2099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677306642809206e-05,
+      "loss": 0.8176,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676884208842323e-05,
+      "loss": 0.7868,
+      "step": 2101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967646150309508e-05,
+      "loss": 0.8405,
+      "step": 2102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676038525579356e-05,
+      "loss": 0.8391,
+      "step": 2103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675615276307026e-05,
+      "loss": 0.7521,
+      "step": 2104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967519175528998e-05,
+      "loss": 0.7181,
+      "step": 2105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674767962540113e-05,
+      "loss": 0.8309,
+      "step": 2106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967434389806932e-05,
+      "loss": 0.7551,
+      "step": 2107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673919561889524e-05,
+      "loss": 0.772,
+      "step": 2108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967349495401263e-05,
+      "loss": 0.7563,
+      "step": 2109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673070074450572e-05,
+      "loss": 0.8444,
+      "step": 2110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967264492321528e-05,
+      "loss": 0.8297,
+      "step": 2111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672219500318692e-05,
+      "loss": 0.7809,
+      "step": 2112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671793805772757e-05,
+      "loss": 0.7205,
+      "step": 2113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671367839589438e-05,
+      "loss": 0.7507,
+      "step": 2114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670941601780687e-05,
+      "loss": 0.8546,
+      "step": 2115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670515092358486e-05,
+      "loss": 0.8102,
+      "step": 2116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670088311334807e-05,
+      "loss": 0.7646,
+      "step": 2117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966966125872164e-05,
+      "loss": 0.7174,
+      "step": 2118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669233934530975e-05,
+      "loss": 0.7652,
+      "step": 2119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668806338774815e-05,
+      "loss": 0.7926,
+      "step": 2120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668378471465172e-05,
+      "loss": 0.6978,
+      "step": 2121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667950332614055e-05,
+      "loss": 0.7671,
+      "step": 2122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667521922233503e-05,
+      "loss": 0.7955,
+      "step": 2123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667093240335534e-05,
+      "loss": 0.7282,
+      "step": 2124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666664286932198e-05,
+      "loss": 0.7571,
+      "step": 2125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966623506203553e-05,
+      "loss": 0.7801,
+      "step": 2126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665805565657602e-05,
+      "loss": 0.8281,
+      "step": 2127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966537579781046e-05,
+      "loss": 0.8256,
+      "step": 2128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664945758506188e-05,
+      "loss": 0.7823,
+      "step": 2129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664515447756855e-05,
+      "loss": 0.7319,
+      "step": 2130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966408486557455e-05,
+      "loss": 0.7285,
+      "step": 2131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966365401197136e-05,
+      "loss": 0.84,
+      "step": 2132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663222886959396e-05,
+      "loss": 0.7329,
+      "step": 2133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662791490550755e-05,
+      "loss": 0.7251,
+      "step": 2134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662359822757567e-05,
+      "loss": 0.7312,
+      "step": 2135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661927883591946e-05,
+      "loss": 0.8127,
+      "step": 2136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966149567306602e-05,
+      "loss": 0.782,
+      "step": 2137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661063191191937e-05,
+      "loss": 0.7766,
+      "step": 2138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966063043798184e-05,
+      "loss": 0.7781,
+      "step": 2139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966019741344788e-05,
+      "loss": 0.7526,
+      "step": 2140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659764117602223e-05,
+      "loss": 0.7299,
+      "step": 2141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659330550457038e-05,
+      "loss": 0.7619,
+      "step": 2142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96588967120245e-05,
+      "loss": 0.8421,
+      "step": 2143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965846260231679e-05,
+      "loss": 0.7522,
+      "step": 2144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658028221346108e-05,
+      "loss": 0.7725,
+      "step": 2145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965759356912465e-05,
+      "loss": 0.8248,
+      "step": 2146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657158645664624e-05,
+      "loss": 0.7313,
+      "step": 2147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965672345097824e-05,
+      "loss": 0.8061,
+      "step": 2148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965628798507773e-05,
+      "loss": 0.775,
+      "step": 2149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655852247975317e-05,
+      "loss": 0.85,
+      "step": 2150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655416239683245e-05,
+      "loss": 0.799,
+      "step": 2151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654979960213754e-05,
+      "loss": 0.7833,
+      "step": 2152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96545434095791e-05,
+      "loss": 0.7057,
+      "step": 2153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965410658779154e-05,
+      "loss": 0.7759,
+      "step": 2154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653669494863346e-05,
+      "loss": 0.72,
+      "step": 2155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653232130806798e-05,
+      "loss": 0.7853,
+      "step": 2156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965279449563417e-05,
+      "loss": 0.7814,
+      "step": 2157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652356589357757e-05,
+      "loss": 0.7801,
+      "step": 2158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965191841198986e-05,
+      "loss": 0.7977,
+      "step": 2159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965147996354279e-05,
+      "loss": 0.805,
+      "step": 2160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651041244028848e-05,
+      "loss": 0.7865,
+      "step": 2161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650602253460366e-05,
+      "loss": 0.8251,
+      "step": 2162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650162991849672e-05,
+      "loss": 0.7698,
+      "step": 2163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96497234592091e-05,
+      "loss": 0.7487,
+      "step": 2164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649283655550994e-05,
+      "loss": 0.9306,
+      "step": 2165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964884358088771e-05,
+      "loss": 0.7794,
+      "step": 2166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96484032352316e-05,
+      "loss": 0.7782,
+      "step": 2167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964796261859504e-05,
+      "loss": 0.8605,
+      "step": 2168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647521730990408e-05,
+      "loss": 0.8,
+      "step": 2169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647080572430072e-05,
+      "loss": 0.8172,
+      "step": 2170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646639142926435e-05,
+      "loss": 0.7386,
+      "step": 2171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964619744249189e-05,
+      "loss": 0.8378,
+      "step": 2172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645755471138842e-05,
+      "loss": 0.794,
+      "step": 2173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964531322887971e-05,
+      "loss": 0.6994,
+      "step": 2174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96448707157269e-05,
+      "loss": 0.7874,
+      "step": 2175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644427931692855e-05,
+      "loss": 0.8295,
+      "step": 2176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643984876790004e-05,
+      "loss": 0.8229,
+      "step": 2177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964354155103079e-05,
+      "loss": 0.7677,
+      "step": 2178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964309795442767e-05,
+      "loss": 0.7841,
+      "step": 2179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642654086993098e-05,
+      "loss": 0.8073,
+      "step": 2180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964220994873954e-05,
+      "loss": 0.7776,
+      "step": 2181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641765539679466e-05,
+      "loss": 0.8448,
+      "step": 2182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641320859825367e-05,
+      "loss": 0.3107,
+      "step": 2183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640875909189728e-05,
+      "loss": 0.7644,
+      "step": 2184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640430687785042e-05,
+      "loss": 0.7425,
+      "step": 2185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639985195623815e-05,
+      "loss": 0.86,
+      "step": 2186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639539432718566e-05,
+      "loss": 0.7782,
+      "step": 2187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96390933990818e-05,
+      "loss": 0.7853,
+      "step": 2188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638647094726057e-05,
+      "loss": 0.7798,
+      "step": 2189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638200519663866e-05,
+      "loss": 0.8474,
+      "step": 2190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963775367390777e-05,
+      "loss": 0.7697,
+      "step": 2191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637306557470325e-05,
+      "loss": 0.7333,
+      "step": 2192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963685917036408e-05,
+      "loss": 0.7978,
+      "step": 2193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96364115126016e-05,
+      "loss": 0.723,
+      "step": 2194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963596358419546e-05,
+      "loss": 0.8206,
+      "step": 2195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635515385158244e-05,
+      "loss": 0.8487,
+      "step": 2196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963506691550254e-05,
+      "loss": 0.8292,
+      "step": 2197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634618175240935e-05,
+      "loss": 0.8349,
+      "step": 2198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963416916438604e-05,
+      "loss": 0.7476,
+      "step": 2199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633719882950458e-05,
+      "loss": 0.8112,
+      "step": 2200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963327033094682e-05,
+      "loss": 0.8339,
+      "step": 2201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632820508387743e-05,
+      "loss": 0.7733,
+      "step": 2202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963237041528586e-05,
+      "loss": 0.817,
+      "step": 2203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631920051653813e-05,
+      "loss": 0.7462,
+      "step": 2204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631469417504253e-05,
+      "loss": 0.8171,
+      "step": 2205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631018512849835e-05,
+      "loss": 0.7927,
+      "step": 2206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630567337703225e-05,
+      "loss": 0.847,
+      "step": 2207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963011589207709e-05,
+      "loss": 0.861,
+      "step": 2208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962966417598411e-05,
+      "loss": 0.7636,
+      "step": 2209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962921218943698e-05,
+      "loss": 0.8086,
+      "step": 2210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962875993244838e-05,
+      "loss": 0.8328,
+      "step": 2211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962830740503102e-05,
+      "loss": 0.8022,
+      "step": 2212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962785460719761e-05,
+      "loss": 0.7181,
+      "step": 2213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627401538960868e-05,
+      "loss": 0.7889,
+      "step": 2214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626948200333515e-05,
+      "loss": 0.9152,
+      "step": 2215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626494591328286e-05,
+      "loss": 0.7759,
+      "step": 2216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962604071195792e-05,
+      "loss": 0.687,
+      "step": 2217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962558656223516e-05,
+      "loss": 0.8474,
+      "step": 2218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962513214217277e-05,
+      "loss": 0.7709,
+      "step": 2219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624677451783504e-05,
+      "loss": 0.7791,
+      "step": 2220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962422249108014e-05,
+      "loss": 0.745,
+      "step": 2221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962376726007545e-05,
+      "loss": 0.829,
+      "step": 2222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623311758782217e-05,
+      "loss": 0.7273,
+      "step": 2223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622855987213243e-05,
+      "loss": 0.7573,
+      "step": 2224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962239994538132e-05,
+      "loss": 0.6957,
+      "step": 2225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621943633299264e-05,
+      "loss": 0.7704,
+      "step": 2226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621487050979884e-05,
+      "loss": 0.3082,
+      "step": 2227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621030198436007e-05,
+      "loss": 0.8212,
+      "step": 2228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620573075680462e-05,
+      "loss": 0.8097,
+      "step": 2229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620115682726092e-05,
+      "loss": 0.7811,
+      "step": 2230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619658019585736e-05,
+      "loss": 0.8815,
+      "step": 2231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961920008627225e-05,
+      "loss": 0.8946,
+      "step": 2232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96187418827985e-05,
+      "loss": 0.8943,
+      "step": 2233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961828340917735e-05,
+      "loss": 0.7892,
+      "step": 2234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961782466542168e-05,
+      "loss": 0.7608,
+      "step": 2235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617365651544367e-05,
+      "loss": 0.779,
+      "step": 2236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616906367558314e-05,
+      "loss": 0.7706,
+      "step": 2237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961644681347641e-05,
+      "loss": 0.776,
+      "step": 2238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615986989311567e-05,
+      "loss": 0.7063,
+      "step": 2239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615526895076697e-05,
+      "loss": 0.7984,
+      "step": 2240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615066530784724e-05,
+      "loss": 0.8423,
+      "step": 2241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961460589644858e-05,
+      "loss": 0.8044,
+      "step": 2242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614144992081193e-05,
+      "loss": 0.7424,
+      "step": 2243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613683817695516e-05,
+      "loss": 0.8476,
+      "step": 2244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613222373304497e-05,
+      "loss": 0.7761,
+      "step": 2245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96127606589211e-05,
+      "loss": 0.7589,
+      "step": 2246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961229867455829e-05,
+      "loss": 0.8213,
+      "step": 2247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611836420229043e-05,
+      "loss": 0.8006,
+      "step": 2248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611373895946338e-05,
+      "loss": 0.7648,
+      "step": 2249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610911101723174e-05,
+      "loss": 0.7806,
+      "step": 2250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610448037572537e-05,
+      "loss": 0.7481,
+      "step": 2251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609984703507443e-05,
+      "loss": 0.7808,
+      "step": 2252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96095210995409e-05,
+      "loss": 0.806,
+      "step": 2253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609057225685926e-05,
+      "loss": 0.7653,
+      "step": 2254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608593081955555e-05,
+      "loss": 0.7402,
+      "step": 2255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960812866836282e-05,
+      "loss": 0.7887,
+      "step": 2256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960766398492076e-05,
+      "loss": 0.8685,
+      "step": 2257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607199031642435e-05,
+      "loss": 0.7173,
+      "step": 2258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606733808540897e-05,
+      "loss": 0.7501,
+      "step": 2259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960626831562921e-05,
+      "loss": 0.7825,
+      "step": 2260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605802552920455e-05,
+      "loss": 0.7953,
+      "step": 2261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960533652042771e-05,
+      "loss": 0.7959,
+      "step": 2262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604870218164063e-05,
+      "loss": 0.7904,
+      "step": 2263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604403646142607e-05,
+      "loss": 0.7476,
+      "step": 2264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960393680437645e-05,
+      "loss": 0.8677,
+      "step": 2265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603469692878705e-05,
+      "loss": 0.7827,
+      "step": 2266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960300231166249e-05,
+      "loss": 0.8248,
+      "step": 2267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960253466074093e-05,
+      "loss": 0.7562,
+      "step": 2268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602066740127157e-05,
+      "loss": 0.8057,
+      "step": 2269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601598549834318e-05,
+      "loss": 0.812,
+      "step": 2270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960113008987556e-05,
+      "loss": 0.8208,
+      "step": 2271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600661360264038e-05,
+      "loss": 0.7916,
+      "step": 2272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960019236101292e-05,
+      "loss": 0.7912,
+      "step": 2273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599723092135376e-05,
+      "loss": 0.3049,
+      "step": 2274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599253553644585e-05,
+      "loss": 0.7921,
+      "step": 2275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598783745553738e-05,
+      "loss": 0.8353,
+      "step": 2276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598313667876027e-05,
+      "loss": 0.7626,
+      "step": 2277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597843320624652e-05,
+      "loss": 0.8591,
+      "step": 2278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959737270381283e-05,
+      "loss": 0.8423,
+      "step": 2279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959690181745377e-05,
+      "loss": 0.7932,
+      "step": 2280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596430661560704e-05,
+      "loss": 0.8286,
+      "step": 2281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959595923614686e-05,
+      "loss": 0.7785,
+      "step": 2282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959548754122548e-05,
+      "loss": 0.7544,
+      "step": 2283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595015576809812e-05,
+      "loss": 0.7243,
+      "step": 2284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959454334291311e-05,
+      "loss": 0.8005,
+      "step": 2285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959407083954864e-05,
+      "loss": 0.7827,
+      "step": 2286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593598066729672e-05,
+      "loss": 0.749,
+      "step": 2287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959312502446948e-05,
+      "loss": 0.7601,
+      "step": 2288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959265171278135e-05,
+      "loss": 0.7123,
+      "step": 2289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959217813167858e-05,
+      "loss": 0.8875,
+      "step": 2290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591704281174472e-05,
+      "loss": 0.7579,
+      "step": 2291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959123016128233e-05,
+      "loss": 0.7585,
+      "step": 2292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590755772015467e-05,
+      "loss": 0.8,
+      "step": 2293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590281113387214e-05,
+      "loss": 0.7132,
+      "step": 2294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589806185410897e-05,
+      "loss": 0.8224,
+      "step": 2295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958933098809985e-05,
+      "loss": 0.8048,
+      "step": 2296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588855521467433e-05,
+      "loss": 0.7963,
+      "step": 2297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588379785526984e-05,
+      "loss": 0.3065,
+      "step": 2298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587903780291878e-05,
+      "loss": 0.8277,
+      "step": 2299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587427505775474e-05,
+      "loss": 0.8325,
+      "step": 2300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586950961991155e-05,
+      "loss": 0.8129,
+      "step": 2301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95864741489523e-05,
+      "loss": 0.7639,
+      "step": 2302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585997066672305e-05,
+      "loss": 0.7957,
+      "step": 2303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585519715164564e-05,
+      "loss": 0.7967,
+      "step": 2304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585042094442488e-05,
+      "loss": 0.7625,
+      "step": 2305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958456420451949e-05,
+      "loss": 0.8301,
+      "step": 2306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584086045408992e-05,
+      "loss": 0.6748,
+      "step": 2307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958360761712442e-05,
+      "loss": 0.7894,
+      "step": 2308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583128919679218e-05,
+      "loss": 0.7252,
+      "step": 2309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958264995308682e-05,
+      "loss": 0.9199,
+      "step": 2310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582170717360692e-05,
+      "loss": 0.7886,
+      "step": 2311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958169121251428e-05,
+      "loss": 0.772,
+      "step": 2312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581211438561058e-05,
+      "loss": 0.7386,
+      "step": 2313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95807313955145e-05,
+      "loss": 0.7577,
+      "step": 2314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958025108338809e-05,
+      "loss": 0.815,
+      "step": 2315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579770502195314e-05,
+      "loss": 0.7958,
+      "step": 2316
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579289651949673e-05,
+      "loss": 0.7377,
+      "step": 2317
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578808532664663e-05,
+      "loss": 0.7504,
+      "step": 2318
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957832714435381e-05,
+      "loss": 0.7303,
+      "step": 2319
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957784548703063e-05,
+      "loss": 0.7927,
+      "step": 2320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577363560708644e-05,
+      "loss": 0.7495,
+      "step": 2321
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957688136540139e-05,
+      "loss": 0.7597,
+      "step": 2322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576398901122414e-05,
+      "loss": 0.8456,
+      "step": 2323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575916167885265e-05,
+      "loss": 0.8262,
+      "step": 2324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95754331657035e-05,
+      "loss": 0.699,
+      "step": 2325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957494989459068e-05,
+      "loss": 0.3116,
+      "step": 2326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574466354560386e-05,
+      "loss": 0.7693,
+      "step": 2327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573982545626197e-05,
+      "loss": 0.8118,
+      "step": 2328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573498467801697e-05,
+      "loss": 0.7797,
+      "step": 2329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573014121100482e-05,
+      "loss": 0.7278,
+      "step": 2330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957252950553616e-05,
+      "loss": 0.8098,
+      "step": 2331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572044621122334e-05,
+      "loss": 0.869,
+      "step": 2332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571559467872633e-05,
+      "loss": 0.8008,
+      "step": 2333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571074045800673e-05,
+      "loss": 0.799,
+      "step": 2334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570588354920085e-05,
+      "loss": 0.8479,
+      "step": 2335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570102395244523e-05,
+      "loss": 0.7475,
+      "step": 2336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569616166787625e-05,
+      "loss": 0.8125,
+      "step": 2337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956912966956305e-05,
+      "loss": 0.8067,
+      "step": 2338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568642903584464e-05,
+      "loss": 0.8036,
+      "step": 2339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956815586886553e-05,
+      "loss": 0.8086,
+      "step": 2340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567668565419935e-05,
+      "loss": 0.7592,
+      "step": 2341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567180993261364e-05,
+      "loss": 0.7603,
+      "step": 2342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566693152403505e-05,
+      "loss": 0.671,
+      "step": 2343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956620504286007e-05,
+      "loss": 0.7973,
+      "step": 2344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565716664644756e-05,
+      "loss": 0.7687,
+      "step": 2345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565228017771283e-05,
+      "loss": 0.7763,
+      "step": 2346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564739102253383e-05,
+      "loss": 0.854,
+      "step": 2347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564249918104777e-05,
+      "loss": 0.741,
+      "step": 2348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563760465339208e-05,
+      "loss": 0.7276,
+      "step": 2349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956327074397042e-05,
+      "loss": 0.8923,
+      "step": 2350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562780754012172e-05,
+      "loss": 0.7237,
+      "step": 2351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562290495478224e-05,
+      "loss": 0.7221,
+      "step": 2352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561799968382343e-05,
+      "loss": 0.8417,
+      "step": 2353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956130917273831e-05,
+      "loss": 0.7514,
+      "step": 2354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95608181085599e-05,
+      "loss": 0.8093,
+      "step": 2355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560326775860913e-05,
+      "loss": 0.7926,
+      "step": 2356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955983517465515e-05,
+      "loss": 0.7675,
+      "step": 2357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955934330495641e-05,
+      "loss": 0.7793,
+      "step": 2358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558851166778513e-05,
+      "loss": 0.7816,
+      "step": 2359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558358760135282e-05,
+      "loss": 0.7871,
+      "step": 2360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955786608504054e-05,
+      "loss": 0.7583,
+      "step": 2361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557373141508133e-05,
+      "loss": 0.6841,
+      "step": 2362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556879929551898e-05,
+      "loss": 0.8123,
+      "step": 2363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556386449185688e-05,
+      "loss": 0.766,
+      "step": 2364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555892700423367e-05,
+      "loss": 0.2924,
+      "step": 2365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95553986832788e-05,
+      "loss": 0.7154,
+      "step": 2366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554904397765858e-05,
+      "loss": 0.7204,
+      "step": 2367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955440984389843e-05,
+      "loss": 0.7794,
+      "step": 2368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553915021690403e-05,
+      "loss": 0.7925,
+      "step": 2369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955341993115567e-05,
+      "loss": 0.7623,
+      "step": 2370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552924572308144e-05,
+      "loss": 0.7385,
+      "step": 2371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955242894516173e-05,
+      "loss": 0.7727,
+      "step": 2372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551933049730353e-05,
+      "loss": 0.8469,
+      "step": 2373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551436886027938e-05,
+      "loss": 0.7561,
+      "step": 2374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955094045406842e-05,
+      "loss": 0.8328,
+      "step": 2375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550443753865743e-05,
+      "loss": 0.7612,
+      "step": 2376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954994678543385e-05,
+      "loss": 0.7623,
+      "step": 2377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549449548786718e-05,
+      "loss": 0.8179,
+      "step": 2378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548952043938286e-05,
+      "loss": 0.8488,
+      "step": 2379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548454270902545e-05,
+      "loss": 0.8322,
+      "step": 2380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954795622969347e-05,
+      "loss": 0.8053,
+      "step": 2381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954745792032505e-05,
+      "loss": 0.8499,
+      "step": 2382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954695934281128e-05,
+      "loss": 0.8448,
+      "step": 2383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954646049716616e-05,
+      "loss": 0.7892,
+      "step": 2384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95459613834037e-05,
+      "loss": 0.7585,
+      "step": 2385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545462001537918e-05,
+      "loss": 0.7731,
+      "step": 2386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544962351582844e-05,
+      "loss": 0.7835,
+      "step": 2387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544462433552508e-05,
+      "loss": 0.7911,
+      "step": 2388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543962247460954e-05,
+      "loss": 0.7471,
+      "step": 2389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543461793322222e-05,
+      "loss": 0.6915,
+      "step": 2390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542961071150374e-05,
+      "loss": 0.8318,
+      "step": 2391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542460080959472e-05,
+      "loss": 0.7954,
+      "step": 2392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541958822763585e-05,
+      "loss": 0.8581,
+      "step": 2393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541457296576794e-05,
+      "loss": 0.7299,
+      "step": 2394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954095550241318e-05,
+      "loss": 0.8806,
+      "step": 2395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540453440286837e-05,
+      "loss": 0.8031,
+      "step": 2396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953995111021187e-05,
+      "loss": 0.831,
+      "step": 2397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539448512202384e-05,
+      "loss": 0.7886,
+      "step": 2398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538945646272495e-05,
+      "loss": 0.7461,
+      "step": 2399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953844251243633e-05,
+      "loss": 0.7762,
+      "step": 2400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537939110708014e-05,
+      "loss": 0.8232,
+      "step": 2401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537435441101683e-05,
+      "loss": 0.7878,
+      "step": 2402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536931503631493e-05,
+      "loss": 0.7617,
+      "step": 2403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536427298311592e-05,
+      "loss": 0.7309,
+      "step": 2404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535922825156137e-05,
+      "loss": 0.7747,
+      "step": 2405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535418084179302e-05,
+      "loss": 0.7891,
+      "step": 2406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953491307539526e-05,
+      "loss": 0.7758,
+      "step": 2407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95344077988182e-05,
+      "loss": 0.7677,
+      "step": 2408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533902254462306e-05,
+      "loss": 0.806,
+      "step": 2409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533396442341775e-05,
+      "loss": 0.7571,
+      "step": 2410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953289036247082e-05,
+      "loss": 0.7472,
+      "step": 2411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953238401486365e-05,
+      "loss": 0.7818,
+      "step": 2412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953187739953449e-05,
+      "loss": 0.804,
+      "step": 2413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531370516497562e-05,
+      "loss": 0.7836,
+      "step": 2414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953086336576711e-05,
+      "loss": 0.7212,
+      "step": 2415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953035594735737e-05,
+      "loss": 0.8334,
+      "step": 2416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529848261282602e-05,
+      "loss": 0.7828,
+      "step": 2417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529340307557056e-05,
+      "loss": 0.7506,
+      "step": 2418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528832086195e-05,
+      "loss": 0.7677,
+      "step": 2419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952832359721071e-05,
+      "loss": 0.7728,
+      "step": 2420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527814840618466e-05,
+      "loss": 0.8037,
+      "step": 2421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952730581643256e-05,
+      "loss": 0.7964,
+      "step": 2422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952679652466728e-05,
+      "loss": 0.7836,
+      "step": 2423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526286965336943e-05,
+      "loss": 0.7954,
+      "step": 2424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525777138455845e-05,
+      "loss": 0.7859,
+      "step": 2425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525267044038318e-05,
+      "loss": 0.807,
+      "step": 2426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524756682098678e-05,
+      "loss": 0.8356,
+      "step": 2427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524246052651264e-05,
+      "loss": 0.738,
+      "step": 2428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523735155710416e-05,
+      "loss": 0.8845,
+      "step": 2429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523223991290484e-05,
+      "loss": 0.8005,
+      "step": 2430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522712559405822e-05,
+      "loss": 0.77,
+      "step": 2431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522200860070794e-05,
+      "loss": 0.7543,
+      "step": 2432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521688893299778e-05,
+      "loss": 0.7858,
+      "step": 2433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952117665910714e-05,
+      "loss": 0.8141,
+      "step": 2434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952066415750728e-05,
+      "loss": 0.834,
+      "step": 2435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520151388514585e-05,
+      "loss": 0.7798,
+      "step": 2436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519638352143454e-05,
+      "loss": 0.8557,
+      "step": 2437
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95191250484083e-05,
+      "loss": 0.7022,
+      "step": 2438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951861147732354e-05,
+      "loss": 0.7811,
+      "step": 2439
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95180976389036e-05,
+      "loss": 0.7978,
+      "step": 2440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95175835331629e-05,
+      "loss": 0.8096,
+      "step": 2441
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951706916011589e-05,
+      "loss": 0.7441,
+      "step": 2442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516554519777014e-05,
+      "loss": 0.7604,
+      "step": 2443
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516039612160727e-05,
+      "loss": 0.7311,
+      "step": 2444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515524437281486e-05,
+      "loss": 0.8182,
+      "step": 2445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515008995153762e-05,
+      "loss": 0.8589,
+      "step": 2446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514493285792032e-05,
+      "loss": 0.735,
+      "step": 2447
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951397730921078e-05,
+      "loss": 0.8159,
+      "step": 2448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95134610654245e-05,
+      "loss": 0.777,
+      "step": 2449
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951294455444768e-05,
+      "loss": 0.7325,
+      "step": 2450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951242777629484e-05,
+      "loss": 0.8026,
+      "step": 2451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951191073098049e-05,
+      "loss": 0.8036,
+      "step": 2452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511393418519154e-05,
+      "loss": 0.7438,
+      "step": 2453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510875838925353e-05,
+      "loss": 0.7371,
+      "step": 2454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510357992213626e-05,
+      "loss": 0.7374,
+      "step": 2455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950983987839852e-05,
+      "loss": 0.7865,
+      "step": 2456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509321497494587e-05,
+      "loss": 0.7963,
+      "step": 2457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508802849516384e-05,
+      "loss": 0.7103,
+      "step": 2458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950828393447848e-05,
+      "loss": 0.7897,
+      "step": 2459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950776475239544e-05,
+      "loss": 0.8351,
+      "step": 2460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507245303281856e-05,
+      "loss": 0.8026,
+      "step": 2461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506725587152315e-05,
+      "loss": 0.8245,
+      "step": 2462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506205604021407e-05,
+      "loss": 0.7703,
+      "step": 2463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505685353903746e-05,
+      "loss": 0.7985,
+      "step": 2464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505164836813938e-05,
+      "loss": 0.8346,
+      "step": 2465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95046440527666e-05,
+      "loss": 0.7888,
+      "step": 2466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504123001776358e-05,
+      "loss": 0.7691,
+      "step": 2467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503601683857854e-05,
+      "loss": 0.8006,
+      "step": 2468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503080099025722e-05,
+      "loss": 0.7709,
+      "step": 2469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502558247294613e-05,
+      "loss": 0.8593,
+      "step": 2470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502036128679184e-05,
+      "loss": 0.8064,
+      "step": 2471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95015137431941e-05,
+      "loss": 0.7573,
+      "step": 2472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950099109085403e-05,
+      "loss": 0.8286,
+      "step": 2473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500468171673653e-05,
+      "loss": 0.7627,
+      "step": 2474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499944985667656e-05,
+      "loss": 0.849,
+      "step": 2475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499421532850737e-05,
+      "loss": 0.7977,
+      "step": 2476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949889781323759e-05,
+      "loss": 0.7698,
+      "step": 2477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949837382684293e-05,
+      "loss": 0.8499,
+      "step": 2478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497849573681472e-05,
+      "loss": 0.7404,
+      "step": 2479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949732505376794e-05,
+      "loss": 0.7259,
+      "step": 2480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949680026711706e-05,
+      "loss": 0.7512,
+      "step": 2481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949627521374358e-05,
+      "loss": 0.7399,
+      "step": 2482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949574989366224e-05,
+      "loss": 0.7797,
+      "step": 2483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495224306887797e-05,
+      "loss": 0.7672,
+      "step": 2484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949469845343501e-05,
+      "loss": 0.836,
+      "step": 2485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949417233331865e-05,
+      "loss": 0.7926,
+      "step": 2486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493645946553493e-05,
+      "loss": 0.8281,
+      "step": 2487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493119293154322e-05,
+      "loss": 0.8703,
+      "step": 2488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949259237313593e-05,
+      "loss": 0.8238,
+      "step": 2489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949206518651311e-05,
+      "loss": 0.6877,
+      "step": 2490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491537733300677e-05,
+      "loss": 0.6777,
+      "step": 2491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949101001351344e-05,
+      "loss": 0.8023,
+      "step": 2492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949048202716622e-05,
+      "loss": 0.7926,
+      "step": 2493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489953774273847e-05,
+      "loss": 0.7033,
+      "step": 2494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948942525485116e-05,
+      "loss": 0.8308,
+      "step": 2495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488896468912992e-05,
+      "loss": 0.8508,
+      "step": 2496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948836741647421e-05,
+      "loss": 0.8009,
+      "step": 2497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487838097549658e-05,
+      "loss": 0.7342,
+      "step": 2498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487308512154213e-05,
+      "loss": 0.7451,
+      "step": 2499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486778660302743e-05,
+      "loss": 0.7681,
+      "step": 2500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948624854201013e-05,
+      "loss": 0.7962,
+      "step": 2501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485718157291265e-05,
+      "loss": 0.7636,
+      "step": 2502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948518750616104e-05,
+      "loss": 0.696,
+      "step": 2503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484656588634362e-05,
+      "loss": 0.8674,
+      "step": 2504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484125404726142e-05,
+      "loss": 0.7651,
+      "step": 2505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483593954451296e-05,
+      "loss": 0.759,
+      "step": 2506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483062237824753e-05,
+      "loss": 0.847,
+      "step": 2507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482530254861447e-05,
+      "loss": 0.818,
+      "step": 2508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481998005576318e-05,
+      "loss": 0.7982,
+      "step": 2509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481465489984312e-05,
+      "loss": 0.7698,
+      "step": 2510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480932708100387e-05,
+      "loss": 0.8398,
+      "step": 2511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480399659939505e-05,
+      "loss": 0.8345,
+      "step": 2512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479866345516638e-05,
+      "loss": 0.8028,
+      "step": 2513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479332764846766e-05,
+      "loss": 0.8188,
+      "step": 2514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478798917944874e-05,
+      "loss": 0.7523,
+      "step": 2515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478264804825954e-05,
+      "loss": 0.8238,
+      "step": 2516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947773042550501e-05,
+      "loss": 0.7746,
+      "step": 2517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477195779997042e-05,
+      "loss": 0.7591,
+      "step": 2518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476660868317076e-05,
+      "loss": 0.7642,
+      "step": 2519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476125690480133e-05,
+      "loss": 0.7762,
+      "step": 2520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475590246501238e-05,
+      "loss": 0.8073,
+      "step": 2521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475054536395435e-05,
+      "loss": 0.7466,
+      "step": 2522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947451856017777e-05,
+      "loss": 0.7009,
+      "step": 2523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947398231786329e-05,
+      "loss": 0.7611,
+      "step": 2524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473445809467062e-05,
+      "loss": 0.8216,
+      "step": 2525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472909035004147e-05,
+      "loss": 0.7676,
+      "step": 2526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947237199448963e-05,
+      "loss": 0.776,
+      "step": 2527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471834687938593e-05,
+      "loss": 0.7343,
+      "step": 2528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471297115366114e-05,
+      "loss": 0.8397,
+      "step": 2529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947075927678731e-05,
+      "loss": 0.7909,
+      "step": 2530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470221172217268e-05,
+      "loss": 0.7479,
+      "step": 2531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469682801671113e-05,
+      "loss": 0.3004,
+      "step": 2532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946914416516396e-05,
+      "loss": 0.7314,
+      "step": 2533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946860526271094e-05,
+      "loss": 0.3009,
+      "step": 2534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468066094327186e-05,
+      "loss": 0.7765,
+      "step": 2535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467526660027843e-05,
+      "loss": 0.7401,
+      "step": 2536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466986959828063e-05,
+      "loss": 0.733,
+      "step": 2537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466446993742997e-05,
+      "loss": 0.7313,
+      "step": 2538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465906761787815e-05,
+      "loss": 0.8312,
+      "step": 2539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465366263977688e-05,
+      "loss": 0.7416,
+      "step": 2540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94648255003278e-05,
+      "loss": 0.7719,
+      "step": 2541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946428447085333e-05,
+      "loss": 0.7682,
+      "step": 2542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946374317556948e-05,
+      "loss": 0.8271,
+      "step": 2543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946320161449146e-05,
+      "loss": 0.7001,
+      "step": 2544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946265978763446e-05,
+      "loss": 0.7616,
+      "step": 2545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462117695013716e-05,
+      "loss": 0.7693,
+      "step": 2546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461575336644444e-05,
+      "loss": 0.8112,
+      "step": 2547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461032712541876e-05,
+      "loss": 0.7681,
+      "step": 2548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460489822721256e-05,
+      "loss": 0.7325,
+      "step": 2549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945994666719783e-05,
+      "loss": 0.7786,
+      "step": 2550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945940324598685e-05,
+      "loss": 0.7049,
+      "step": 2551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945885955910358e-05,
+      "loss": 0.838,
+      "step": 2552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458315606563293e-05,
+      "loss": 0.7913,
+      "step": 2553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945777138838126e-05,
+      "loss": 0.8,
+      "step": 2554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945722690457277e-05,
+      "loss": 0.7878,
+      "step": 2555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945668215515312e-05,
+      "loss": 0.8222,
+      "step": 2556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94561371401376e-05,
+      "loss": 0.7673,
+      "step": 2557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945559185954152e-05,
+      "loss": 0.7902,
+      "step": 2558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455046313380193e-05,
+      "loss": 0.7671,
+      "step": 2559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454500501668946e-05,
+      "loss": 0.8389,
+      "step": 2560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453954424423105e-05,
+      "loss": 0.3276,
+      "step": 2561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945340808165801e-05,
+      "loss": 0.7794,
+      "step": 2562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452861473389e-05,
+      "loss": 0.769,
+      "step": 2563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945231459963143e-05,
+      "loss": 0.8011,
+      "step": 2564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451767460400658e-05,
+      "loss": 0.7101,
+      "step": 2565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451220055712053e-05,
+      "loss": 0.802,
+      "step": 2566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450672385580985e-05,
+      "loss": 0.7969,
+      "step": 2567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450124450022843e-05,
+      "loss": 0.8029,
+      "step": 2568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449576249053008e-05,
+      "loss": 0.8272,
+      "step": 2569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449027782686883e-05,
+      "loss": 0.7332,
+      "step": 2570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448479050939865e-05,
+      "loss": 0.7765,
+      "step": 2571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447930053827368e-05,
+      "loss": 0.7881,
+      "step": 2572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447380791364815e-05,
+      "loss": 0.7579,
+      "step": 2573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944683126356763e-05,
+      "loss": 0.8085,
+      "step": 2574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446281470451245e-05,
+      "loss": 0.765,
+      "step": 2575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445731412031104e-05,
+      "loss": 0.8023,
+      "step": 2576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445181088322652e-05,
+      "loss": 0.7496,
+      "step": 2577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444630499341347e-05,
+      "loss": 0.752,
+      "step": 2578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444079645102656e-05,
+      "loss": 0.7779,
+      "step": 2579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443528525622044e-05,
+      "loss": 0.7861,
+      "step": 2580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944297714091499e-05,
+      "loss": 0.8286,
+      "step": 2581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442425490996987e-05,
+      "loss": 0.8393,
+      "step": 2582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441873575883522e-05,
+      "loss": 0.8644,
+      "step": 2583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94413213955901e-05,
+      "loss": 0.7914,
+      "step": 2584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440768950132223e-05,
+      "loss": 0.787,
+      "step": 2585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944021623952541e-05,
+      "loss": 0.6792,
+      "step": 2586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439663263785187e-05,
+      "loss": 0.7859,
+      "step": 2587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943911002292708e-05,
+      "loss": 0.8023,
+      "step": 2588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943855651696663e-05,
+      "loss": 0.7512,
+      "step": 2589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438002745919383e-05,
+      "loss": 0.6982,
+      "step": 2590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943744870980089e-05,
+      "loss": 0.7915,
+      "step": 2591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436894408626712e-05,
+      "loss": 0.7469,
+      "step": 2592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943633984241242e-05,
+      "loss": 0.8182,
+      "step": 2593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943578501117358e-05,
+      "loss": 0.8436,
+      "step": 2594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435229914925785e-05,
+      "loss": 0.7628,
+      "step": 2595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434674553684622e-05,
+      "loss": 0.8338,
+      "step": 2596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434118927465685e-05,
+      "loss": 0.7233,
+      "step": 2597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433563036284583e-05,
+      "loss": 0.7249,
+      "step": 2598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433006880156928e-05,
+      "loss": 0.8128,
+      "step": 2599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432450459098336e-05,
+      "loss": 0.7533,
+      "step": 2600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943189377312444e-05,
+      "loss": 0.8,
+      "step": 2601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431336822250874e-05,
+      "loss": 0.7991,
+      "step": 2602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430779606493276e-05,
+      "loss": 0.8074,
+      "step": 2603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430222125867303e-05,
+      "loss": 0.6928,
+      "step": 2604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94296643803886e-05,
+      "loss": 0.7872,
+      "step": 2605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429106370072845e-05,
+      "loss": 0.7748,
+      "step": 2606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428548094935703e-05,
+      "loss": 0.8133,
+      "step": 2607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9427989554992854e-05,
+      "loss": 0.7517,
+      "step": 2608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9427430750259987e-05,
+      "loss": 0.8107,
+      "step": 2609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942687168075279e-05,
+      "loss": 0.7544,
+      "step": 2610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426312346486972e-05,
+      "loss": 0.7224,
+      "step": 2611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425752747478243e-05,
+      "loss": 0.7585,
+      "step": 2612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942519288374231e-05,
+      "loss": 0.838,
+      "step": 2613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942463275529491e-05,
+      "loss": 0.8861,
+      "step": 2614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942407236215176e-05,
+      "loss": 0.7974,
+      "step": 2615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423511704328615e-05,
+      "loss": 0.8309,
+      "step": 2616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422950781841212e-05,
+      "loss": 0.8193,
+      "step": 2617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94223895947053e-05,
+      "loss": 0.816,
+      "step": 2618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421828142936655e-05,
+      "loss": 0.7571,
+      "step": 2619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942126642655103e-05,
+      "loss": 0.7791,
+      "step": 2620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942070444556421e-05,
+      "loss": 0.7939,
+      "step": 2621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420142199991976e-05,
+      "loss": 0.6983,
+      "step": 2622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941957968985012e-05,
+      "loss": 0.782,
+      "step": 2623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941901691515444e-05,
+      "loss": 0.7706,
+      "step": 2624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418453875920744e-05,
+      "loss": 0.7135,
+      "step": 2625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941789057216484e-05,
+      "loss": 0.7336,
+      "step": 2626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417327003902554e-05,
+      "loss": 0.8193,
+      "step": 2627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416763171149708e-05,
+      "loss": 0.7613,
+      "step": 2628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416199073922144e-05,
+      "loss": 0.7884,
+      "step": 2629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94156347122357e-05,
+      "loss": 0.8665,
+      "step": 2630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941507008610623e-05,
+      "loss": 0.8301,
+      "step": 2631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414505195549593e-05,
+      "loss": 0.8276,
+      "step": 2632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941394004058165e-05,
+      "loss": 0.61,
+      "step": 2633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413374621218275e-05,
+      "loss": 0.7981,
+      "step": 2634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412808937475348e-05,
+      "loss": 0.7183,
+      "step": 2635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941224298936876e-05,
+      "loss": 0.7642,
+      "step": 2636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411676776914398e-05,
+      "loss": 0.7496,
+      "step": 2637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941111030012817e-05,
+      "loss": 0.7776,
+      "step": 2638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410543559025988e-05,
+      "loss": 0.7974,
+      "step": 2639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409976553623767e-05,
+      "loss": 0.7397,
+      "step": 2640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409409283937427e-05,
+      "loss": 0.8005,
+      "step": 2641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940884174998291e-05,
+      "loss": 0.7622,
+      "step": 2642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408273951776145e-05,
+      "loss": 0.7714,
+      "step": 2643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407705889333082e-05,
+      "loss": 0.7367,
+      "step": 2644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940713756266968e-05,
+      "loss": 0.8401,
+      "step": 2645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406568971801897e-05,
+      "loss": 0.8002,
+      "step": 2646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94060001167457e-05,
+      "loss": 0.6996,
+      "step": 2647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405430997517072e-05,
+      "loss": 0.7635,
+      "step": 2648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404861614131993e-05,
+      "loss": 0.8162,
+      "step": 2649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404291966606456e-05,
+      "loss": 0.7157,
+      "step": 2650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403722054956455e-05,
+      "loss": 0.8606,
+      "step": 2651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403151879198e-05,
+      "loss": 0.8022,
+      "step": 2652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402581439347107e-05,
+      "loss": 0.777,
+      "step": 2653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402010735419794e-05,
+      "loss": 0.7643,
+      "step": 2654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940143976743209e-05,
+      "loss": 0.7798,
+      "step": 2655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940086853540003e-05,
+      "loss": 0.72,
+      "step": 2656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940029703933966e-05,
+      "loss": 0.8063,
+      "step": 2657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939972527926703e-05,
+      "loss": 0.7421,
+      "step": 2658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399153255198193e-05,
+      "loss": 0.7886,
+      "step": 2659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398580967149222e-05,
+      "loss": 0.7796,
+      "step": 2660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939800841513619e-05,
+      "loss": 0.8383,
+      "step": 2661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939743559917517e-05,
+      "loss": 0.7994,
+      "step": 2662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396862519282256e-05,
+      "loss": 0.696,
+      "step": 2663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939628917547354e-05,
+      "loss": 0.772,
+      "step": 2664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395715567765126e-05,
+      "loss": 0.8069,
+      "step": 2665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939514169617313e-05,
+      "loss": 0.7896,
+      "step": 2666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394567560713655e-05,
+      "loss": 0.7856,
+      "step": 2667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939399316140284e-05,
+      "loss": 0.827,
+      "step": 2668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939341849825681e-05,
+      "loss": 0.7499,
+      "step": 2669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392843571291708e-05,
+      "loss": 0.802,
+      "step": 2670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939226838052368e-05,
+      "loss": 0.7939,
+      "step": 2671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939169292596888e-05,
+      "loss": 0.8571,
+      "step": 2672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939111720764347e-05,
+      "loss": 0.7614,
+      "step": 2673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939054122556362e-05,
+      "loss": 0.6761,
+      "step": 2674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389964979745504e-05,
+      "loss": 0.7738,
+      "step": 2675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938938847020531e-05,
+      "loss": 0.7569,
+      "step": 2676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938881169695923e-05,
+      "loss": 0.7355,
+      "step": 2677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938823466002346e-05,
+      "loss": 0.8191,
+      "step": 2678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387657359414206e-05,
+      "loss": 0.7115,
+      "step": 2679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387079795147685e-05,
+      "loss": 0.7757,
+      "step": 2680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386501967240113e-05,
+      "loss": 0.7589,
+      "step": 2681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385923875707726e-05,
+      "loss": 0.8113,
+      "step": 2682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385345520566753e-05,
+      "loss": 0.7756,
+      "step": 2683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384766901833446e-05,
+      "loss": 0.8094,
+      "step": 2684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384188019524045e-05,
+      "loss": 0.7931,
+      "step": 2685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383608873654817e-05,
+      "loss": 0.7654,
+      "step": 2686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938302946424202e-05,
+      "loss": 0.7596,
+      "step": 2687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938244979130193e-05,
+      "loss": 0.7969,
+      "step": 2688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381869854850834e-05,
+      "loss": 0.7641,
+      "step": 2689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381289654905014e-05,
+      "loss": 0.7868,
+      "step": 2690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380709191480763e-05,
+      "loss": 0.7169,
+      "step": 2691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380128464594388e-05,
+      "loss": 0.7491,
+      "step": 2692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379547474262195e-05,
+      "loss": 0.7556,
+      "step": 2693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378966220500503e-05,
+      "loss": 0.7234,
+      "step": 2694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937838470332564e-05,
+      "loss": 0.7968,
+      "step": 2695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377802922753933e-05,
+      "loss": 0.7645,
+      "step": 2696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377220878801727e-05,
+      "loss": 0.8065,
+      "step": 2697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376638571485365e-05,
+      "loss": 0.7097,
+      "step": 2698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93760560008212e-05,
+      "loss": 0.8436,
+      "step": 2699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375473166825603e-05,
+      "loss": 0.6938,
+      "step": 2700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374890069514933e-05,
+      "loss": 0.7615,
+      "step": 2701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374306708905572e-05,
+      "loss": 0.8007,
+      "step": 2702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93737230850139e-05,
+      "loss": 0.7574,
+      "step": 2703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937313919785631e-05,
+      "loss": 0.7708,
+      "step": 2704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372555047449202e-05,
+      "loss": 0.7491,
+      "step": 2705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371970633808982e-05,
+      "loss": 0.7683,
+      "step": 2706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937138595695206e-05,
+      "loss": 0.7645,
+      "step": 2707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370801016894865e-05,
+      "loss": 0.7587,
+      "step": 2708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370215813653822e-05,
+      "loss": 0.6881,
+      "step": 2709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369630347245362e-05,
+      "loss": 0.8166,
+      "step": 2710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936904461768593e-05,
+      "loss": 0.7706,
+      "step": 2711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936845862499198e-05,
+      "loss": 0.8616,
+      "step": 2712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936787236917997e-05,
+      "loss": 0.3406,
+      "step": 2713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936728585026636e-05,
+      "loss": 0.7856,
+      "step": 2714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936669906826762e-05,
+      "loss": 0.8455,
+      "step": 2715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366112023200246e-05,
+      "loss": 0.6872,
+      "step": 2716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365524715080706e-05,
+      "loss": 0.8362,
+      "step": 2717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936493714392551e-05,
+      "loss": 0.8008,
+      "step": 2718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364349309751158e-05,
+      "loss": 0.8113,
+      "step": 2719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936376121257415e-05,
+      "loss": 0.787,
+      "step": 2720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363172852411008e-05,
+      "loss": 0.7386,
+      "step": 2721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936258422927826e-05,
+      "loss": 0.7548,
+      "step": 2722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361995343192436e-05,
+      "loss": 0.808,
+      "step": 2723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361406194170072e-05,
+      "loss": 0.8654,
+      "step": 2724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936081678222772e-05,
+      "loss": 0.7604,
+      "step": 2725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936022710738193e-05,
+      "loss": 0.7024,
+      "step": 2726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359637169649262e-05,
+      "loss": 0.7739,
+      "step": 2727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935904696904629e-05,
+      "loss": 0.8578,
+      "step": 2728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358456505589586e-05,
+      "loss": 0.8582,
+      "step": 2729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357865779295733e-05,
+      "loss": 0.825,
+      "step": 2730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357274790181323e-05,
+      "loss": 0.777,
+      "step": 2731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356683538262957e-05,
+      "loss": 0.7495,
+      "step": 2732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356092023557237e-05,
+      "loss": 0.8228,
+      "step": 2733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355500246080775e-05,
+      "loss": 0.8443,
+      "step": 2734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354908205850197e-05,
+      "loss": 0.7441,
+      "step": 2735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935431590288213e-05,
+      "loss": 0.7316,
+      "step": 2736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353723337193198e-05,
+      "loss": 0.7307,
+      "step": 2737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935313050880006e-05,
+      "loss": 0.7675,
+      "step": 2738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352537417719356e-05,
+      "loss": 0.7972,
+      "step": 2739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351944063967742e-05,
+      "loss": 0.7711,
+      "step": 2740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351350447561888e-05,
+      "loss": 0.7481,
+      "step": 2741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350756568518466e-05,
+      "loss": 0.8533,
+      "step": 2742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350162426854152e-05,
+      "loss": 0.8212,
+      "step": 2743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934956802258563e-05,
+      "loss": 0.8375,
+      "step": 2744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348973355729603e-05,
+      "loss": 0.7194,
+      "step": 2745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348378426302764e-05,
+      "loss": 0.8519,
+      "step": 2746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347783234321827e-05,
+      "loss": 0.7408,
+      "step": 2747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934718777980351e-05,
+      "loss": 0.8237,
+      "step": 2748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346592062764532e-05,
+      "loss": 0.7762,
+      "step": 2749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345996083221624e-05,
+      "loss": 0.8032,
+      "step": 2750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345399841191527e-05,
+      "loss": 0.7415,
+      "step": 2751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344803336690985e-05,
+      "loss": 0.8046,
+      "step": 2752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344206569736752e-05,
+      "loss": 0.739,
+      "step": 2753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934360954034559e-05,
+      "loss": 0.8268,
+      "step": 2754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343012248534263e-05,
+      "loss": 0.7924,
+      "step": 2755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934241469431955e-05,
+      "loss": 0.7184,
+      "step": 2756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341816877718232e-05,
+      "loss": 0.7214,
+      "step": 2757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93412187987471e-05,
+      "loss": 0.7392,
+      "step": 2758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934062045742295e-05,
+      "loss": 0.7511,
+      "step": 2759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340021853762585e-05,
+      "loss": 0.8391,
+      "step": 2760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933942298778282e-05,
+      "loss": 0.7687,
+      "step": 2761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338823859500474e-05,
+      "loss": 0.8331,
+      "step": 2762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338224468932377e-05,
+      "loss": 0.7928,
+      "step": 2763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933762481609536e-05,
+      "loss": 0.7488,
+      "step": 2764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933702490100626e-05,
+      "loss": 0.8085,
+      "step": 2765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933642472368193e-05,
+      "loss": 0.7786,
+      "step": 2766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933582428413923e-05,
+      "loss": 0.7337,
+      "step": 2767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335223582395017e-05,
+      "loss": 0.7601,
+      "step": 2768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334622618466168e-05,
+      "loss": 0.7227,
+      "step": 2769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933402139236956e-05,
+      "loss": 0.8483,
+      "step": 2770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333419904122076e-05,
+      "loss": 0.7271,
+      "step": 2771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332818153740608e-05,
+      "loss": 0.67,
+      "step": 2772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933221614124206e-05,
+      "loss": 0.7754,
+      "step": 2773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933161386664334e-05,
+      "loss": 0.8379,
+      "step": 2774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933101132996136e-05,
+      "loss": 0.7918,
+      "step": 2775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933040853121305e-05,
+      "loss": 0.2806,
+      "step": 2776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932980547041533e-05,
+      "loss": 0.7924,
+      "step": 2777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9329202147585144e-05,
+      "loss": 0.8165,
+      "step": 2778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328598562739435e-05,
+      "loss": 0.7697,
+      "step": 2779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932799471589515e-05,
+      "loss": 0.7568,
+      "step": 2780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327390607069255e-05,
+      "loss": 0.7905,
+      "step": 2781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326786236278717e-05,
+      "loss": 0.7919,
+      "step": 2782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326181603540505e-05,
+      "loss": 0.7766,
+      "step": 2783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325576708871605e-05,
+      "loss": 0.8196,
+      "step": 2784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324971552289e-05,
+      "loss": 0.7697,
+      "step": 2785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324366133809692e-05,
+      "loss": 0.7688,
+      "step": 2786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323760453450686e-05,
+      "loss": 0.7773,
+      "step": 2787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323154511228987e-05,
+      "loss": 0.7756,
+      "step": 2788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322548307161612e-05,
+      "loss": 0.7822,
+      "step": 2789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321941841265596e-05,
+      "loss": 0.8276,
+      "step": 2790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321335113557963e-05,
+      "loss": 0.7138,
+      "step": 2791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320728124055757e-05,
+      "loss": 0.7899,
+      "step": 2792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932012087277602e-05,
+      "loss": 0.7399,
+      "step": 2793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319513359735817e-05,
+      "loss": 0.6626,
+      "step": 2794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318905584952205e-05,
+      "loss": 0.8143,
+      "step": 2795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318297548442252e-05,
+      "loss": 0.7707,
+      "step": 2796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317689250223036e-05,
+      "loss": 0.7234,
+      "step": 2797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9317080690311646e-05,
+      "loss": 0.3154,
+      "step": 2798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316471868725167e-05,
+      "loss": 0.828,
+      "step": 2799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93158627854807e-05,
+      "loss": 0.8151,
+      "step": 2800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315253440595354e-05,
+      "loss": 0.843,
+      "step": 2801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9314643834086242e-05,
+      "loss": 0.7433,
+      "step": 2802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314033965970483e-05,
+      "loss": 0.7961,
+      "step": 2803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313423836265207e-05,
+      "loss": 0.7851,
+      "step": 2804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312813444987554e-05,
+      "loss": 0.8041,
+      "step": 2805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931220279215466e-05,
+      "loss": 0.7898,
+      "step": 2806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311591877783674e-05,
+      "loss": 0.7393,
+      "step": 2807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931098070189176e-05,
+      "loss": 0.7925,
+      "step": 2808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931036926449608e-05,
+      "loss": 0.8673,
+      "step": 2809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930975756561381e-05,
+      "loss": 0.8229,
+      "step": 2810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930914560526213e-05,
+      "loss": 0.8081,
+      "step": 2811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308533383458222e-05,
+      "loss": 0.7775,
+      "step": 2812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307920900219284e-05,
+      "loss": 0.8654,
+      "step": 2813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930730815556252e-05,
+      "loss": 0.7278,
+      "step": 2814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306695149505133e-05,
+      "loss": 0.7995,
+      "step": 2815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306081882064346e-05,
+      "loss": 0.7277,
+      "step": 2816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305468353257377e-05,
+      "loss": 0.8019,
+      "step": 2817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304854563101462e-05,
+      "loss": 0.7806,
+      "step": 2818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930424051161384e-05,
+      "loss": 0.7699,
+      "step": 2819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303626198811752e-05,
+      "loss": 0.7777,
+      "step": 2820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303011624712455e-05,
+      "loss": 0.811,
+      "step": 2821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930239678933321e-05,
+      "loss": 0.7531,
+      "step": 2822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301781692691283e-05,
+      "loss": 0.7327,
+      "step": 2823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301166334803953e-05,
+      "loss": 0.7575,
+      "step": 2824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93005507156885e-05,
+      "loss": 0.7413,
+      "step": 2825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929993483536221e-05,
+      "loss": 0.8141,
+      "step": 2826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929931869384239e-05,
+      "loss": 0.7659,
+      "step": 2827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298702291146333e-05,
+      "loss": 0.8354,
+      "step": 2828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929808562729136e-05,
+      "loss": 0.755,
+      "step": 2829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297468702294785e-05,
+      "loss": 0.8025,
+      "step": 2830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296851516173937e-05,
+      "loss": 0.8383,
+      "step": 2831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929623406894615e-05,
+      "loss": 0.6924,
+      "step": 2832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929561636062877e-05,
+      "loss": 0.874,
+      "step": 2833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294998391239133e-05,
+      "loss": 0.7616,
+      "step": 2834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294380160794607e-05,
+      "loss": 0.8068,
+      "step": 2835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929376166931255e-05,
+      "loss": 0.7885,
+      "step": 2836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293142916810337e-05,
+      "loss": 0.8017,
+      "step": 2837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929252390330534e-05,
+      "loss": 0.8454,
+      "step": 2838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291904628814948e-05,
+      "loss": 0.7284,
+      "step": 2839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291285093356555e-05,
+      "loss": 0.7823,
+      "step": 2840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290665296947555e-05,
+      "loss": 0.7802,
+      "step": 2841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929004523960536e-05,
+      "loss": 0.8281,
+      "step": 2842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289424921347384e-05,
+      "loss": 0.802,
+      "step": 2843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288804342191046e-05,
+      "loss": 0.7363,
+      "step": 2844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288183502153782e-05,
+      "loss": 0.8226,
+      "step": 2845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287562401253023e-05,
+      "loss": 0.7378,
+      "step": 2846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928694103950622e-05,
+      "loss": 0.7935,
+      "step": 2847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286319416930813e-05,
+      "loss": 0.6944,
+      "step": 2848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285697533544264e-05,
+      "loss": 0.8098,
+      "step": 2849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285075389364045e-05,
+      "loss": 0.8265,
+      "step": 2850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284452984407624e-05,
+      "loss": 0.7321,
+      "step": 2851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283830318692486e-05,
+      "loss": 0.7718,
+      "step": 2852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928320739223611e-05,
+      "loss": 0.8809,
+      "step": 2853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282584205056004e-05,
+      "loss": 0.7427,
+      "step": 2854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928196075716966e-05,
+      "loss": 0.7635,
+      "step": 2855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281337048594594e-05,
+      "loss": 0.765,
+      "step": 2856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280713079348317e-05,
+      "loss": 0.7917,
+      "step": 2857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280088849448362e-05,
+      "loss": 0.7959,
+      "step": 2858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279464358912253e-05,
+      "loss": 0.8525,
+      "step": 2859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278839607757533e-05,
+      "loss": 0.8359,
+      "step": 2860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278214596001746e-05,
+      "loss": 0.6836,
+      "step": 2861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277589323662453e-05,
+      "loss": 0.8003,
+      "step": 2862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276963790757207e-05,
+      "loss": 0.7156,
+      "step": 2863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276337997303577e-05,
+      "loss": 0.7193,
+      "step": 2864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275711943319142e-05,
+      "loss": 0.8261,
+      "step": 2865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275085628821485e-05,
+      "loss": 0.7937,
+      "step": 2866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274459053828193e-05,
+      "loss": 0.8306,
+      "step": 2867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273832218356867e-05,
+      "loss": 0.7226,
+      "step": 2868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273205122425108e-05,
+      "loss": 0.8022,
+      "step": 2869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272577766050536e-05,
+      "loss": 0.717,
+      "step": 2870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927195014925076e-05,
+      "loss": 0.7911,
+      "step": 2871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927132227204342e-05,
+      "loss": 0.7787,
+      "step": 2872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927069413444614e-05,
+      "loss": 0.7994,
+      "step": 2873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927006573647656e-05,
+      "loss": 0.746,
+      "step": 2874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269437078152342e-05,
+      "loss": 0.6974,
+      "step": 2875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926880815949113e-05,
+      "loss": 0.7081,
+      "step": 2876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268178980510587e-05,
+      "loss": 0.7855,
+      "step": 2877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267549541228394e-05,
+      "loss": 0.7726,
+      "step": 2878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926691984166222e-05,
+      "loss": 0.8057,
+      "step": 2879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266289881829757e-05,
+      "loss": 0.7813,
+      "step": 2880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926565966174869e-05,
+      "loss": 0.7539,
+      "step": 2881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265029181436726e-05,
+      "loss": 0.6987,
+      "step": 2882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926439844091157e-05,
+      "loss": 0.7856,
+      "step": 2883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263767440190934e-05,
+      "loss": 0.7674,
+      "step": 2884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263136179292545e-05,
+      "loss": 0.7446,
+      "step": 2885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262504658234126e-05,
+      "loss": 0.798,
+      "step": 2886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926187287703342e-05,
+      "loss": 0.7874,
+      "step": 2887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261240835708168e-05,
+      "loss": 0.83,
+      "step": 2888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260608534276123e-05,
+      "loss": 0.8194,
+      "step": 2889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925997597275504e-05,
+      "loss": 0.7925,
+      "step": 2890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259343151162687e-05,
+      "loss": 0.8173,
+      "step": 2891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925871006951684e-05,
+      "loss": 0.8491,
+      "step": 2892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258076727835275e-05,
+      "loss": 0.8045,
+      "step": 2893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257443126135776e-05,
+      "loss": 0.7636,
+      "step": 2894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256809264436147e-05,
+      "loss": 0.7534,
+      "step": 2895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9256175142754188e-05,
+      "loss": 0.3151,
+      "step": 2896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255540761107704e-05,
+      "loss": 0.7632,
+      "step": 2897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254906119514517e-05,
+      "loss": 0.8256,
+      "step": 2898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925427121799245e-05,
+      "loss": 0.7104,
+      "step": 2899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253636056559334e-05,
+      "loss": 0.7802,
+      "step": 2900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253000635233006e-05,
+      "loss": 0.7067,
+      "step": 2901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925236495403131e-05,
+      "loss": 0.7375,
+      "step": 2902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925172901297211e-05,
+      "loss": 0.7432,
+      "step": 2903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925109281207326e-05,
+      "loss": 0.8376,
+      "step": 2904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250456351352622e-05,
+      "loss": 0.8373,
+      "step": 2905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924981963082808e-05,
+      "loss": 0.7599,
+      "step": 2906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924918265051752e-05,
+      "loss": 0.6733,
+      "step": 2907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248545410438817e-05,
+      "loss": 0.785,
+      "step": 2908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247907910609883e-05,
+      "loss": 0.7626,
+      "step": 2909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247270151048613e-05,
+      "loss": 0.7931,
+      "step": 2910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246632131772923e-05,
+      "loss": 0.7729,
+      "step": 2911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924599385280073e-05,
+      "loss": 0.7563,
+      "step": 2912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924535531414997e-05,
+      "loss": 0.7932,
+      "step": 2913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244716515838558e-05,
+      "loss": 0.8462,
+      "step": 2914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244077457884454e-05,
+      "loss": 0.7894,
+      "step": 2915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243438140305594e-05,
+      "loss": 0.7191,
+      "step": 2916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242798563119937e-05,
+      "loss": 0.8074,
+      "step": 2917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242158726345445e-05,
+      "loss": 0.8285,
+      "step": 2918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241518630000092e-05,
+      "loss": 0.7209,
+      "step": 2919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240878274101854e-05,
+      "loss": 0.7391,
+      "step": 2920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240237658668714e-05,
+      "loss": 0.842,
+      "step": 2921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239596783718665e-05,
+      "loss": 0.7941,
+      "step": 2922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238955649269704e-05,
+      "loss": 0.7701,
+      "step": 2923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923831425533984e-05,
+      "loss": 0.6976,
+      "step": 2924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923767260194709e-05,
+      "loss": 0.7777,
+      "step": 2925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237030689109466e-05,
+      "loss": 0.7413,
+      "step": 2926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236388516845005e-05,
+      "loss": 0.7574,
+      "step": 2927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923574608517174e-05,
+      "loss": 0.7723,
+      "step": 2928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235103394107716e-05,
+      "loss": 0.7472,
+      "step": 2929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234460443670976e-05,
+      "loss": 0.7432,
+      "step": 2930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233817233879585e-05,
+      "loss": 0.8254,
+      "step": 2931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233173764751608e-05,
+      "loss": 0.7897,
+      "step": 2932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923253003630511e-05,
+      "loss": 0.7415,
+      "step": 2933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231886048558183e-05,
+      "loss": 0.7858,
+      "step": 2934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231241801528896e-05,
+      "loss": 0.8691,
+      "step": 2935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923059729523536e-05,
+      "loss": 0.7509,
+      "step": 2936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229952529695668e-05,
+      "loss": 0.7214,
+      "step": 2937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922930750492793e-05,
+      "loss": 0.7404,
+      "step": 2938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922866222095026e-05,
+      "loss": 0.7477,
+      "step": 2939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228016677780785e-05,
+      "loss": 0.7469,
+      "step": 2940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922737087543763e-05,
+      "loss": 0.8624,
+      "step": 2941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226724813938942e-05,
+      "loss": 0.7623,
+      "step": 2942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226078493302856e-05,
+      "loss": 0.7512,
+      "step": 2943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225431913547526e-05,
+      "loss": 0.7413,
+      "step": 2944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922478507469112e-05,
+      "loss": 0.7385,
+      "step": 2945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224137976751797e-05,
+      "loss": 0.7908,
+      "step": 2946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223490619747728e-05,
+      "loss": 0.7825,
+      "step": 2947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92228430036971e-05,
+      "loss": 0.7933,
+      "step": 2948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222195128618108e-05,
+      "loss": 0.767,
+      "step": 2949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922154699452893e-05,
+      "loss": 0.804,
+      "step": 2950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220898601447788e-05,
+      "loss": 0.7998,
+      "step": 2951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922024994939288e-05,
+      "loss": 0.6825,
+      "step": 2952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219601038382428e-05,
+      "loss": 0.7958,
+      "step": 2953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218951868434657e-05,
+      "loss": 0.7755,
+      "step": 2954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218302439567803e-05,
+      "loss": 0.7755,
+      "step": 2955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217652751800097e-05,
+      "loss": 0.7953,
+      "step": 2956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217002805149795e-05,
+      "loss": 0.7967,
+      "step": 2957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216352599635142e-05,
+      "loss": 0.8081,
+      "step": 2958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921570213527441e-05,
+      "loss": 0.8473,
+      "step": 2959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215051412085857e-05,
+      "loss": 0.7601,
+      "step": 2960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214400430087762e-05,
+      "loss": 0.6907,
+      "step": 2961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213749189298416e-05,
+      "loss": 0.7628,
+      "step": 2962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92130976897361e-05,
+      "loss": 0.7872,
+      "step": 2963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212445931419112e-05,
+      "loss": 0.6516,
+      "step": 2964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921179391436576e-05,
+      "loss": 0.7795,
+      "step": 2965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211141638594363e-05,
+      "loss": 0.819,
+      "step": 2966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210489104123228e-05,
+      "loss": 0.7299,
+      "step": 2967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209836310970686e-05,
+      "loss": 0.7921,
+      "step": 2968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209183259155076e-05,
+      "loss": 0.7923,
+      "step": 2969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208529948694733e-05,
+      "loss": 0.7742,
+      "step": 2970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920787637960801e-05,
+      "loss": 0.7701,
+      "step": 2971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207222551913263e-05,
+      "loss": 0.7198,
+      "step": 2972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920656846562885e-05,
+      "loss": 0.7255,
+      "step": 2973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205914120773146e-05,
+      "loss": 0.7368,
+      "step": 2974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920525951736453e-05,
+      "loss": 0.7158,
+      "step": 2975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204604655421376e-05,
+      "loss": 0.7501,
+      "step": 2976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920394953496209e-05,
+      "loss": 0.8383,
+      "step": 2977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203294156005068e-05,
+      "loss": 0.7469,
+      "step": 2978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920263851856871e-05,
+      "loss": 0.7674,
+      "step": 2979
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201982622671437e-05,
+      "loss": 0.7425,
+      "step": 2980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201326468331665e-05,
+      "loss": 0.7653,
+      "step": 2981
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200670055567826e-05,
+      "loss": 0.8059,
+      "step": 2982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200013384398356e-05,
+      "loss": 0.7679,
+      "step": 2983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919935645484169e-05,
+      "loss": 0.6981,
+      "step": 2984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919869926691629e-05,
+      "loss": 0.752,
+      "step": 2985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919804182064061e-05,
+      "loss": 0.7651,
+      "step": 2986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919738411603311e-05,
+      "loss": 0.7438,
+      "step": 2987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196726153112264e-05,
+      "loss": 0.8013,
+      "step": 2988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196067931896555e-05,
+      "loss": 0.822,
+      "step": 2989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195409452404465e-05,
+      "loss": 0.8211,
+      "step": 2990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919475071465449e-05,
+      "loss": 0.7695,
+      "step": 2991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194091718665126e-05,
+      "loss": 0.7691,
+      "step": 2992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919343246445489e-05,
+      "loss": 0.8156,
+      "step": 2993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192772952042292e-05,
+      "loss": 0.7286,
+      "step": 2994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192113181445856e-05,
+      "loss": 0.811,
+      "step": 2995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191453152684115e-05,
+      "loss": 0.7712,
+      "step": 2996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190792865775595e-05,
+      "loss": 0.7517,
+      "step": 2997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190132320738856e-05,
+      "loss": 0.7236,
+      "step": 2998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918947151759244e-05,
+      "loss": 0.6776,
+      "step": 2999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188810456354907e-05,
+      "loss": 0.7811,
+      "step": 3000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188149137044826e-05,
+      "loss": 0.8461,
+      "step": 3001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187487559680767e-05,
+      "loss": 0.776,
+      "step": 3002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186825724281314e-05,
+      "loss": 0.7496,
+      "step": 3003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186163630865055e-05,
+      "loss": 0.8214,
+      "step": 3004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185501279450583e-05,
+      "loss": 0.7732,
+      "step": 3005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184838670056503e-05,
+      "loss": 0.7376,
+      "step": 3006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918417580270142e-05,
+      "loss": 0.7903,
+      "step": 3007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183512677403957e-05,
+      "loss": 0.7686,
+      "step": 3008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182849294182734e-05,
+      "loss": 0.7115,
+      "step": 3009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182185653056385e-05,
+      "loss": 0.7797,
+      "step": 3010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918152175404355e-05,
+      "loss": 0.8257,
+      "step": 3011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180857597162866e-05,
+      "loss": 0.7252,
+      "step": 3012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180193182432998e-05,
+      "loss": 0.8514,
+      "step": 3013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91795285098726e-05,
+      "loss": 0.7743,
+      "step": 3014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917886357950034e-05,
+      "loss": 0.7397,
+      "step": 3015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178198391334893e-05,
+      "loss": 0.7638,
+      "step": 3016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177532945394945e-05,
+      "loss": 0.8109,
+      "step": 3017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176867241699184e-05,
+      "loss": 0.8051,
+      "step": 3018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91762012802663e-05,
+      "loss": 0.7909,
+      "step": 3019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175535061115e-05,
+      "loss": 0.7642,
+      "step": 3020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174868584264002e-05,
+      "loss": 0.7823,
+      "step": 3021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174201849732017e-05,
+      "loss": 0.7474,
+      "step": 3022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173534857537777e-05,
+      "loss": 0.7908,
+      "step": 3023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172867607700004e-05,
+      "loss": 0.3209,
+      "step": 3024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917220010023745e-05,
+      "loss": 0.7434,
+      "step": 3025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171532335168855e-05,
+      "loss": 0.8044,
+      "step": 3026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170864312512977e-05,
+      "loss": 0.7243,
+      "step": 3027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170196032288575e-05,
+      "loss": 0.771,
+      "step": 3028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916952749451442e-05,
+      "loss": 0.8015,
+      "step": 3029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916885869920929e-05,
+      "loss": 0.8217,
+      "step": 3030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916818964639196e-05,
+      "loss": 0.7144,
+      "step": 3031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167520336081234e-05,
+      "loss": 0.8342,
+      "step": 3032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166850768295896e-05,
+      "loss": 0.828,
+      "step": 3033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166180943054762e-05,
+      "loss": 0.7645,
+      "step": 3034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916551086037664e-05,
+      "loss": 0.8238,
+      "step": 3035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916484052028035e-05,
+      "loss": 0.7499,
+      "step": 3036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164169922784717e-05,
+      "loss": 0.8234,
+      "step": 3037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916349906790858e-05,
+      "loss": 0.7159,
+      "step": 3038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162827955670773e-05,
+      "loss": 0.6842,
+      "step": 3039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916215658609015e-05,
+      "loss": 0.775,
+      "step": 3040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161484959185567e-05,
+      "loss": 0.8145,
+      "step": 3041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160813074975887e-05,
+      "loss": 0.855,
+      "step": 3042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160140933479974e-05,
+      "loss": 0.707,
+      "step": 3043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915946853471671e-05,
+      "loss": 0.847,
+      "step": 3044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158795878704986e-05,
+      "loss": 0.8212,
+      "step": 3045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158122965463683e-05,
+      "loss": 0.8007,
+      "step": 3046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157449795011705e-05,
+      "loss": 0.7437,
+      "step": 3047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915677636736796e-05,
+      "loss": 0.7956,
+      "step": 3048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156102682551356e-05,
+      "loss": 0.7524,
+      "step": 3049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915542874058082e-05,
+      "loss": 0.8096,
+      "step": 3050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915475454147528e-05,
+      "loss": 0.695,
+      "step": 3051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154080085253665e-05,
+      "loss": 0.7681,
+      "step": 3052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915340537193492e-05,
+      "loss": 0.8542,
+      "step": 3053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152730401538e-05,
+      "loss": 0.7921,
+      "step": 3054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152055174081855e-05,
+      "loss": 0.8256,
+      "step": 3055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151379689585457e-05,
+      "loss": 0.7309,
+      "step": 3056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150703948067767e-05,
+      "loss": 0.7605,
+      "step": 3057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150027949547768e-05,
+      "loss": 0.7716,
+      "step": 3058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914935169404445e-05,
+      "loss": 0.6837,
+      "step": 3059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91486751815768e-05,
+      "loss": 0.8779,
+      "step": 3060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147998412163825e-05,
+      "loss": 0.7826,
+      "step": 3061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147321385824523e-05,
+      "loss": 0.2986,
+      "step": 3062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146644102577917e-05,
+      "loss": 0.7239,
+      "step": 3063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145966562443023e-05,
+      "loss": 0.7894,
+      "step": 3064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145288765438875e-05,
+      "loss": 0.8185,
+      "step": 3065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144610711584505e-05,
+      "loss": 0.7986,
+      "step": 3066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914393240089896e-05,
+      "loss": 0.7992,
+      "step": 3067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914325383340129e-05,
+      "loss": 0.7988,
+      "step": 3068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914257500911055e-05,
+      "loss": 0.814,
+      "step": 3069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914189592804581e-05,
+      "loss": 0.7968,
+      "step": 3070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141216590226137e-05,
+      "loss": 0.7762,
+      "step": 3071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140536995670617e-05,
+      "loss": 0.7839,
+      "step": 3072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913985714439833e-05,
+      "loss": 0.7877,
+      "step": 3073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913917703642837e-05,
+      "loss": 0.7548,
+      "step": 3074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138496671779847e-05,
+      "loss": 0.8275,
+      "step": 3075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913781605047186e-05,
+      "loss": 0.7595,
+      "step": 3076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913713517252353e-05,
+      "loss": 0.7748,
+      "step": 3077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136454037953974e-05,
+      "loss": 0.7507,
+      "step": 3078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913577264678233e-05,
+      "loss": 0.7322,
+      "step": 3079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135090999027733e-05,
+      "loss": 0.85,
+      "step": 3080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913440909470932e-05,
+      "loss": 0.7596,
+      "step": 3081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133726933846252e-05,
+      "loss": 0.7615,
+      "step": 3082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133044516457684e-05,
+      "loss": 0.7603,
+      "step": 3083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913236184256278e-05,
+      "loss": 0.749,
+      "step": 3084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913167891218072e-05,
+      "loss": 0.8591,
+      "step": 3085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130995725330674e-05,
+      "loss": 0.7652,
+      "step": 3086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913031228203184e-05,
+      "loss": 0.8481,
+      "step": 3087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129628582303405e-05,
+      "loss": 0.8,
+      "step": 3088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128944626164576e-05,
+      "loss": 0.7106,
+      "step": 3089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128260413634562e-05,
+      "loss": 0.7519,
+      "step": 3090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127575944732575e-05,
+      "loss": 0.7627,
+      "step": 3091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126891219477845e-05,
+      "loss": 0.7307,
+      "step": 3092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126206237889596e-05,
+      "loss": 0.7512,
+      "step": 3093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125520999987073e-05,
+      "loss": 0.7997,
+      "step": 3094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912483550578952e-05,
+      "loss": 0.7491,
+      "step": 3095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124149755316185e-05,
+      "loss": 0.7813,
+      "step": 3096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912346374858633e-05,
+      "loss": 0.8563,
+      "step": 3097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912277748561922e-05,
+      "loss": 0.7997,
+      "step": 3098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122090966434134e-05,
+      "loss": 0.7586,
+      "step": 3099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912140419105035e-05,
+      "loss": 0.7086,
+      "step": 3100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120717159487157e-05,
+      "loss": 0.7918,
+      "step": 3101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912002987176385e-05,
+      "loss": 0.7124,
+      "step": 3102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119342327899732e-05,
+      "loss": 0.7509,
+      "step": 3103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118654527914112e-05,
+      "loss": 0.7552,
+      "step": 3104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911796647182631e-05,
+      "loss": 0.8064,
+      "step": 3105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911727815965565e-05,
+      "loss": 0.7722,
+      "step": 3106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911658959142146e-05,
+      "loss": 0.7926,
+      "step": 3107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115900767143084e-05,
+      "loss": 0.7653,
+      "step": 3108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115211686839863e-05,
+      "loss": 0.7761,
+      "step": 3109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114522350531156e-05,
+      "loss": 0.6813,
+      "step": 3110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113832758236316e-05,
+      "loss": 0.8867,
+      "step": 3111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113142909974715e-05,
+      "loss": 0.8322,
+      "step": 3112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112452805765728e-05,
+      "loss": 0.7749,
+      "step": 3113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111762445628738e-05,
+      "loss": 0.8319,
+      "step": 3114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911107182958313e-05,
+      "loss": 0.7887,
+      "step": 3115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91103809576483e-05,
+      "loss": 0.6807,
+      "step": 3116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109689829843658e-05,
+      "loss": 0.794,
+      "step": 3117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108998446188612e-05,
+      "loss": 0.7057,
+      "step": 3118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108306806702576e-05,
+      "loss": 0.7689,
+      "step": 3119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107614911404976e-05,
+      "loss": 0.8462,
+      "step": 3120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910692276031525e-05,
+      "loss": 0.8111,
+      "step": 3121
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910623035345283e-05,
+      "loss": 0.7526,
+      "step": 3122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105537690837168e-05,
+      "loss": 0.7794,
+      "step": 3123
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104844772487716e-05,
+      "loss": 0.8847,
+      "step": 3124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910415159842393e-05,
+      "loss": 0.6679,
+      "step": 3125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910345816866529e-05,
+      "loss": 0.8369,
+      "step": 3126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910276448323126e-05,
+      "loss": 0.7617,
+      "step": 3127
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910207054214133e-05,
+      "loss": 0.7189,
+      "step": 3128
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101376345414985e-05,
+      "loss": 0.7939,
+      "step": 3129
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910068189307172e-05,
+      "loss": 0.7044,
+      "step": 3130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099987185131046e-05,
+      "loss": 0.7629,
+      "step": 3131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909929222161247e-05,
+      "loss": 0.7198,
+      "step": 3132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098597002535507e-05,
+      "loss": 0.738,
+      "step": 3133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909790152791969e-05,
+      "loss": 0.7485,
+      "step": 3134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097205797784552e-05,
+      "loss": 0.7781,
+      "step": 3135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096509812149624e-05,
+      "loss": 0.7388,
+      "step": 3136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909581357103446e-05,
+      "loss": 0.7965,
+      "step": 3137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095117074458614e-05,
+      "loss": 0.8141,
+      "step": 3138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094420322441644e-05,
+      "loss": 0.7559,
+      "step": 3139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909372331500312e-05,
+      "loss": 0.7612,
+      "step": 3140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093026052162623e-05,
+      "loss": 0.7727,
+      "step": 3141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092328533939727e-05,
+      "loss": 0.7607,
+      "step": 3142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909163076035403e-05,
+      "loss": 0.8361,
+      "step": 3143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090932731425128e-05,
+      "loss": 0.8021,
+      "step": 3144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909023444717262e-05,
+      "loss": 0.8214,
+      "step": 3145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908953590761612e-05,
+      "loss": 0.7783,
+      "step": 3146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088837112775255e-05,
+      "loss": 0.7369,
+      "step": 3147
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908813806266964e-05,
+      "loss": 0.7686,
+      "step": 3148
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908743875731891e-05,
+      "loss": 0.7628,
+      "step": 3149
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908673919674271e-05,
+      "loss": 0.7697,
+      "step": 3150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086039380960684e-05,
+      "loss": 0.7182,
+      "step": 3151
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908533930999249e-05,
+      "loss": 0.8048,
+      "step": 3152
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908463898385779e-05,
+      "loss": 0.7518,
+      "step": 3153
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083938402576246e-05,
+      "loss": 0.7535,
+      "step": 3154
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908323756616754e-05,
+      "loss": 0.746,
+      "step": 3155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082536474651355e-05,
+      "loss": 0.7876,
+      "step": 3156
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081835128047383e-05,
+      "loss": 0.7466,
+      "step": 3157
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908113352637532e-05,
+      "loss": 0.6938,
+      "step": 3158
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080431669654866e-05,
+      "loss": 0.833,
+      "step": 3159
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907972955790574e-05,
+      "loss": 0.7715,
+      "step": 3160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907902719114766e-05,
+      "loss": 0.7339,
+      "step": 3161
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078324569400347e-05,
+      "loss": 0.7434,
+      "step": 3162
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077621692683543e-05,
+      "loss": 0.805,
+      "step": 3163
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076918561016985e-05,
+      "loss": 0.7804,
+      "step": 3164
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076215174420414e-05,
+      "loss": 0.7696,
+      "step": 3165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075511532913595e-05,
+      "loss": 0.7532,
+      "step": 3166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074807636516286e-05,
+      "loss": 0.8541,
+      "step": 3167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074103485248258e-05,
+      "loss": 0.7516,
+      "step": 3168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073399079129283e-05,
+      "loss": 0.8087,
+      "step": 3169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072694418179148e-05,
+      "loss": 0.7355,
+      "step": 3170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071989502417643e-05,
+      "loss": 0.7777,
+      "step": 3171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907128433186457e-05,
+      "loss": 0.6916,
+      "step": 3172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070578906539722e-05,
+      "loss": 0.7653,
+      "step": 3173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069873226462927e-05,
+      "loss": 0.768,
+      "step": 3174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069167291653993e-05,
+      "loss": 0.7447,
+      "step": 3175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906846110213275e-05,
+      "loss": 0.7814,
+      "step": 3176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067754657919037e-05,
+      "loss": 0.7697,
+      "step": 3177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067047959032686e-05,
+      "loss": 0.7679,
+      "step": 3178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906634100549355e-05,
+      "loss": 0.7501,
+      "step": 3179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065633797321484e-05,
+      "loss": 0.8444,
+      "step": 3180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064926334536346e-05,
+      "loss": 0.7582,
+      "step": 3181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064218617158013e-05,
+      "loss": 0.8208,
+      "step": 3182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063510645206357e-05,
+      "loss": 0.7962,
+      "step": 3183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906280241870126e-05,
+      "loss": 0.7792,
+      "step": 3184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062093937662616e-05,
+      "loss": 0.7693,
+      "step": 3185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061385202110324e-05,
+      "loss": 0.7249,
+      "step": 3186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060676212064283e-05,
+      "loss": 0.8279,
+      "step": 3187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9059966967544415e-05,
+      "loss": 0.8066,
+      "step": 3188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059257468570632e-05,
+      "loss": 0.7766,
+      "step": 3189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058547715162866e-05,
+      "loss": 0.7719,
+      "step": 3190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057837707341046e-05,
+      "loss": 0.7112,
+      "step": 3191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057127445125112e-05,
+      "loss": 0.7406,
+      "step": 3192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056416928535016e-05,
+      "loss": 0.7513,
+      "step": 3193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055706157590715e-05,
+      "loss": 0.7621,
+      "step": 3194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054995132312167e-05,
+      "loss": 0.7718,
+      "step": 3195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054283852719342e-05,
+      "loss": 0.785,
+      "step": 3196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053572318832217e-05,
+      "loss": 0.7339,
+      "step": 3197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905286053067078e-05,
+      "loss": 0.7665,
+      "step": 3198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052148488255013e-05,
+      "loss": 0.7742,
+      "step": 3199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905143619160492e-05,
+      "loss": 0.7625,
+      "step": 3200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050723640740512e-05,
+      "loss": 0.7609,
+      "step": 3201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905001083568179e-05,
+      "loss": 0.8287,
+      "step": 3202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904929777644878e-05,
+      "loss": 0.7437,
+      "step": 3203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048584463061507e-05,
+      "loss": 0.8303,
+      "step": 3204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047870895540002e-05,
+      "loss": 0.7582,
+      "step": 3205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904715707390431e-05,
+      "loss": 0.7784,
+      "step": 3206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904644299817448e-05,
+      "loss": 0.8998,
+      "step": 3207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045728668370567e-05,
+      "loss": 0.7653,
+      "step": 3208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904501408451263e-05,
+      "loss": 0.752,
+      "step": 3209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044299246620738e-05,
+      "loss": 0.7114,
+      "step": 3210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904358415471497e-05,
+      "loss": 0.684,
+      "step": 3211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904286880881541e-05,
+      "loss": 0.7699,
+      "step": 3212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042153208942147e-05,
+      "loss": 0.7331,
+      "step": 3213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904143735511528e-05,
+      "loss": 0.838,
+      "step": 3214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040721247354918e-05,
+      "loss": 0.7918,
+      "step": 3215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040004885681165e-05,
+      "loss": 0.7808,
+      "step": 3216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039288270114148e-05,
+      "loss": 0.7111,
+      "step": 3217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038571400673986e-05,
+      "loss": 0.7419,
+      "step": 3218
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903785427738082e-05,
+      "loss": 0.8182,
+      "step": 3219
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037136900254788e-05,
+      "loss": 0.8192,
+      "step": 3220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036419269316038e-05,
+      "loss": 0.7556,
+      "step": 3221
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035701384584723e-05,
+      "loss": 0.7912,
+      "step": 3222
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034983246081008e-05,
+      "loss": 0.7716,
+      "step": 3223
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903426485382506e-05,
+      "loss": 0.728,
+      "step": 3224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903354620783706e-05,
+      "loss": 0.8151,
+      "step": 3225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032827308137185e-05,
+      "loss": 0.8661,
+      "step": 3226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032108154745634e-05,
+      "loss": 0.8463,
+      "step": 3227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031388747682592e-05,
+      "loss": 0.839,
+      "step": 3228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030669086968277e-05,
+      "loss": 0.7242,
+      "step": 3229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029949172622895e-05,
+      "loss": 0.8163,
+      "step": 3230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029229004666664e-05,
+      "loss": 0.7813,
+      "step": 3231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902850858311981e-05,
+      "loss": 0.8109,
+      "step": 3232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027787908002574e-05,
+      "loss": 0.7289,
+      "step": 3233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902706697933519e-05,
+      "loss": 0.8118,
+      "step": 3234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026345797137903e-05,
+      "loss": 0.8032,
+      "step": 3235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025624361430973e-05,
+      "loss": 0.7901,
+      "step": 3236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024902672234665e-05,
+      "loss": 0.7285,
+      "step": 3237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024180729569236e-05,
+      "loss": 0.8055,
+      "step": 3238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023458533454974e-05,
+      "loss": 0.7947,
+      "step": 3239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022736083912158e-05,
+      "loss": 0.744,
+      "step": 3240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022013380961078e-05,
+      "loss": 0.7589,
+      "step": 3241
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902129042462203e-05,
+      "loss": 0.7457,
+      "step": 3242
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020567214915324e-05,
+      "loss": 0.7737,
+      "step": 3243
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019843751861265e-05,
+      "loss": 0.8337,
+      "step": 3244
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019120035480174e-05,
+      "loss": 0.7596,
+      "step": 3245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901839606579238e-05,
+      "loss": 0.7351,
+      "step": 3246
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017671842818213e-05,
+      "loss": 0.7326,
+      "step": 3247
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016947366578014e-05,
+      "loss": 0.7743,
+      "step": 3248
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901622263709213e-05,
+      "loss": 0.8045,
+      "step": 3249
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015497654380917e-05,
+      "loss": 0.7474,
+      "step": 3250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901477241846473e-05,
+      "loss": 0.7697,
+      "step": 3251
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901404692936395e-05,
+      "loss": 0.7585,
+      "step": 3252
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013321187098947e-05,
+      "loss": 0.7263,
+      "step": 3253
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012595191690096e-05,
+      "loss": 0.7768,
+      "step": 3254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011868943157796e-05,
+      "loss": 0.7989,
+      "step": 3255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011142441522442e-05,
+      "loss": 0.802,
+      "step": 3256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010415686804435e-05,
+      "loss": 0.7529,
+      "step": 3257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.759,
+      "step": 3258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900896141820213e-05,
+      "loss": 0.771,
+      "step": 3259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900823390435867e-05,
+      "loss": 0.7359,
+      "step": 3260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007506137514248e-05,
+      "loss": 0.7744,
+      "step": 3261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006778117689304e-05,
+      "loss": 0.7618,
+      "step": 3262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900604984490428e-05,
+      "loss": 0.7196,
+      "step": 3263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900532131917964e-05,
+      "loss": 0.7726,
+      "step": 3264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004592540535836e-05,
+      "loss": 0.8004,
+      "step": 3265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003863508993338e-05,
+      "loss": 0.7845,
+      "step": 3266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003134224572623e-05,
+      "loss": 0.79,
+      "step": 3267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002404687294176e-05,
+      "loss": 0.7512,
+      "step": 3268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900167489717848e-05,
+      "loss": 0.8275,
+      "step": 3269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000944854246036e-05,
+      "loss": 0.7184,
+      "step": 3270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000214558517343e-05,
+      "loss": 0.742,
+      "step": 3271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999484010012918e-05,
+      "loss": 0.8107,
+      "step": 3272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998753208753275e-05,
+      "loss": 0.7991,
+      "step": 3273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899802215475894e-05,
+      "loss": 0.7312,
+      "step": 3274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899729084805045e-05,
+      "loss": 0.715,
+      "step": 3275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996559288648332e-05,
+      "loss": 0.6601,
+      "step": 3276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899582747657314e-05,
+      "loss": 0.7019,
+      "step": 3277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995095411845427e-05,
+      "loss": 0.7745,
+      "step": 3278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994363094485754e-05,
+      "loss": 0.8439,
+      "step": 3279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899363052451469e-05,
+      "loss": 0.8018,
+      "step": 3280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992897701952803e-05,
+      "loss": 0.8033,
+      "step": 3281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899216462682068e-05,
+      "loss": 0.8032,
+      "step": 3282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899143129913891e-05,
+      "loss": 0.7812,
+      "step": 3283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990697718928084e-05,
+      "loss": 0.7434,
+      "step": 3284
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898996388620881e-05,
+      "loss": 0.7485,
+      "step": 3285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989229801001698e-05,
+      "loss": 0.7369,
+      "step": 3286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898849546332736e-05,
+      "loss": 0.7853,
+      "step": 3287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987760873206427e-05,
+      "loss": 0.7856,
+      "step": 3288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987026030659527e-05,
+      "loss": 0.7163,
+      "step": 3289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89862909357073e-05,
+      "loss": 0.8052,
+      "step": 3290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985555588370387e-05,
+      "loss": 0.7747,
+      "step": 3291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984819988669447e-05,
+      "loss": 0.7128,
+      "step": 3292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984084136625137e-05,
+      "loss": 0.8457,
+      "step": 3293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898334803225812e-05,
+      "loss": 0.7042,
+      "step": 3294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982611675589074e-05,
+      "loss": 0.7163,
+      "step": 3295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898187506663868e-05,
+      "loss": 0.7829,
+      "step": 3296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898113820542763e-05,
+      "loss": 0.789,
+      "step": 3297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898040109197661e-05,
+      "loss": 0.8023,
+      "step": 3298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979663726306323e-05,
+      "loss": 0.7764,
+      "step": 3299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978926108437487e-05,
+      "loss": 0.7922,
+      "step": 3300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897818823839081e-05,
+      "loss": 0.8065,
+      "step": 3301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897745011618702e-05,
+      "loss": 0.7346,
+      "step": 3302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897671174184685e-05,
+      "loss": 0.8419,
+      "step": 3303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897597311539103e-05,
+      "loss": 0.7579,
+      "step": 3304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975234236840306e-05,
+      "loss": 0.7515,
+      "step": 3305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974495106215437e-05,
+      "loss": 0.7211,
+      "step": 3306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973755723537178e-05,
+      "loss": 0.7012,
+      "step": 3307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973016088826292e-05,
+      "loss": 0.7095,
+      "step": 3308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897227620210355e-05,
+      "loss": 0.6926,
+      "step": 3309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971536063389745e-05,
+      "loss": 0.868,
+      "step": 3310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897079567270565e-05,
+      "loss": 0.7365,
+      "step": 3311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897005503007207e-05,
+      "loss": 0.8689,
+      "step": 3312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969314135509794e-05,
+      "loss": 0.7642,
+      "step": 3313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896857298903964e-05,
+      "loss": 0.8531,
+      "step": 3314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967831590682423e-05,
+      "loss": 0.7784,
+      "step": 3315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967089940458964e-05,
+      "loss": 0.8112,
+      "step": 3316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896634803839009e-05,
+      "loss": 0.8017,
+      "step": 3317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965605884496642e-05,
+      "loss": 0.3529,
+      "step": 3318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896486347879946e-05,
+      "loss": 0.7638,
+      "step": 3319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964120821319397e-05,
+      "loss": 0.7732,
+      "step": 3320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896337791207731e-05,
+      "loss": 0.8132,
+      "step": 3321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962634751094065e-05,
+      "loss": 0.717,
+      "step": 3322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961891338390534e-05,
+      "loss": 0.7413,
+      "step": 3323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961147673987598e-05,
+      "loss": 0.7806,
+      "step": 3324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896040375790614e-05,
+      "loss": 0.7347,
+      "step": 3325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959659590167054e-05,
+      "loss": 0.7587,
+      "step": 3326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958915170791234e-05,
+      "loss": 0.8235,
+      "step": 3327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958170499799602e-05,
+      "loss": 0.7747,
+      "step": 3328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895742557721306e-05,
+      "loss": 0.6331,
+      "step": 3329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956680403052537e-05,
+      "loss": 0.7802,
+      "step": 3330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955934977338957e-05,
+      "loss": 0.7117,
+      "step": 3331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955189300093257e-05,
+      "loss": 0.7951,
+      "step": 3332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954443371336378e-05,
+      "loss": 0.8354,
+      "step": 3333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953697191089277e-05,
+      "loss": 0.8344,
+      "step": 3334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952950759372904e-05,
+      "loss": 0.7402,
+      "step": 3335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952204076208224e-05,
+      "loss": 0.7936,
+      "step": 3336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951457141616206e-05,
+      "loss": 0.8118,
+      "step": 3337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950709955617834e-05,
+      "loss": 0.81,
+      "step": 3338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949962518234086e-05,
+      "loss": 0.835,
+      "step": 3339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894921482948596e-05,
+      "loss": 0.8964,
+      "step": 3340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948466889394454e-05,
+      "loss": 0.793,
+      "step": 3341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947718697980574e-05,
+      "loss": 0.6975,
+      "step": 3342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894697025526533e-05,
+      "loss": 0.7035,
+      "step": 3343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946221561269746e-05,
+      "loss": 0.7824,
+      "step": 3344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894547261601485e-05,
+      "loss": 0.8152,
+      "step": 3345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944723419521674e-05,
+      "loss": 0.7637,
+      "step": 3346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894397397181126e-05,
+      "loss": 0.7363,
+      "step": 3347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894322427290466e-05,
+      "loss": 0.8047,
+      "step": 3348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942474322822924e-05,
+      "loss": 0.779,
+      "step": 3349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894172412158712e-05,
+      "loss": 0.8162,
+      "step": 3350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940973669218316e-05,
+      "loss": 0.6989,
+      "step": 3351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940222965737583e-05,
+      "loss": 0.8074,
+      "step": 3352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939472011166016e-05,
+      "loss": 0.7243,
+      "step": 3353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89387208055247e-05,
+      "loss": 0.7202,
+      "step": 3354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893796934883473e-05,
+      "loss": 0.7095,
+      "step": 3355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937217641117217e-05,
+      "loss": 0.7643,
+      "step": 3356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893646568239327e-05,
+      "loss": 0.7561,
+      "step": 3357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935713472684007e-05,
+      "loss": 0.7927,
+      "step": 3358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893496101201056e-05,
+      "loss": 0.7342,
+      "step": 3359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934208300394052e-05,
+      "loss": 0.8598,
+      "step": 3360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933455337855633e-05,
+      "loss": 0.8293,
+      "step": 3361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932702124416446e-05,
+      "loss": 0.7313,
+      "step": 3362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931948660097647e-05,
+      "loss": 0.7836,
+      "step": 3363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931194944920394e-05,
+      "loss": 0.7986,
+      "step": 3364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893044097890586e-05,
+      "loss": 0.8214,
+      "step": 3365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892968676207522e-05,
+      "loss": 0.7203,
+      "step": 3366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892893229444965e-05,
+      "loss": 0.6732,
+      "step": 3367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892817757605035e-05,
+      "loss": 0.707,
+      "step": 3368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927422606898508e-05,
+      "loss": 0.7846,
+      "step": 3369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926667387015333e-05,
+      "loss": 0.8021,
+      "step": 3370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892591191642203e-05,
+      "loss": 0.7718,
+      "step": 3371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925156195139826e-05,
+      "loss": 0.6994,
+      "step": 3372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924400223189937e-05,
+      "loss": 0.8091,
+      "step": 3373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89236440005936e-05,
+      "loss": 0.7873,
+      "step": 3374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922887527372053e-05,
+      "loss": 0.8337,
+      "step": 3375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892213080354654e-05,
+      "loss": 0.7987,
+      "step": 3376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921373829138313e-05,
+      "loss": 0.7927,
+      "step": 3377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920616604168635e-05,
+      "loss": 0.7516,
+      "step": 3378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919859128658775e-05,
+      "loss": 0.8381,
+      "step": 3379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919101402630002e-05,
+      "loss": 0.8055,
+      "step": 3380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89183434261036e-05,
+      "loss": 0.7474,
+      "step": 3381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8917585199100856e-05,
+      "loss": 0.6935,
+      "step": 3382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916826721643068e-05,
+      "loss": 0.8068,
+      "step": 3383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916067993751535e-05,
+      "loss": 0.8052,
+      "step": 3384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915309015447566e-05,
+      "loss": 0.7593,
+      "step": 3385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891454978675248e-05,
+      "loss": 0.7165,
+      "step": 3386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89137903076876e-05,
+      "loss": 0.7008,
+      "step": 3387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913030578274254e-05,
+      "loss": 0.7954,
+      "step": 3388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891227059853378e-05,
+      "loss": 0.801,
+      "step": 3389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911510368487526e-05,
+      "loss": 0.3581,
+      "step": 3390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910749888156833e-05,
+      "loss": 0.768,
+      "step": 3391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909989157563077e-05,
+      "loss": 0.7387,
+      "step": 3392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890922817672761e-05,
+      "loss": 0.8373,
+      "step": 3393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908466945671805e-05,
+      "loss": 0.7849,
+      "step": 3394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907705464417047e-05,
+      "loss": 0.7213,
+      "step": 3395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890694373298472e-05,
+      "loss": 0.7557,
+      "step": 3396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890618175139622e-05,
+      "loss": 0.7881,
+      "step": 3397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905419519672944e-05,
+      "loss": 0.8696,
+      "step": 3398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89046570378363e-05,
+      "loss": 0.7698,
+      "step": 3399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903894305907705e-05,
+      "loss": 0.785,
+      "step": 3400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890313132390858e-05,
+      "loss": 0.8284,
+      "step": 3401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902368091860353e-05,
+      "loss": 0.8228,
+      "step": 3402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890160460978446e-05,
+      "loss": 0.8391,
+      "step": 3403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900840877702343e-05,
+      "loss": 0.6973,
+      "step": 3404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900076895635457e-05,
+      "loss": 0.7745,
+      "step": 3405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899312663605252e-05,
+      "loss": 0.8054,
+      "step": 3406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898548181633195e-05,
+      "loss": 0.754,
+      "step": 3407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897783449740757e-05,
+      "loss": 0.8,
+      "step": 3408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897018467949415e-05,
+      "loss": 0.7577,
+      "step": 3409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889625323628065e-05,
+      "loss": 0.7054,
+      "step": 3410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895487754755965e-05,
+      "loss": 0.7668,
+      "step": 3411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894722023396846e-05,
+      "loss": 0.802,
+      "step": 3412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889395604222481e-05,
+      "loss": 0.7671,
+      "step": 3413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893189811261364e-05,
+      "loss": 0.7675,
+      "step": 3414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892423330528026e-05,
+      "loss": 0.8084,
+      "step": 3415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889165660004633e-05,
+      "loss": 0.8118,
+      "step": 3416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890889619837805e-05,
+      "loss": 0.7315,
+      "step": 3417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889012238992399e-05,
+      "loss": 0.7983,
+      "step": 3418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888935491032644e-05,
+      "loss": 0.7873,
+      "step": 3419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888587181066705e-05,
+      "loss": 0.7463,
+      "step": 3420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888781920216635e-05,
+      "loss": 0.7467,
+      "step": 3421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887050973646938e-05,
+      "loss": 0.757,
+      "step": 3422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886282495530053e-05,
+      "loss": 0.7902,
+      "step": 3423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885513767837277e-05,
+      "loss": 0.829,
+      "step": 3424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888474479059019e-05,
+      "loss": 0.7494,
+      "step": 3425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883975563810404e-05,
+      "loss": 0.3451,
+      "step": 3426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883206087519512e-05,
+      "loss": 0.7611,
+      "step": 3427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888243636173913e-05,
+      "loss": 0.768,
+      "step": 3428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881666386490874e-05,
+      "loss": 0.7738,
+      "step": 3429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880896161796374e-05,
+      "loss": 0.7958,
+      "step": 3430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888012568767726e-05,
+      "loss": 0.7802,
+      "step": 3431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879354964155166e-05,
+      "loss": 0.3441,
+      "step": 3432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887858399125174e-05,
+      "loss": 0.7968,
+      "step": 3433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877812768988643e-05,
+      "loss": 0.8506,
+      "step": 3434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877041297387526e-05,
+      "loss": 0.7946,
+      "step": 3435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876269576470057e-05,
+      "loss": 0.8124,
+      "step": 3436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875497606257917e-05,
+      "loss": 0.8751,
+      "step": 3437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874725386772783e-05,
+      "loss": 0.7968,
+      "step": 3438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887395291803634e-05,
+      "loss": 0.775,
+      "step": 3439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887318020007029e-05,
+      "loss": 0.7292,
+      "step": 3440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872407232896325e-05,
+      "loss": 0.8536,
+      "step": 3441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871634016536167e-05,
+      "loss": 0.7592,
+      "step": 3442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870860551011524e-05,
+      "loss": 0.7578,
+      "step": 3443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870086836344122e-05,
+      "loss": 0.8286,
+      "step": 3444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886931287255569e-05,
+      "loss": 0.7892,
+      "step": 3445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886853865966796e-05,
+      "loss": 0.7767,
+      "step": 3446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867764197702688e-05,
+      "loss": 0.7941,
+      "step": 3447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866989486681615e-05,
+      "loss": 0.7733,
+      "step": 3448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866214526626507e-05,
+      "loss": 0.7662,
+      "step": 3449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886543931755912e-05,
+      "loss": 0.7894,
+      "step": 3450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864663859501236e-05,
+      "loss": 0.8051,
+      "step": 3451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863888152474626e-05,
+      "loss": 0.735,
+      "step": 3452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886311219650108e-05,
+      "loss": 0.7629,
+      "step": 3453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8862335991602392e-05,
+      "loss": 0.8052,
+      "step": 3454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886155953780036e-05,
+      "loss": 0.7186,
+      "step": 3455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886078283511679e-05,
+      "loss": 0.8555,
+      "step": 3456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88600058835735e-05,
+      "loss": 0.762,
+      "step": 3457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859228683192307e-05,
+      "loss": 0.8228,
+      "step": 3458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858451233995042e-05,
+      "loss": 0.7878,
+      "step": 3459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885767353600354e-05,
+      "loss": 0.7532,
+      "step": 3460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856895589239644e-05,
+      "loss": 0.8007,
+      "step": 3461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88561173937252e-05,
+      "loss": 0.7277,
+      "step": 3462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855338949482066e-05,
+      "loss": 0.7449,
+      "step": 3463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 0.7408,
+      "step": 3464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885378131489718e-05,
+      "loss": 0.8288,
+      "step": 3465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885300212459918e-05,
+      "loss": 0.7604,
+      "step": 3466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885222268565998e-05,
+      "loss": 0.7317,
+      "step": 3467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851442998101475e-05,
+      "loss": 0.7678,
+      "step": 3468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850663061945564e-05,
+      "loss": 0.7325,
+      "step": 3469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884988287721415e-05,
+      "loss": 0.7691,
+      "step": 3470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849102443929145e-05,
+      "loss": 0.7454,
+      "step": 3471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848321762112464e-05,
+      "loss": 0.7416,
+      "step": 3472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884754083178604e-05,
+      "loss": 0.7512,
+      "step": 3473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846759652971802e-05,
+      "loss": 0.7186,
+      "step": 3474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845978225691694e-05,
+      "loss": 0.7952,
+      "step": 3475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845196549967653e-05,
+      "loss": 0.7399,
+      "step": 3476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884441462582164e-05,
+      "loss": 0.7365,
+      "step": 3477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884363245327562e-05,
+      "loss": 0.7469,
+      "step": 3478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884285003235155e-05,
+      "loss": 0.7543,
+      "step": 3479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884206736307141e-05,
+      "loss": 0.7853,
+      "step": 3480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841284445457187e-05,
+      "loss": 0.752,
+      "step": 3481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840501279530857e-05,
+      "loss": 0.7699,
+      "step": 3482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883971786531443e-05,
+      "loss": 0.8284,
+      "step": 3483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88389342028299e-05,
+      "loss": 0.81,
+      "step": 3484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838150292099273e-05,
+      "loss": 0.7329,
+      "step": 3485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837366133144574e-05,
+      "loss": 0.8179,
+      "step": 3486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836581725987826e-05,
+      "loss": 0.777,
+      "step": 3487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835797070651054e-05,
+      "loss": 0.7557,
+      "step": 3488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88350121671563e-05,
+      "loss": 0.7476,
+      "step": 3489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834227015525606e-05,
+      "loss": 0.7984,
+      "step": 3490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833441615781024e-05,
+      "loss": 0.7081,
+      "step": 3491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832655967944607e-05,
+      "loss": 0.7666,
+      "step": 3492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831870072038434e-05,
+      "loss": 0.7498,
+      "step": 3493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831083928084565e-05,
+      "loss": 0.7717,
+      "step": 3494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830297536105086e-05,
+      "loss": 0.7379,
+      "step": 3495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829510896122075e-05,
+      "loss": 0.7567,
+      "step": 3496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828724008157632e-05,
+      "loss": 0.7345,
+      "step": 3497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882793687223386e-05,
+      "loss": 0.7462,
+      "step": 3498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882714948837286e-05,
+      "loss": 0.7869,
+      "step": 3499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826361856596745e-05,
+      "loss": 0.7355,
+      "step": 3500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825573976927646e-05,
+      "loss": 0.784,
+      "step": 3501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882478584938768e-05,
+      "loss": 0.8051,
+      "step": 3502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823997473998987e-05,
+      "loss": 0.8436,
+      "step": 3503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882320885078371e-05,
+      "loss": 0.7093,
+      "step": 3504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882241997976399e-05,
+      "loss": 0.8572,
+      "step": 3505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882163086096199e-05,
+      "loss": 0.7592,
+      "step": 3506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820841494399873e-05,
+      "loss": 0.7507,
+      "step": 3507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882005188009981e-05,
+      "loss": 0.733,
+      "step": 3508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881926201808397e-05,
+      "loss": 0.7628,
+      "step": 3509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818471908374548e-05,
+      "loss": 0.7023,
+      "step": 3510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881768155099373e-05,
+      "loss": 0.6955,
+      "step": 3511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88168909459637e-05,
+      "loss": 0.8295,
+      "step": 3512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816100093306687e-05,
+      "loss": 0.8284,
+      "step": 3513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815308993044884e-05,
+      "loss": 0.6423,
+      "step": 3514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814517645200518e-05,
+      "loss": 0.7948,
+      "step": 3515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813726049795807e-05,
+      "loss": 0.7749,
+      "step": 3516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812934206852995e-05,
+      "loss": 0.7379,
+      "step": 3517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812142116394308e-05,
+      "loss": 0.7632,
+      "step": 3518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811349778442006e-05,
+      "loss": 0.8234,
+      "step": 3519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810557193018332e-05,
+      "loss": 0.7591,
+      "step": 3520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809764360145547e-05,
+      "loss": 0.7965,
+      "step": 3521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808971279845923e-05,
+      "loss": 0.8386,
+      "step": 3522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8808177952141734e-05,
+      "loss": 0.7488,
+      "step": 3523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807384377055255e-05,
+      "loss": 0.7422,
+      "step": 3524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806590554608778e-05,
+      "loss": 0.7721,
+      "step": 3525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805796484824598e-05,
+      "loss": 0.7605,
+      "step": 3526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805002167725018e-05,
+      "loss": 0.8184,
+      "step": 3527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804207603332344e-05,
+      "loss": 0.739,
+      "step": 3528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8803412791668895e-05,
+      "loss": 0.8392,
+      "step": 3529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802617732756986e-05,
+      "loss": 0.7489,
+      "step": 3530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801822426618956e-05,
+      "loss": 0.7319,
+      "step": 3531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801026873277138e-05,
+      "loss": 0.6984,
+      "step": 3532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800231072753874e-05,
+      "loss": 0.8175,
+      "step": 3533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799435025071515e-05,
+      "loss": 0.7917,
+      "step": 3534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798638730252424e-05,
+      "loss": 0.8173,
+      "step": 3535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797842188318958e-05,
+      "loss": 0.7805,
+      "step": 3536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879704539929349e-05,
+      "loss": 0.7536,
+      "step": 3537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87962483631984e-05,
+      "loss": 0.7367,
+      "step": 3538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879545108005607e-05,
+      "loss": 0.3319,
+      "step": 3539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794653549888896e-05,
+      "loss": 0.7504,
+      "step": 3540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793855772719277e-05,
+      "loss": 0.8035,
+      "step": 3541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793057748569613e-05,
+      "loss": 0.7337,
+      "step": 3542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792259477462326e-05,
+      "loss": 0.7797,
+      "step": 3543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791460959419832e-05,
+      "loss": 0.698,
+      "step": 3544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790662194464554e-05,
+      "loss": 0.7833,
+      "step": 3545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878986318261893e-05,
+      "loss": 0.7471,
+      "step": 3546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87890639239054e-05,
+      "loss": 0.735,
+      "step": 3547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878826441834641e-05,
+      "loss": 0.7581,
+      "step": 3548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787464665964416e-05,
+      "loss": 0.7815,
+      "step": 3549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786664666781882e-05,
+      "loss": 0.8349,
+      "step": 3550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878586442082127e-05,
+      "loss": 0.7792,
+      "step": 3551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878506392810506e-05,
+      "loss": 0.7649,
+      "step": 3552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784263188655733e-05,
+      "loss": 0.7518,
+      "step": 3553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783462202495776e-05,
+      "loss": 0.7615,
+      "step": 3554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782660969647688e-05,
+      "loss": 0.7981,
+      "step": 3555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781859490133977e-05,
+      "loss": 0.8163,
+      "step": 3556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878105776397714e-05,
+      "loss": 0.6724,
+      "step": 3557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780255791199704e-05,
+      "loss": 0.7449,
+      "step": 3558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877945357182419e-05,
+      "loss": 0.7679,
+      "step": 3559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877865110587313e-05,
+      "loss": 0.8584,
+      "step": 3560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877784839336906e-05,
+      "loss": 0.7523,
+      "step": 3561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777045434334527e-05,
+      "loss": 0.7949,
+      "step": 3562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776242228792076e-05,
+      "loss": 0.8039,
+      "step": 3563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775438776764273e-05,
+      "loss": 0.7323,
+      "step": 3564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774635078273678e-05,
+      "loss": 0.7768,
+      "step": 3565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877383113334287e-05,
+      "loss": 0.7541,
+      "step": 3566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877302694199442e-05,
+      "loss": 0.7513,
+      "step": 3567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877222250425092e-05,
+      "loss": 0.7994,
+      "step": 3568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771417820134963e-05,
+      "loss": 0.7322,
+      "step": 3569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770612889669146e-05,
+      "loss": 0.8101,
+      "step": 3570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769807712876075e-05,
+      "loss": 0.754,
+      "step": 3571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876900228977837e-05,
+      "loss": 0.756,
+      "step": 3572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768196620398644e-05,
+      "loss": 0.7589,
+      "step": 3573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8767390704759525e-05,
+      "loss": 0.7893,
+      "step": 3574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8766584542883657e-05,
+      "loss": 0.7696,
+      "step": 3575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765778134793675e-05,
+      "loss": 0.7385,
+      "step": 3576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876497148051222e-05,
+      "loss": 0.7824,
+      "step": 3577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764164580061963e-05,
+      "loss": 0.7875,
+      "step": 3578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763357433465556e-05,
+      "loss": 0.768,
+      "step": 3579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876255004074567e-05,
+      "loss": 0.797,
+      "step": 3580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876174240192498e-05,
+      "loss": 0.8444,
+      "step": 3581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876093451702617e-05,
+      "loss": 0.7189,
+      "step": 3582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760126386071937e-05,
+      "loss": 0.7679,
+      "step": 3583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759318009084965e-05,
+      "loss": 0.705,
+      "step": 3584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758509386087963e-05,
+      "loss": 0.7624,
+      "step": 3585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757700517103645e-05,
+      "loss": 0.8592,
+      "step": 3586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875689140215472e-05,
+      "loss": 0.8027,
+      "step": 3587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756082041263924e-05,
+      "loss": 0.744,
+      "step": 3588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755272434453983e-05,
+      "loss": 0.8124,
+      "step": 3589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754462581747633e-05,
+      "loss": 0.7052,
+      "step": 3590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753652483167622e-05,
+      "loss": 0.7349,
+      "step": 3591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87528421387367e-05,
+      "loss": 0.7773,
+      "step": 3592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752031548477626e-05,
+      "loss": 0.7228,
+      "step": 3593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751220712413166e-05,
+      "loss": 0.6695,
+      "step": 3594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750409630566095e-05,
+      "loss": 0.7266,
+      "step": 3595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749598302959195e-05,
+      "loss": 0.7685,
+      "step": 3596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748786729615242e-05,
+      "loss": 0.3367,
+      "step": 3597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874797491055704e-05,
+      "loss": 0.7787,
+      "step": 3598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747162845807384e-05,
+      "loss": 0.8467,
+      "step": 3599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746350535389082e-05,
+      "loss": 0.7217,
+      "step": 3600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874553797932495e-05,
+      "loss": 0.7365,
+      "step": 3601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744725177637807e-05,
+      "loss": 0.8243,
+      "step": 3602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743912130350484e-05,
+      "loss": 0.7333,
+      "step": 3603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743098837485813e-05,
+      "loss": 0.7511,
+      "step": 3604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742285299066636e-05,
+      "loss": 0.8068,
+      "step": 3605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741471515115807e-05,
+      "loss": 0.7751,
+      "step": 3606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874065748565617e-05,
+      "loss": 0.8318,
+      "step": 3607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739843210710597e-05,
+      "loss": 0.7406,
+      "step": 3608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739028690301953e-05,
+      "loss": 0.7577,
+      "step": 3609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873821392445312e-05,
+      "loss": 0.7256,
+      "step": 3610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873739891318697e-05,
+      "loss": 0.7517,
+      "step": 3611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736583656526404e-05,
+      "loss": 0.7498,
+      "step": 3612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735768154494317e-05,
+      "loss": 0.3368,
+      "step": 3613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734952407113607e-05,
+      "loss": 0.8318,
+      "step": 3614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734136414407193e-05,
+      "loss": 0.8551,
+      "step": 3615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8733320176397983e-05,
+      "loss": 0.7294,
+      "step": 3616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873250369310891e-05,
+      "loss": 0.7076,
+      "step": 3617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87316869645629e-05,
+      "loss": 0.758,
+      "step": 3618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873086999078289e-05,
+      "loss": 0.7864,
+      "step": 3619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873005277179183e-05,
+      "loss": 0.7542,
+      "step": 3620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8729235307612672e-05,
+      "loss": 0.7492,
+      "step": 3621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728417598268375e-05,
+      "loss": 0.7088,
+      "step": 3622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727599643781897e-05,
+      "loss": 0.777,
+      "step": 3623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872678144417622e-05,
+      "loss": 0.7856,
+      "step": 3624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872596299947432e-05,
+      "loss": 0.7287,
+      "step": 3625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725144309699184e-05,
+      "loss": 0.7655,
+      "step": 3626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724325374873805e-05,
+      "loss": 0.6692,
+      "step": 3627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723506195021182e-05,
+      "loss": 0.7791,
+      "step": 3628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722686770164326e-05,
+      "loss": 0.8373,
+      "step": 3629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721867100326242e-05,
+      "loss": 0.7986,
+      "step": 3630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872104718552996e-05,
+      "loss": 0.7696,
+      "step": 3631
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720227025798507e-05,
+      "loss": 0.7623,
+      "step": 3632
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719406621154914e-05,
+      "loss": 0.7345,
+      "step": 3633
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718585971622223e-05,
+      "loss": 0.7831,
+      "step": 3634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717765077223485e-05,
+      "loss": 0.7608,
+      "step": 3635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716943937981753e-05,
+      "loss": 0.7317,
+      "step": 3636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716122553920088e-05,
+      "loss": 0.801,
+      "step": 3637
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871530092506156e-05,
+      "loss": 0.7307,
+      "step": 3638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871447905142925e-05,
+      "loss": 0.7804,
+      "step": 3639
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713656933046232e-05,
+      "loss": 0.7866,
+      "step": 3640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87128345699356e-05,
+      "loss": 0.8068,
+      "step": 3641
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712011962120455e-05,
+      "loss": 0.8048,
+      "step": 3642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711189109623893e-05,
+      "loss": 0.8022,
+      "step": 3643
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710366012469026e-05,
+      "loss": 0.7423,
+      "step": 3644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870954267067897e-05,
+      "loss": 0.7437,
+      "step": 3645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708719084276856e-05,
+      "loss": 0.7346,
+      "step": 3646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707895253285806e-05,
+      "loss": 0.7683,
+      "step": 3647
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707071177728962e-05,
+      "loss": 0.7877,
+      "step": 3648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870624685762947e-05,
+      "loss": 0.6842,
+      "step": 3649
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870542229301048e-05,
+      "loss": 0.6898,
+      "step": 3650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870459748389515e-05,
+      "loss": 0.8403,
+      "step": 3651
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703772430306644e-05,
+      "loss": 0.7779,
+      "step": 3652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702947132268137e-05,
+      "loss": 0.7842,
+      "step": 3653
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702121589802806e-05,
+      "loss": 0.8307,
+      "step": 3654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701295802933837e-05,
+      "loss": 0.7876,
+      "step": 3655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870046977168442e-05,
+      "loss": 0.7971,
+      "step": 3656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869964349607776e-05,
+      "loss": 0.7517,
+      "step": 3657
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869881697613706e-05,
+      "loss": 0.7946,
+      "step": 3658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869799021188553e-05,
+      "loss": 0.8156,
+      "step": 3659
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697163203346397e-05,
+      "loss": 0.8629,
+      "step": 3660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696335950542886e-05,
+      "loss": 0.7836,
+      "step": 3661
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869550845349823e-05,
+      "loss": 0.7646,
+      "step": 3662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694680712235668e-05,
+      "loss": 0.6938,
+      "step": 3663
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869385272677845e-05,
+      "loss": 0.8345,
+      "step": 3664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869302449714983e-05,
+      "loss": 0.7725,
+      "step": 3665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692196023373067e-05,
+      "loss": 0.7862,
+      "step": 3666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869136730547143e-05,
+      "loss": 0.8671,
+      "step": 3667
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690538343468195e-05,
+      "loss": 0.8612,
+      "step": 3668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689709137386643e-05,
+      "loss": 0.8132,
+      "step": 3669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688879687250067e-05,
+      "loss": 0.7819,
+      "step": 3670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688049993081755e-05,
+      "loss": 0.7705,
+      "step": 3671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687220054905015e-05,
+      "loss": 0.7328,
+      "step": 3672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686389872743156e-05,
+      "loss": 0.7458,
+      "step": 3673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868555944661949e-05,
+      "loss": 0.817,
+      "step": 3674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868472877655734e-05,
+      "loss": 0.7576,
+      "step": 3675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868389786258004e-05,
+      "loss": 0.7227,
+      "step": 3676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683066704710932e-05,
+      "loss": 0.7861,
+      "step": 3677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682235302973346e-05,
+      "loss": 0.7311,
+      "step": 3678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681403657390638e-05,
+      "loss": 0.7354,
+      "step": 3679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680571767986165e-05,
+      "loss": 0.8342,
+      "step": 3680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86797396347833e-05,
+      "loss": 0.7954,
+      "step": 3681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678907257805398e-05,
+      "loss": 0.7926,
+      "step": 3682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678074637075847e-05,
+      "loss": 0.7423,
+      "step": 3683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867724177261803e-05,
+      "loss": 0.8268,
+      "step": 3684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676408664455336e-05,
+      "loss": 0.7645,
+      "step": 3685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675575312611166e-05,
+      "loss": 0.7148,
+      "step": 3686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674741717108923e-05,
+      "loss": 0.6934,
+      "step": 3687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673907877972025e-05,
+      "loss": 0.7188,
+      "step": 3688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867307379522388e-05,
+      "loss": 0.8041,
+      "step": 3689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672239468887923e-05,
+      "loss": 0.7772,
+      "step": 3690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671404898987583e-05,
+      "loss": 0.7139,
+      "step": 3691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670570085546294e-05,
+      "loss": 0.7873,
+      "step": 3692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669735028587515e-05,
+      "loss": 0.7255,
+      "step": 3693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668899728134686e-05,
+      "loss": 0.8108,
+      "step": 3694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668064184211275e-05,
+      "loss": 0.8304,
+      "step": 3695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667228396840746e-05,
+      "loss": 0.7357,
+      "step": 3696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666392366046572e-05,
+      "loss": 0.7191,
+      "step": 3697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665556091852235e-05,
+      "loss": 0.7445,
+      "step": 3698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866471957428122e-05,
+      "loss": 0.7271,
+      "step": 3699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866388281335702e-05,
+      "loss": 0.801,
+      "step": 3700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866304580910314e-05,
+      "loss": 0.7681,
+      "step": 3701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662208561543092e-05,
+      "loss": 0.7643,
+      "step": 3702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8661371070700374e-05,
+      "loss": 0.7459,
+      "step": 3703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660533336598525e-05,
+      "loss": 0.7953,
+      "step": 3704
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659695359261062e-05,
+      "loss": 0.8101,
+      "step": 3705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658857138711526e-05,
+      "loss": 0.7489,
+      "step": 3706
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658018674973457e-05,
+      "loss": 0.7088,
+      "step": 3707
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657179968070402e-05,
+      "loss": 0.7792,
+      "step": 3708
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865634101802592e-05,
+      "loss": 0.7668,
+      "step": 3709
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655501824863568e-05,
+      "loss": 0.7184,
+      "step": 3710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654662388606918e-05,
+      "loss": 0.7209,
+      "step": 3711
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653822709279553e-05,
+      "loss": 0.8726,
+      "step": 3712
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652982786905043e-05,
+      "loss": 0.7027,
+      "step": 3713
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652142621506985e-05,
+      "loss": 0.7276,
+      "step": 3714
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865130221310898e-05,
+      "loss": 0.8573,
+      "step": 3715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865046156173462e-05,
+      "loss": 0.8323,
+      "step": 3716
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864962066740752e-05,
+      "loss": 0.6626,
+      "step": 3717
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86487795301513e-05,
+      "loss": 0.8057,
+      "step": 3718
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647938149989584e-05,
+      "loss": 0.7455,
+      "step": 3719
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647096526946e-05,
+      "loss": 0.7636,
+      "step": 3720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646254661044182e-05,
+      "loss": 0.7192,
+      "step": 3721
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864541255230778e-05,
+      "loss": 0.7337,
+      "step": 3722
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644570200760445e-05,
+      "loss": 0.8159,
+      "step": 3723
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864372760642583e-05,
+      "loss": 0.7687,
+      "step": 3724
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642884769327602e-05,
+      "loss": 0.7889,
+      "step": 3725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642041689489435e-05,
+      "loss": 0.6747,
+      "step": 3726
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641198366935006e-05,
+      "loss": 0.7199,
+      "step": 3727
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640354801688e-05,
+      "loss": 0.7621,
+      "step": 3728
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639510993772102e-05,
+      "loss": 0.7402,
+      "step": 3729
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638666943211026e-05,
+      "loss": 0.7579,
+      "step": 3730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637822650028466e-05,
+      "loss": 0.7352,
+      "step": 3731
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636978114248134e-05,
+      "loss": 0.7138,
+      "step": 3732
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636133335893757e-05,
+      "loss": 0.7895,
+      "step": 3733
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635288314989054e-05,
+      "loss": 0.6893,
+      "step": 3734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634443051557764e-05,
+      "loss": 0.7152,
+      "step": 3735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863359754562362e-05,
+      "loss": 0.7643,
+      "step": 3736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632751797210373e-05,
+      "loss": 0.839,
+      "step": 3737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631905806341773e-05,
+      "loss": 0.7173,
+      "step": 3738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631059573041583e-05,
+      "loss": 0.7839,
+      "step": 3739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863021309733357e-05,
+      "loss": 0.7648,
+      "step": 3740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629366379241508e-05,
+      "loss": 0.7606,
+      "step": 3741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8628519418789174e-05,
+      "loss": 0.7442,
+      "step": 3742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862767221600036e-05,
+      "loss": 0.7788,
+      "step": 3743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626824770898856e-05,
+      "loss": 0.7515,
+      "step": 3744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862597708350846e-05,
+      "loss": 0.7964,
+      "step": 3745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862512915385299e-05,
+      "loss": 0.8027,
+      "step": 3746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624280981956252e-05,
+      "loss": 0.7539,
+      "step": 3747
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623432567842075e-05,
+      "loss": 0.7017,
+      "step": 3748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622583911534276e-05,
+      "loss": 0.8015,
+      "step": 3749
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86217350130567e-05,
+      "loss": 0.6932,
+      "step": 3750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620885872433184e-05,
+      "loss": 0.7253,
+      "step": 3751
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620036489687578e-05,
+      "loss": 0.8213,
+      "step": 3752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619186864843737e-05,
+      "loss": 0.7645,
+      "step": 3753
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618336997925524e-05,
+      "loss": 0.773,
+      "step": 3754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617486888956807e-05,
+      "loss": 0.7448,
+      "step": 3755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861663653796146e-05,
+      "loss": 0.7989,
+      "step": 3756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861578594496337e-05,
+      "loss": 0.7878,
+      "step": 3757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861493510998642e-05,
+      "loss": 0.7625,
+      "step": 3758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614084033054514e-05,
+      "loss": 0.8436,
+      "step": 3759
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613232714191548e-05,
+      "loss": 0.8134,
+      "step": 3760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612381153421437e-05,
+      "loss": 0.8069,
+      "step": 3761
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611529350768092e-05,
+      "loss": 0.7535,
+      "step": 3762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610677306255443e-05,
+      "loss": 0.724,
+      "step": 3763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609825019907413e-05,
+      "loss": 0.7689,
+      "step": 3764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608972491747946e-05,
+      "loss": 0.7064,
+      "step": 3765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860811972180098e-05,
+      "loss": 0.7546,
+      "step": 3766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860726671009047e-05,
+      "loss": 0.749,
+      "step": 3767
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860641345664037e-05,
+      "loss": 0.715,
+      "step": 3768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605559961474642e-05,
+      "loss": 0.7247,
+      "step": 3769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604706224617262e-05,
+      "loss": 0.7944,
+      "step": 3770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603852246092208e-05,
+      "loss": 0.8067,
+      "step": 3771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860299802592346e-05,
+      "loss": 0.7411,
+      "step": 3772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860214356413501e-05,
+      "loss": 0.8014,
+      "step": 3773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8601288860750857e-05,
+      "loss": 0.7121,
+      "step": 3774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600433915795007e-05,
+      "loss": 0.752,
+      "step": 3775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599578729291468e-05,
+      "loss": 0.7166,
+      "step": 3776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598723301264264e-05,
+      "loss": 0.7276,
+      "step": 3777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859786763173742e-05,
+      "loss": 0.7941,
+      "step": 3778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859701172073496e-05,
+      "loss": 0.7354,
+      "step": 3779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859615556828093e-05,
+      "loss": 0.6995,
+      "step": 3780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859529917439937e-05,
+      "loss": 0.8412,
+      "step": 3781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859444253911434e-05,
+      "loss": 0.8372,
+      "step": 3782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593585662449892e-05,
+      "loss": 0.7811,
+      "step": 3783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859272854443009e-05,
+      "loss": 0.6938,
+      "step": 3784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591871185079018e-05,
+      "loss": 0.7264,
+      "step": 3785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591013584420747e-05,
+      "loss": 0.7805,
+      "step": 3786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590155742479365e-05,
+      "loss": 0.7701,
+      "step": 3787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589297659278964e-05,
+      "loss": 0.7969,
+      "step": 3788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858843933484364e-05,
+      "loss": 0.7965,
+      "step": 3789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587580769197507e-05,
+      "loss": 0.7954,
+      "step": 3790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586721962364676e-05,
+      "loss": 0.8154,
+      "step": 3791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858586291436926e-05,
+      "loss": 0.7884,
+      "step": 3792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85850036252354e-05,
+      "loss": 0.7821,
+      "step": 3793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584144094987214e-05,
+      "loss": 0.7289,
+      "step": 3794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583284323648855e-05,
+      "loss": 0.8108,
+      "step": 3795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858242431124446e-05,
+      "loss": 0.8108,
+      "step": 3796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581564057798194e-05,
+      "loss": 0.735,
+      "step": 3797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580703563334208e-05,
+      "loss": 0.7696,
+      "step": 3798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579842827876673e-05,
+      "loss": 0.7134,
+      "step": 3799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578981851449764e-05,
+      "loss": 0.7578,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857812063407766e-05,
+      "loss": 0.7474,
+      "step": 3801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857725917578455e-05,
+      "loss": 0.8149,
+      "step": 3802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576397476594627e-05,
+      "loss": 0.8001,
+      "step": 3803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575535536532093e-05,
+      "loss": 0.8283,
+      "step": 3804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857467335562116e-05,
+      "loss": 0.7105,
+      "step": 3805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573810933886038e-05,
+      "loss": 0.7543,
+      "step": 3806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857294827135095e-05,
+      "loss": 0.7785,
+      "step": 3807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572085368040126e-05,
+      "loss": 0.7512,
+      "step": 3808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8571222223977795e-05,
+      "loss": 0.7261,
+      "step": 3809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570358839188202e-05,
+      "loss": 0.8038,
+      "step": 3810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569495213695603e-05,
+      "loss": 0.7282,
+      "step": 3811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856863134752424e-05,
+      "loss": 0.7244,
+      "step": 3812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856776724069839e-05,
+      "loss": 0.745,
+      "step": 3813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856690289324231e-05,
+      "loss": 0.7832,
+      "step": 3814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856603830518028e-05,
+      "loss": 0.9067,
+      "step": 3815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565173476536585e-05,
+      "loss": 0.7592,
+      "step": 3816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564308407335508e-05,
+      "loss": 0.8156,
+      "step": 3817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856344309760135e-05,
+      "loss": 0.8089,
+      "step": 3818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856257754735841e-05,
+      "loss": 0.7459,
+      "step": 3819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561711756631004e-05,
+      "loss": 0.7714,
+      "step": 3820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560845725443438e-05,
+      "loss": 0.7821,
+      "step": 3821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559979453820044e-05,
+      "loss": 0.8022,
+      "step": 3822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855911294178515e-05,
+      "loss": 0.8176,
+      "step": 3823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558246189363086e-05,
+      "loss": 0.7584,
+      "step": 3824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8557379196578202e-05,
+      "loss": 0.7342,
+      "step": 3825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855651196345485e-05,
+      "loss": 0.7877,
+      "step": 3826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855564449001738e-05,
+      "loss": 0.8196,
+      "step": 3827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554776776290153e-05,
+      "loss": 0.7834,
+      "step": 3828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855390882229755e-05,
+      "loss": 0.7318,
+      "step": 3829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855304062806394e-05,
+      "loss": 0.7776,
+      "step": 3830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552172193613705e-05,
+      "loss": 0.3176,
+      "step": 3831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551303518971247e-05,
+      "loss": 0.7892,
+      "step": 3832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855043460416095e-05,
+      "loss": 0.7879,
+      "step": 3833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549565449207224e-05,
+      "loss": 0.7022,
+      "step": 3834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548696054134482e-05,
+      "loss": 0.7485,
+      "step": 3835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547826418967138e-05,
+      "loss": 0.8261,
+      "step": 3836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546956543729616e-05,
+      "loss": 0.6982,
+      "step": 3837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546086428446345e-05,
+      "loss": 0.825,
+      "step": 3838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545216073141766e-05,
+      "loss": 0.806,
+      "step": 3839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544345477840323e-05,
+      "loss": 0.7234,
+      "step": 3840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8543474642566467e-05,
+      "loss": 0.753,
+      "step": 3841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542603567344656e-05,
+      "loss": 0.8057,
+      "step": 3842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541732252199357e-05,
+      "loss": 0.7824,
+      "step": 3843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540860697155036e-05,
+      "loss": 0.8354,
+      "step": 3844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853998890223617e-05,
+      "loss": 0.6812,
+      "step": 3845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539116867467252e-05,
+      "loss": 0.7617,
+      "step": 3846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538244592872768e-05,
+      "loss": 0.7524,
+      "step": 3847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537372078477216e-05,
+      "loss": 0.7847,
+      "step": 3848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536499324305102e-05,
+      "loss": 0.7676,
+      "step": 3849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853562633038094e-05,
+      "loss": 0.8387,
+      "step": 3850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534753096729244e-05,
+      "loss": 0.7469,
+      "step": 3851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533879623374543e-05,
+      "loss": 0.6975,
+      "step": 3852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853300591034137e-05,
+      "loss": 0.7538,
+      "step": 3853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532131957654257e-05,
+      "loss": 0.7449,
+      "step": 3854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531257765337757e-05,
+      "loss": 0.7531,
+      "step": 3855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853038333341642e-05,
+      "loss": 0.7273,
+      "step": 3856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85295086619148e-05,
+      "loss": 0.7662,
+      "step": 3857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528633750857466e-05,
+      "loss": 0.73,
+      "step": 3858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527758600268994e-05,
+      "loss": 0.8086,
+      "step": 3859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526883210173957e-05,
+      "loss": 0.7423,
+      "step": 3860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526007580596946e-05,
+      "loss": 0.7728,
+      "step": 3861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852513171156255e-05,
+      "loss": 0.7339,
+      "step": 3862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524255603095368e-05,
+      "loss": 0.7922,
+      "step": 3863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852337925522001e-05,
+      "loss": 0.7793,
+      "step": 3864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522502667961085e-05,
+      "loss": 0.8115,
+      "step": 3865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521625841343214e-05,
+      "loss": 0.7571,
+      "step": 3866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520748775391024e-05,
+      "loss": 0.8336,
+      "step": 3867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519871470129143e-05,
+      "loss": 0.7477,
+      "step": 3868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518993925582217e-05,
+      "loss": 0.7632,
+      "step": 3869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518116141774886e-05,
+      "loss": 0.8182,
+      "step": 3870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517238118731807e-05,
+      "loss": 0.7001,
+      "step": 3871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851635985647764e-05,
+      "loss": 0.6783,
+      "step": 3872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515481355037052e-05,
+      "loss": 0.8333,
+      "step": 3873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514602614434714e-05,
+      "loss": 0.7899,
+      "step": 3874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513723634695308e-05,
+      "loss": 0.6898,
+      "step": 3875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512844415843514e-05,
+      "loss": 0.7756,
+      "step": 3876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511964957904034e-05,
+      "loss": 0.7804,
+      "step": 3877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511085260901566e-05,
+      "loss": 0.7297,
+      "step": 3878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851020532486081e-05,
+      "loss": 0.742,
+      "step": 3879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850932514980649e-05,
+      "loss": 0.8381,
+      "step": 3880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508444735763315e-05,
+      "loss": 0.747,
+      "step": 3881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507564082756024e-05,
+      "loss": 0.767,
+      "step": 3882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506683190809344e-05,
+      "loss": 0.7399,
+      "step": 3883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505802059948012e-05,
+      "loss": 0.7955,
+      "step": 3884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504920690196782e-05,
+      "loss": 0.7515,
+      "step": 3885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85040390815804e-05,
+      "loss": 0.803,
+      "step": 3886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8503157234123635e-05,
+      "loss": 0.7728,
+      "step": 3887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850227514785125e-05,
+      "loss": 0.7671,
+      "step": 3888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850139282278802e-05,
+      "loss": 0.7814,
+      "step": 3889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850051025895872e-05,
+      "loss": 0.7151,
+      "step": 3890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499627456388144e-05,
+      "loss": 0.7501,
+      "step": 3891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498744415101083e-05,
+      "loss": 0.7371,
+      "step": 3892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849786113512234e-05,
+      "loss": 0.7356,
+      "step": 3893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849697761647672e-05,
+      "loss": 0.7827,
+      "step": 3894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496093859189037e-05,
+      "loss": 0.7567,
+      "step": 3895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495209863284115e-05,
+      "loss": 0.7936,
+      "step": 3896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849432562878678e-05,
+      "loss": 0.7649,
+      "step": 3897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849344115572186e-05,
+      "loss": 0.7502,
+      "step": 3898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492556444114204e-05,
+      "loss": 0.7577,
+      "step": 3899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491671493988658e-05,
+      "loss": 0.8299,
+      "step": 3900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490786305370073e-05,
+      "loss": 0.6475,
+      "step": 3901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848990087828331e-05,
+      "loss": 0.7867,
+      "step": 3902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489015212753246e-05,
+      "loss": 0.8043,
+      "step": 3903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488129308804742e-05,
+      "loss": 0.7981,
+      "step": 3904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487243166462688e-05,
+      "loss": 0.7283,
+      "step": 3905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486356785751966e-05,
+      "loss": 0.7721,
+      "step": 3906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485470166697476e-05,
+      "loss": 0.7557,
+      "step": 3907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848458330932412e-05,
+      "loss": 0.7887,
+      "step": 3908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483696213656793e-05,
+      "loss": 0.8137,
+      "step": 3909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482808879720427e-05,
+      "loss": 0.7895,
+      "step": 3910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848192130753993e-05,
+      "loss": 0.7531,
+      "step": 3911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848103349714024e-05,
+      "loss": 0.7296,
+      "step": 3912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480145448546286e-05,
+      "loss": 0.7858,
+      "step": 3913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847925716178301e-05,
+      "loss": 0.7882,
+      "step": 3914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478368636875357e-05,
+      "loss": 0.7045,
+      "step": 3915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847747987384829e-05,
+      "loss": 0.7617,
+      "step": 3916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847659087272676e-05,
+      "loss": 0.7262,
+      "step": 3917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847570163353574e-05,
+      "loss": 0.7847,
+      "step": 3918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847481215630021e-05,
+      "loss": 0.6685,
+      "step": 3919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473922441045142e-05,
+      "loss": 0.8194,
+      "step": 3920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847303248779553e-05,
+      "loss": 0.7597,
+      "step": 3921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472142296576366e-05,
+      "loss": 0.8005,
+      "step": 3922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471251867412655e-05,
+      "loss": 0.7494,
+      "step": 3923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470361200329403e-05,
+      "loss": 0.7354,
+      "step": 3924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469470295351625e-05,
+      "loss": 0.7699,
+      "step": 3925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846857915250434e-05,
+      "loss": 0.7775,
+      "step": 3926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846768777181258e-05,
+      "loss": 0.7764,
+      "step": 3927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466796153301376e-05,
+      "loss": 0.695,
+      "step": 3928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465904296995775e-05,
+      "loss": 0.8233,
+      "step": 3929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465012202920823e-05,
+      "loss": 0.7169,
+      "step": 3930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846411987110157e-05,
+      "loss": 0.686,
+      "step": 3931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463227301563086e-05,
+      "loss": 0.7637,
+      "step": 3932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462334494330435e-05,
+      "loss": 0.8213,
+      "step": 3933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846144144942869e-05,
+      "loss": 0.8145,
+      "step": 3934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460548166882938e-05,
+      "loss": 0.8218,
+      "step": 3935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845965464671826e-05,
+      "loss": 0.7288,
+      "step": 3936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458760888959758e-05,
+      "loss": 0.6979,
+      "step": 3937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457866893632528e-05,
+      "loss": 0.7947,
+      "step": 3938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845697266076168e-05,
+      "loss": 0.8006,
+      "step": 3939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456078190372333e-05,
+      "loss": 0.7581,
+      "step": 3940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8455183482489604e-05,
+      "loss": 0.8103,
+      "step": 3941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454288537138626e-05,
+      "loss": 0.7894,
+      "step": 3942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845339335434453e-05,
+      "loss": 0.7752,
+      "step": 3943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452497934132453e-05,
+      "loss": 0.7129,
+      "step": 3944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8451602276527553e-05,
+      "loss": 0.7924,
+      "step": 3945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450706381554982e-05,
+      "loss": 0.6879,
+      "step": 3946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84498102492399e-05,
+      "loss": 0.6998,
+      "step": 3947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448913879607477e-05,
+      "loss": 0.808,
+      "step": 3948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448017272682884e-05,
+      "loss": 0.7047,
+      "step": 3949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844712042849131e-05,
+      "loss": 0.7178,
+      "step": 3950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446223347057936e-05,
+      "loss": 0.7754,
+      "step": 3951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445326028407963e-05,
+      "loss": 0.7814,
+      "step": 3952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444428472566585e-05,
+      "loss": 0.8297,
+      "step": 3953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844353067955902e-05,
+      "loss": 0.8028,
+      "step": 3954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442632649410473e-05,
+      "loss": 0.8611,
+      "step": 3955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441734382146175e-05,
+      "loss": 0.7716,
+      "step": 3956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440835877791347e-05,
+      "loss": 0.7523,
+      "step": 3957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843993713637123e-05,
+      "loss": 0.7454,
+      "step": 3958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843903815791106e-05,
+      "loss": 0.7762,
+      "step": 3959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843813894243609e-05,
+      "loss": 0.7748,
+      "step": 3960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8437239489971572e-05,
+      "loss": 0.7704,
+      "step": 3961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436339800542766e-05,
+      "loss": 0.788,
+      "step": 3962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435439874174945e-05,
+      "loss": 0.7709,
+      "step": 3963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843453971089338e-05,
+      "loss": 0.7223,
+      "step": 3964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843363931072336e-05,
+      "loss": 0.7842,
+      "step": 3965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843273867369016e-05,
+      "loss": 0.7087,
+      "step": 3966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431837799819083e-05,
+      "loss": 0.7601,
+      "step": 3967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430936689135433e-05,
+      "loss": 0.7452,
+      "step": 3968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430035341664513e-05,
+      "loss": 0.7643,
+      "step": 3969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842913375743164e-05,
+      "loss": 0.7309,
+      "step": 3970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428231936462136e-05,
+      "loss": 0.7931,
+      "step": 3971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427329878781326e-05,
+      "loss": 0.734,
+      "step": 3972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842642758441455e-05,
+      "loss": 0.7957,
+      "step": 3973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425525053387145e-05,
+      "loss": 0.7318,
+      "step": 3974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424622285724457e-05,
+      "loss": 0.8268,
+      "step": 3975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842371928145185e-05,
+      "loss": 0.7726,
+      "step": 3976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422816040594677e-05,
+      "loss": 0.7383,
+      "step": 3977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842191256317831e-05,
+      "loss": 0.8353,
+      "step": 3978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842100884922812e-05,
+      "loss": 0.7914,
+      "step": 3979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842010489876949e-05,
+      "loss": 0.7554,
+      "step": 3980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419200711827808e-05,
+      "loss": 0.7852,
+      "step": 3981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418296288428473e-05,
+      "loss": 0.6959,
+      "step": 3982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841739162859688e-05,
+      "loss": 0.8118,
+      "step": 3983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416486732358436e-05,
+      "loss": 0.7681,
+      "step": 3984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841558159973856e-05,
+      "loss": 0.7468,
+      "step": 3985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414676230762674e-05,
+      "loss": 0.7701,
+      "step": 3986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84137706254562e-05,
+      "loss": 0.728,
+      "step": 3987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412864783844577e-05,
+      "loss": 0.3311,
+      "step": 3988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411958705953248e-05,
+      "loss": 0.7256,
+      "step": 3989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411052391807652e-05,
+      "loss": 0.742,
+      "step": 3990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410145841433255e-05,
+      "loss": 0.8178,
+      "step": 3991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409239054855507e-05,
+      "loss": 0.7892,
+      "step": 3992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840833203209988e-05,
+      "loss": 0.7278,
+      "step": 3993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840742477319185e-05,
+      "loss": 0.8165,
+      "step": 3994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406517278156896e-05,
+      "loss": 0.7199,
+      "step": 3995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405609547020508e-05,
+      "loss": 0.7725,
+      "step": 3996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404701579808174e-05,
+      "loss": 0.7933,
+      "step": 3997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84037933765454e-05,
+      "loss": 0.7523,
+      "step": 3998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402884937257693e-05,
+      "loss": 0.8445,
+      "step": 3999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401976261970565e-05,
+      "loss": 0.7195,
+      "step": 4000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840106735070954e-05,
+      "loss": 0.7183,
+      "step": 4001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840015820350014e-05,
+      "loss": 0.7631,
+      "step": 4002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399248820367902e-05,
+      "loss": 0.7063,
+      "step": 4003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8398339201338366e-05,
+      "loss": 0.7681,
+      "step": 4004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397429346437082e-05,
+      "loss": 0.6973,
+      "step": 4005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83965192556896e-05,
+      "loss": 0.7375,
+      "step": 4006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8395608929121477e-05,
+      "loss": 0.698,
+      "step": 4007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839469836675829e-05,
+      "loss": 0.8035,
+      "step": 4008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393787568625602e-05,
+      "loss": 0.7099,
+      "step": 4009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392876534749002e-05,
+      "loss": 0.7469,
+      "step": 4010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391965265154076e-05,
+      "loss": 0.7815,
+      "step": 4011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391053759866412e-05,
+      "loss": 0.7761,
+      "step": 4012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839014201891161e-05,
+      "loss": 0.736,
+      "step": 4013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838923004231528e-05,
+      "loss": 0.7815,
+      "step": 4014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388317830103037e-05,
+      "loss": 0.7352,
+      "step": 4015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83874053823005e-05,
+      "loss": 0.7542,
+      "step": 4016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386492698933294e-05,
+      "loss": 0.7146,
+      "step": 4017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838557978002705e-05,
+      "loss": 0.7416,
+      "step": 4018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838466662560741e-05,
+      "loss": 0.7338,
+      "step": 4019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383753235700026e-05,
+      "loss": 0.7201,
+      "step": 4020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382839610330542e-05,
+      "loss": 0.6874,
+      "step": 4021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838192574952462e-05,
+      "loss": 0.7266,
+      "step": 4022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381011653307933e-05,
+      "loss": 0.7578,
+      "step": 4023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838009732170614e-05,
+      "loss": 0.7677,
+      "step": 4024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379182754744936e-05,
+      "loss": 0.7246,
+      "step": 4025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837826795245e-05,
+      "loss": 0.3523,
+      "step": 4026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837735291484702e-05,
+      "loss": 0.7462,
+      "step": 4027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376437641961704e-05,
+      "loss": 0.7258,
+      "step": 4028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375522133819754e-05,
+      "loss": 0.7742,
+      "step": 4029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374606390446883e-05,
+      "loss": 0.8111,
+      "step": 4030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837369041186881e-05,
+      "loss": 0.7536,
+      "step": 4031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837277419811126e-05,
+      "loss": 0.7678,
+      "step": 4032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371857749199963e-05,
+      "loss": 0.7532,
+      "step": 4033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370941065160662e-05,
+      "loss": 0.8121,
+      "step": 4034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370024146019103e-05,
+      "loss": 0.7781,
+      "step": 4035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836910699180104e-05,
+      "loss": 0.7139,
+      "step": 4036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836818960253222e-05,
+      "loss": 0.7552,
+      "step": 4037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367271978238422e-05,
+      "loss": 0.7828,
+      "step": 4038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836635411894541e-05,
+      "loss": 0.7826,
+      "step": 4039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836543602467897e-05,
+      "loss": 0.6715,
+      "step": 4040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364517695464875e-05,
+      "loss": 0.7776,
+      "step": 4041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836359913132893e-05,
+      "loss": 0.7871,
+      "step": 4042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362680332296925e-05,
+      "loss": 0.7033,
+      "step": 4043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836176129839467e-05,
+      "loss": 0.369,
+      "step": 4044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360842029647966e-05,
+      "loss": 0.7609,
+      "step": 4045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359922526082647e-05,
+      "loss": 0.8201,
+      "step": 4046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835900278772453e-05,
+      "loss": 0.31,
+      "step": 4047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835808281459944e-05,
+      "loss": 0.7715,
+      "step": 4048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357162606733224e-05,
+      "loss": 0.7529,
+      "step": 4049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356242164151726e-05,
+      "loss": 0.798,
+      "step": 4050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835532148688079e-05,
+      "loss": 0.7819,
+      "step": 4051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835440057494628e-05,
+      "loss": 0.7268,
+      "step": 4052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835347942837406e-05,
+      "loss": 0.6865,
+      "step": 4053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835255804719e-05,
+      "loss": 0.7372,
+      "step": 4054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351636431419976e-05,
+      "loss": 0.7715,
+      "step": 4055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350714581089876e-05,
+      "loss": 0.7339,
+      "step": 4056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349792496225585e-05,
+      "loss": 0.7641,
+      "step": 4057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348870176853003e-05,
+      "loss": 0.7803,
+      "step": 4058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834794762299803e-05,
+      "loss": 0.7636,
+      "step": 4059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8347024834686586e-05,
+      "loss": 0.7892,
+      "step": 4060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834610181194458e-05,
+      "loss": 0.7502,
+      "step": 4061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834517855479794e-05,
+      "loss": 0.7493,
+      "step": 4062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834425506327259e-05,
+      "loss": 0.7612,
+      "step": 4063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343331337394472e-05,
+      "loss": 0.8137,
+      "step": 4064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834240737718953e-05,
+      "loss": 0.809,
+      "step": 4065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834148318268371e-05,
+      "loss": 0.7132,
+      "step": 4066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834055875390297e-05,
+      "loss": 0.7553,
+      "step": 4067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339634090873275e-05,
+      "loss": 0.7856,
+      "step": 4068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833870919362059e-05,
+      "loss": 0.7312,
+      "step": 4069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83377840621709e-05,
+      "loss": 0.7266,
+      "step": 4070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336858696550177e-05,
+      "loss": 0.8135,
+      "step": 4071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335933096784416e-05,
+      "loss": 0.715,
+      "step": 4072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335007262899615e-05,
+      "loss": 0.8177,
+      "step": 4073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334081194921772e-05,
+      "loss": 0.7477,
+      "step": 4074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333154892876898e-05,
+      "loss": 0.729,
+      "step": 4075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833222835679101e-05,
+      "loss": 0.7466,
+      "step": 4076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833130158669013e-05,
+      "loss": 0.719,
+      "step": 4077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330374582600285e-05,
+      "loss": 0.7314,
+      "step": 4078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329447344547507e-05,
+      "loss": 0.7816,
+      "step": 4079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328519872557846e-05,
+      "loss": 0.7709,
+      "step": 4080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327592166657346e-05,
+      "loss": 0.7576,
+      "step": 4081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326664226872063e-05,
+      "loss": 0.8198,
+      "step": 4082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325736053228064e-05,
+      "loss": 0.7478,
+      "step": 4083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324807645751404e-05,
+      "loss": 0.8079,
+      "step": 4084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832387900446817e-05,
+      "loss": 0.7008,
+      "step": 4085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322950129404432e-05,
+      "loss": 0.7942,
+      "step": 4086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322021020586295e-05,
+      "loss": 0.7993,
+      "step": 4087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321091678039837e-05,
+      "loss": 0.7705,
+      "step": 4088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320162101791165e-05,
+      "loss": 0.8239,
+      "step": 4089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8319232291866388e-05,
+      "loss": 0.766,
+      "step": 4090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831830224829162e-05,
+      "loss": 0.7805,
+      "step": 4091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317371971092983e-05,
+      "loss": 0.7299,
+      "step": 4092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83164414602966e-05,
+      "loss": 0.6696,
+      "step": 4093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315510715928607e-05,
+      "loss": 0.7255,
+      "step": 4094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314579738015147e-05,
+      "loss": 0.7967,
+      "step": 4095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313648526582363e-05,
+      "loss": 0.6774,
+      "step": 4096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831271708165641e-05,
+      "loss": 0.7068,
+      "step": 4097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831178540326345e-05,
+      "loss": 0.7784,
+      "step": 4098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310853491429647e-05,
+      "loss": 0.778,
+      "step": 4099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309921346181173e-05,
+      "loss": 0.773,
+      "step": 4100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308988967544215e-05,
+      "loss": 0.822,
+      "step": 4101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830805635554495e-05,
+      "loss": 0.7748,
+      "step": 4102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830712351020958e-05,
+      "loss": 0.7436,
+      "step": 4103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8306190431564293e-05,
+      "loss": 0.7278,
+      "step": 4104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305257119635308e-05,
+      "loss": 0.7089,
+      "step": 4105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304323574448827e-05,
+      "loss": 0.8154,
+      "step": 4106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303389796031076e-05,
+      "loss": 0.7579,
+      "step": 4107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830245578440828e-05,
+      "loss": 0.7543,
+      "step": 4108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301521539606667e-05,
+      "loss": 0.7222,
+      "step": 4109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300587061652476e-05,
+      "loss": 0.7672,
+      "step": 4110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299652350571957e-05,
+      "loss": 0.7828,
+      "step": 4111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298717406391356e-05,
+      "loss": 0.7064,
+      "step": 4112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297782229136937e-05,
+      "loss": 0.7072,
+      "step": 4113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296846818834963e-05,
+      "loss": 0.7822,
+      "step": 4114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295911175511704e-05,
+      "loss": 0.7729,
+      "step": 4115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829497529919344e-05,
+      "loss": 0.7015,
+      "step": 4116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294039189906455e-05,
+      "loss": 0.7902,
+      "step": 4117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829310284767704e-05,
+      "loss": 0.8412,
+      "step": 4118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829216627253149e-05,
+      "loss": 0.8133,
+      "step": 4119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291229464496112e-05,
+      "loss": 0.7966,
+      "step": 4120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290292423597214e-05,
+      "loss": 0.7537,
+      "step": 4121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828935514986112e-05,
+      "loss": 0.7543,
+      "step": 4122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828841764331415e-05,
+      "loss": 0.8283,
+      "step": 4123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828747990398263e-05,
+      "loss": 0.7758,
+      "step": 4124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8286541931892903e-05,
+      "loss": 0.8427,
+      "step": 4125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828560372707131e-05,
+      "loss": 0.7321,
+      "step": 4126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284665289544202e-05,
+      "loss": 0.7524,
+      "step": 4127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283726619337935e-05,
+      "loss": 0.8003,
+      "step": 4128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828278771647887e-05,
+      "loss": 0.8392,
+      "step": 4129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281848580993376e-05,
+      "loss": 0.7786,
+      "step": 4130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280909212907834e-05,
+      "loss": 0.8137,
+      "step": 4131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279969612248628e-05,
+      "loss": 0.7241,
+      "step": 4132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279029779042136e-05,
+      "loss": 0.747,
+      "step": 4133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278089713314763e-05,
+      "loss": 0.7679,
+      "step": 4134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277149415092913e-05,
+      "loss": 0.6978,
+      "step": 4135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827620888440299e-05,
+      "loss": 0.7356,
+      "step": 4136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275268121271405e-05,
+      "loss": 0.6945,
+      "step": 4137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274327125724588e-05,
+      "loss": 0.7065,
+      "step": 4138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273385897788967e-05,
+      "loss": 0.7908,
+      "step": 4139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827244443749097e-05,
+      "loss": 0.7299,
+      "step": 4140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271502744857045e-05,
+      "loss": 0.751,
+      "step": 4141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270560819913637e-05,
+      "loss": 0.7529,
+      "step": 4142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269618662687202e-05,
+      "loss": 0.7451,
+      "step": 4143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268676273204197e-05,
+      "loss": 0.8157,
+      "step": 4144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267733651491097e-05,
+      "loss": 0.783,
+      "step": 4145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826679079757437e-05,
+      "loss": 0.6786,
+      "step": 4146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265847711480495e-05,
+      "loss": 0.8097,
+      "step": 4147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264904393235967e-05,
+      "loss": 0.831,
+      "step": 4148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263960842867274e-05,
+      "loss": 0.7847,
+      "step": 4149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263017060400914e-05,
+      "loss": 0.7816,
+      "step": 4150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262073045863397e-05,
+      "loss": 0.7519,
+      "step": 4151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261128799281235e-05,
+      "loss": 0.8318,
+      "step": 4152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8260184320680953e-05,
+      "loss": 0.6764,
+      "step": 4153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8259239610089066e-05,
+      "loss": 0.7782,
+      "step": 4154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.825829466753212e-05,
+      "loss": 0.8055,
+      "step": 4155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825734949303664e-05,
+      "loss": 0.734,
+      "step": 4156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825640408662918e-05,
+      "loss": 0.7463,
+      "step": 4157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8255458448336297e-05,
+      "loss": 0.7859,
+      "step": 4158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8254512578184543e-05,
+      "loss": 0.8315,
+      "step": 4159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825356647620048e-05,
+      "loss": 0.6821,
+      "step": 4160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252620142410687e-05,
+      "loss": 0.6732,
+      "step": 4161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825167357684174e-05,
+      "loss": 0.8455,
+      "step": 4162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250726779520223e-05,
+      "loss": 0.8,
+      "step": 4163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249779750472725e-05,
+      "loss": 0.7671,
+      "step": 4164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824883248972585e-05,
+      "loss": 0.7827,
+      "step": 4165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247884997306196e-05,
+      "loss": 0.7762,
+      "step": 4166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246937273240375e-05,
+      "loss": 0.7467,
+      "step": 4167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245989317555008e-05,
+      "loss": 0.775,
+      "step": 4168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245041130276714e-05,
+      "loss": 0.7991,
+      "step": 4169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824409271143213e-05,
+      "loss": 0.8024,
+      "step": 4170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243144061047886e-05,
+      "loss": 0.7266,
+      "step": 4171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824219517915063e-05,
+      "loss": 0.7432,
+      "step": 4172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241246065767006e-05,
+      "loss": 0.7657,
+      "step": 4173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240296720923678e-05,
+      "loss": 0.6739,
+      "step": 4174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239347144647303e-05,
+      "loss": 0.7806,
+      "step": 4175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238397336964553e-05,
+      "loss": 0.7301,
+      "step": 4176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237447297902105e-05,
+      "loss": 0.7541,
+      "step": 4177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823649702748664e-05,
+      "loss": 0.7266,
+      "step": 4178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235546525744844e-05,
+      "loss": 0.783,
+      "step": 4179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234595792703413e-05,
+      "loss": 0.7584,
+      "step": 4180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233644828389056e-05,
+      "loss": 0.7608,
+      "step": 4181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232693632828472e-05,
+      "loss": 0.7626,
+      "step": 4182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823174220604838e-05,
+      "loss": 0.8054,
+      "step": 4183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230790548075502e-05,
+      "loss": 0.3279,
+      "step": 4184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229838658936566e-05,
+      "loss": 0.809,
+      "step": 4185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228886538658303e-05,
+      "loss": 0.7913,
+      "step": 4186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227934187267457e-05,
+      "loss": 0.726,
+      "step": 4187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226981604790775e-05,
+      "loss": 0.7477,
+      "step": 4188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822602879125501e-05,
+      "loss": 0.7566,
+      "step": 4189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225075746686924e-05,
+      "loss": 0.7125,
+      "step": 4190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822412247111328e-05,
+      "loss": 0.7794,
+      "step": 4191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223168964560853e-05,
+      "loss": 0.7809,
+      "step": 4192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222215227056422e-05,
+      "loss": 0.6805,
+      "step": 4193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221261258626777e-05,
+      "loss": 0.7483,
+      "step": 4194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220307059298708e-05,
+      "loss": 0.7433,
+      "step": 4195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219352629099013e-05,
+      "loss": 0.7648,
+      "step": 4196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82183979680545e-05,
+      "loss": 0.778,
+      "step": 4197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821744307619198e-05,
+      "loss": 0.7074,
+      "step": 4198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821648795353827e-05,
+      "loss": 0.7892,
+      "step": 4199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215532600120197e-05,
+      "loss": 0.7331,
+      "step": 4200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214577015964593e-05,
+      "loss": 0.7697,
+      "step": 4201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8213621201098295e-05,
+      "loss": 0.7625,
+      "step": 4202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821266515554815e-05,
+      "loss": 0.7465,
+      "step": 4203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211708879341005e-05,
+      "loss": 0.7999,
+      "step": 4204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821075237250372e-05,
+      "loss": 0.6816,
+      "step": 4205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209795635063158e-05,
+      "loss": 0.7274,
+      "step": 4206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8208838667046193e-05,
+      "loss": 0.7493,
+      "step": 4207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207881468479697e-05,
+      "loss": 0.7183,
+      "step": 4208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206924039390555e-05,
+      "loss": 0.8546,
+      "step": 4209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820596637980566e-05,
+      "loss": 0.4264,
+      "step": 4210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82050084897519e-05,
+      "loss": 0.7465,
+      "step": 4211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204050369256187e-05,
+      "loss": 0.7806,
+      "step": 4212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203092018345426e-05,
+      "loss": 0.7579,
+      "step": 4213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202133437046538e-05,
+      "loss": 0.7558,
+      "step": 4214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201174625386434e-05,
+      "loss": 0.8027,
+      "step": 4215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200215583392052e-05,
+      "loss": 0.7612,
+      "step": 4216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199256311090326e-05,
+      "loss": 0.7592,
+      "step": 4217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81982968085082e-05,
+      "loss": 0.6677,
+      "step": 4218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197337075672613e-05,
+      "loss": 0.6693,
+      "step": 4219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196377112610524e-05,
+      "loss": 0.7938,
+      "step": 4220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81954169193489e-05,
+      "loss": 0.745,
+      "step": 4221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194456495914704e-05,
+      "loss": 0.7818,
+      "step": 4222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193495842334907e-05,
+      "loss": 0.7121,
+      "step": 4223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192534958636493e-05,
+      "loss": 0.8002,
+      "step": 4224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191573844846447e-05,
+      "loss": 0.7484,
+      "step": 4225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190612500991766e-05,
+      "loss": 0.7311,
+      "step": 4226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189650927099448e-05,
+      "loss": 0.749,
+      "step": 4227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188689123196498e-05,
+      "loss": 0.7603,
+      "step": 4228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187727089309927e-05,
+      "loss": 0.7428,
+      "step": 4229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186764825466758e-05,
+      "loss": 0.7888,
+      "step": 4230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185802331694016e-05,
+      "loss": 0.8643,
+      "step": 4231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184839608018734e-05,
+      "loss": 0.7628,
+      "step": 4232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183876654467947e-05,
+      "loss": 0.6829,
+      "step": 4233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81829134710687e-05,
+      "loss": 0.7991,
+      "step": 4234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818195005784805e-05,
+      "loss": 0.3256,
+      "step": 4235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180986414833054e-05,
+      "loss": 0.7513,
+      "step": 4236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180022542050768e-05,
+      "loss": 0.7073,
+      "step": 4237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179058439528275e-05,
+      "loss": 0.8066,
+      "step": 4238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178094107292642e-05,
+      "loss": 0.7598,
+      "step": 4239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177129545370957e-05,
+      "loss": 0.7772,
+      "step": 4240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176164753790312e-05,
+      "loss": 0.7571,
+      "step": 4241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8175199732577802e-05,
+      "loss": 0.7317,
+      "step": 4242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174234481760532e-05,
+      "loss": 0.7481,
+      "step": 4243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173269001365606e-05,
+      "loss": 0.7722,
+      "step": 4244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172303291420144e-05,
+      "loss": 0.7201,
+      "step": 4245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171337351951268e-05,
+      "loss": 0.7609,
+      "step": 4246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817037118298611e-05,
+      "loss": 0.7756,
+      "step": 4247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81694047845518e-05,
+      "loss": 0.7704,
+      "step": 4248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168438156675482e-05,
+      "loss": 0.7638,
+      "step": 4249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167471299384306e-05,
+      "loss": 0.7629,
+      "step": 4250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8166504212705424e-05,
+      "loss": 0.7096,
+      "step": 4251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165536896665997e-05,
+      "loss": 0.7643,
+      "step": 4252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8164569351293197e-05,
+      "loss": 0.7743,
+      "step": 4253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163601576614195e-05,
+      "loss": 0.6635,
+      "step": 4254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162633572656168e-05,
+      "loss": 0.7606,
+      "step": 4255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816166533944631e-05,
+      "loss": 0.7769,
+      "step": 4256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160696877011808e-05,
+      "loss": 0.7273,
+      "step": 4257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159728185379867e-05,
+      "loss": 0.8226,
+      "step": 4258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158759264577684e-05,
+      "loss": 0.7471,
+      "step": 4259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157790114632486e-05,
+      "loss": 0.7479,
+      "step": 4260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156820735571482e-05,
+      "loss": 0.8033,
+      "step": 4261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81558511274219e-05,
+      "loss": 0.7685,
+      "step": 4262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815488129021097e-05,
+      "loss": 0.758,
+      "step": 4263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153911223965932e-05,
+      "loss": 0.7112,
+      "step": 4264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152940928714035e-05,
+      "loss": 0.6872,
+      "step": 4265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151970404482518e-05,
+      "loss": 0.7662,
+      "step": 4266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150999651298656e-05,
+      "loss": 0.7225,
+      "step": 4267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81500286691897e-05,
+      "loss": 0.6716,
+      "step": 4268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149057458182924e-05,
+      "loss": 0.7464,
+      "step": 4269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148086018305607e-05,
+      "loss": 0.7604,
+      "step": 4270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814711434958503e-05,
+      "loss": 0.7325,
+      "step": 4271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146142452048484e-05,
+      "loss": 0.7619,
+      "step": 4272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8145170325723262e-05,
+      "loss": 0.8044,
+      "step": 4273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144197970636674e-05,
+      "loss": 0.7809,
+      "step": 4274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143225386816025e-05,
+      "loss": 0.7753,
+      "step": 4275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814225257428863e-05,
+      "loss": 0.7492,
+      "step": 4276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814127953308181e-05,
+      "loss": 0.7463,
+      "step": 4277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140306263222894e-05,
+      "loss": 0.7032,
+      "step": 4278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139332764739216e-05,
+      "loss": 0.8484,
+      "step": 4279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138359037658123e-05,
+      "loss": 0.7491,
+      "step": 4280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137385082006954e-05,
+      "loss": 0.7078,
+      "step": 4281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813641089781307e-05,
+      "loss": 0.7158,
+      "step": 4282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813543648510383e-05,
+      "loss": 0.775,
+      "step": 4283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134461843906597e-05,
+      "loss": 0.7689,
+      "step": 4284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133486974248748e-05,
+      "loss": 0.7594,
+      "step": 4285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813251187615766e-05,
+      "loss": 0.7105,
+      "step": 4286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131536549660722e-05,
+      "loss": 0.7396,
+      "step": 4287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130560994785325e-05,
+      "loss": 0.7143,
+      "step": 4288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812958521155887e-05,
+      "loss": 0.75,
+      "step": 4289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812860920000876e-05,
+      "loss": 0.7332,
+      "step": 4290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812763296016241e-05,
+      "loss": 0.7327,
+      "step": 4291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812665649204723e-05,
+      "loss": 0.7825,
+      "step": 4292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125679795690656e-05,
+      "loss": 0.8141,
+      "step": 4293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812470287112011e-05,
+      "loss": 0.7928,
+      "step": 4294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123725718363034e-05,
+      "loss": 0.7393,
+      "step": 4295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812274833744687e-05,
+      "loss": 0.3315,
+      "step": 4296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121770728399072e-05,
+      "loss": 0.7904,
+      "step": 4297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812079289124709e-05,
+      "loss": 0.6807,
+      "step": 4298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119814826018392e-05,
+      "loss": 0.7496,
+      "step": 4299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118836532740446e-05,
+      "loss": 0.801,
+      "step": 4300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117858011440728e-05,
+      "loss": 0.8522,
+      "step": 4301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811687926214672e-05,
+      "loss": 0.7181,
+      "step": 4302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115900284885912e-05,
+      "loss": 0.8215,
+      "step": 4303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81149210796858e-05,
+      "loss": 0.665,
+      "step": 4304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8113941646573878e-05,
+      "loss": 0.8144,
+      "step": 4305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112961985577665e-05,
+      "loss": 0.795,
+      "step": 4306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111982096724664e-05,
+      "loss": 0.7812,
+      "step": 4307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111001980042405e-05,
+      "loss": 0.7559,
+      "step": 4308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811002163555841e-05,
+      "loss": 0.7834,
+      "step": 4309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810904106330022e-05,
+      "loss": 0.7437,
+      "step": 4310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108060263295364e-05,
+      "loss": 0.734,
+      "step": 4311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107079235571393e-05,
+      "loss": 0.7376,
+      "step": 4312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106097980155862e-05,
+      "loss": 0.7313,
+      "step": 4313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8105116497076326e-05,
+      "loss": 0.7682,
+      "step": 4314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810413478636035e-05,
+      "loss": 0.662,
+      "step": 4315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810315284803551e-05,
+      "loss": 0.7797,
+      "step": 4316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102170682129387e-05,
+      "loss": 0.7654,
+      "step": 4317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101188288669556e-05,
+      "loss": 0.7998,
+      "step": 4318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100205667683614e-05,
+      "loss": 0.7343,
+      "step": 4319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809922281919916e-05,
+      "loss": 0.6694,
+      "step": 4320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098239743243794e-05,
+      "loss": 0.834,
+      "step": 4321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097256439845128e-05,
+      "loss": 0.842,
+      "step": 4322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809627290903078e-05,
+      "loss": 0.7332,
+      "step": 4323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8095289150828368e-05,
+      "loss": 0.7132,
+      "step": 4324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094305165265528e-05,
+      "loss": 0.7291,
+      "step": 4325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093320952369886e-05,
+      "loss": 0.7747,
+      "step": 4326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092336512169098e-05,
+      "loss": 0.7664,
+      "step": 4327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80913518446908e-05,
+      "loss": 0.7102,
+      "step": 4328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090366949962655e-05,
+      "loss": 0.6891,
+      "step": 4329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089381828012316e-05,
+      "loss": 0.7958,
+      "step": 4330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808839647886746e-05,
+      "loss": 0.7712,
+      "step": 4331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8087410902555757e-05,
+      "loss": 0.7578,
+      "step": 4332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086425099104886e-05,
+      "loss": 0.8248,
+      "step": 4333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085439068542538e-05,
+      "loss": 0.7513,
+      "step": 4334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80844528108964e-05,
+      "loss": 0.7682,
+      "step": 4335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083466326194178e-05,
+      "loss": 0.7469,
+      "step": 4336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808247961446357e-05,
+      "loss": 0.7982,
+      "step": 4337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80814926757323e-05,
+      "loss": 0.7909,
+      "step": 4338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080505510028073e-05,
+      "loss": 0.7878,
+      "step": 4339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079518117378622e-05,
+      "loss": 0.712,
+      "step": 4340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807853049781168e-05,
+      "loss": 0.7228,
+      "step": 4341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807754265135498e-05,
+      "loss": 0.7434,
+      "step": 4342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807655457803627e-05,
+      "loss": 0.7663,
+      "step": 4343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8075566277883298e-05,
+      "loss": 0.7129,
+      "step": 4344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807457775092382e-05,
+      "loss": 0.7611,
+      "step": 4345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8073588997185603e-05,
+      "loss": 0.7731,
+      "step": 4346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072600016696416e-05,
+      "loss": 0.7411,
+      "step": 4347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8071610809484033e-05,
+      "loss": 0.7938,
+      "step": 4348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070621375576235e-05,
+      "loss": 0.772,
+      "step": 4349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8069631715000815e-05,
+      "loss": 0.7871,
+      "step": 4350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068641827785564e-05,
+      "loss": 0.869,
+      "step": 4351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067651713958287e-05,
+      "loss": 0.8325,
+      "step": 4352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066661373546793e-05,
+      "loss": 0.753,
+      "step": 4353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806567080657889e-05,
+      "loss": 0.7944,
+      "step": 4354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064680013082406e-05,
+      "loss": 0.7644,
+      "step": 4355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063688993085163e-05,
+      "loss": 0.7939,
+      "step": 4356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062697746614993e-05,
+      "loss": 0.8024,
+      "step": 4357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806170627369974e-05,
+      "loss": 0.769,
+      "step": 4358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8060714574367248e-05,
+      "loss": 0.7759,
+      "step": 4359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805972264864537e-05,
+      "loss": 0.7596,
+      "step": 4360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805873049656196e-05,
+      "loss": 0.7292,
+      "step": 4361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057738118144893e-05,
+      "loss": 0.7098,
+      "step": 4362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056745513422034e-05,
+      "loss": 0.7248,
+      "step": 4363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055752682421262e-05,
+      "loss": 0.7883,
+      "step": 4364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805475962517046e-05,
+      "loss": 0.7868,
+      "step": 4365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805376634169752e-05,
+      "loss": 0.7275,
+      "step": 4366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052772832030338e-05,
+      "loss": 0.7838,
+      "step": 4367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805177909619682e-05,
+      "loss": 0.7515,
+      "step": 4368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050785134224868e-05,
+      "loss": 0.799,
+      "step": 4369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049790946142406e-05,
+      "loss": 0.7672,
+      "step": 4370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048796531977357e-05,
+      "loss": 0.7391,
+      "step": 4371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047801891757642e-05,
+      "loss": 0.7624,
+      "step": 4372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80468070255112e-05,
+      "loss": 0.7856,
+      "step": 4373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8045811933265973e-05,
+      "loss": 0.7363,
+      "step": 4374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804481661504991e-05,
+      "loss": 0.7505,
+      "step": 4375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043821070890958e-05,
+      "loss": 0.7592,
+      "step": 4376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8042825300817085e-05,
+      "loss": 0.3338,
+      "step": 4377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041829304856258e-05,
+      "loss": 0.8262,
+      "step": 4378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040833083036448e-05,
+      "loss": 0.7536,
+      "step": 4379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803983663538563e-05,
+      "loss": 0.7769,
+      "step": 4380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038839961931795e-05,
+      "loss": 0.8108,
+      "step": 4381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037843062702935e-05,
+      "loss": 0.7725,
+      "step": 4382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036845937727045e-05,
+      "loss": 0.7506,
+      "step": 4383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803584858703213e-05,
+      "loss": 0.8752,
+      "step": 4384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034851010646206e-05,
+      "loss": 0.7813,
+      "step": 4385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033853208597288e-05,
+      "loss": 0.733,
+      "step": 4386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8032855180913398e-05,
+      "loss": 0.7254,
+      "step": 4387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031856927622568e-05,
+      "loss": 0.7839,
+      "step": 4388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803085844875283e-05,
+      "loss": 0.7349,
+      "step": 4389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029859744332238e-05,
+      "loss": 0.7522,
+      "step": 4390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028860814388826e-05,
+      "loss": 0.8365,
+      "step": 4391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802786165895066e-05,
+      "loss": 0.7066,
+      "step": 4392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80268622780458e-05,
+      "loss": 0.7969,
+      "step": 4393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025862671702313e-05,
+      "loss": 0.7608,
+      "step": 4394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024862839948272e-05,
+      "loss": 0.7229,
+      "step": 4395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802386278281176e-05,
+      "loss": 0.7596,
+      "step": 4396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022862500320863e-05,
+      "loss": 0.771,
+      "step": 4397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021861992503678e-05,
+      "loss": 0.7853,
+      "step": 4398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020861259388298e-05,
+      "loss": 0.7526,
+      "step": 4399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019860301002835e-05,
+      "loss": 0.7893,
+      "step": 4400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018859117375398e-05,
+      "loss": 0.7907,
+      "step": 4401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017857708534107e-05,
+      "loss": 0.7726,
+      "step": 4402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016856074507088e-05,
+      "loss": 0.7851,
+      "step": 4403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015854215322472e-05,
+      "loss": 0.7521,
+      "step": 4404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014852131008395e-05,
+      "loss": 0.7193,
+      "step": 4405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013849821593005e-05,
+      "loss": 0.7613,
+      "step": 4406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012847287104447e-05,
+      "loss": 0.8276,
+      "step": 4407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801184452757088e-05,
+      "loss": 0.6637,
+      "step": 4408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8010841543020472e-05,
+      "loss": 0.8081,
+      "step": 4409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8009838333481383e-05,
+      "loss": 0.7966,
+      "step": 4410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008834898981795e-05,
+      "loss": 0.8058,
+      "step": 4411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007831239549888e-05,
+      "loss": 0.7578,
+      "step": 4412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8006827355213853e-05,
+      "loss": 0.8129,
+      "step": 4413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800582324600188e-05,
+      "loss": 0.6659,
+      "step": 4414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004818911942173e-05,
+      "loss": 0.6928,
+      "step": 4415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800381435306294e-05,
+      "loss": 0.7858,
+      "step": 4416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800280956939239e-05,
+      "loss": 0.7383,
+      "step": 4417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001804560958745e-05,
+      "loss": 0.8091,
+      "step": 4418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000799327790237e-05,
+      "loss": 0.7191,
+      "step": 4419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999793869915093e-05,
+      "loss": 0.8206,
+      "step": 4420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998788187361552e-05,
+      "loss": 0.8617,
+      "step": 4421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997782280157858e-05,
+      "loss": 0.7329,
+      "step": 4422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996776148332265e-05,
+      "loss": 0.8255,
+      "step": 4423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799576979191303e-05,
+      "loss": 0.6859,
+      "step": 4424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994763210928417e-05,
+      "loss": 0.7955,
+      "step": 4425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993756405406696e-05,
+      "loss": 0.7784,
+      "step": 4426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799274937537614e-05,
+      "loss": 0.7664,
+      "step": 4427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799174212086504e-05,
+      "loss": 0.7224,
+      "step": 4428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990734641901683e-05,
+      "loss": 0.7358,
+      "step": 4429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798972693851436e-05,
+      "loss": 0.7576,
+      "step": 4430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988719010731377e-05,
+      "loss": 0.8276,
+      "step": 4431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987710858581037e-05,
+      "loss": 0.7162,
+      "step": 4432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986702482091662e-05,
+      "loss": 0.7722,
+      "step": 4433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985693881291565e-05,
+      "loss": 0.7509,
+      "step": 4434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984685056209082e-05,
+      "loss": 0.7044,
+      "step": 4435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798367600687254e-05,
+      "loss": 0.7603,
+      "step": 4436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798266673331028e-05,
+      "loss": 0.756,
+      "step": 4437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981657235550644e-05,
+      "loss": 0.7568,
+      "step": 4438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980647513621993e-05,
+      "loss": 0.7868,
+      "step": 4439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797963756755268e-05,
+      "loss": 0.7596,
+      "step": 4440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978627397371072e-05,
+      "loss": 0.7744,
+      "step": 4441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977617003105536e-05,
+      "loss": 0.8274,
+      "step": 4442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976606384784455e-05,
+      "loss": 0.8136,
+      "step": 4443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975595542436207e-05,
+      "loss": 0.7129,
+      "step": 4444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974584476089188e-05,
+      "loss": 0.7459,
+      "step": 4445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797357318577179e-05,
+      "loss": 0.793,
+      "step": 4446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972561671512418e-05,
+      "loss": 0.7382,
+      "step": 4447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971549933339482e-05,
+      "loss": 0.7567,
+      "step": 4448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970537971281393e-05,
+      "loss": 0.8296,
+      "step": 4449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969525785366573e-05,
+      "loss": 0.7391,
+      "step": 4450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968513375623457e-05,
+      "loss": 0.8047,
+      "step": 4451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796750074208047e-05,
+      "loss": 0.7509,
+      "step": 4452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966487884766052e-05,
+      "loss": 0.766,
+      "step": 4453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965474803708657e-05,
+      "loss": 0.7539,
+      "step": 4454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964461498936735e-05,
+      "loss": 0.6942,
+      "step": 4455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963447970478746e-05,
+      "loss": 0.832,
+      "step": 4456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962434218363155e-05,
+      "loss": 0.7897,
+      "step": 4457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961420242618427e-05,
+      "loss": 0.7241,
+      "step": 4458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796040604327305e-05,
+      "loss": 0.7515,
+      "step": 4459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959391620355504e-05,
+      "loss": 0.7215,
+      "step": 4460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795837697389428e-05,
+      "loss": 0.7227,
+      "step": 4461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957362103917876e-05,
+      "loss": 0.7598,
+      "step": 4462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795634701045479e-05,
+      "loss": 0.8223,
+      "step": 4463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7955331693533537e-05,
+      "loss": 0.6964,
+      "step": 4464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795431615318263e-05,
+      "loss": 0.726,
+      "step": 4465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953300389430594e-05,
+      "loss": 0.7574,
+      "step": 4466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952284402305956e-05,
+      "loss": 0.7667,
+      "step": 4467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951268191837246e-05,
+      "loss": 0.7132,
+      "step": 4468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950251758053013e-05,
+      "loss": 0.7113,
+      "step": 4469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949235100981797e-05,
+      "loss": 0.7632,
+      "step": 4470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948218220652155e-05,
+      "loss": 0.7161,
+      "step": 4471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947201117092645e-05,
+      "loss": 0.7354,
+      "step": 4472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946183790331835e-05,
+      "loss": 0.711,
+      "step": 4473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7945166240398294e-05,
+      "loss": 0.803,
+      "step": 4474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79441484673206e-05,
+      "loss": 0.7531,
+      "step": 4475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943130471127345e-05,
+      "loss": 0.7952,
+      "step": 4476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794211225184711e-05,
+      "loss": 0.7206,
+      "step": 4477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941093809508498e-05,
+      "loss": 0.7213,
+      "step": 4478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794007514414011e-05,
+      "loss": 0.7154,
+      "step": 4479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7939056255770562e-05,
+      "loss": 0.8472,
+      "step": 4480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938037144428463e-05,
+      "loss": 0.8002,
+      "step": 4481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937017810142437e-05,
+      "loss": 0.7009,
+      "step": 4482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935998252941114e-05,
+      "loss": 0.7463,
+      "step": 4483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934978472853127e-05,
+      "loss": 0.8215,
+      "step": 4484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7933958469907116e-05,
+      "loss": 0.7218,
+      "step": 4485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793293824413173e-05,
+      "loss": 0.7218,
+      "step": 4486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931917795555628e-05,
+      "loss": 0.7354,
+      "step": 4487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930897124207458e-05,
+      "loss": 0.7296,
+      "step": 4488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929876230115897e-05,
+      "loss": 0.7507,
+      "step": 4489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792885511330961e-05,
+      "loss": 0.7609,
+      "step": 4490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792783377381728e-05,
+      "loss": 0.7787,
+      "step": 4491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926812211667593e-05,
+      "loss": 0.7507,
+      "step": 4492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925790426889237e-05,
+      "loss": 0.8148,
+      "step": 4493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924768419510906e-05,
+      "loss": 0.7735,
+      "step": 4494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792374618956131e-05,
+      "loss": 0.7986,
+      "step": 4495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922723737069158e-05,
+      "loss": 0.7426,
+      "step": 4496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792170106206316e-05,
+      "loss": 0.7612,
+      "step": 4497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920678164572046e-05,
+      "loss": 0.7497,
+      "step": 4498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791965504462454e-05,
+      "loss": 0.6988,
+      "step": 4499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791863170224938e-05,
+      "loss": 0.7102,
+      "step": 4500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917608137475305e-05,
+      "loss": 0.7275,
+      "step": 4501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916584350331062e-05,
+      "loss": 0.7298,
+      "step": 4502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791556034084541e-05,
+      "loss": 0.8076,
+      "step": 4503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914536109047097e-05,
+      "loss": 0.6625,
+      "step": 4504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79135116549649e-05,
+      "loss": 0.7804,
+      "step": 4505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791248697862759e-05,
+      "loss": 0.7814,
+      "step": 4506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791146208006394e-05,
+      "loss": 0.7129,
+      "step": 4507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791043695930274e-05,
+      "loss": 0.7453,
+      "step": 4508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909411616372776e-05,
+      "loss": 0.7948,
+      "step": 4509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790838605130285e-05,
+      "loss": 0.7386,
+      "step": 4510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907360264121764e-05,
+      "loss": 0.7522,
+      "step": 4511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906334254858327e-05,
+      "loss": 0.7894,
+      "step": 4512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905308023541358e-05,
+      "loss": 0.7515,
+      "step": 4513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790428157019967e-05,
+      "loss": 0.7764,
+      "step": 4514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790325489486211e-05,
+      "loss": 0.8088,
+      "step": 4515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902227997557492e-05,
+      "loss": 0.7139,
+      "step": 4516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901200878314667e-05,
+      "loss": 0.7605,
+      "step": 4517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900173537162486e-05,
+      "loss": 0.7666,
+      "step": 4518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899145974129793e-05,
+      "loss": 0.7626,
+      "step": 4519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898118189245456e-05,
+      "loss": 0.7562,
+      "step": 4520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897090182538335e-05,
+      "loss": 0.8503,
+      "step": 4521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7896061954037305e-05,
+      "loss": 0.3752,
+      "step": 4522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789503350377124e-05,
+      "loss": 0.6881,
+      "step": 4523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894004831769037e-05,
+      "loss": 0.7295,
+      "step": 4524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892975938059568e-05,
+      "loss": 0.7399,
+      "step": 4525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7891946822671747e-05,
+      "loss": 0.8228,
+      "step": 4526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789091748563447e-05,
+      "loss": 0.7545,
+      "step": 4527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889887926976643e-05,
+      "loss": 0.7937,
+      "step": 4528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888858146727187e-05,
+      "loss": 0.74,
+      "step": 4529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887828144915023e-05,
+      "loss": 0.7563,
+      "step": 4530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788679792156908e-05,
+      "loss": 0.7803,
+      "step": 4531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885767476718285e-05,
+      "loss": 0.7354,
+      "step": 4532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884736810391586e-05,
+      "loss": 0.783,
+      "step": 4533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7883705922617936e-05,
+      "loss": 0.7535,
+      "step": 4534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788267481342627e-05,
+      "loss": 0.7648,
+      "step": 4535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881643482845564e-05,
+      "loss": 0.8475,
+      "step": 4536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7880611930904774e-05,
+      "loss": 0.7368,
+      "step": 4537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879580157632875e-05,
+      "loss": 0.7398,
+      "step": 4538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878548163058846e-05,
+      "loss": 0.86,
+      "step": 4539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787751594721167e-05,
+      "loss": 0.7653,
+      "step": 4540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787648351012034e-05,
+      "loss": 0.7429,
+      "step": 4541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787545085181384e-05,
+      "loss": 0.7645,
+      "step": 4542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874417972321188e-05,
+      "loss": 0.8251,
+      "step": 4543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873384871671388e-05,
+      "loss": 0.7282,
+      "step": 4544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872351549893452e-05,
+      "loss": 0.7694,
+      "step": 4545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871318007016406e-05,
+      "loss": 0.7567,
+      "step": 4546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870284243069274e-05,
+      "loss": 0.7459,
+      "step": 4547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786925025808109e-05,
+      "loss": 0.8026,
+      "step": 4548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868216052080898e-05,
+      "loss": 0.7566,
+      "step": 4549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867181625097737e-05,
+      "loss": 0.6638,
+      "step": 4550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866146977160668e-05,
+      "loss": 0.7043,
+      "step": 4551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865112108298745e-05,
+      "loss": 0.8141,
+      "step": 4552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786407701854103e-05,
+      "loss": 0.7523,
+      "step": 4553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78630417079166e-05,
+      "loss": 0.758,
+      "step": 4554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786200617645453e-05,
+      "loss": 0.7894,
+      "step": 4555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78609704241839e-05,
+      "loss": 0.6898,
+      "step": 4556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859934451133808e-05,
+      "loss": 0.6967,
+      "step": 4557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785889825733334e-05,
+      "loss": 0.7636,
+      "step": 4558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857861842811603e-05,
+      "loss": 0.7527,
+      "step": 4559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856825207597706e-05,
+      "loss": 0.7484,
+      "step": 4560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785578835172076e-05,
+      "loss": 0.7015,
+      "step": 4561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854751275209888e-05,
+      "loss": 0.8055,
+      "step": 4562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785371397809422e-05,
+      "loss": 0.7462,
+      "step": 4563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852676460402885e-05,
+      "loss": 0.8159,
+      "step": 4564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785163872216502e-05,
+      "loss": 0.6998,
+      "step": 4565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850600763409777e-05,
+      "loss": 0.7767,
+      "step": 4566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849562584166303e-05,
+      "loss": 0.726,
+      "step": 4567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848524184463755e-05,
+      "loss": 0.6738,
+      "step": 4568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78474855643313e-05,
+      "loss": 0.7977,
+      "step": 4569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846446723798105e-05,
+      "loss": 0.7075,
+      "step": 4570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784540766289335e-05,
+      "loss": 0.7654,
+      "step": 4571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844368381646216e-05,
+      "loss": 0.8127,
+      "step": 4572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784332888008589e-05,
+      "loss": 0.7171,
+      "step": 4573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842289158241574e-05,
+      "loss": 0.7765,
+      "step": 4574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841249216142458e-05,
+      "loss": 0.7211,
+      "step": 4575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840209053817757e-05,
+      "loss": 0.8251,
+      "step": 4576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839168671296683e-05,
+      "loss": 0.7611,
+      "step": 4577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7838128068608454e-05,
+      "loss": 0.7148,
+      "step": 4578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78370872457823e-05,
+      "loss": 0.7145,
+      "step": 4579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836046202847446e-05,
+      "loss": 0.6954,
+      "step": 4580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7835004939833137e-05,
+      "loss": 0.7384,
+      "step": 4581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833963456768613e-05,
+      "loss": 0.7471,
+      "step": 4582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832921753683127e-05,
+      "loss": 0.762,
+      "step": 4583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783187983060594e-05,
+      "loss": 0.7092,
+      "step": 4584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7830837687566303e-05,
+      "loss": 0.7781,
+      "step": 4585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829795324593494e-05,
+      "loss": 0.7493,
+      "step": 4586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782875274171679e-05,
+      "loss": 0.7412,
+      "step": 4587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782770993896547e-05,
+      "loss": 0.696,
+      "step": 4588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7826666916368815e-05,
+      "loss": 0.7158,
+      "step": 4589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825623673956127e-05,
+      "loss": 0.723,
+      "step": 4590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824580211756703e-05,
+      "loss": 0.7452,
+      "step": 4591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782353652979985e-05,
+      "loss": 0.7452,
+      "step": 4592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782249262811488e-05,
+      "loss": 0.7945,
+      "step": 4593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821448506731115e-05,
+      "loss": 0.7802,
+      "step": 4594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820404165677873e-05,
+      "loss": 0.757,
+      "step": 4595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781935960498449e-05,
+      "loss": 0.7667,
+      "step": 4596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.7817,
+      "step": 4597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817269824794648e-05,
+      "loss": 0.7697,
+      "step": 4598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816224605356883e-05,
+      "loss": 0.741,
+      "step": 4599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781517916639636e-05,
+      "loss": 0.7124,
+      "step": 4600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814133507942443e-05,
+      "loss": 0.7705,
+      "step": 4601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7813087630024494e-05,
+      "loss": 0.7404,
+      "step": 4602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812041532671892e-05,
+      "loss": 0.7946,
+      "step": 4603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810995215914017e-05,
+      "loss": 0.7822,
+      "step": 4604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809948679780252e-05,
+      "loss": 0.7806,
+      "step": 4605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78089019243e-05,
+      "loss": 0.8361,
+      "step": 4606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807854949502643e-05,
+      "loss": 0.7697,
+      "step": 4607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806807755417595e-05,
+      "loss": 0.722,
+      "step": 4608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780576034207427e-05,
+      "loss": 0.789,
+      "step": 4609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804712709502083e-05,
+      "loss": 0.7882,
+      "step": 4610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803664857730453e-05,
+      "loss": 0.7482,
+      "step": 4611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802616786788815e-05,
+      "loss": 0.6355,
+      "step": 4612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78015684967066e-05,
+      "loss": 0.719,
+      "step": 4613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800519987513252e-05,
+      "loss": 0.8073,
+      "step": 4614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7799471259238217e-05,
+      "loss": 0.7177,
+      "step": 4615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798422311910954e-05,
+      "loss": 0.8215,
+      "step": 4616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797373145560917e-05,
+      "loss": 0.7076,
+      "step": 4617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796323760217576e-05,
+      "loss": 0.7971,
+      "step": 4618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795274155910408e-05,
+      "loss": 0.7318,
+      "step": 4619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794224332668882e-05,
+      "loss": 0.7114,
+      "step": 4620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779317429052249e-05,
+      "loss": 0.8117,
+      "step": 4621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779212402950072e-05,
+      "loss": 0.796,
+      "step": 4622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791073549633065e-05,
+      "loss": 0.7703,
+      "step": 4623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779002285094904e-05,
+      "loss": 0.7012,
+      "step": 4624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778897193347814e-05,
+      "loss": 0.6777,
+      "step": 4625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787920797249895e-05,
+      "loss": 0.7684,
+      "step": 4626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786869442293814e-05,
+      "loss": 0.7436,
+      "step": 4627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7785817868639435e-05,
+      "loss": 0.7737,
+      "step": 4628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784766076316286e-05,
+      "loss": 0.8348,
+      "step": 4629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7783714065353907e-05,
+      "loss": 0.7259,
+      "step": 4630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782661835781848e-05,
+      "loss": 0.7448,
+      "step": 4631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781609387629653e-05,
+      "loss": 0.7745,
+      "step": 4632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7780556720926896e-05,
+      "loss": 0.75,
+      "step": 4633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779503835703126e-05,
+      "loss": 0.7556,
+      "step": 4634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7778450731987925e-05,
+      "loss": 0.7488,
+      "step": 4635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777397409810863e-05,
+      "loss": 0.7674,
+      "step": 4636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776343869201523e-05,
+      "loss": 0.7172,
+      "step": 4637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77752901101895e-05,
+      "loss": 0.732,
+      "step": 4638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774236132804387e-05,
+      "loss": 0.7386,
+      "step": 4639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773181937075784e-05,
+      "loss": 0.7247,
+      "step": 4640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772127523033297e-05,
+      "loss": 0.6973,
+      "step": 4641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7771072890706545e-05,
+      "loss": 0.8122,
+      "step": 4642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770018040125144e-05,
+      "loss": 0.7305,
+      "step": 4643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776896297131872e-05,
+      "loss": 0.7619,
+      "step": 4644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767907684316904e-05,
+      "loss": 0.7223,
+      "step": 4645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766852179149338e-05,
+      "loss": 0.7363,
+      "step": 4646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776579645584567e-05,
+      "loss": 0.6988,
+      "step": 4647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7764740514435536e-05,
+      "loss": 0.7502,
+      "step": 4648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763684354948608e-05,
+      "loss": 0.7831,
+      "step": 4649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776262797741454e-05,
+      "loss": 0.8236,
+      "step": 4650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7761571381863005e-05,
+      "loss": 0.7572,
+      "step": 4651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760514568323677e-05,
+      "loss": 0.7861,
+      "step": 4652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759457536826237e-05,
+      "loss": 0.7968,
+      "step": 4653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758400287400372e-05,
+      "loss": 0.7564,
+      "step": 4654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7757342820075773e-05,
+      "loss": 0.7241,
+      "step": 4655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756285134882147e-05,
+      "loss": 0.7932,
+      "step": 4656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755227231849194e-05,
+      "loss": 0.7569,
+      "step": 4657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775416911100662e-05,
+      "loss": 0.7704,
+      "step": 4658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7753110772384156e-05,
+      "loss": 0.716,
+      "step": 4659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752052216011516e-05,
+      "loss": 0.7917,
+      "step": 4660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750993441918435e-05,
+      "loss": 0.7457,
+      "step": 4661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774993445013465e-05,
+      "loss": 0.7278,
+      "step": 4662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748875240689898e-05,
+      "loss": 0.7374,
+      "step": 4663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774781581361393e-05,
+      "loss": 0.7672,
+      "step": 4664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746756168936504e-05,
+      "loss": 0.7133,
+      "step": 4665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745696306687375e-05,
+      "loss": 0.7873,
+      "step": 4666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744636226896316e-05,
+      "loss": 0.6967,
+      "step": 4667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743575929593092e-05,
+      "loss": 0.816,
+      "step": 4668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742515414807487e-05,
+      "loss": 0.8113,
+      "step": 4669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741454682569287e-05,
+      "loss": 0.368,
+      "step": 4670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740393732908283e-05,
+      "loss": 0.7928,
+      "step": 4671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739332565854267e-05,
+      "loss": 0.7701,
+      "step": 4672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773827118143705e-05,
+      "loss": 0.7773,
+      "step": 4673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773720957968644e-05,
+      "loss": 0.7377,
+      "step": 4674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773614776063225e-05,
+      "loss": 0.7736,
+      "step": 4675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77350857243043e-05,
+      "loss": 0.7074,
+      "step": 4676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734023470732417e-05,
+      "loss": 0.7341,
+      "step": 4677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773296099994644e-05,
+      "loss": 0.6902,
+      "step": 4678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731898311976214e-05,
+      "loss": 0.6991,
+      "step": 4679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730835406851574e-05,
+      "loss": 0.743,
+      "step": 4680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729772284602373e-05,
+      "loss": 0.7815,
+      "step": 4681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728708945258476e-05,
+      "loss": 0.7134,
+      "step": 4682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727645388849745e-05,
+      "loss": 0.756,
+      "step": 4683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772658161540605e-05,
+      "loss": 0.7772,
+      "step": 4684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725517624957268e-05,
+      "loss": 0.7048,
+      "step": 4685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772445341753328e-05,
+      "loss": 0.7522,
+      "step": 4686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772338899316398e-05,
+      "loss": 0.7853,
+      "step": 4687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722324351879252e-05,
+      "loss": 0.7282,
+      "step": 4688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772125949370901e-05,
+      "loss": 0.7858,
+      "step": 4689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772019441868315e-05,
+      "loss": 0.7327,
+      "step": 4690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7719129126831597e-05,
+      "loss": 0.8562,
+      "step": 4691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771806361818426e-05,
+      "loss": 0.6292,
+      "step": 4692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716997892771068e-05,
+      "loss": 0.7176,
+      "step": 4693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715931950621955e-05,
+      "loss": 0.7103,
+      "step": 4694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714865791766852e-05,
+      "loss": 0.7647,
+      "step": 4695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713799416235707e-05,
+      "loss": 0.8237,
+      "step": 4696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771273282405847e-05,
+      "loss": 0.8061,
+      "step": 4697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77116660152651e-05,
+      "loss": 0.7482,
+      "step": 4698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7710598989885553e-05,
+      "loss": 0.7038,
+      "step": 4699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709531747949796e-05,
+      "loss": 0.7889,
+      "step": 4700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770846428948781e-05,
+      "loss": 0.789,
+      "step": 4701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707396614529567e-05,
+      "loss": 0.7838,
+      "step": 4702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770632872310506e-05,
+      "loss": 0.749,
+      "step": 4703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770526061524428e-05,
+      "loss": 0.7004,
+      "step": 4704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770419229097722e-05,
+      "loss": 0.711,
+      "step": 4705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703123750333894e-05,
+      "loss": 0.7053,
+      "step": 4706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702054993344308e-05,
+      "loss": 0.7706,
+      "step": 4707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700986020038472e-05,
+      "loss": 0.7051,
+      "step": 4708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699916830446416e-05,
+      "loss": 0.6788,
+      "step": 4709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698847424598168e-05,
+      "loss": 0.7733,
+      "step": 4710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769777780252376e-05,
+      "loss": 0.7808,
+      "step": 4711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696707964253236e-05,
+      "loss": 0.7317,
+      "step": 4712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769563790981664e-05,
+      "loss": 0.6983,
+      "step": 4713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769456763924403e-05,
+      "loss": 0.7854,
+      "step": 4714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769349715256546e-05,
+      "loss": 0.3548,
+      "step": 4715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7692426449810997e-05,
+      "loss": 0.7132,
+      "step": 4716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7691355531010713e-05,
+      "loss": 0.6776,
+      "step": 4717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690284396194686e-05,
+      "loss": 0.7141,
+      "step": 4718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689213045392996e-05,
+      "loss": 0.8517,
+      "step": 4719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688141478635733e-05,
+      "loss": 0.7118,
+      "step": 4720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687069695952997e-05,
+      "loss": 0.7143,
+      "step": 4721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7685997697374885e-05,
+      "loss": 0.7635,
+      "step": 4722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684925482931507e-05,
+      "loss": 0.7029,
+      "step": 4723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683853052652974e-05,
+      "loss": 0.7614,
+      "step": 4724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768278040656941e-05,
+      "loss": 0.7404,
+      "step": 4725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681707544710937e-05,
+      "loss": 0.7316,
+      "step": 4726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768063446710769e-05,
+      "loss": 0.7692,
+      "step": 4727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7679561173789805e-05,
+      "loss": 0.8466,
+      "step": 4728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678487664787427e-05,
+      "loss": 0.7375,
+      "step": 4729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677413940130704e-05,
+      "loss": 0.376,
+      "step": 4730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676339999849796e-05,
+      "loss": 0.7667,
+      "step": 4731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675265843974858e-05,
+      "loss": 0.7442,
+      "step": 4732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7674191472536066e-05,
+      "loss": 0.7794,
+      "step": 4733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767311688556359e-05,
+      "loss": 0.7395,
+      "step": 4734
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767204208308761e-05,
+      "loss": 0.7108,
+      "step": 4735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7670967065138315e-05,
+      "loss": 0.739,
+      "step": 4736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76698918317459e-05,
+      "loss": 0.7349,
+      "step": 4737
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668816382940556e-05,
+      "loss": 0.7746,
+      "step": 4738
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667740718752494e-05,
+      "loss": 0.803,
+      "step": 4739
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766666483921192e-05,
+      "loss": 0.7537,
+      "step": 4740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7665588744349054e-05,
+      "loss": 0.8111,
+      "step": 4741
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664512434194113e-05,
+      "loss": 0.7509,
+      "step": 4742
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663435908777336e-05,
+      "loss": 0.7763,
+      "step": 4743
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662359168128952e-05,
+      "loss": 0.7502,
+      "step": 4744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76612822122792e-05,
+      "loss": 0.7694,
+      "step": 4745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660205041258325e-05,
+      "loss": 0.6547,
+      "step": 4746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659127655096585e-05,
+      "loss": 0.7733,
+      "step": 4747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658050053824237e-05,
+      "loss": 0.7178,
+      "step": 4748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765697223747155e-05,
+      "loss": 0.6909,
+      "step": 4749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765589420606879e-05,
+      "loss": 0.8182,
+      "step": 4750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7654815959646235e-05,
+      "loss": 0.7141,
+      "step": 4751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653737498234166e-05,
+      "loss": 0.6747,
+      "step": 4752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765265882186288e-05,
+      "loss": 0.7656,
+      "step": 4753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651579930562662e-05,
+      "loss": 0.6493,
+      "step": 4754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765050082436382e-05,
+      "loss": 0.8323,
+      "step": 4755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649421503296662e-05,
+      "loss": 0.748,
+      "step": 4756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648341967391495e-05,
+      "loss": 0.7738,
+      "step": 4757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647262216678643e-05,
+      "loss": 0.6925,
+      "step": 4758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764618225118843e-05,
+      "loss": 0.7071,
+      "step": 4759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764510207095119e-05,
+      "loss": 0.7719,
+      "step": 4760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764402167599726e-05,
+      "loss": 0.7109,
+      "step": 4761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642941066356977e-05,
+      "loss": 0.8253,
+      "step": 4762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641860242060697e-05,
+      "loss": 0.723,
+      "step": 4763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7640779203138777e-05,
+      "loss": 0.7526,
+      "step": 4764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763969794962157e-05,
+      "loss": 0.7057,
+      "step": 4765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763861648153945e-05,
+      "loss": 0.7667,
+      "step": 4766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7637534798922792e-05,
+      "loss": 0.7312,
+      "step": 4767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763645290180197e-05,
+      "loss": 0.8226,
+      "step": 4768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635370790207376e-05,
+      "loss": 0.7749,
+      "step": 4769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7634288464169395e-05,
+      "loss": 0.7499,
+      "step": 4770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763320592371843e-05,
+      "loss": 0.8088,
+      "step": 4771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763212316888488e-05,
+      "loss": 0.7198,
+      "step": 4772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631040199699165e-05,
+      "loss": 0.729,
+      "step": 4773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629957016191683e-05,
+      "loss": 0.7735,
+      "step": 4774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628873618392872e-05,
+      "loss": 0.682,
+      "step": 4775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627790006333154e-05,
+      "loss": 0.6912,
+      "step": 4776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626706180042963e-05,
+      "loss": 0.7723,
+      "step": 4777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625622139552737e-05,
+      "loss": 0.7988,
+      "step": 4778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624537884892923e-05,
+      "loss": 0.8095,
+      "step": 4779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762345341609397e-05,
+      "loss": 0.7979,
+      "step": 4780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622368733186347e-05,
+      "loss": 0.777,
+      "step": 4781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621283836200504e-05,
+      "loss": 0.7176,
+      "step": 4782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7620198725166918e-05,
+      "loss": 0.7605,
+      "step": 4783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619113400116065e-05,
+      "loss": 0.7382,
+      "step": 4784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618027861078427e-05,
+      "loss": 0.785,
+      "step": 4785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761694210808449e-05,
+      "loss": 0.7935,
+      "step": 4786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615856141164744e-05,
+      "loss": 0.7479,
+      "step": 4787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614769960349697e-05,
+      "loss": 0.7747,
+      "step": 4788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613683565669853e-05,
+      "loss": 0.7886,
+      "step": 4789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612596957155718e-05,
+      "loss": 0.7913,
+      "step": 4790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611510134837816e-05,
+      "loss": 0.7384,
+      "step": 4791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761042309874667e-05,
+      "loss": 0.8856,
+      "step": 4792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760933584891281e-05,
+      "loss": 0.738,
+      "step": 4793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608248385366774e-05,
+      "loss": 0.743,
+      "step": 4794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607160708139097e-05,
+      "loss": 0.7658,
+      "step": 4795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606072817260333e-05,
+      "loss": 0.8144,
+      "step": 4796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760498471276103e-05,
+      "loss": 0.7417,
+      "step": 4797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760389639467176e-05,
+      "loss": 0.739,
+      "step": 4798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7602807863023075e-05,
+      "loss": 0.7877,
+      "step": 4799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601719117845556e-05,
+      "loss": 0.8467,
+      "step": 4800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760063015916978e-05,
+      "loss": 0.7875,
+      "step": 4801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759954098702633e-05,
+      "loss": 0.7674,
+      "step": 4802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759845160144579e-05,
+      "loss": 0.7495,
+      "step": 4803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597362002458762e-05,
+      "loss": 0.6842,
+      "step": 4804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759627219009585e-05,
+      "loss": 0.7462,
+      "step": 4805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595182164387656e-05,
+      "loss": 0.6842,
+      "step": 4806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75940919253648e-05,
+      "loss": 0.7634,
+      "step": 4807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593001473057894e-05,
+      "loss": 0.7371,
+      "step": 4808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591910807497573e-05,
+      "loss": 0.7981,
+      "step": 4809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590819928714465e-05,
+      "loss": 0.6772,
+      "step": 4810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589728836739205e-05,
+      "loss": 0.7392,
+      "step": 4811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588637531602442e-05,
+      "loss": 0.6974,
+      "step": 4812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587546013334822e-05,
+      "loss": 0.7665,
+      "step": 4813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586454281967005e-05,
+      "loss": 0.7942,
+      "step": 4814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585362337529645e-05,
+      "loss": 0.7677,
+      "step": 4815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758427018005342e-05,
+      "loss": 0.7752,
+      "step": 4816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583177809568997e-05,
+      "loss": 0.7859,
+      "step": 4817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582085226107054e-05,
+      "loss": 0.3345,
+      "step": 4818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580992429698284e-05,
+      "loss": 0.7369,
+      "step": 4819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579899420373373e-05,
+      "loss": 0.719,
+      "step": 4820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7578806198163024e-05,
+      "loss": 0.7702,
+      "step": 4821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577712763097937e-05,
+      "loss": 0.783,
+      "step": 4822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7576619115208816e-05,
+      "loss": 0.6949,
+      "step": 4823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757552525452639e-05,
+      "loss": 0.7743,
+      "step": 4824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574431181081374e-05,
+      "loss": 0.6799,
+      "step": 4825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757333689490449e-05,
+      "loss": 0.7456,
+      "step": 4826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757224239602648e-05,
+      "loss": 0.7602,
+      "step": 4827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757114768447808e-05,
+      "loss": 0.7788,
+      "step": 4828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7570052760290036e-05,
+      "loss": 0.7414,
+      "step": 4829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75689576234931e-05,
+      "loss": 0.7858,
+      "step": 4830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567862274118027e-05,
+      "loss": 0.7515,
+      "step": 4831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566766712195585e-05,
+      "loss": 0.7431,
+      "step": 4832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7565670937756537e-05,
+      "loss": 0.7564,
+      "step": 4833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564574950831665e-05,
+      "loss": 0.7513,
+      "step": 4834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563478751451746e-05,
+      "loss": 0.7876,
+      "step": 4835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756238233964757e-05,
+      "loss": 0.3557,
+      "step": 4836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756128571544993e-05,
+      "loss": 0.7637,
+      "step": 4837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756018887888962e-05,
+      "loss": 0.6485,
+      "step": 4838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559091829997456e-05,
+      "loss": 0.7639,
+      "step": 4839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755799456880424e-05,
+      "loss": 0.719,
+      "step": 4840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755689709534079e-05,
+      "loss": 0.7641,
+      "step": 4841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555799409637936e-05,
+      "loss": 0.7093,
+      "step": 4842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75547015117265e-05,
+      "loss": 0.723,
+      "step": 4843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553603401637317e-05,
+      "loss": 0.7102,
+      "step": 4844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7552505079401233e-05,
+      "loss": 0.7682,
+      "step": 4845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755140654504909e-05,
+      "loss": 0.6942,
+      "step": 4846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550307798611746e-05,
+      "loss": 0.7094,
+      "step": 4847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7549208840120054e-05,
+      "loss": 0.7821,
+      "step": 4848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754810966960488e-05,
+      "loss": 0.809,
+      "step": 4849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547010287097102e-05,
+      "loss": 0.6894,
+      "step": 4850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545910692627594e-05,
+      "loss": 0.7995,
+      "step": 4851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544810886227228e-05,
+      "loss": 0.7658,
+      "step": 4852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543710867926906e-05,
+      "loss": 0.7579,
+      "step": 4853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542610637757514e-05,
+      "loss": 0.791,
+      "step": 4854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541510195749957e-05,
+      "loss": 0.779,
+      "step": 4855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754040954193514e-05,
+      "loss": 0.7647,
+      "step": 4856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539308676343972e-05,
+      "loss": 0.791,
+      "step": 4857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538207599007382e-05,
+      "loss": 0.7074,
+      "step": 4858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537106309956286e-05,
+      "loss": 0.763,
+      "step": 4859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753600480922161e-05,
+      "loss": 0.7482,
+      "step": 4860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534903096834303e-05,
+      "loss": 0.783,
+      "step": 4861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533801172825297e-05,
+      "loss": 0.7702,
+      "step": 4862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532699037225543e-05,
+      "loss": 0.915,
+      "step": 4863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531596690065998e-05,
+      "loss": 0.6866,
+      "step": 4864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530494131377613e-05,
+      "loss": 0.7159,
+      "step": 4865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529391361191366e-05,
+      "loss": 0.6891,
+      "step": 4866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752828837953822e-05,
+      "loss": 0.7029,
+      "step": 4867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752718518644916e-05,
+      "loss": 0.7499,
+      "step": 4868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752608178195516e-05,
+      "loss": 0.7294,
+      "step": 4869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752497816608722e-05,
+      "loss": 0.7764,
+      "step": 4870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752387433887633e-05,
+      "loss": 0.7435,
+      "step": 4871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522770300353488e-05,
+      "loss": 0.7798,
+      "step": 4872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752166605054971e-05,
+      "loss": 0.7336,
+      "step": 4873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520561589496008e-05,
+      "loss": 0.7686,
+      "step": 4874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519456917223393e-05,
+      "loss": 0.71,
+      "step": 4875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75183520337629e-05,
+      "loss": 0.6886,
+      "step": 4876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751724693914556e-05,
+      "loss": 0.7149,
+      "step": 4877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75161416334024e-05,
+      "loss": 0.3259,
+      "step": 4878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751503611656447e-05,
+      "loss": 0.8682,
+      "step": 4879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751393038866282e-05,
+      "loss": 0.7228,
+      "step": 4880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512824449728506e-05,
+      "loss": 0.7467,
+      "step": 4881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511718299792586e-05,
+      "loss": 0.795,
+      "step": 4882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510611938886123e-05,
+      "loss": 0.7532,
+      "step": 4883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75095053670402e-05,
+      "loss": 0.7852,
+      "step": 4884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508398584285883e-05,
+      "loss": 0.817,
+      "step": 4885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507291590654268e-05,
+      "loss": 0.7421,
+      "step": 4886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750618438617644e-05,
+      "loss": 0.6973,
+      "step": 4887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505076970883493e-05,
+      "loss": 0.7684,
+      "step": 4888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503969344806536e-05,
+      "loss": 0.6771,
+      "step": 4889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7502861507976675e-05,
+      "loss": 0.7493,
+      "step": 4890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750175346042502e-05,
+      "loss": 0.7594,
+      "step": 4891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500645202182695e-05,
+      "loss": 0.7963,
+      "step": 4892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499536733280823e-05,
+      "loss": 0.7164,
+      "step": 4893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749842805375054e-05,
+      "loss": 0.6765,
+      "step": 4894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749731916362298e-05,
+      "loss": 0.68,
+      "step": 4895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496210062929287e-05,
+      "loss": 0.7579,
+      "step": 4896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495100751700617e-05,
+      "loss": 0.7647,
+      "step": 4897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493991229968118e-05,
+      "loss": 0.7923,
+      "step": 4898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492881497762958e-05,
+      "loss": 0.7667,
+      "step": 4899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491771555116298e-05,
+      "loss": 0.7393,
+      "step": 4900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749066140205931e-05,
+      "loss": 0.7503,
+      "step": 4901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489551038623183e-05,
+      "loss": 0.7817,
+      "step": 4902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488440464839097e-05,
+      "loss": 0.7941,
+      "step": 4903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748732968073824e-05,
+      "loss": 0.7946,
+      "step": 4904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748621868635181e-05,
+      "loss": 0.6948,
+      "step": 4905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 0.8165,
+      "step": 4906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483996066847054e-05,
+      "loss": 0.7701,
+      "step": 4907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482884441791153e-05,
+      "loss": 0.712,
+      "step": 4908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481772606574526e-05,
+      "loss": 0.737,
+      "step": 4909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74806605612284e-05,
+      "loss": 0.8279,
+      "step": 4910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747954830578401e-05,
+      "loss": 0.7903,
+      "step": 4911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747843584027259e-05,
+      "loss": 0.7103,
+      "step": 4912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477323164725386e-05,
+      "loss": 0.7581,
+      "step": 4913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747621027917365e-05,
+      "loss": 0.7135,
+      "step": 4914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747509718364864e-05,
+      "loss": 0.7237,
+      "step": 4915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7473983878181607e-05,
+      "loss": 0.7411,
+      "step": 4916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747287036280383e-05,
+      "loss": 0.7408,
+      "step": 4917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747175663754658e-05,
+      "loss": 0.7507,
+      "step": 4918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470642702441133e-05,
+      "loss": 0.7126,
+      "step": 4919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746952855751878e-05,
+      "loss": 0.8344,
+      "step": 4920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746841420281081e-05,
+      "loss": 0.7962,
+      "step": 4921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467299638348514e-05,
+      "loss": 0.7313,
+      "step": 4922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7466184864163205e-05,
+      "loss": 0.7504,
+      "step": 4923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746506988028619e-05,
+      "loss": 0.7702,
+      "step": 4924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7463954686748776e-05,
+      "loss": 0.7176,
+      "step": 4925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462839283582292e-05,
+      "loss": 0.776,
+      "step": 4926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746172367081806e-05,
+      "loss": 0.6752,
+      "step": 4927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7460607848487417e-05,
+      "loss": 0.8134,
+      "step": 4928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74594918166217e-05,
+      "loss": 0.6974,
+      "step": 4929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458375575252245e-05,
+      "loss": 0.757,
+      "step": 4930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457259124410418e-05,
+      "loss": 0.8523,
+      "step": 4931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456142464127562e-05,
+      "loss": 0.7332,
+      "step": 4932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455025594435047e-05,
+      "loss": 0.7512,
+      "step": 4933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453908515364238e-05,
+      "loss": 0.7826,
+      "step": 4934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452791226946502e-05,
+      "loss": 0.6564,
+      "step": 4935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7451673729213233e-05,
+      "loss": 0.7334,
+      "step": 4936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450556022195802e-05,
+      "loss": 0.7767,
+      "step": 4937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744943810592561e-05,
+      "loss": 0.7145,
+      "step": 4938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744831998043405e-05,
+      "loss": 0.7021,
+      "step": 4939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7447201645752526e-05,
+      "loss": 0.6858,
+      "step": 4940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744608310191245e-05,
+      "loss": 0.8212,
+      "step": 4941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7444964348945233e-05,
+      "loss": 0.761,
+      "step": 4942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7443845386882296e-05,
+      "loss": 0.7149,
+      "step": 4943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442726215755066e-05,
+      "loss": 0.7191,
+      "step": 4944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441606835594976e-05,
+      "loss": 0.723,
+      "step": 4945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440487246433466e-05,
+      "loss": 0.8498,
+      "step": 4946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439367448301976e-05,
+      "loss": 0.7438,
+      "step": 4947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743824744123196e-05,
+      "loss": 0.7808,
+      "step": 4948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437127225254877e-05,
+      "loss": 0.7644,
+      "step": 4949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436006800402178e-05,
+      "loss": 0.7739,
+      "step": 4950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743488616670534e-05,
+      "loss": 0.6956,
+      "step": 4951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433765324195836e-05,
+      "loss": 0.6986,
+      "step": 4952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432644272905142e-05,
+      "loss": 0.7014,
+      "step": 4953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7431523012864743e-05,
+      "loss": 0.7318,
+      "step": 4954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430401544106136e-05,
+      "loss": 0.7286,
+      "step": 4955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742927986666081e-05,
+      "loss": 0.7855,
+      "step": 4956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428157980560274e-05,
+      "loss": 0.6948,
+      "step": 4957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742703588583603e-05,
+      "loss": 0.7137,
+      "step": 4958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7425913582519602e-05,
+      "loss": 0.728,
+      "step": 4959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424791070642505e-05,
+      "loss": 0.7213,
+      "step": 4960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423668350236265e-05,
+      "loss": 0.723,
+      "step": 4961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422545421332414e-05,
+      "loss": 0.7175,
+      "step": 4962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421422283962496e-05,
+      "loss": 0.7304,
+      "step": 4963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7420298938158044e-05,
+      "loss": 0.6635,
+      "step": 4964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741917538395062e-05,
+      "loss": 0.7314,
+      "step": 4965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7418051621371767e-05,
+      "loss": 0.7088,
+      "step": 4966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416927650453056e-05,
+      "loss": 0.701,
+      "step": 4967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741580347122605e-05,
+      "loss": 0.7725,
+      "step": 4968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741467908372233e-05,
+      "loss": 0.7808,
+      "step": 4969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741355448797346e-05,
+      "loss": 0.7231,
+      "step": 4970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412429684011037e-05,
+      "loss": 0.8468,
+      "step": 4971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411304671866647e-05,
+      "loss": 0.7303,
+      "step": 4972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410179451571886e-05,
+      "loss": 0.7569,
+      "step": 4973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409054023158362e-05,
+      "loss": 0.7146,
+      "step": 4974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407928386657675e-05,
+      "loss": 0.765,
+      "step": 4975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406802542101445e-05,
+      "loss": 0.7258,
+      "step": 4976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405676489521292e-05,
+      "loss": 0.7394,
+      "step": 4977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404550228948837e-05,
+      "loss": 0.7189,
+      "step": 4978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403423760415717e-05,
+      "loss": 0.6641,
+      "step": 4979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402297083953565e-05,
+      "loss": 0.7685,
+      "step": 4980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401170199594027e-05,
+      "loss": 0.7594,
+      "step": 4981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7400043107368755e-05,
+      "loss": 0.6883,
+      "step": 4982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7398915807309397e-05,
+      "loss": 0.6766,
+      "step": 4983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739778829944762e-05,
+      "loss": 0.7413,
+      "step": 4984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396660583815083e-05,
+      "loss": 0.6891,
+      "step": 4985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739553266044347e-05,
+      "loss": 0.77,
+      "step": 4986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7394404529364447e-05,
+      "loss": 0.6732,
+      "step": 4987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393276190609704e-05,
+      "loss": 0.7922,
+      "step": 4988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392147644210933e-05,
+      "loss": 0.7314,
+      "step": 4989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739101889019983e-05,
+      "loss": 0.7634,
+      "step": 4990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738988992860809e-05,
+      "loss": 0.7331,
+      "step": 4991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388760759467426e-05,
+      "loss": 0.761,
+      "step": 4992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738763138280955e-05,
+      "loss": 0.6371,
+      "step": 4993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386501798666183e-05,
+      "loss": 0.7935,
+      "step": 4994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385372007069047e-05,
+      "loss": 0.7371,
+      "step": 4995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384242008049876e-05,
+      "loss": 0.6952,
+      "step": 4996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73831118016404e-05,
+      "loss": 0.8143,
+      "step": 4997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738198138787237e-05,
+      "loss": 0.7752,
+      "step": 4998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738085076677753e-05,
+      "loss": 0.8646,
+      "step": 4999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7379719938387634e-05,
+      "loss": 0.7629,
+      "step": 5000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737858890273444e-05,
+      "loss": 0.7715,
+      "step": 5001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377457659849718e-05,
+      "loss": 0.8053,
+      "step": 5002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737632620976524e-05,
+      "loss": 0.8138,
+      "step": 5003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375194552512777e-05,
+      "loss": 0.7968,
+      "step": 5004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374062688124118e-05,
+      "loss": 0.8055,
+      "step": 5005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737293061663105e-05,
+      "loss": 0.7705,
+      "step": 5006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371798338065367e-05,
+      "loss": 0.8718,
+      "step": 5007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737066585245887e-05,
+      "loss": 0.7304,
+      "step": 5008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369533159843368e-05,
+      "loss": 0.7276,
+      "step": 5009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736840026025067e-05,
+      "loss": 0.698,
+      "step": 5010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367267153712596e-05,
+      "loss": 0.7859,
+      "step": 5011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366133840260973e-05,
+      "loss": 0.7595,
+      "step": 5012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365000319927624e-05,
+      "loss": 0.6683,
+      "step": 5013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7363866592744385e-05,
+      "loss": 0.7422,
+      "step": 5014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73627326587431e-05,
+      "loss": 0.7389,
+      "step": 5015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736159851795562e-05,
+      "loss": 0.828,
+      "step": 5016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360464170413792e-05,
+      "loss": 0.6953,
+      "step": 5017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359329616149473e-05,
+      "loss": 0.7623,
+      "step": 5018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358194855194536e-05,
+      "loss": 0.7302,
+      "step": 5019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357059887580846e-05,
+      "loss": 0.7898,
+      "step": 5020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735592471334028e-05,
+      "loss": 0.769,
+      "step": 5021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354789332504717e-05,
+      "loss": 0.7435,
+      "step": 5022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735365374510605e-05,
+      "loss": 0.6831,
+      "step": 5023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352517951176167e-05,
+      "loss": 0.7092,
+      "step": 5024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735138195074697e-05,
+      "loss": 0.7299,
+      "step": 5025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350245743850367e-05,
+      "loss": 0.7097,
+      "step": 5026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7349109330518267e-05,
+      "loss": 0.7889,
+      "step": 5027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347972710782586e-05,
+      "loss": 0.7749,
+      "step": 5028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346835884675248e-05,
+      "loss": 0.789,
+      "step": 5029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345698852228173e-05,
+      "loss": 0.7577,
+      "step": 5030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734456161347331e-05,
+      "loss": 0.7476,
+      "step": 5031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343424168442588e-05,
+      "loss": 0.8067,
+      "step": 5032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7342286517167957e-05,
+      "loss": 0.7447,
+      "step": 5033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734114865968137e-05,
+      "loss": 0.7406,
+      "step": 5034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340010596014778e-05,
+      "loss": 0.7482,
+      "step": 5035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338872326200148e-05,
+      "loss": 0.7781,
+      "step": 5036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733773385026945e-05,
+      "loss": 0.7579,
+      "step": 5037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7336595168254657e-05,
+      "loss": 0.8188,
+      "step": 5038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335456280187752e-05,
+      "loss": 0.7496,
+      "step": 5039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733431718610072e-05,
+      "loss": 0.8519,
+      "step": 5040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333177886025546e-05,
+      "loss": 0.8106,
+      "step": 5041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733203837999424e-05,
+      "loss": 0.7767,
+      "step": 5042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330898668038797e-05,
+      "loss": 0.7432,
+      "step": 5043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329758750191233e-05,
+      "loss": 0.8213,
+      "step": 5044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328618626483557e-05,
+      "loss": 0.6788,
+      "step": 5045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732747829694779e-05,
+      "loss": 0.7706,
+      "step": 5046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7326337761615965e-05,
+      "loss": 0.7426,
+      "step": 5047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325197020520108e-05,
+      "loss": 0.7905,
+      "step": 5048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732405607369226e-05,
+      "loss": 0.7363,
+      "step": 5049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732291492116447e-05,
+      "loss": 0.7058,
+      "step": 5050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732177356296878e-05,
+      "loss": 0.7769,
+      "step": 5051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7320631999137245e-05,
+      "loss": 0.8152,
+      "step": 5052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319490229701934e-05,
+      "loss": 0.6541,
+      "step": 5053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318348254694907e-05,
+      "loss": 0.7067,
+      "step": 5054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317206074148245e-05,
+      "loss": 0.7492,
+      "step": 5055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731606368809402e-05,
+      "loss": 0.692,
+      "step": 5056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7314921096564314e-05,
+      "loss": 0.7521,
+      "step": 5057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313778299591227e-05,
+      "loss": 0.7255,
+      "step": 5058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312635297206846e-05,
+      "loss": 0.8495,
+      "step": 5059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731149208944328e-05,
+      "loss": 0.7537,
+      "step": 5060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310348676332634e-05,
+      "loss": 0.6733,
+      "step": 5061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730920505790702e-05,
+      "loss": 0.7858,
+      "step": 5062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308061234198553e-05,
+      "loss": 0.7693,
+      "step": 5063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306917205239365e-05,
+      "loss": 0.743,
+      "step": 5064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305772971061583e-05,
+      "loss": 0.7902,
+      "step": 5065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304628531697347e-05,
+      "loss": 0.7421,
+      "step": 5066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303483887178797e-05,
+      "loss": 0.7015,
+      "step": 5067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7302339037538076e-05,
+      "loss": 0.7774,
+      "step": 5068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730119398280735e-05,
+      "loss": 0.7888,
+      "step": 5069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300048723018764e-05,
+      "loss": 0.692,
+      "step": 5070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298903258204492e-05,
+      "loss": 0.7845,
+      "step": 5071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297757588396703e-05,
+      "loss": 0.7308,
+      "step": 5072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7296611713627575e-05,
+      "loss": 0.7333,
+      "step": 5073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729546563392929e-05,
+      "loss": 0.7477,
+      "step": 5074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294319349334033e-05,
+      "loss": 0.7792,
+      "step": 5075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7293172859874003e-05,
+      "loss": 0.7586,
+      "step": 5076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7292026165581396e-05,
+      "loss": 0.7359,
+      "step": 5077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729087926648842e-05,
+      "loss": 0.75,
+      "step": 5078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7289732162627286e-05,
+      "loss": 0.6818,
+      "step": 5079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288584854030207e-05,
+      "loss": 0.7347,
+      "step": 5080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728743734072941e-05,
+      "loss": 0.7565,
+      "step": 5081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286289622757123e-05,
+      "loss": 0.7003,
+      "step": 5082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728514170014558e-05,
+      "loss": 0.7377,
+      "step": 5083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283993572927017e-05,
+      "loss": 0.7048,
+      "step": 5084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7282845241133688e-05,
+      "loss": 0.7677,
+      "step": 5085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281696704797836e-05,
+      "loss": 0.7641,
+      "step": 5086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280547963951724e-05,
+      "loss": 0.7495,
+      "step": 5087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279399018627613e-05,
+      "loss": 0.7405,
+      "step": 5088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278249868857772e-05,
+      "loss": 0.7616,
+      "step": 5089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277100514674475e-05,
+      "loss": 0.7561,
+      "step": 5090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275950956110003e-05,
+      "loss": 0.7596,
+      "step": 5091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274801193196644e-05,
+      "loss": 0.7461,
+      "step": 5092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273651225966682e-05,
+      "loss": 0.7723,
+      "step": 5093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727250105445242e-05,
+      "loss": 0.7562,
+      "step": 5094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271350678686164e-05,
+      "loss": 0.7427,
+      "step": 5095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727020009870022e-05,
+      "loss": 0.7599,
+      "step": 5096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72690493145269e-05,
+      "loss": 0.6858,
+      "step": 5097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726789832619853e-05,
+      "loss": 0.8023,
+      "step": 5098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726674713374743e-05,
+      "loss": 0.806,
+      "step": 5099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7265595737205933e-05,
+      "loss": 0.788,
+      "step": 5100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726444413660638e-05,
+      "loss": 0.7231,
+      "step": 5101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263292331981115e-05,
+      "loss": 0.6792,
+      "step": 5102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262140323362483e-05,
+      "loss": 0.7262,
+      "step": 5103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260988110782836e-05,
+      "loss": 0.7601,
+      "step": 5104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259835694274542e-05,
+      "loss": 0.7745,
+      "step": 5105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258683073869965e-05,
+      "loss": 0.6276,
+      "step": 5106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257530249601474e-05,
+      "loss": 0.7868,
+      "step": 5107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256377221501454e-05,
+      "loss": 0.7327,
+      "step": 5108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255223989602277e-05,
+      "loss": 0.707,
+      "step": 5109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7254070553936338e-05,
+      "loss": 0.7736,
+      "step": 5110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252916914536038e-05,
+      "loss": 0.8063,
+      "step": 5111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251763071433767e-05,
+      "loss": 0.7335,
+      "step": 5112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7250609024661935e-05,
+      "loss": 0.7422,
+      "step": 5113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249454774252958e-05,
+      "loss": 0.7684,
+      "step": 5114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724830032023925e-05,
+      "loss": 0.7222,
+      "step": 5115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247145662653234e-05,
+      "loss": 0.6866,
+      "step": 5116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724599080152734e-05,
+      "loss": 0.7737,
+      "step": 5117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244835736894002e-05,
+      "loss": 0.699,
+      "step": 5118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243680468785662e-05,
+      "loss": 0.7477,
+      "step": 5119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7242524997234767e-05,
+      "loss": 0.8013,
+      "step": 5120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7241369322273772e-05,
+      "loss": 0.3593,
+      "step": 5121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240213443935125e-05,
+      "loss": 0.7418,
+      "step": 5122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239057362251297e-05,
+      "loss": 0.7455,
+      "step": 5123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237901077254754e-05,
+      "loss": 0.7311,
+      "step": 5124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236744588977977e-05,
+      "loss": 0.7711,
+      "step": 5125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235587897453436e-05,
+      "loss": 0.7713,
+      "step": 5126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234431002713626e-05,
+      "loss": 0.6752,
+      "step": 5127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233273904791036e-05,
+      "loss": 0.7904,
+      "step": 5128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232116603718167e-05,
+      "loss": 0.7151,
+      "step": 5129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230959099527512e-05,
+      "loss": 0.6897,
+      "step": 5130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722980139225159e-05,
+      "loss": 0.6658,
+      "step": 5131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228643481922916e-05,
+      "loss": 0.7256,
+      "step": 5132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227485368574006e-05,
+      "loss": 0.7621,
+      "step": 5133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226327052237388e-05,
+      "loss": 0.8151,
+      "step": 5134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225168532945593e-05,
+      "loss": 0.3352,
+      "step": 5135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7224009810731163e-05,
+      "loss": 0.7257,
+      "step": 5136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222850885626632e-05,
+      "loss": 0.7212,
+      "step": 5137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7221691757664558e-05,
+      "loss": 0.7678,
+      "step": 5138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722053242687749e-05,
+      "loss": 0.6949,
+      "step": 5139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7219372893297995e-05,
+      "loss": 0.7587,
+      "step": 5140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218213156958633e-05,
+      "loss": 0.7424,
+      "step": 5141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7217053217891975e-05,
+      "loss": 0.7654,
+      "step": 5142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7215893076130603e-05,
+      "loss": 0.8355,
+      "step": 5143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72147327317071e-05,
+      "loss": 0.7934,
+      "step": 5144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213572184654054e-05,
+      "loss": 0.7849,
+      "step": 5145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212411435004055e-05,
+      "loss": 0.7872,
+      "step": 5146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721125048278971e-05,
+      "loss": 0.7693,
+      "step": 5147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210089328043622e-05,
+      "loss": 0.7248,
+      "step": 5148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72089279707984e-05,
+      "loss": 0.7344,
+      "step": 5149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7207766411086667e-05,
+      "loss": 0.7917,
+      "step": 5150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206604648941045e-05,
+      "loss": 0.7638,
+      "step": 5151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720544268439416e-05,
+      "loss": 0.3376,
+      "step": 5152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720428051747864e-05,
+      "loss": 0.6936,
+      "step": 5153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720311814822714e-05,
+      "loss": 0.8634,
+      "step": 5154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201955576672298e-05,
+      "loss": 0.7004,
+      "step": 5155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200792802846763e-05,
+      "loss": 0.826,
+      "step": 5156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199629826783195e-05,
+      "loss": 0.7378,
+      "step": 5157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7198466648514254e-05,
+      "loss": 0.7856,
+      "step": 5158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197303268072613e-05,
+      "loss": 0.7009,
+      "step": 5159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7196139685490943e-05,
+      "loss": 0.7149,
+      "step": 5160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194975900801923e-05,
+      "loss": 0.7222,
+      "step": 5161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719381191403825e-05,
+      "loss": 0.6727,
+      "step": 5162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7192647725232595e-05,
+      "loss": 0.6931,
+      "step": 5163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719148333441767e-05,
+      "loss": 0.6547,
+      "step": 5164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190318741626168e-05,
+      "loss": 0.7687,
+      "step": 5165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7189153946890806e-05,
+      "loss": 0.6963,
+      "step": 5166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187988950244294e-05,
+      "loss": 0.7026,
+      "step": 5167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186823751719348e-05,
+      "loss": 0.7525,
+      "step": 5168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185658351348703e-05,
+      "loss": 0.7141,
+      "step": 5169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718449274916508e-05,
+      "loss": 0.7399,
+      "step": 5170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183326945201216e-05,
+      "loss": 0.7197,
+      "step": 5171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718216093948986e-05,
+      "loss": 0.7896,
+      "step": 5172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180994732063756e-05,
+      "loss": 0.7078,
+      "step": 5173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179828322955656e-05,
+      "loss": 0.7622,
+      "step": 5174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717866171219832e-05,
+      "loss": 0.7357,
+      "step": 5175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717749489982452e-05,
+      "loss": 0.7578,
+      "step": 5176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176327885867013e-05,
+      "loss": 0.7458,
+      "step": 5177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175160670358586e-05,
+      "loss": 0.7705,
+      "step": 5178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173993253332016e-05,
+      "loss": 0.7127,
+      "step": 5179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7172825634820095e-05,
+      "loss": 0.758,
+      "step": 5180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717165781485561e-05,
+      "loss": 0.7337,
+      "step": 5181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7170489793471365e-05,
+      "loss": 0.7354,
+      "step": 5182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716932157070016e-05,
+      "loss": 0.7456,
+      "step": 5183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716815314657481e-05,
+      "loss": 0.8392,
+      "step": 5184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166984521128127e-05,
+      "loss": 0.6771,
+      "step": 5185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165815694392936e-05,
+      "loss": 0.792,
+      "step": 5186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716464666640206e-05,
+      "loss": 0.78,
+      "step": 5187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716347743718833e-05,
+      "loss": 0.7469,
+      "step": 5188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7162308006784594e-05,
+      "loss": 0.7296,
+      "step": 5189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161138375223688e-05,
+      "loss": 0.8242,
+      "step": 5190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7159968542538464e-05,
+      "loss": 0.762,
+      "step": 5191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7158798508761778e-05,
+      "loss": 0.8228,
+      "step": 5192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7157628273926487e-05,
+      "loss": 0.7133,
+      "step": 5193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156457838065464e-05,
+      "loss": 0.7404,
+      "step": 5194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7155287201211578e-05,
+      "loss": 0.7556,
+      "step": 5195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154116363397706e-05,
+      "loss": 0.798,
+      "step": 5196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152945324656734e-05,
+      "loss": 0.7499,
+      "step": 5197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151774085021548e-05,
+      "loss": 0.74,
+      "step": 5198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150602644525046e-05,
+      "loss": 0.7421,
+      "step": 5199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149431003200128e-05,
+      "loss": 0.7701,
+      "step": 5200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7148259161079698e-05,
+      "loss": 0.7894,
+      "step": 5201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147087118196673e-05,
+      "loss": 0.8028,
+      "step": 5202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145914874583964e-05,
+      "loss": 0.7795,
+      "step": 5203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144742430274496e-05,
+      "loss": 0.7801,
+      "step": 5204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71435697853012e-05,
+      "loss": 0.7382,
+      "step": 5205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714239693969701e-05,
+      "loss": 0.6884,
+      "step": 5206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141223893494863e-05,
+      "loss": 0.7812,
+      "step": 5207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140050646727706e-05,
+      "loss": 0.7915,
+      "step": 5208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138877199428493e-05,
+      "loss": 0.8134,
+      "step": 5209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7137703551630175e-05,
+      "loss": 0.693,
+      "step": 5210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136529703365724e-05,
+      "loss": 0.7094,
+      "step": 5211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135355654668098e-05,
+      "loss": 0.7714,
+      "step": 5212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7134181405570272e-05,
+      "loss": 0.7886,
+      "step": 5213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133006956105237e-05,
+      "loss": 0.8084,
+      "step": 5214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131832306305964e-05,
+      "loss": 0.7744,
+      "step": 5215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713065745620545e-05,
+      "loss": 0.7659,
+      "step": 5216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712948240583669e-05,
+      "loss": 0.7663,
+      "step": 5217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128307155232685e-05,
+      "loss": 0.7139,
+      "step": 5218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712713170442645e-05,
+      "loss": 0.6989,
+      "step": 5219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125956053450985e-05,
+      "loss": 0.7995,
+      "step": 5220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712478020233932e-05,
+      "loss": 0.7349,
+      "step": 5221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7123604151124475e-05,
+      "loss": 0.7758,
+      "step": 5222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712242789983948e-05,
+      "loss": 0.7581,
+      "step": 5223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712125144851737e-05,
+      "loss": 0.737,
+      "step": 5224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120074797191185e-05,
+      "loss": 0.7255,
+      "step": 5225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118897945893976e-05,
+      "loss": 0.7614,
+      "step": 5226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117720894658793e-05,
+      "loss": 0.7194,
+      "step": 5227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116543643518695e-05,
+      "loss": 0.7665,
+      "step": 5228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7115366192506746e-05,
+      "loss": 0.732,
+      "step": 5229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711418854165601e-05,
+      "loss": 0.7616,
+      "step": 5230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113010690999572e-05,
+      "loss": 0.695,
+      "step": 5231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7111832640570508e-05,
+      "loss": 0.7424,
+      "step": 5232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110654390401896e-05,
+      "loss": 0.689,
+      "step": 5233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710947594052684e-05,
+      "loss": 0.765,
+      "step": 5234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108297290978435e-05,
+      "loss": 0.7841,
+      "step": 5235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7107118441789776e-05,
+      "loss": 0.3309,
+      "step": 5236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105939392993982e-05,
+      "loss": 0.7928,
+      "step": 5237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7104760144624157e-05,
+      "loss": 0.7553,
+      "step": 5238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103580696713427e-05,
+      "loss": 0.7113,
+      "step": 5239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102401049294922e-05,
+      "loss": 0.7369,
+      "step": 5240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710122120240176e-05,
+      "loss": 0.7109,
+      "step": 5241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710004115606709e-05,
+      "loss": 0.7814,
+      "step": 5242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7098860910324046e-05,
+      "loss": 0.745,
+      "step": 5243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709768046520578e-05,
+      "loss": 0.7727,
+      "step": 5244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7096499820745448e-05,
+      "loss": 0.7678,
+      "step": 5245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095318976976202e-05,
+      "loss": 0.7037,
+      "step": 5246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709413793393121e-05,
+      "loss": 0.8077,
+      "step": 5247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7092956691643642e-05,
+      "loss": 0.731,
+      "step": 5248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091775250146678e-05,
+      "loss": 0.8267,
+      "step": 5249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709059360947349e-05,
+      "loss": 0.7024,
+      "step": 5250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7089411769657273e-05,
+      "loss": 0.7281,
+      "step": 5251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708822973073122e-05,
+      "loss": 0.7734,
+      "step": 5252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708704749272852e-05,
+      "loss": 0.8264,
+      "step": 5253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708586505568239e-05,
+      "loss": 0.7273,
+      "step": 5254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084682419626028e-05,
+      "loss": 0.7322,
+      "step": 5255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083499584592654e-05,
+      "loss": 0.7229,
+      "step": 5256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082316550615486e-05,
+      "loss": 0.6973,
+      "step": 5257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081133317727756e-05,
+      "loss": 0.7631,
+      "step": 5258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079949885962693e-05,
+      "loss": 0.8057,
+      "step": 5259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707876625535353e-05,
+      "loss": 0.6756,
+      "step": 5260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707758242593351e-05,
+      "loss": 0.7735,
+      "step": 5261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707639839773589e-05,
+      "loss": 0.7904,
+      "step": 5262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075214170793918e-05,
+      "loss": 0.7838,
+      "step": 5263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074029745140852e-05,
+      "loss": 0.7639,
+      "step": 5264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707284512080996e-05,
+      "loss": 0.7164,
+      "step": 5265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707166029783451e-05,
+      "loss": 0.3637,
+      "step": 5266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7070475276247785e-05,
+      "loss": 0.8061,
+      "step": 5267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069290056083056e-05,
+      "loss": 0.8022,
+      "step": 5268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068104637373625e-05,
+      "loss": 0.8065,
+      "step": 5269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706691902015277e-05,
+      "loss": 0.7896,
+      "step": 5270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70657332044538e-05,
+      "loss": 0.6965,
+      "step": 5271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706454719031002e-05,
+      "loss": 0.836,
+      "step": 5272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063360977754726e-05,
+      "loss": 0.7949,
+      "step": 5273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706217456682125e-05,
+      "loss": 0.7187,
+      "step": 5274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060987957542906e-05,
+      "loss": 0.7094,
+      "step": 5275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705980114995302e-05,
+      "loss": 0.7025,
+      "step": 5276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7058614144084924e-05,
+      "loss": 0.8203,
+      "step": 5277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705742693997196e-05,
+      "loss": 0.7614,
+      "step": 5278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705623953764746e-05,
+      "loss": 0.7773,
+      "step": 5279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705505193714479e-05,
+      "loss": 0.6753,
+      "step": 5280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053864138497288e-05,
+      "loss": 0.7346,
+      "step": 5281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052676141738324e-05,
+      "loss": 0.7621,
+      "step": 5282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705148794690126e-05,
+      "loss": 0.7327,
+      "step": 5283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050299554019466e-05,
+      "loss": 0.8085,
+      "step": 5284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7049110963126324e-05,
+      "loss": 0.763,
+      "step": 5285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704792217425521e-05,
+      "loss": 0.7911,
+      "step": 5286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046733187439512e-05,
+      "loss": 0.7634,
+      "step": 5287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7045544002712628e-05,
+      "loss": 0.7133,
+      "step": 5288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044354620107955e-05,
+      "loss": 0.7849,
+      "step": 5289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043165039658896e-05,
+      "loss": 0.7543,
+      "step": 5290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041975261398862e-05,
+      "loss": 0.7119,
+      "step": 5291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040785285361267e-05,
+      "loss": 0.6608,
+      "step": 5292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039595111579533e-05,
+      "loss": 0.7531,
+      "step": 5293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703840474008709e-05,
+      "loss": 0.7219,
+      "step": 5294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037214170917366e-05,
+      "loss": 0.683,
+      "step": 5295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036023404103797e-05,
+      "loss": 0.7131,
+      "step": 5296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703483243967983e-05,
+      "loss": 0.7109,
+      "step": 5297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033641277678916e-05,
+      "loss": 0.6639,
+      "step": 5298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032449918134506e-05,
+      "loss": 0.7163,
+      "step": 5299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703125836108006e-05,
+      "loss": 0.7319,
+      "step": 5300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7030066606549043e-05,
+      "loss": 0.7797,
+      "step": 5301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702887465457493e-05,
+      "loss": 0.6493,
+      "step": 5302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027682505191192e-05,
+      "loss": 0.7673,
+      "step": 5303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026490158431316e-05,
+      "loss": 0.791,
+      "step": 5304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025297614328785e-05,
+      "loss": 0.8038,
+      "step": 5305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024104872917095e-05,
+      "loss": 0.7889,
+      "step": 5306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022911934229744e-05,
+      "loss": 0.7608,
+      "step": 5307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702171879830024e-05,
+      "loss": 0.7631,
+      "step": 5308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020525465162084e-05,
+      "loss": 0.7505,
+      "step": 5309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7019331934848803e-05,
+      "loss": 0.7529,
+      "step": 5310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701813820739391e-05,
+      "loss": 0.7014,
+      "step": 5311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7016944282830935e-05,
+      "loss": 0.7943,
+      "step": 5312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015750161193406e-05,
+      "loss": 0.7588,
+      "step": 5313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701455584251486e-05,
+      "loss": 0.7871,
+      "step": 5314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013361326828847e-05,
+      "loss": 0.762,
+      "step": 5315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012166614168913e-05,
+      "loss": 0.6689,
+      "step": 5316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701097170456861e-05,
+      "loss": 0.8104,
+      "step": 5317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009776598061496e-05,
+      "loss": 0.7241,
+      "step": 5318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700858129468114e-05,
+      "loss": 0.6859,
+      "step": 5319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700738579446111e-05,
+      "loss": 0.3437,
+      "step": 5320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700619009743499e-05,
+      "loss": 0.7689,
+      "step": 5321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004994203636344e-05,
+      "loss": 0.751,
+      "step": 5322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7003798113098776e-05,
+      "loss": 0.6738,
+      "step": 5323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700260182585588e-05,
+      "loss": 0.8223,
+      "step": 5324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700140534194124e-05,
+      "loss": 0.73,
+      "step": 5325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000208661388467e-05,
+      "loss": 0.7703,
+      "step": 5326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6999011784231175e-05,
+      "loss": 0.7256,
+      "step": 5327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6997814710502973e-05,
+      "loss": 0.7572,
+      "step": 5328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996617440237482e-05,
+      "loss": 0.7549,
+      "step": 5329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699541997346833e-05,
+      "loss": 0.8015,
+      "step": 5330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994222310229155e-05,
+      "loss": 0.6794,
+      "step": 5331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699302445055358e-05,
+      "loss": 0.7738,
+      "step": 5332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991826394475254e-05,
+      "loss": 0.7142,
+      "step": 5333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6990628142027828e-05,
+      "loss": 0.76,
+      "step": 5334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698942969324495e-05,
+      "loss": 0.7341,
+      "step": 5335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988231048160287e-05,
+      "loss": 0.766,
+      "step": 5336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987032206807498e-05,
+      "loss": 0.7754,
+      "step": 5337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698583316922025e-05,
+      "loss": 0.7834,
+      "step": 5338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984633935432223e-05,
+      "loss": 0.7151,
+      "step": 5339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69834345054771e-05,
+      "loss": 0.7051,
+      "step": 5340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698223487938856e-05,
+      "loss": 0.6962,
+      "step": 5341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6981035057200306e-05,
+      "loss": 0.7245,
+      "step": 5342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979835038946028e-05,
+      "loss": 0.7607,
+      "step": 5343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697863482465943e-05,
+      "loss": 0.7183,
+      "step": 5344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697743441437422e-05,
+      "loss": 0.7156,
+      "step": 5345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6976233808124115e-05,
+      "loss": 0.7639,
+      "step": 5346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975033005942837e-05,
+      "loss": 0.7172,
+      "step": 5347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6973832007864102e-05,
+      "loss": 0.8344,
+      "step": 5348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697263081392165e-05,
+      "loss": 0.7649,
+      "step": 5349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697142942414921e-05,
+      "loss": 0.7739,
+      "step": 5350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697022783858053e-05,
+      "loss": 0.7418,
+      "step": 5351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6969026057249355e-05,
+      "loss": 0.7786,
+      "step": 5352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967824080189435e-05,
+      "loss": 0.7515,
+      "step": 5353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696662190743453e-05,
+      "loss": 0.6687,
+      "step": 5354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965419539018405e-05,
+      "loss": 0.7184,
+      "step": 5355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964216974974824e-05,
+      "loss": 0.6717,
+      "step": 5356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696301421533757e-05,
+      "loss": 0.7833,
+      "step": 5357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961811260140417e-05,
+      "loss": 0.845,
+      "step": 5358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960608109417148e-05,
+      "loss": 0.7742,
+      "step": 5359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6959404763201564e-05,
+      "loss": 0.332,
+      "step": 5360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958201221527457e-05,
+      "loss": 0.8436,
+      "step": 5361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956997484428624e-05,
+      "loss": 0.6822,
+      "step": 5362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955793551938878e-05,
+      "loss": 0.6696,
+      "step": 5363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695458942409203e-05,
+      "loss": 0.3288,
+      "step": 5364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69533851009219e-05,
+      "loss": 0.77,
+      "step": 5365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952180582462313e-05,
+      "loss": 0.7847,
+      "step": 5366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695097586874709e-05,
+      "loss": 0.7516,
+      "step": 5367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949770959810078e-05,
+      "loss": 0.7031,
+      "step": 5368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948565855685114e-05,
+      "loss": 0.7885,
+      "step": 5369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694736055640604e-05,
+      "loss": 0.7056,
+      "step": 5370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946155062006712e-05,
+      "loss": 0.8328,
+      "step": 5371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694494937252098e-05,
+      "loss": 0.8317,
+      "step": 5372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943743487982714e-05,
+      "loss": 0.7608,
+      "step": 5373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694253740842578e-05,
+      "loss": 0.756,
+      "step": 5374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694133113388405e-05,
+      "loss": 0.7406,
+      "step": 5375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940124664391403e-05,
+      "loss": 0.7348,
+      "step": 5376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693891799998172e-05,
+      "loss": 0.758,
+      "step": 5377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69377111406889e-05,
+      "loss": 0.7397,
+      "step": 5378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693650408654683e-05,
+      "loss": 0.8411,
+      "step": 5379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935296837589414e-05,
+      "loss": 0.7946,
+      "step": 5380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934089393850555e-05,
+      "loss": 0.7527,
+      "step": 5381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693288175536417e-05,
+      "loss": 0.752,
+      "step": 5382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931673922164172e-05,
+      "loss": 0.7605,
+      "step": 5383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6930465894284488e-05,
+      "loss": 0.7569,
+      "step": 5384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692925767175904e-05,
+      "loss": 0.7474,
+      "step": 5385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6928049254621765e-05,
+      "loss": 0.8141,
+      "step": 5386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926840642906604e-05,
+      "loss": 0.7507,
+      "step": 5387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69256318366475e-05,
+      "loss": 0.6873,
+      "step": 5388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69244228358784e-05,
+      "loss": 0.7375,
+      "step": 5389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692321364063326e-05,
+      "loss": 0.7255,
+      "step": 5390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922004250946046e-05,
+      "loss": 0.7254,
+      "step": 5391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920794666850722e-05,
+      "loss": 0.7504,
+      "step": 5392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919584888381253e-05,
+      "loss": 0.6691,
+      "step": 5393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6918374915571627e-05,
+      "loss": 0.7374,
+      "step": 5394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6917164748455818e-05,
+      "loss": 0.742,
+      "step": 5395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6915954387067822e-05,
+      "loss": 0.7785,
+      "step": 5396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6914743831441625e-05,
+      "loss": 0.763,
+      "step": 5397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6913533081611228e-05,
+      "loss": 0.7385,
+      "step": 5398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691232213761064e-05,
+      "loss": 0.7584,
+      "step": 5399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691111099947387e-05,
+      "loss": 0.7953,
+      "step": 5400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6909899667234927e-05,
+      "loss": 0.7988,
+      "step": 5401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690868814092784e-05,
+      "loss": 0.7344,
+      "step": 5402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907476420586632e-05,
+      "loss": 0.8169,
+      "step": 5403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906264506245334e-05,
+      "loss": 0.7685,
+      "step": 5404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905052397937983e-05,
+      "loss": 0.7618,
+      "step": 5405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690384009569862e-05,
+      "loss": 0.358,
+      "step": 5406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902627599561298e-05,
+      "loss": 0.7759,
+      "step": 5407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690141490956007e-05,
+      "loss": 0.7701,
+      "step": 5408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900202025728992e-05,
+      "loss": 0.684,
+      "step": 5409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689898894810213e-05,
+      "loss": 0.7174,
+      "step": 5410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897775676713552e-05,
+      "loss": 0.7384,
+      "step": 5411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896562211597337e-05,
+      "loss": 0.7665,
+      "step": 5412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6895348552787564e-05,
+      "loss": 0.6836,
+      "step": 5413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894134700318317e-05,
+      "loss": 0.8478,
+      "step": 5414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892920654223695e-05,
+      "loss": 0.8322,
+      "step": 5415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689170641453778e-05,
+      "loss": 0.7262,
+      "step": 5416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890491981294692e-05,
+      "loss": 0.7978,
+      "step": 5417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688927735452853e-05,
+      "loss": 0.7038,
+      "step": 5418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888062534273412e-05,
+      "loss": 0.7719,
+      "step": 5419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886847520563448e-05,
+      "loss": 0.699,
+      "step": 5420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885632313432772e-05,
+      "loss": 0.707,
+      "step": 5421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688441691291551e-05,
+      "loss": 0.7164,
+      "step": 5422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6883201319045795e-05,
+      "loss": 0.7072,
+      "step": 5423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688198553185777e-05,
+      "loss": 0.748,
+      "step": 5424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688076955138558e-05,
+      "loss": 0.7336,
+      "step": 5425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6879553377663377e-05,
+      "loss": 0.6846,
+      "step": 5426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687833701072532e-05,
+      "loss": 0.7415,
+      "step": 5427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877120450605566e-05,
+      "loss": 0.7116,
+      "step": 5428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687590369733829e-05,
+      "loss": 0.7386,
+      "step": 5429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874686750957657e-05,
+      "loss": 0.7286,
+      "step": 5430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687346961149785e-05,
+      "loss": 0.6776,
+      "step": 5431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872252278993056e-05,
+      "loss": 0.7105,
+      "step": 5432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871034753477458e-05,
+      "loss": 0.7529,
+      "step": 5433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686981703498525e-05,
+      "loss": 0.6803,
+      "step": 5434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6868599123550645e-05,
+      "loss": 0.7784,
+      "step": 5435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867381019207836e-05,
+      "loss": 0.7663,
+      "step": 5436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866162721991037e-05,
+      "loss": 0.7588,
+      "step": 5437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864944231934466e-05,
+      "loss": 0.7491,
+      "step": 5438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6863725549072344e-05,
+      "loss": 0.7462,
+      "step": 5439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862506673438898e-05,
+      "loss": 0.718,
+      "step": 5440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861287605068365e-05,
+      "loss": 0.7715,
+      "step": 5441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6860068343994975e-05,
+      "loss": 0.8109,
+      "step": 5442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685884889025298e-05,
+      "loss": 0.8185,
+      "step": 5443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6857629243876624e-05,
+      "loss": 0.7505,
+      "step": 5444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685640940490016e-05,
+      "loss": 0.782,
+      "step": 5445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855189373357853e-05,
+      "loss": 0.7585,
+      "step": 5446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853969149283967e-05,
+      "loss": 0.6423,
+      "step": 5447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685274873271277e-05,
+      "loss": 0.7852,
+      "step": 5448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851528123678536e-05,
+      "loss": 0.6631,
+      "step": 5449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850307322215554e-05,
+      "loss": 0.7562,
+      "step": 5450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849086328358105e-05,
+      "loss": 0.75,
+      "step": 5451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847865142140486e-05,
+      "loss": 0.664,
+      "step": 5452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846643763596987e-05,
+      "loss": 0.6788,
+      "step": 5453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684542219276192e-05,
+      "loss": 0.7646,
+      "step": 5454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6844200429669582e-05,
+      "loss": 0.7402,
+      "step": 5455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842978474354298e-05,
+      "loss": 0.7076,
+      "step": 5456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684175632685039e-05,
+      "loss": 0.7551,
+      "step": 5457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840533987192166e-05,
+      "loss": 0.7376,
+      "step": 5458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683931145541397e-05,
+      "loss": 0.8181,
+      "step": 5459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838088731550133e-05,
+      "loss": 0.8545,
+      "step": 5460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836865815634997e-05,
+      "loss": 0.7578,
+      "step": 5461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835642707702905e-05,
+      "loss": 0.8289,
+      "step": 5462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6834419407788208e-05,
+      "loss": 0.7389,
+      "step": 5463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6833195915925274e-05,
+      "loss": 0.735,
+      "step": 5464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831972232148452e-05,
+      "loss": 0.7316,
+      "step": 5465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830748356492118e-05,
+      "loss": 0.7453,
+      "step": 5466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682952428899064e-05,
+      "loss": 0.7258,
+      "step": 5467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68283000296784e-05,
+      "loss": 0.7661,
+      "step": 5468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827075578589783e-05,
+      "loss": 0.7162,
+      "step": 5469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6825850935759175e-05,
+      "loss": 0.7104,
+      "step": 5470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6824626101220972e-05,
+      "loss": 0.7648,
+      "step": 5471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682340107500957e-05,
+      "loss": 0.8232,
+      "step": 5472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822175857159383e-05,
+      "loss": 0.77,
+      "step": 5473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682095044770482e-05,
+      "loss": 0.7566,
+      "step": 5474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681972484668029e-05,
+      "loss": 0.694,
+      "step": 5475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681849905412022e-05,
+      "loss": 0.6491,
+      "step": 5476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681727307005904e-05,
+      "loss": 0.7969,
+      "step": 5477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6816046894531176e-05,
+      "loss": 0.7487,
+      "step": 5478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814820527571068e-05,
+      "loss": 0.7765,
+      "step": 5479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813593969213166e-05,
+      "loss": 0.7809,
+      "step": 5480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812367219491907e-05,
+      "loss": 0.7535,
+      "step": 5481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6811140278441748e-05,
+      "loss": 0.7348,
+      "step": 5482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6809913146097153e-05,
+      "loss": 0.7224,
+      "step": 5483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808685822492586e-05,
+      "loss": 0.7531,
+      "step": 5484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6807458307662512e-05,
+      "loss": 0.801,
+      "step": 5485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680623060164141e-05,
+      "loss": 0.7538,
+      "step": 5486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805002704463765e-05,
+      "loss": 0.7006,
+      "step": 5487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803774616164054e-05,
+      "loss": 0.7449,
+      "step": 5488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802546336776772e-05,
+      "loss": 0.7014,
+      "step": 5489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6801317866336423e-05,
+      "loss": 0.7291,
+      "step": 5490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800089204877495e-05,
+      "loss": 0.8721,
+      "step": 5491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798860352434512e-05,
+      "loss": 0.832,
+      "step": 5492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797631309041975e-05,
+      "loss": 0.6477,
+      "step": 5493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796402074734404e-05,
+      "loss": 0.7332,
+      "step": 5494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6795172649546326e-05,
+      "loss": 0.7668,
+      "step": 5495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793943033512272e-05,
+      "loss": 0.7094,
+      "step": 5496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679271322666677e-05,
+      "loss": 0.3507,
+      "step": 5497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6791483229044363e-05,
+      "loss": 0.718,
+      "step": 5498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790253040679596e-05,
+      "loss": 0.7658,
+      "step": 5499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789022661607018e-05,
+      "loss": 0.7178,
+      "step": 5500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678779209186119e-05,
+      "loss": 0.7805,
+      "step": 5501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678656133147667e-05,
+      "loss": 0.7867,
+      "step": 5502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6785330380488022e-05,
+      "loss": 0.6744,
+      "step": 5503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784099238929822e-05,
+      "loss": 0.758,
+      "step": 5504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678286790683664e-05,
+      "loss": 0.6953,
+      "step": 5505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678163638424307e-05,
+      "loss": 0.7616,
+      "step": 5506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678040467118369e-05,
+      "loss": 0.6855,
+      "step": 5507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6779172767693096e-05,
+      "loss": 0.8303,
+      "step": 5508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777940673805894e-05,
+      "loss": 0.6761,
+      "step": 5509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776708389556676e-05,
+      "loss": 0.7613,
+      "step": 5510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677547591498006e-05,
+      "loss": 0.7017,
+      "step": 5511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774243250110656e-05,
+      "loss": 0.7415,
+      "step": 5512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773010394983085e-05,
+      "loss": 0.7043,
+      "step": 5513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677177734963198e-05,
+      "loss": 0.7712,
+      "step": 5514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677054411409196e-05,
+      "loss": 0.6787,
+      "step": 5515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769310688397665e-05,
+      "loss": 0.7695,
+      "step": 5516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676807707258374e-05,
+      "loss": 0.7454,
+      "step": 5517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766843266684828e-05,
+      "loss": 0.6929,
+      "step": 5518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765609270735585e-05,
+      "loss": 0.7949,
+      "step": 5519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764375084770657e-05,
+      "loss": 0.8068,
+      "step": 5520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763140708824724e-05,
+      "loss": 0.7594,
+      "step": 5521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761906142932445e-05,
+      "loss": 0.7356,
+      "step": 5522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676067138712849e-05,
+      "loss": 0.7752,
+      "step": 5523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6759436441447544e-05,
+      "loss": 0.7505,
+      "step": 5524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675820130592429e-05,
+      "loss": 0.7581,
+      "step": 5525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756965980593417e-05,
+      "loss": 0.7697,
+      "step": 5526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755730465489617e-05,
+      "loss": 0.749,
+      "step": 5527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675449476064759e-05,
+      "loss": 0.6966,
+      "step": 5528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6753258866102047e-05,
+      "loss": 0.7576,
+      "step": 5529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752022781887696e-05,
+      "loss": 0.79,
+      "step": 5530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750786508039246e-05,
+      "loss": 0.7051,
+      "step": 5531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749550044591432e-05,
+      "loss": 0.7726,
+      "step": 5532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6748313391578973e-05,
+      "loss": 0.6469,
+      "step": 5533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6747076549036597e-05,
+      "loss": 0.7576,
+      "step": 5534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674583951699905e-05,
+      "loss": 0.6288,
+      "step": 5535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674460229550107e-05,
+      "loss": 0.7343,
+      "step": 5536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743364884577403e-05,
+      "loss": 0.6969,
+      "step": 5537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742127284262804e-05,
+      "loss": 0.732,
+      "step": 5538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740889494592036e-05,
+      "loss": 0.7578,
+      "step": 5539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673965151559986e-05,
+      "loss": 0.7327,
+      "step": 5540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6738413347321043e-05,
+      "loss": 0.7363,
+      "step": 5541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737174989790364e-05,
+      "loss": 0.3192,
+      "step": 5542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735936443042598e-05,
+      "loss": 0.6775,
+      "step": 5543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673469770711253e-05,
+      "loss": 0.7835,
+      "step": 5544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6733458782034956e-05,
+      "loss": 0.772,
+      "step": 5545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673221966784467e-05,
+      "loss": 0.7715,
+      "step": 5546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730980364576474e-05,
+      "loss": 0.7908,
+      "step": 5547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6729740872265167e-05,
+      "loss": 0.7555,
+      "step": 5548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672850119094557e-05,
+      "loss": 0.7791,
+      "step": 5549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727261320652496e-05,
+      "loss": 0.7189,
+      "step": 5550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726021261420767e-05,
+      "loss": 0.7621,
+      "step": 5551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6724781013285215e-05,
+      "loss": 0.7592,
+      "step": 5552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723540576280668e-05,
+      "loss": 0.767,
+      "step": 5553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722299950441963e-05,
+      "loss": 0.7171,
+      "step": 5554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672105913580395e-05,
+      "loss": 0.7216,
+      "step": 5555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719818132401476e-05,
+      "loss": 0.7676,
+      "step": 5556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718576940269388e-05,
+      "loss": 0.7807,
+      "step": 5557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6717335559442557e-05,
+      "loss": 0.7469,
+      "step": 5558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671609398995584e-05,
+      "loss": 0.7324,
+      "step": 5559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671485223184411e-05,
+      "loss": 0.7337,
+      "step": 5560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671361028514224e-05,
+      "loss": 0.7886,
+      "step": 5561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6712368149885115e-05,
+      "loss": 0.7197,
+      "step": 5562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711125826107613e-05,
+      "loss": 0.6799,
+      "step": 5563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709883313844634e-05,
+      "loss": 0.7123,
+      "step": 5564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708640613131074e-05,
+      "loss": 0.7144,
+      "step": 5565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707397724001832e-05,
+      "loss": 0.7955,
+      "step": 5566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706154646491812e-05,
+      "loss": 0.695,
+      "step": 5567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704911380635932e-05,
+      "loss": 0.7393,
+      "step": 5568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670366792646911e-05,
+      "loss": 0.8397,
+      "step": 5569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702424284026265e-05,
+      "loss": 0.7869,
+      "step": 5570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701180453342324e-05,
+      "loss": 0.7801,
+      "step": 5571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669993643445223e-05,
+      "loss": 0.7421,
+      "step": 5572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669869222739091e-05,
+      "loss": 0.7779,
+      "step": 5573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697447832193315e-05,
+      "loss": 0.766,
+      "step": 5574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6696203248894397e-05,
+      "loss": 0.7423,
+      "step": 5575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694958477529102e-05,
+      "loss": 0.7854,
+      "step": 5576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693713518132396e-05,
+      "loss": 0.6375,
+      "step": 5577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669246837073925e-05,
+      "loss": 0.772,
+      "step": 5578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669122303538462e-05,
+      "loss": 0.7636,
+      "step": 5579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689977512103494e-05,
+      "loss": 0.8328,
+      "step": 5580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688731800930848e-05,
+      "loss": 0.7281,
+      "step": 5581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668748590190167e-05,
+      "loss": 0.7122,
+      "step": 5582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6686239815050955e-05,
+      "loss": 0.739,
+      "step": 5583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668499354041369e-05,
+      "loss": 0.7073,
+      "step": 5584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683747078024887e-05,
+      "loss": 0.727,
+      "step": 5585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668250042791955e-05,
+      "loss": 0.716,
+      "step": 5586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668125359013269e-05,
+      "loss": 0.7094,
+      "step": 5587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668000656469933e-05,
+      "loss": 0.7234,
+      "step": 5588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667875935165449e-05,
+      "loss": 0.6723,
+      "step": 5589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66775119510332e-05,
+      "loss": 0.7532,
+      "step": 5590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667626436287049e-05,
+      "loss": 0.799,
+      "step": 5591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6675016587201406e-05,
+      "loss": 0.7665,
+      "step": 5592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6673768624060986e-05,
+      "loss": 0.7912,
+      "step": 5593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672520473484284e-05,
+      "loss": 0.7906,
+      "step": 5594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671272135506353e-05,
+      "loss": 0.6912,
+      "step": 5595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667002361016225e-05,
+      "loss": 0.769,
+      "step": 5596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6668774897487052e-05,
+      "loss": 0.7573,
+      "step": 5597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6667525997515815e-05,
+      "loss": 0.7999,
+      "step": 5598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6666276910283623e-05,
+      "loss": 0.7281,
+      "step": 5599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665027635825557e-05,
+      "loss": 0.7416,
+      "step": 5600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663778174176704e-05,
+      "loss": 0.7315,
+      "step": 5601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6662528525372153e-05,
+      "loss": 0.7225,
+      "step": 5602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661278689447e-05,
+      "loss": 0.7704,
+      "step": 5603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660028666436354e-05,
+      "loss": 0.7399,
+      "step": 5604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658778456375315e-05,
+      "loss": 0.7683,
+      "step": 5605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657528059299e-05,
+      "loss": 0.7281,
+      "step": 5606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656277475242525e-05,
+      "loss": 0.6987,
+      "step": 5607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655026704241017e-05,
+      "loss": 0.774,
+      "step": 5608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653775746329598e-05,
+      "loss": 0.7372,
+      "step": 5609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665252460154341e-05,
+      "loss": 0.7389,
+      "step": 5610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651273269917586e-05,
+      "loss": 0.7678,
+      "step": 5611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665002175148727e-05,
+      "loss": 0.6862,
+      "step": 5612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648770046287617e-05,
+      "loss": 0.7446,
+      "step": 5613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664751815435378e-05,
+      "loss": 0.8595,
+      "step": 5614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646266075720914e-05,
+      "loss": 0.7531,
+      "step": 5615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6645013810424186e-05,
+      "loss": 0.7518,
+      "step": 5616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643761358498775e-05,
+      "loss": 0.6205,
+      "step": 5617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642508719979845e-05,
+      "loss": 0.7636,
+      "step": 5618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664125589490258e-05,
+      "loss": 0.6535,
+      "step": 5619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640002883302172e-05,
+      "loss": 0.7422,
+      "step": 5620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663874968521381e-05,
+      "loss": 0.8036,
+      "step": 5621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637496300672688e-05,
+      "loss": 0.7144,
+      "step": 5622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6636242729714012e-05,
+      "loss": 0.7811,
+      "step": 5623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663498897237298e-05,
+      "loss": 0.6698,
+      "step": 5624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633735028684817e-05,
+      "loss": 0.7668,
+      "step": 5625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6632480898684734e-05,
+      "loss": 0.7408,
+      "step": 5626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631226582407954e-05,
+      "loss": 0.3156,
+      "step": 5627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629972079889703e-05,
+      "loss": 0.7693,
+      "step": 5628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662871739116522e-05,
+      "loss": 0.7873,
+      "step": 5629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627462516269745e-05,
+      "loss": 0.7066,
+      "step": 5630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662620745523851e-05,
+      "loss": 0.6999,
+      "step": 5631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624952208106776e-05,
+      "loss": 0.7501,
+      "step": 5632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623696774909793e-05,
+      "loss": 0.6919,
+      "step": 5633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662244115568282e-05,
+      "loss": 0.7172,
+      "step": 5634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662118535046112e-05,
+      "loss": 0.8432,
+      "step": 5635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619929359279967e-05,
+      "loss": 0.7919,
+      "step": 5636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6618673182174635e-05,
+      "loss": 0.7891,
+      "step": 5637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66174168191804e-05,
+      "loss": 0.7283,
+      "step": 5638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616160270332553e-05,
+      "loss": 0.7284,
+      "step": 5639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6614903535666388e-05,
+      "loss": 0.6916,
+      "step": 5640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661364661521719e-05,
+      "loss": 0.7855,
+      "step": 5641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661238950902027e-05,
+      "loss": 0.7195,
+      "step": 5642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611132217110928e-05,
+      "loss": 0.7383,
+      "step": 5643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660987473952448e-05,
+      "loss": 0.6997,
+      "step": 5644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660861707629624e-05,
+      "loss": 0.6717,
+      "step": 5645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607359227461534e-05,
+      "loss": 0.7806,
+      "step": 5646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606101193055684e-05,
+      "loss": 0.7191,
+      "step": 5647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660484297311403e-05,
+      "loss": 0.7399,
+      "step": 5648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603584567671904e-05,
+      "loss": 0.6665,
+      "step": 5649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660232597676465e-05,
+      "loss": 0.7254,
+      "step": 5650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6601067200427615e-05,
+      "loss": 0.6895,
+      "step": 5651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599808238696153e-05,
+      "loss": 0.754,
+      "step": 5652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659854909160563e-05,
+      "loss": 0.7502,
+      "step": 5653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597289759191393e-05,
+      "loss": 0.717,
+      "step": 5654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596030241488826e-05,
+      "loss": 0.7772,
+      "step": 5655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594770538533303e-05,
+      "loss": 0.6734,
+      "step": 5656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593510650360195e-05,
+      "loss": 0.7072,
+      "step": 5657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659225057700489e-05,
+      "loss": 0.7877,
+      "step": 5658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659099031850278e-05,
+      "loss": 0.7789,
+      "step": 5659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589729874889253e-05,
+      "loss": 0.7258,
+      "step": 5660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658846924619972e-05,
+      "loss": 0.6916,
+      "step": 5661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658720843246958e-05,
+      "loss": 0.7481,
+      "step": 5662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585947433734247e-05,
+      "loss": 0.6797,
+      "step": 5663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584686250029126e-05,
+      "loss": 0.8056,
+      "step": 5664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583424881389657e-05,
+      "loss": 0.6859,
+      "step": 5665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658216332785125e-05,
+      "loss": 0.7393,
+      "step": 5666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580901589449342e-05,
+      "loss": 0.6966,
+      "step": 5667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579639666219368e-05,
+      "loss": 0.7565,
+      "step": 5668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657837755819678e-05,
+      "loss": 0.744,
+      "step": 5669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6577115265417007e-05,
+      "loss": 0.7217,
+      "step": 5670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6575852787915513e-05,
+      "loss": 0.775,
+      "step": 5671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574590125727755e-05,
+      "loss": 0.7538,
+      "step": 5672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573327278889193e-05,
+      "loss": 0.7747,
+      "step": 5673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6572064247435293e-05,
+      "loss": 0.6631,
+      "step": 5674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657080103140153e-05,
+      "loss": 0.7264,
+      "step": 5675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569537630823385e-05,
+      "loss": 0.7262,
+      "step": 5676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568274045736334e-05,
+      "loss": 0.7282,
+      "step": 5677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656701027617587e-05,
+      "loss": 0.7009,
+      "step": 5678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656574632217749e-05,
+      "loss": 0.7591,
+      "step": 5679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564482183776688e-05,
+      "loss": 0.709,
+      "step": 5680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656321786100897e-05,
+      "loss": 0.7786,
+      "step": 5681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561953353909843e-05,
+      "loss": 0.6829,
+      "step": 5682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560688662514822e-05,
+      "loss": 0.6855,
+      "step": 5683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559423786859427e-05,
+      "loss": 0.7502,
+      "step": 5684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655815872697919e-05,
+      "loss": 0.73,
+      "step": 5685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6556893482909626e-05,
+      "loss": 0.3556,
+      "step": 5686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555628054686278e-05,
+      "loss": 0.6632,
+      "step": 5687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655436244234469e-05,
+      "loss": 0.708,
+      "step": 5688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65530966459204e-05,
+      "loss": 0.7391,
+      "step": 5689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6551830665448963e-05,
+      "loss": 0.7279,
+      "step": 5690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6550564500965933e-05,
+      "loss": 0.7364,
+      "step": 5691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6549298152506875e-05,
+      "loss": 0.7734,
+      "step": 5692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654803162010735e-05,
+      "loss": 0.7602,
+      "step": 5693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654676490380293e-05,
+      "loss": 0.7019,
+      "step": 5694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654549800362919e-05,
+      "loss": 0.7351,
+      "step": 5695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544230919621718e-05,
+      "loss": 0.6967,
+      "step": 5696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542963651816094e-05,
+      "loss": 0.7254,
+      "step": 5697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6541696200247913e-05,
+      "loss": 0.7902,
+      "step": 5698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654042856495277e-05,
+      "loss": 0.7316,
+      "step": 5699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6539160745966273e-05,
+      "loss": 0.6802,
+      "step": 5700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.653789274332402e-05,
+      "loss": 0.7723,
+      "step": 5701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536624557061632e-05,
+      "loss": 0.7374,
+      "step": 5702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6535356187214723e-05,
+      "loss": 0.75,
+      "step": 5703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534087633818914e-05,
+      "loss": 0.7332,
+      "step": 5704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532818896909836e-05,
+      "loss": 0.7688,
+      "step": 5705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6531549976523123e-05,
+      "loss": 0.6867,
+      "step": 5706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530280872694405e-05,
+      "loss": 0.8022,
+      "step": 5707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652901158545934e-05,
+      "loss": 0.7424,
+      "step": 5708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527742114853564e-05,
+      "loss": 0.7455,
+      "step": 5709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6526472460912736e-05,
+      "loss": 0.7166,
+      "step": 5710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525202623672515e-05,
+      "loss": 0.7676,
+      "step": 5711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6523932603168567e-05,
+      "loss": 0.7316,
+      "step": 5712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522662399436553e-05,
+      "loss": 0.7099,
+      "step": 5713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652139201251216e-05,
+      "loss": 0.7465,
+      "step": 5714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520121442431053e-05,
+      "loss": 0.6985,
+      "step": 5715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6518850689228926e-05,
+      "loss": 0.7407,
+      "step": 5716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6517579752941463e-05,
+      "loss": 0.7556,
+      "step": 5717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651630863360437e-05,
+      "loss": 0.7035,
+      "step": 5718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515037331253333e-05,
+      "loss": 0.7366,
+      "step": 5719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513765845924065e-05,
+      "loss": 0.7336,
+      "step": 5720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6512494177652277e-05,
+      "loss": 0.6068,
+      "step": 5721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511222326473683e-05,
+      "loss": 0.8529,
+      "step": 5722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509950292424003e-05,
+      "loss": 0.7499,
+      "step": 5723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650867807553896e-05,
+      "loss": 0.7593,
+      "step": 5724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507405675854295e-05,
+      "loss": 0.78,
+      "step": 5725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650613309340573e-05,
+      "loss": 0.7854,
+      "step": 5726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504860328229013e-05,
+      "loss": 0.3888,
+      "step": 5727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503587380359894e-05,
+      "loss": 0.7849,
+      "step": 5728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502314249834117e-05,
+      "loss": 0.725,
+      "step": 5729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6501040936687444e-05,
+      "loss": 0.7831,
+      "step": 5730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499767440955636e-05,
+      "loss": 0.6831,
+      "step": 5731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6498493762674455e-05,
+      "loss": 0.7266,
+      "step": 5732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497219901879676e-05,
+      "loss": 0.7443,
+      "step": 5733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495945858607082e-05,
+      "loss": 0.7723,
+      "step": 5734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494671632892448e-05,
+      "loss": 0.7501,
+      "step": 5735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493397224771562e-05,
+      "loss": 0.3375,
+      "step": 5736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6492122634280217e-05,
+      "loss": 0.7358,
+      "step": 5737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490847861454212e-05,
+      "loss": 0.7449,
+      "step": 5738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489572906329345e-05,
+      "loss": 0.7112,
+      "step": 5739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6488297768941434e-05,
+      "loss": 0.7371,
+      "step": 5740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487022449326282e-05,
+      "loss": 0.7603,
+      "step": 5741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485746947519708e-05,
+      "loss": 0.8265,
+      "step": 5742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648447126355754e-05,
+      "loss": 0.6761,
+      "step": 5743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6483195397475606e-05,
+      "loss": 0.6873,
+      "step": 5744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481919349309734e-05,
+      "loss": 0.75,
+      "step": 5745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480643119095763e-05,
+      "loss": 0.7391,
+      "step": 5746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6479366706869547e-05,
+      "loss": 0.7052,
+      "step": 5747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647809011266692e-05,
+      "loss": 0.7718,
+      "step": 5748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6476813336523744e-05,
+      "loss": 0.763,
+      "step": 5749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647553637847588e-05,
+      "loss": 0.7265,
+      "step": 5750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6474259238559186e-05,
+      "loss": 0.7665,
+      "step": 5751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647298191680953e-05,
+      "loss": 0.7933,
+      "step": 5752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6471704413262795e-05,
+      "loss": 0.7077,
+      "step": 5753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647042672795486e-05,
+      "loss": 0.7566,
+      "step": 5754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469148860921597e-05,
+      "loss": 0.85,
+      "step": 5755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467870812198907e-05,
+      "loss": 0.7356,
+      "step": 5756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646659258182268e-05,
+      "loss": 0.7164,
+      "step": 5757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465314169828816e-05,
+      "loss": 0.6645,
+      "step": 5758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646403557625322e-05,
+      "loss": 0.7571,
+      "step": 5759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646275680113181e-05,
+      "loss": 0.7178,
+      "step": 5760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461477844500487e-05,
+      "loss": 0.7828,
+      "step": 5761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460198706395178e-05,
+      "loss": 0.817,
+      "step": 5762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645891938685181e-05,
+      "loss": 0.7448,
+      "step": 5763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457639885906308e-05,
+      "loss": 0.7199,
+      "step": 5764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456360203594616e-05,
+      "loss": 0.7505,
+      "step": 5765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6455080339952668e-05,
+      "loss": 0.7163,
+      "step": 5766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645380029501641e-05,
+      "loss": 0.6931,
+      "step": 5767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6452520068821796e-05,
+      "loss": 0.7649,
+      "step": 5768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645123966140478e-05,
+      "loss": 0.7774,
+      "step": 5769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644995907280132e-05,
+      "loss": 0.7341,
+      "step": 5770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448678303047386e-05,
+      "loss": 0.8094,
+      "step": 5771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447397352178945e-05,
+      "loss": 0.7066,
+      "step": 5772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446116220231984e-05,
+      "loss": 0.8232,
+      "step": 5773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644483490724247e-05,
+      "loss": 0.787,
+      "step": 5774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6443553413246393e-05,
+      "loss": 0.6935,
+      "step": 5775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442271738279752e-05,
+      "loss": 0.6873,
+      "step": 5776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644098988237854e-05,
+      "loss": 0.7377,
+      "step": 5777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643970784557875e-05,
+      "loss": 0.7634,
+      "step": 5778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64384256279164e-05,
+      "loss": 0.7294,
+      "step": 5779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64371432294275e-05,
+      "loss": 0.6767,
+      "step": 5780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643586065014806e-05,
+      "loss": 0.7618,
+      "step": 5781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643457789011411e-05,
+      "loss": 0.7309,
+      "step": 5782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6433294949361674e-05,
+      "loss": 0.8085,
+      "step": 5783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643201182792678e-05,
+      "loss": 0.7571,
+      "step": 5784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643072852584547e-05,
+      "loss": 0.7418,
+      "step": 5785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642944504315378e-05,
+      "loss": 0.7994,
+      "step": 5786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428161379887768e-05,
+      "loss": 0.7156,
+      "step": 5787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6426877536083477e-05,
+      "loss": 0.781,
+      "step": 5788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642559351177697e-05,
+      "loss": 0.7469,
+      "step": 5789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6424309307004306e-05,
+      "loss": 0.7498,
+      "step": 5790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6423024921801553e-05,
+      "loss": 0.7894,
+      "step": 5791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421740356204786e-05,
+      "loss": 0.693,
+      "step": 5792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420455610250077e-05,
+      "loss": 0.7435,
+      "step": 5793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419170683973516e-05,
+      "loss": 0.7081,
+      "step": 5794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641788557741119e-05,
+      "loss": 0.6933,
+      "step": 5795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6416600290599187e-05,
+      "loss": 0.7271,
+      "step": 5796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415314823573608e-05,
+      "loss": 0.6717,
+      "step": 5797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6414029176370557e-05,
+      "loss": 0.7695,
+      "step": 5798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641274334902614e-05,
+      "loss": 0.7368,
+      "step": 5799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411457341576467e-05,
+      "loss": 0.7094,
+      "step": 5800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6410171154057663e-05,
+      "loss": 0.7144,
+      "step": 5801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640888478650585e-05,
+      "loss": 0.7279,
+      "step": 5802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407598238957152e-05,
+      "loss": 0.7549,
+      "step": 5803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406311511447705e-05,
+      "loss": 0.7949,
+      "step": 5804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405024604013647e-05,
+      "loss": 0.7701,
+      "step": 5805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403737516691122e-05,
+      "loss": 0.7585,
+      "step": 5806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6402450249516283e-05,
+      "loss": 0.7358,
+      "step": 5807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6401162802525273e-05,
+      "loss": 0.7251,
+      "step": 5808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399875175754258e-05,
+      "loss": 0.7546,
+      "step": 5809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63985873692394e-05,
+      "loss": 0.8209,
+      "step": 5810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397299383016867e-05,
+      "loss": 0.7319,
+      "step": 5811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639601121712284e-05,
+      "loss": 0.7508,
+      "step": 5812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639472287159348e-05,
+      "loss": 0.7461,
+      "step": 5813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639343434646499e-05,
+      "loss": 0.797,
+      "step": 5814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392145641773552e-05,
+      "loss": 0.7458,
+      "step": 5815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390856757555357e-05,
+      "loss": 0.7261,
+      "step": 5816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6389567693846605e-05,
+      "loss": 0.618,
+      "step": 5817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63882784506835e-05,
+      "loss": 0.7812,
+      "step": 5818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386989028102257e-05,
+      "loss": 0.7021,
+      "step": 5819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385699426139083e-05,
+      "loss": 0.5955,
+      "step": 5820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63844096448302e-05,
+      "loss": 0.7042,
+      "step": 5821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638311968421183e-05,
+      "loss": 0.8051,
+      "step": 5822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381829544320203e-05,
+      "loss": 0.8199,
+      "step": 5823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6380539225191556e-05,
+      "loss": 0.7791,
+      "step": 5824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637924872686213e-05,
+      "loss": 0.7184,
+      "step": 5825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637795804936816e-05,
+      "loss": 0.7281,
+      "step": 5826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376667192745903e-05,
+      "loss": 0.7354,
+      "step": 5827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375376157031614e-05,
+      "loss": 0.6893,
+      "step": 5828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6374084942261547e-05,
+      "loss": 0.7258,
+      "step": 5829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637279354847197e-05,
+      "loss": 0.796,
+      "step": 5830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637150197569915e-05,
+      "loss": 0.7102,
+      "step": 5831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370210223979366e-05,
+      "loss": 0.3466,
+      "step": 5832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6368918293348893e-05,
+      "loss": 0.7061,
+      "step": 5833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636762618384402e-05,
+      "loss": 0.6997,
+      "step": 5834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366333895501033e-05,
+      "loss": 0.7534,
+      "step": 5835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6365041428356226e-05,
+      "loss": 0.7672,
+      "step": 5836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63637487824459e-05,
+      "loss": 0.6934,
+      "step": 5837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6362455957806362e-05,
+      "loss": 0.8057,
+      "step": 5838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6361162954473918e-05,
+      "loss": 0.7248,
+      "step": 5839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6359869772484884e-05,
+      "loss": 0.7991,
+      "step": 5840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635857641187558e-05,
+      "loss": 0.7459,
+      "step": 5841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357282872682332e-05,
+      "loss": 0.726,
+      "step": 5842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355989154941466e-05,
+      "loss": 0.8156,
+      "step": 5843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635469525868932e-05,
+      "loss": 0.6884,
+      "step": 5844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353401183962235e-05,
+      "loss": 0.7178,
+      "step": 5845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635210693079655e-05,
+      "loss": 0.7778,
+      "step": 5846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350812499228618e-05,
+      "loss": 0.7791,
+      "step": 5847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349517889294795e-05,
+      "loss": 0.6647,
+      "step": 5848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348223101031444e-05,
+      "loss": 0.7824,
+      "step": 5849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634692813447492e-05,
+      "loss": 0.7534,
+      "step": 5850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345632989661606e-05,
+      "loss": 0.3605,
+      "step": 5851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344337666627862e-05,
+      "loss": 0.7723,
+      "step": 5852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343042165410075e-05,
+      "loss": 0.6811,
+      "step": 5853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6341746486044634e-05,
+      "loss": 0.7556,
+      "step": 5854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6340450628567924e-05,
+      "loss": 0.7747,
+      "step": 5855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339154593016337e-05,
+      "loss": 0.7104,
+      "step": 5856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337858379426278e-05,
+      "loss": 0.7325,
+      "step": 5857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336561987834155e-05,
+      "loss": 0.7616,
+      "step": 5858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633526541827637e-05,
+      "loss": 0.6964,
+      "step": 5859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633396867078934e-05,
+      "loss": 0.7471,
+      "step": 5860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6332671745409486e-05,
+      "loss": 0.6754,
+      "step": 5861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633137464217323e-05,
+      "loss": 0.838,
+      "step": 5862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633007736111701e-05,
+      "loss": 0.8105,
+      "step": 5863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6328779902277254e-05,
+      "loss": 0.8195,
+      "step": 5864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63274822656904e-05,
+      "loss": 0.5903,
+      "step": 5865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63261844513929e-05,
+      "loss": 0.8142,
+      "step": 5866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63248864594212e-05,
+      "loss": 0.7772,
+      "step": 5867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6323588289811748e-05,
+      "loss": 0.7227,
+      "step": 5868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322289942601016e-05,
+      "loss": 0.8859,
+      "step": 5869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632099141782546e-05,
+      "loss": 0.774,
+      "step": 5870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6319692715521552e-05,
+      "loss": 0.7568,
+      "step": 5871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318393835725773e-05,
+      "loss": 0.6813,
+      "step": 5872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6317094778474592e-05,
+      "loss": 0.7401,
+      "step": 5873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63157955438045e-05,
+      "loss": 0.761,
+      "step": 5874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314496131751987e-05,
+      "loss": 0.7312,
+      "step": 5875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6313196542353547e-05,
+      "loss": 0.7342,
+      "step": 5876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311896775645676e-05,
+      "loss": 0.6664,
+      "step": 5877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6310596831664885e-05,
+      "loss": 0.7097,
+      "step": 5878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309296710447674e-05,
+      "loss": 0.7199,
+      "step": 5879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630799641203057e-05,
+      "loss": 0.6959,
+      "step": 5880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306695936450086e-05,
+      "loss": 0.7782,
+      "step": 5881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305395283742743e-05,
+      "loss": 0.7543,
+      "step": 5882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304094453945077e-05,
+      "loss": 0.7336,
+      "step": 5883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630279344709362e-05,
+      "loss": 0.6957,
+      "step": 5884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630149226322491e-05,
+      "loss": 0.7441,
+      "step": 5885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300190902375494e-05,
+      "loss": 0.7374,
+      "step": 5886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298889364581917e-05,
+      "loss": 0.7275,
+      "step": 5887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297587649880742e-05,
+      "loss": 0.7426,
+      "step": 5888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629628575830852e-05,
+      "loss": 0.7508,
+      "step": 5889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629498368990182e-05,
+      "loss": 0.6984,
+      "step": 5890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.62936814446972e-05,
+      "loss": 0.7329,
+      "step": 5891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6292379022731257e-05,
+      "loss": 0.7302,
+      "step": 5892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291076424040548e-05,
+      "loss": 0.6965,
+      "step": 5893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6289773648661667e-05,
+      "loss": 0.7169,
+      "step": 5894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288470696631205e-05,
+      "loss": 0.655,
+      "step": 5895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6287167567985754e-05,
+      "loss": 0.6884,
+      "step": 5896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628586426276191e-05,
+      "loss": 0.6926,
+      "step": 5897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628456078099628e-05,
+      "loss": 0.7419,
+      "step": 5898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628325712272547e-05,
+      "loss": 0.7303,
+      "step": 5899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6281953287986102e-05,
+      "loss": 0.7324,
+      "step": 5900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628064927681478e-05,
+      "loss": 0.754,
+      "step": 5901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279345089248143e-05,
+      "loss": 0.7654,
+      "step": 5902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278040725322815e-05,
+      "loss": 0.7266,
+      "step": 5903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6276736185075426e-05,
+      "loss": 0.7762,
+      "step": 5904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275431468542616e-05,
+      "loss": 0.723,
+      "step": 5905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274126575761032e-05,
+      "loss": 0.7403,
+      "step": 5906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627282150676732e-05,
+      "loss": 0.689,
+      "step": 5907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271516261598132e-05,
+      "loss": 0.7213,
+      "step": 5908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627021084029013e-05,
+      "loss": 0.7922,
+      "step": 5909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268905242879975e-05,
+      "loss": 0.733,
+      "step": 5910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626759946940434e-05,
+      "loss": 0.7249,
+      "step": 5911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266293519899886e-05,
+      "loss": 0.7137,
+      "step": 5912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626498739440331e-05,
+      "loss": 0.783,
+      "step": 5913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626368109295128e-05,
+      "loss": 0.7697,
+      "step": 5914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626237461558049e-05,
+      "loss": 0.7225,
+      "step": 5915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626106796232763e-05,
+      "loss": 0.734,
+      "step": 5916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259761133229402e-05,
+      "loss": 0.7424,
+      "step": 5917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258454128322508e-05,
+      "loss": 0.7116,
+      "step": 5918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6257146947643656e-05,
+      "loss": 0.7633,
+      "step": 5919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6255839591229557e-05,
+      "loss": 0.7828,
+      "step": 5920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625453205911693e-05,
+      "loss": 0.7794,
+      "step": 5921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6253224351342495e-05,
+      "loss": 0.6595,
+      "step": 5922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625191646794299e-05,
+      "loss": 0.7874,
+      "step": 5923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250608408955132e-05,
+      "loss": 0.7006,
+      "step": 5924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249300174415668e-05,
+      "loss": 0.6803,
+      "step": 5925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624799176436134e-05,
+      "loss": 0.6988,
+      "step": 5926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246683178828894e-05,
+      "loss": 0.7501,
+      "step": 5927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245374417855082e-05,
+      "loss": 0.7648,
+      "step": 5928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624406548147666e-05,
+      "loss": 0.7941,
+      "step": 5929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242756369730396e-05,
+      "loss": 0.7104,
+      "step": 5930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241447082653053e-05,
+      "loss": 0.7259,
+      "step": 5931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62401376202814e-05,
+      "loss": 0.7265,
+      "step": 5932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623882798265222e-05,
+      "loss": 0.6921,
+      "step": 5933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623751816980229e-05,
+      "loss": 0.7574,
+      "step": 5934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236208181768396e-05,
+      "loss": 0.7668,
+      "step": 5935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.774,
+      "step": 5936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6233587680295905e-05,
+      "loss": 0.7861,
+      "step": 5937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232277166930896e-05,
+      "loss": 0.8163,
+      "step": 5938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230966478529127e-05,
+      "loss": 0.7565,
+      "step": 5939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6229655615127403e-05,
+      "loss": 0.7531,
+      "step": 5940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622834457676254e-05,
+      "loss": 0.8197,
+      "step": 5941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6227033363471364e-05,
+      "loss": 0.7812,
+      "step": 5942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225721975290697e-05,
+      "loss": 0.7958,
+      "step": 5943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224410412257367e-05,
+      "loss": 0.7268,
+      "step": 5944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6223098674408214e-05,
+      "loss": 0.7273,
+      "step": 5945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221786761780083e-05,
+      "loss": 0.7715,
+      "step": 5946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220474674409815e-05,
+      "loss": 0.6875,
+      "step": 5947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6219162412334258e-05,
+      "loss": 0.735,
+      "step": 5948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217849975590275e-05,
+      "loss": 0.7743,
+      "step": 5949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621653736421472e-05,
+      "loss": 0.6774,
+      "step": 5950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215224578244456e-05,
+      "loss": 0.7415,
+      "step": 5951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213911617716365e-05,
+      "loss": 0.7418,
+      "step": 5952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621259848266731e-05,
+      "loss": 0.7081,
+      "step": 5953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6211285173134176e-05,
+      "loss": 0.7126,
+      "step": 5954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6209971689153852e-05,
+      "loss": 0.6848,
+      "step": 5955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620865803076322e-05,
+      "loss": 0.7459,
+      "step": 5956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207344197999183e-05,
+      "loss": 0.6887,
+      "step": 5957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6206030190898633e-05,
+      "loss": 0.6644,
+      "step": 5958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620471600949848e-05,
+      "loss": 0.7552,
+      "step": 5959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203401653835632e-05,
+      "loss": 0.7661,
+      "step": 5960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202087123947002e-05,
+      "loss": 0.728,
+      "step": 5961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200772419869508e-05,
+      "loss": 0.7719,
+      "step": 5962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619945754164008e-05,
+      "loss": 0.7069,
+      "step": 5963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198142489295638e-05,
+      "loss": 0.7477,
+      "step": 5964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196827262873122e-05,
+      "loss": 0.7183,
+      "step": 5965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6195511862409473e-05,
+      "loss": 0.7552,
+      "step": 5966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619419628794163e-05,
+      "loss": 0.6946,
+      "step": 5967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6192880539506547e-05,
+      "loss": 0.7556,
+      "step": 5968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619156461714117e-05,
+      "loss": 0.7292,
+      "step": 5969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619024852088246e-05,
+      "loss": 0.7928,
+      "step": 5970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188932250767385e-05,
+      "loss": 0.7101,
+      "step": 5971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187615806832904e-05,
+      "loss": 0.7064,
+      "step": 5972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186299189116e-05,
+      "loss": 0.7554,
+      "step": 5973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184982397653646e-05,
+      "loss": 0.8149,
+      "step": 5974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618366543248283e-05,
+      "loss": 0.7256,
+      "step": 5975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618234829364053e-05,
+      "loss": 0.7667,
+      "step": 5976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618103098116374e-05,
+      "loss": 0.7115,
+      "step": 5977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179713495089466e-05,
+      "loss": 0.7762,
+      "step": 5978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178395835454702e-05,
+      "loss": 0.6964,
+      "step": 5979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617707800229646e-05,
+      "loss": 0.6414,
+      "step": 5980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175759995651752e-05,
+      "loss": 0.7393,
+      "step": 5981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174441815557594e-05,
+      "loss": 0.7795,
+      "step": 5982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6173123462051003e-05,
+      "loss": 0.6778,
+      "step": 5983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617180493516901e-05,
+      "loss": 0.7474,
+      "step": 5984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617048623494865e-05,
+      "loss": 0.8083,
+      "step": 5985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616916736142695e-05,
+      "loss": 0.7371,
+      "step": 5986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616784831464096e-05,
+      "loss": 0.7177,
+      "step": 5987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6166529094627727e-05,
+      "loss": 0.7801,
+      "step": 5988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616520970142429e-05,
+      "loss": 0.7605,
+      "step": 5989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163890135067717e-05,
+      "loss": 0.7235,
+      "step": 5990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6162570395595064e-05,
+      "loss": 0.7979,
+      "step": 5991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161250483043395e-05,
+      "loss": 0.7542,
+      "step": 5992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159930397449786e-05,
+      "loss": 0.7614,
+      "step": 5993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6158610138851308e-05,
+      "loss": 0.6954,
+      "step": 5994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615728970728504e-05,
+      "loss": 0.6795,
+      "step": 5995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155969102788073e-05,
+      "loss": 0.7346,
+      "step": 5996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154648325397486e-05,
+      "loss": 0.7098,
+      "step": 5997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153327375150385e-05,
+      "loss": 0.7272,
+      "step": 5998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152006252083866e-05,
+      "loss": 0.6821,
+      "step": 5999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615068495623503e-05,
+      "loss": 0.6931,
+      "step": 6000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614936348764099e-05,
+      "loss": 0.6667,
+      "step": 6001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614804184633886e-05,
+      "loss": 0.7014,
+      "step": 6002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146720032365756e-05,
+      "loss": 0.6884,
+      "step": 6003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145398045758804e-05,
+      "loss": 0.799,
+      "step": 6004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144075886555134e-05,
+      "loss": 0.7474,
+      "step": 6005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6142753554791878e-05,
+      "loss": 0.7842,
+      "step": 6006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141431050506175e-05,
+      "loss": 0.6516,
+      "step": 6007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6140108373735165e-05,
+      "loss": 0.7083,
+      "step": 6008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6138785524516e-05,
+      "loss": 0.794,
+      "step": 6009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613746250288583e-05,
+      "loss": 0.8013,
+      "step": 6010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6136139308881816e-05,
+      "loss": 0.743,
+      "step": 6011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613481594254112e-05,
+      "loss": 0.8206,
+      "step": 6012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133492403900907e-05,
+      "loss": 0.7445,
+      "step": 6013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6132168692998353e-05,
+      "loss": 0.7083,
+      "step": 6014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613084480987063e-05,
+      "loss": 0.7383,
+      "step": 6015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612952075455492e-05,
+      "loss": 0.7284,
+      "step": 6016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128196527088418e-05,
+      "loss": 0.7618,
+      "step": 6017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126872127508308e-05,
+      "loss": 0.7721,
+      "step": 6018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6125547555851787e-05,
+      "loss": 0.7582,
+      "step": 6019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612422281215606e-05,
+      "loss": 0.7455,
+      "step": 6020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122897896458332e-05,
+      "loss": 0.7069,
+      "step": 6021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612157280879581e-05,
+      "loss": 0.6812,
+      "step": 6022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120247549205714e-05,
+      "loss": 0.7006,
+      "step": 6023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118922117725266e-05,
+      "loss": 0.7358,
+      "step": 6024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117596514391686e-05,
+      "loss": 0.7715,
+      "step": 6025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611627073924221e-05,
+      "loss": 0.6801,
+      "step": 6026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114944792314067e-05,
+      "loss": 0.7782,
+      "step": 6027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6113618673644497e-05,
+      "loss": 0.7083,
+      "step": 6028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6112292383270753e-05,
+      "loss": 0.7848,
+      "step": 6029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611096592123008e-05,
+      "loss": 0.6892,
+      "step": 6030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610963928755973e-05,
+      "loss": 0.7076,
+      "step": 6031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610831248229696e-05,
+      "loss": 0.7171,
+      "step": 6032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106985505479043e-05,
+      "loss": 0.6432,
+      "step": 6033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105658357143242e-05,
+      "loss": 0.7593,
+      "step": 6034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610433103732683e-05,
+      "loss": 0.8064,
+      "step": 6035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6103003546067087e-05,
+      "loss": 0.7314,
+      "step": 6036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101675883401296e-05,
+      "loss": 0.7195,
+      "step": 6037
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6100348049366744e-05,
+      "loss": 0.7077,
+      "step": 6038
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609902004400073e-05,
+      "loss": 0.7283,
+      "step": 6039
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097691867340547e-05,
+      "loss": 0.8326,
+      "step": 6040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096363519423493e-05,
+      "loss": 0.7767,
+      "step": 6041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609503500028688e-05,
+      "loss": 0.7546,
+      "step": 6042
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093706309968022e-05,
+      "loss": 0.7033,
+      "step": 6043
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092377448504233e-05,
+      "loss": 0.7511,
+      "step": 6044
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6091048415932835e-05,
+      "loss": 0.7236,
+      "step": 6045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6089719212291158e-05,
+      "loss": 0.854,
+      "step": 6046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6088389837616526e-05,
+      "loss": 0.826,
+      "step": 6047
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087060291946285e-05,
+      "loss": 0.8188,
+      "step": 6048
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085730575317767e-05,
+      "loss": 0.7289,
+      "step": 6049
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084400687768325e-05,
+      "loss": 0.7426,
+      "step": 6050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083070629335305e-05,
+      "loss": 0.3854,
+      "step": 6051
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608174040005606e-05,
+      "loss": 0.8038,
+      "step": 6052
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6080409999967962e-05,
+      "loss": 0.7769,
+      "step": 6053
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079079429108357e-05,
+      "loss": 0.7816,
+      "step": 6054
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077748687514632e-05,
+      "loss": 0.6956,
+      "step": 6055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6076417775224154e-05,
+      "loss": 0.6615,
+      "step": 6056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075086692274305e-05,
+      "loss": 0.7584,
+      "step": 6057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6073755438702465e-05,
+      "loss": 0.764,
+      "step": 6058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607242401454603e-05,
+      "loss": 0.7559,
+      "step": 6059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071092419842384e-05,
+      "loss": 0.7337,
+      "step": 6060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069760654628933e-05,
+      "loss": 0.7199,
+      "step": 6061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606842871894308e-05,
+      "loss": 0.7549,
+      "step": 6062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067096612822228e-05,
+      "loss": 0.7377,
+      "step": 6063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065764336303796e-05,
+      "loss": 0.723,
+      "step": 6064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064431889425198e-05,
+      "loss": 0.7665,
+      "step": 6065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6063099272223856e-05,
+      "loss": 0.6806,
+      "step": 6066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60617664847372e-05,
+      "loss": 0.7162,
+      "step": 6067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6060433527002663e-05,
+      "loss": 0.7492,
+      "step": 6068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059100399057676e-05,
+      "loss": 0.7697,
+      "step": 6069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057767100939683e-05,
+      "loss": 0.7309,
+      "step": 6070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605643363268613e-05,
+      "loss": 0.6944,
+      "step": 6071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6055099994334476e-05,
+      "loss": 0.726,
+      "step": 6072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605376618592217e-05,
+      "loss": 0.6881,
+      "step": 6073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6052432207486666e-05,
+      "loss": 0.7287,
+      "step": 6074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605109805906544e-05,
+      "loss": 0.7214,
+      "step": 6075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6049763740695966e-05,
+      "loss": 0.7748,
+      "step": 6076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6048429252415703e-05,
+      "loss": 0.6827,
+      "step": 6077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6047094594262142e-05,
+      "loss": 0.7563,
+      "step": 6078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604575976627277e-05,
+      "loss": 0.809,
+      "step": 6079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604442476848507e-05,
+      "loss": 0.8038,
+      "step": 6080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6043089600936532e-05,
+      "loss": 0.8035,
+      "step": 6081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041754263664663e-05,
+      "loss": 0.6469,
+      "step": 6082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6040418756706966e-05,
+      "loss": 0.761,
+      "step": 6083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603908308010095e-05,
+      "loss": 0.7283,
+      "step": 6084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603774723388412e-05,
+      "loss": 0.7379,
+      "step": 6085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6036411218094007e-05,
+      "loss": 0.7111,
+      "step": 6086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603507503276812e-05,
+      "loss": 0.7632,
+      "step": 6087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033738677943994e-05,
+      "loss": 0.7734,
+      "step": 6088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6032402153659164e-05,
+      "loss": 0.7056,
+      "step": 6089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031065459951163e-05,
+      "loss": 0.6993,
+      "step": 6090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602972859685753e-05,
+      "loss": 0.681,
+      "step": 6091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602839156441582e-05,
+      "loss": 0.797,
+      "step": 6092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6027054362663572e-05,
+      "loss": 0.7406,
+      "step": 6093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025716991638353e-05,
+      "loss": 0.7572,
+      "step": 6094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602437945137772e-05,
+      "loss": 0.7172,
+      "step": 6095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6023041741919233e-05,
+      "loss": 0.7285,
+      "step": 6096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602170386330047e-05,
+      "loss": 0.7055,
+      "step": 6097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020365815559003e-05,
+      "loss": 0.7457,
+      "step": 6098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6019027598732412e-05,
+      "loss": 0.7489,
+      "step": 6099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017689212858283e-05,
+      "loss": 0.7059,
+      "step": 6100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6016350657974205e-05,
+      "loss": 0.7289,
+      "step": 6101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601501193411777e-05,
+      "loss": 0.8036,
+      "step": 6102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013673041326577e-05,
+      "loss": 0.7627,
+      "step": 6103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012333979638225e-05,
+      "loss": 0.8047,
+      "step": 6104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6010994749090338e-05,
+      "loss": 0.7816,
+      "step": 6105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600965534972051e-05,
+      "loss": 0.6977,
+      "step": 6106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600831578156637e-05,
+      "loss": 0.7499,
+      "step": 6107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6006976044665537e-05,
+      "loss": 0.3599,
+      "step": 6108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600563613905564e-05,
+      "loss": 0.8106,
+      "step": 6109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004296064774308e-05,
+      "loss": 0.8201,
+      "step": 6110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002955821859178e-05,
+      "loss": 0.7425,
+      "step": 6111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60016154103479e-05,
+      "loss": 0.7697,
+      "step": 6112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000274830278106e-05,
+      "loss": 0.3224,
+      "step": 6113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998934081687458e-05,
+      "loss": 0.7599,
+      "step": 6114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5997593164613604e-05,
+      "loss": 0.7008,
+      "step": 6115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599625207909421e-05,
+      "loss": 0.7115,
+      "step": 6116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599491082516694e-05,
+      "loss": 0.7386,
+      "step": 6117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599356940286946e-05,
+      "loss": 0.7965,
+      "step": 6118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599222781223945e-05,
+      "loss": 0.7986,
+      "step": 6119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990886053314586e-05,
+      "loss": 0.7319,
+      "step": 6120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989544126132554e-05,
+      "loss": 0.7201,
+      "step": 6121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988202030731036e-05,
+      "loss": 0.6626,
+      "step": 6122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986859767147734e-05,
+      "loss": 0.715,
+      "step": 6123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985517335420346e-05,
+      "loss": 0.7539,
+      "step": 6124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598417473558657e-05,
+      "loss": 0.7914,
+      "step": 6125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982831967684114e-05,
+      "loss": 0.7693,
+      "step": 6126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598148903175069e-05,
+      "loss": 0.7332,
+      "step": 6127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598014592782402e-05,
+      "loss": 0.7465,
+      "step": 6128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978802655941822e-05,
+      "loss": 0.7389,
+      "step": 6129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977459216141824e-05,
+      "loss": 0.7372,
+      "step": 6130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976115608461756e-05,
+      "loss": 0.6883,
+      "step": 6131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5974771832939355e-05,
+      "loss": 0.7514,
+      "step": 6132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973427889612357e-05,
+      "loss": 0.7446,
+      "step": 6133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972083778518518e-05,
+      "loss": 0.7593,
+      "step": 6134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970739499695577e-05,
+      "loss": 0.7472,
+      "step": 6135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969395053181295e-05,
+      "loss": 0.821,
+      "step": 6136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5968050439013426e-05,
+      "loss": 0.7092,
+      "step": 6137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596670565722974e-05,
+      "loss": 0.6375,
+      "step": 6138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965360707868004e-05,
+      "loss": 0.7999,
+      "step": 6139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596401559096599e-05,
+      "loss": 0.727,
+      "step": 6140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596267030656148e-05,
+      "loss": 0.6992,
+      "step": 6141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5961324854692254e-05,
+      "loss": 0.7774,
+      "step": 6142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59599792353961e-05,
+      "loss": 0.7029,
+      "step": 6143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595863344871081e-05,
+      "loss": 0.6558,
+      "step": 6144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5957287494674185e-05,
+      "loss": 0.743,
+      "step": 6145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595594137332402e-05,
+      "loss": 0.7461,
+      "step": 6146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595459508469813e-05,
+      "loss": 0.7823,
+      "step": 6147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953248628834315e-05,
+      "loss": 0.7801,
+      "step": 6148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59519020057704e-05,
+      "loss": 0.6756,
+      "step": 6149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950555215544206e-05,
+      "loss": 0.8088,
+      "step": 6150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949208258193554e-05,
+      "loss": 0.7295,
+      "step": 6151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947861133756274e-05,
+      "loss": 0.7242,
+      "step": 6152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59465138422702e-05,
+      "loss": 0.6955,
+      "step": 6153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945166383773178e-05,
+      "loss": 0.7357,
+      "step": 6154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594381875830304e-05,
+      "loss": 0.788,
+      "step": 6155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594247096589765e-05,
+      "loss": 0.7546,
+      "step": 6156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594112300659485e-05,
+      "loss": 0.7452,
+      "step": 6157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5939774880432505e-05,
+      "loss": 0.6926,
+      "step": 6158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938426587448473e-05,
+      "loss": 0.7443,
+      "step": 6159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593707812768062e-05,
+      "loss": 0.8364,
+      "step": 6160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935729501166826e-05,
+      "loss": 0.7411,
+      "step": 6161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593438070794496e-05,
+      "loss": 0.7435,
+      "step": 6162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593303174805291e-05,
+      "loss": 0.7468,
+      "step": 6163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931682621528558e-05,
+      "loss": 0.6742,
+      "step": 6164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5930333328409794e-05,
+      "loss": 0.7667,
+      "step": 6165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928983868734517e-05,
+      "loss": 0.7369,
+      "step": 6166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592763424254063e-05,
+      "loss": 0.7375,
+      "step": 6167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592628444986603e-05,
+      "loss": 0.7125,
+      "step": 6168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924934490748627e-05,
+      "loss": 0.7351,
+      "step": 6169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592358436522634e-05,
+      "loss": 0.6894,
+      "step": 6170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592223407333709e-05,
+      "loss": 0.7431,
+      "step": 6171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5920883615118795e-05,
+      "loss": 0.7773,
+      "step": 6172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591953299060939e-05,
+      "loss": 0.8018,
+      "step": 6173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59181821998468e-05,
+      "loss": 0.7079,
+      "step": 6174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916831242868965e-05,
+      "loss": 0.7286,
+      "step": 6175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915480119713832e-05,
+      "loss": 0.6971,
+      "step": 6176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914128830419345e-05,
+      "loss": 0.7281,
+      "step": 6177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912777375023453e-05,
+      "loss": 0.768,
+      "step": 6178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5911425753564115e-05,
+      "loss": 0.7003,
+      "step": 6179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910073966079293e-05,
+      "loss": 0.7927,
+      "step": 6180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590872201260695e-05,
+      "loss": 0.7736,
+      "step": 6181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907369893185057e-05,
+      "loss": 0.7922,
+      "step": 6182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906017607851595e-05,
+      "loss": 0.7303,
+      "step": 6183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5904665156644532e-05,
+      "loss": 0.7709,
+      "step": 6184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590331253960186e-05,
+      "loss": 0.7509,
+      "step": 6185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590195975676157e-05,
+      "loss": 0.7491,
+      "step": 6186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900606808161648e-05,
+      "loss": 0.7338,
+      "step": 6187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899253693840097e-05,
+      "loss": 0.7566,
+      "step": 6188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589790041383492e-05,
+      "loss": 0.763,
+      "step": 6189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589654696818413e-05,
+      "loss": 0.7184,
+      "step": 6190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895193356925726e-05,
+      "loss": 0.7739,
+      "step": 6191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589383958009773e-05,
+      "loss": 0.7892,
+      "step": 6192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892485637738177e-05,
+      "loss": 0.7109,
+      "step": 6193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589113152988507e-05,
+      "loss": 0.7838,
+      "step": 6194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5889777256576456e-05,
+      "loss": 0.7849,
+      "step": 6195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888422817850367e-05,
+      "loss": 0.7368,
+      "step": 6196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887068213744842e-05,
+      "loss": 0.7007,
+      "step": 6197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885713444297924e-05,
+      "loss": 0.7428,
+      "step": 6198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884358509547667e-05,
+      "loss": 0.7199,
+      "step": 6199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883003409532124e-05,
+      "loss": 0.7513,
+      "step": 6200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881648144289347e-05,
+      "loss": 0.7233,
+      "step": 6201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880292713857405e-05,
+      "loss": 0.7481,
+      "step": 6202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5878937118274367e-05,
+      "loss": 0.7308,
+      "step": 6203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877581357578306e-05,
+      "loss": 0.7201,
+      "step": 6204
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876225431807294e-05,
+      "loss": 0.7399,
+      "step": 6205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5874869340999423e-05,
+      "loss": 0.6916,
+      "step": 6206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5873513085192763e-05,
+      "loss": 0.7347,
+      "step": 6207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587215666442542e-05,
+      "loss": 0.7399,
+      "step": 6208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5870800078735484e-05,
+      "loss": 0.7497,
+      "step": 6209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5869443328161057e-05,
+      "loss": 0.7294,
+      "step": 6210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868086412740242e-05,
+      "loss": 0.7582,
+      "step": 6211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586672933251115e-05,
+      "loss": 0.7327,
+      "step": 6212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5865372087511896e-05,
+      "loss": 0.7785,
+      "step": 6213
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864014677780597e-05,
+      "loss": 0.7388,
+      "step": 6214
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862657103355376e-05,
+      "loss": 0.7442,
+      "step": 6215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5861299364274367e-05,
+      "loss": 0.7103,
+      "step": 6216
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5859941460575696e-05,
+      "loss": 0.7659,
+      "step": 6217
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5858583392297503e-05,
+      "loss": 0.6953,
+      "step": 6218
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585722515947793e-05,
+      "loss": 0.7508,
+      "step": 6219
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855866762155122e-05,
+      "loss": 0.3274,
+      "step": 6220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854508200367237e-05,
+      "loss": 0.7124,
+      "step": 6221
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585314947415242e-05,
+      "loss": 0.6792,
+      "step": 6222
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851790583548843e-05,
+      "loss": 0.7061,
+      "step": 6223
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850431528594663e-05,
+      "loss": 0.7454,
+      "step": 6224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849072309328054e-05,
+      "loss": 0.7737,
+      "step": 6225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5847712925787188e-05,
+      "loss": 0.7372,
+      "step": 6226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584635337801024e-05,
+      "loss": 0.7119,
+      "step": 6227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5844993666035406e-05,
+      "loss": 0.7904,
+      "step": 6228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843633789900862e-05,
+      "loss": 0.7547,
+      "step": 6229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842273749644808e-05,
+      "loss": 0.737,
+      "step": 6230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5840913545305437e-05,
+      "loss": 0.771,
+      "step": 6231
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5839553176920956e-05,
+      "loss": 0.706,
+      "step": 6232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838192644529567e-05,
+      "loss": 0.7211,
+      "step": 6233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836831948169483e-05,
+      "loss": 0.3279,
+      "step": 6234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583547108787892e-05,
+      "loss": 0.6235,
+      "step": 6235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834110063696097e-05,
+      "loss": 0.7121,
+      "step": 6236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583274887565924e-05,
+      "loss": 0.7483,
+      "step": 6237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5831387523806584e-05,
+      "loss": 0.7278,
+      "step": 6238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830026008176352e-05,
+      "loss": 0.6332,
+      "step": 6239
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582866432880679e-05,
+      "loss": 0.775,
+      "step": 6240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582730248573614e-05,
+      "loss": 0.7449,
+      "step": 6241
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825940479002654e-05,
+      "loss": 0.6807,
+      "step": 6242
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5824578308644583e-05,
+      "loss": 0.7426,
+      "step": 6243
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823215974700178e-05,
+      "loss": 0.7787,
+      "step": 6244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582185347720771e-05,
+      "loss": 0.6807,
+      "step": 6245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582049081620544e-05,
+      "loss": 0.7211,
+      "step": 6246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819127991731633e-05,
+      "loss": 0.7554,
+      "step": 6247
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817765003824578e-05,
+      "loss": 0.7123,
+      "step": 6248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581640185252255e-05,
+      "loss": 0.7405,
+      "step": 6249
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5815038537863832e-05,
+      "loss": 0.7687,
+      "step": 6250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5813675059886712e-05,
+      "loss": 0.6871,
+      "step": 6251
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812311418629488e-05,
+      "loss": 0.7731,
+      "step": 6252
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5810947614130454e-05,
+      "loss": 0.6777,
+      "step": 6253
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809583646427917e-05,
+      "loss": 0.7715,
+      "step": 6254
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5808219515560184e-05,
+      "loss": 0.7318,
+      "step": 6255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806855221565567e-05,
+      "loss": 0.733,
+      "step": 6256
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805490764482382e-05,
+      "loss": 0.7444,
+      "step": 6257
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5804126144348954e-05,
+      "loss": 0.7263,
+      "step": 6258
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802761361203604e-05,
+      "loss": 0.7892,
+      "step": 6259
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5801396415084664e-05,
+      "loss": 0.7188,
+      "step": 6260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800031306030476e-05,
+      "loss": 0.7048,
+      "step": 6261
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579866603407937e-05,
+      "loss": 0.7292,
+      "step": 6262
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5797300599269693e-05,
+      "loss": 0.6849,
+      "step": 6263
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795935001639796e-05,
+      "loss": 0.6742,
+      "step": 6264
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5794569241228032e-05,
+      "loss": 0.7141,
+      "step": 6265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5793203318072762e-05,
+      "loss": 0.7543,
+      "step": 6266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791837232212342e-05,
+      "loss": 0.7034,
+      "step": 6267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5790470983685147e-05,
+      "loss": 0.6899,
+      "step": 6268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578910457252954e-05,
+      "loss": 0.7006,
+      "step": 6269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5787737998783903e-05,
+      "loss": 0.7767,
+      "step": 6270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578637126248662e-05,
+      "loss": 0.724,
+      "step": 6271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785004363676072e-05,
+      "loss": 0.7772,
+      "step": 6272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783637302390644e-05,
+      "loss": 0.6776,
+      "step": 6273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578227007866874e-05,
+      "loss": 0.7602,
+      "step": 6274
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5780902692548756e-05,
+      "loss": 0.7247,
+      "step": 6275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5779535144069096e-05,
+      "loss": 0.8165,
+      "step": 6276
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778167433268165e-05,
+      "loss": 0.6716,
+      "step": 6277
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577679956018438e-05,
+      "loss": 0.8088,
+      "step": 6278
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5775431524856156e-05,
+      "loss": 0.7169,
+      "step": 6279
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5774063327321914e-05,
+      "loss": 0.6935,
+      "step": 6280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772694967620087e-05,
+      "loss": 0.7047,
+      "step": 6281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5771326445789097e-05,
+      "loss": 0.7375,
+      "step": 6282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769957761867385e-05,
+      "loss": 0.3574,
+      "step": 6283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576858891589339e-05,
+      "loss": 0.7336,
+      "step": 6284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576721990790556e-05,
+      "loss": 0.6999,
+      "step": 6285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5765850737942338e-05,
+      "loss": 0.7106,
+      "step": 6286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764481406042183e-05,
+      "loss": 0.7076,
+      "step": 6287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5763111912243552e-05,
+      "loss": 0.6428,
+      "step": 6288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5761742256584908e-05,
+      "loss": 0.6881,
+      "step": 6289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760372439104715e-05,
+      "loss": 0.7233,
+      "step": 6290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575900245984145e-05,
+      "loss": 0.7305,
+      "step": 6291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757632318833587e-05,
+      "loss": 0.7014,
+      "step": 6292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575626201611961e-05,
+      "loss": 0.7049,
+      "step": 6293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754891551738003e-05,
+      "loss": 0.6313,
+      "step": 6294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753520925727254e-05,
+      "loss": 0.7063,
+      "step": 6295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752150138125857e-05,
+      "loss": 0.7619,
+      "step": 6296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5750779188972316e-05,
+      "loss": 0.7341,
+      "step": 6297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5749408078305134e-05,
+      "loss": 0.7493,
+      "step": 6298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5748036806162816e-05,
+      "loss": 0.7265,
+      "step": 6299
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746665372583878e-05,
+      "loss": 0.6474,
+      "step": 6300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5745293777606834e-05,
+      "loss": 0.7051,
+      "step": 6301
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574392202127021e-05,
+      "loss": 0.7902,
+      "step": 6302
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574255010361253e-05,
+      "loss": 0.775,
+      "step": 6303
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5741178024672326e-05,
+      "loss": 0.7309,
+      "step": 6304
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573980578448813e-05,
+      "loss": 0.7271,
+      "step": 6305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738433383098493e-05,
+      "loss": 0.7313,
+      "step": 6306
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737060820541947e-05,
+      "loss": 0.7356,
+      "step": 6307
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735688096857048e-05,
+      "loss": 0.7039,
+      "step": 6308
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5734315212082346e-05,
+      "loss": 0.6699,
+      "step": 6309
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57329421662564e-05,
+      "loss": 0.7853,
+      "step": 6310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731568959417776e-05,
+      "loss": 0.7655,
+      "step": 6311
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5730195591605036e-05,
+      "loss": 0.7908,
+      "step": 6312
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5728822062856758e-05,
+      "loss": 0.7243,
+      "step": 6313
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727448373211515e-05,
+      "loss": 0.7419,
+      "step": 6314
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5726074522707885e-05,
+      "loss": 0.75,
+      "step": 6315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724700511384455e-05,
+      "loss": 0.7619,
+      "step": 6316
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572332633927982e-05,
+      "loss": 0.3448,
+      "step": 6317
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572195200643257e-05,
+      "loss": 0.755,
+      "step": 6318
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57205775128813e-05,
+      "loss": 0.7367,
+      "step": 6319
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5719202858664623e-05,
+      "loss": 0.7113,
+      "step": 6320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5717828043821138e-05,
+      "loss": 0.7098,
+      "step": 6321
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571645306838946e-05,
+      "loss": 0.8119,
+      "step": 6322
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571507793240821e-05,
+      "loss": 0.7199,
+      "step": 6323
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5713702635916003e-05,
+      "loss": 0.8075,
+      "step": 6324
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712327178951467e-05,
+      "loss": 0.7429,
+      "step": 6325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5710951561553237e-05,
+      "loss": 0.7353,
+      "step": 6326
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570957578375994e-05,
+      "loss": 0.726,
+      "step": 6327
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708199845610224e-05,
+      "loss": 0.7574,
+      "step": 6328
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5706823747142724e-05,
+      "loss": 0.6609,
+      "step": 6329
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5705447488396096e-05,
+      "loss": 0.6805,
+      "step": 6330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570407106940899e-05,
+      "loss": 0.7073,
+      "step": 6331
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702694490220066e-05,
+      "loss": 0.6492,
+      "step": 6332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570131775086798e-05,
+      "loss": 0.7706,
+      "step": 6333
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56999408513914e-05,
+      "loss": 0.7551,
+      "step": 6334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698563791829006e-05,
+      "loss": 0.7455,
+      "step": 6335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569718657221946e-05,
+      "loss": 0.8288,
+      "step": 6336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569580919260145e-05,
+      "loss": 0.7699,
+      "step": 6337
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569443165301366e-05,
+      "loss": 0.7254,
+      "step": 6338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693053953494776e-05,
+      "loss": 0.6336,
+      "step": 6339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5691676094083495e-05,
+      "loss": 0.7263,
+      "step": 6340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5690298074818507e-05,
+      "loss": 0.7279,
+      "step": 6341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688919895738524e-05,
+      "loss": 0.7634,
+      "step": 6342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687541556882252e-05,
+      "loss": 0.7117,
+      "step": 6343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686163058288397e-05,
+      "loss": 0.7341,
+      "step": 6344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568478439999568e-05,
+      "loss": 0.735,
+      "step": 6345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5683405582042816e-05,
+      "loss": 0.6838,
+      "step": 6346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682026604468533e-05,
+      "loss": 0.7158,
+      "step": 6347
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 0.6202,
+      "step": 6348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567926817061063e-05,
+      "loss": 0.711,
+      "step": 6349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5677888714404483e-05,
+      "loss": 0.7047,
+      "step": 6350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5676509098731857e-05,
+      "loss": 0.7876,
+      "step": 6351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567512932363151e-05,
+      "loss": 0.7945,
+      "step": 6352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5673749389142182e-05,
+      "loss": 0.7766,
+      "step": 6353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672369295302636e-05,
+      "loss": 0.7354,
+      "step": 6354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5670989042151625e-05,
+      "loss": 0.7431,
+      "step": 6355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5669608629727927e-05,
+      "loss": 0.7592,
+      "step": 6356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56682280580703e-05,
+      "loss": 0.8016,
+      "step": 6357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5666847327217524e-05,
+      "loss": 0.8189,
+      "step": 6358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5665466437208373e-05,
+      "loss": 0.6926,
+      "step": 6359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664085388081637e-05,
+      "loss": 0.7436,
+      "step": 6360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5662704179876092e-05,
+      "loss": 0.7433,
+      "step": 6361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5661322812630544e-05,
+      "loss": 0.7198,
+      "step": 6362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5659941286383782e-05,
+      "loss": 0.7603,
+      "step": 6363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658559601174605e-05,
+      "loss": 0.6532,
+      "step": 6364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657177757041823e-05,
+      "loss": 0.7651,
+      "step": 6365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5655795754024244e-05,
+      "loss": 0.7337,
+      "step": 6366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565441359216068e-05,
+      "loss": 0.7656,
+      "step": 6367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653031271489952e-05,
+      "loss": 0.7761,
+      "step": 6368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5651648792050886e-05,
+      "loss": 0.8015,
+      "step": 6369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5650266153882304e-05,
+      "loss": 0.7802,
+      "step": 6370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564888335702304e-05,
+      "loss": 0.7496,
+      "step": 6371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647500401511932e-05,
+      "loss": 0.7955,
+      "step": 6372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646117287387822e-05,
+      "loss": 0.7622,
+      "step": 6373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564473401468955e-05,
+      "loss": 0.7246,
+      "step": 6374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5643350583455976e-05,
+      "loss": 0.6992,
+      "step": 6375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641966993725942e-05,
+      "loss": 0.72,
+      "step": 6376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640583245538318e-05,
+      "loss": 0.6971,
+      "step": 6377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5639199338931958e-05,
+      "loss": 0.7503,
+      "step": 6378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637815273945735e-05,
+      "loss": 0.6731,
+      "step": 6379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636431050618523e-05,
+      "loss": 0.7623,
+      "step": 6380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5635046668989194e-05,
+      "loss": 0.7012,
+      "step": 6381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563366212909663e-05,
+      "loss": 0.7108,
+      "step": 6382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632277430979718e-05,
+      "loss": 0.7964,
+      "step": 6383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5630892574677348e-05,
+      "loss": 0.7151,
+      "step": 6384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5629507560228414e-05,
+      "loss": 0.7894,
+      "step": 6385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5628122387671815e-05,
+      "loss": 0.7621,
+      "step": 6386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562673705704645e-05,
+      "loss": 0.6931,
+      "step": 6387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625351568391234e-05,
+      "loss": 0.7585,
+      "step": 6388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5623965921745077e-05,
+      "loss": 0.7979,
+      "step": 6389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622580117146893e-05,
+      "loss": 0.7052,
+      "step": 6390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5621194154635604e-05,
+      "loss": 0.7762,
+      "step": 6391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5619808034250137e-05,
+      "loss": 0.7487,
+      "step": 6392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561842175602942e-05,
+      "loss": 0.6884,
+      "step": 6393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617035320012394e-05,
+      "loss": 0.7634,
+      "step": 6394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5615648726237983e-05,
+      "loss": 0.7164,
+      "step": 6395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614261974745147e-05,
+      "loss": 0.82,
+      "step": 6396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561287506557282e-05,
+      "loss": 0.7723,
+      "step": 6397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611487998759966e-05,
+      "loss": 0.7339,
+      "step": 6398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561010077434553e-05,
+      "loss": 0.6692,
+      "step": 6399
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560871339236848e-05,
+      "loss": 0.7756,
+      "step": 6400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560732585286778e-05,
+      "loss": 0.6894,
+      "step": 6401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56059381558824e-05,
+      "loss": 0.6848,
+      "step": 6402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604550301451314e-05,
+      "loss": 0.7385,
+      "step": 6403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603162289613503e-05,
+      "loss": 0.6785,
+      "step": 6404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560177412040794e-05,
+      "loss": 0.7647,
+      "step": 6405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600385793873627e-05,
+      "loss": 0.6896,
+      "step": 6406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5598997310049545e-05,
+      "loss": 0.7554,
+      "step": 6407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55976086689747e-05,
+      "loss": 0.7758,
+      "step": 6408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596219870688077e-05,
+      "loss": 0.7385,
+      "step": 6409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5594830915228695e-05,
+      "loss": 0.7272,
+      "step": 6410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593441802635564e-05,
+      "loss": 0.7256,
+      "step": 6411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5592052532947687e-05,
+      "loss": 0.6228,
+      "step": 6412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5590663106204093e-05,
+      "loss": 0.7195,
+      "step": 6413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55892735224438e-05,
+      "loss": 0.7542,
+      "step": 6414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587883781705833e-05,
+      "loss": 0.751,
+      "step": 6415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5586493884029227e-05,
+      "loss": 0.7449,
+      "step": 6416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5585103829453018e-05,
+      "loss": 0.7791,
+      "step": 6417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583713618016245e-05,
+      "loss": 0.7138,
+      "step": 6418
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5582323249757955e-05,
+      "loss": 0.7123,
+      "step": 6419
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558093272471719e-05,
+      "loss": 0.7441,
+      "step": 6420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557954204293302e-05,
+      "loss": 0.7283,
+      "step": 6421
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5578151204444485e-05,
+      "loss": 0.7852,
+      "step": 6422
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576760209290654e-05,
+      "loss": 0.759,
+      "step": 6423
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575369057510598e-05,
+      "loss": 0.7479,
+      "step": 6424
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573977749143387e-05,
+      "loss": 0.7503,
+      "step": 6425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557258628422809e-05,
+      "loss": 0.6887,
+      "step": 6426
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571194662803794e-05,
+      "loss": 0.7519,
+      "step": 6427
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556980288490958e-05,
+      "loss": 0.6778,
+      "step": 6428
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5568410950584537e-05,
+      "loss": 0.6903,
+      "step": 6429
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556701885986776e-05,
+      "loss": 0.6973,
+      "step": 6430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5565626612798346e-05,
+      "loss": 0.799,
+      "step": 6431
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564234209415397e-05,
+      "loss": 0.71,
+      "step": 6432
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5562841649758017e-05,
+      "loss": 0.7005,
+      "step": 6433
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561448933865318e-05,
+      "loss": 0.7015,
+      "step": 6434
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556005606177642e-05,
+      "loss": 0.7033,
+      "step": 6435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558663033530432e-05,
+      "loss": 0.7711,
+      "step": 6436
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555726984916649e-05,
+      "loss": 0.6941,
+      "step": 6437
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5555876508723715e-05,
+      "loss": 0.7018,
+      "step": 6438
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555448301224124e-05,
+      "loss": 0.69,
+      "step": 6439
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553089359758208e-05,
+      "loss": 0.7188,
+      "step": 6440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555169555131375e-05,
+      "loss": 0.7924,
+      "step": 6441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550301586947017e-05,
+      "loss": 0.8226,
+      "step": 6442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554890746669716e-05,
+      "loss": 0.7498,
+      "step": 6443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547513190603344e-05,
+      "loss": 0.741,
+      "step": 6444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546118758704703e-05,
+      "loss": 0.6822,
+      "step": 6445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544724171040424e-05,
+      "loss": 0.6938,
+      "step": 6446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5543329427649662e-05,
+      "loss": 0.7563,
+      "step": 6447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5541934528571596e-05,
+      "loss": 0.7709,
+      "step": 6448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540539473845394e-05,
+      "loss": 0.7393,
+      "step": 6449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539144263510248e-05,
+      "loss": 0.8127,
+      "step": 6450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553774889760533e-05,
+      "loss": 0.8143,
+      "step": 6451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536353376169843e-05,
+      "loss": 0.7688,
+      "step": 6452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5534957699242973e-05,
+      "loss": 0.6699,
+      "step": 6453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533561866863916e-05,
+      "loss": 0.7603,
+      "step": 6454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5532165879071885e-05,
+      "loss": 0.7191,
+      "step": 6455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5530769735906076e-05,
+      "loss": 0.7493,
+      "step": 6456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529373437405707e-05,
+      "loss": 0.8165,
+      "step": 6457
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527976983609995e-05,
+      "loss": 0.6805,
+      "step": 6458
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5526580374558152e-05,
+      "loss": 0.721,
+      "step": 6459
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5525183610289412e-05,
+      "loss": 0.7072,
+      "step": 6460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5523786690842996e-05,
+      "loss": 0.6966,
+      "step": 6461
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552238961625814e-05,
+      "loss": 0.6998,
+      "step": 6462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5520992386574084e-05,
+      "loss": 0.7384,
+      "step": 6463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5519595001830068e-05,
+      "loss": 0.7856,
+      "step": 6464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518197462065338e-05,
+      "loss": 0.7954,
+      "step": 6465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5516799767319145e-05,
+      "loss": 0.6895,
+      "step": 6466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5515401917630745e-05,
+      "loss": 0.7812,
+      "step": 6467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514003913039395e-05,
+      "loss": 0.8098,
+      "step": 6468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551260575358436e-05,
+      "loss": 0.7861,
+      "step": 6469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551120743930491e-05,
+      "loss": 0.7171,
+      "step": 6470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5509808970240314e-05,
+      "loss": 0.7443,
+      "step": 6471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550841034642985e-05,
+      "loss": 0.7106,
+      "step": 6472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507011567912802e-05,
+      "loss": 0.7703,
+      "step": 6473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550561263472845e-05,
+      "loss": 0.7447,
+      "step": 6474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5504213546916088e-05,
+      "loss": 0.7126,
+      "step": 6475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502814304515007e-05,
+      "loss": 0.7142,
+      "step": 6476
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5501414907564505e-05,
+      "loss": 0.7768,
+      "step": 6477
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550001535610389e-05,
+      "loss": 0.6646,
+      "step": 6478
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5498615650172467e-05,
+      "loss": 0.7729,
+      "step": 6479
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549721578980955e-05,
+      "loss": 0.715,
+      "step": 6480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495815775054445e-05,
+      "loss": 0.7975,
+      "step": 6481
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549441560594648e-05,
+      "loss": 0.7413,
+      "step": 6482
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493015282524977e-05,
+      "loss": 0.6342,
+      "step": 6483
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491614804829267e-05,
+      "loss": 0.7499,
+      "step": 6484
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549021417289868e-05,
+      "loss": 0.6984,
+      "step": 6485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548881338677256e-05,
+      "loss": 0.7144,
+      "step": 6486
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487412446490236e-05,
+      "loss": 0.7339,
+      "step": 6487
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5486011352091068e-05,
+      "loss": 0.7778,
+      "step": 6488
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54846101036144e-05,
+      "loss": 0.7395,
+      "step": 6489
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548320870109959e-05,
+      "loss": 0.6747,
+      "step": 6490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548180714458599e-05,
+      "loss": 0.7688,
+      "step": 6491
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5480405434112973e-05,
+      "loss": 0.7144,
+      "step": 6492
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5479003569719897e-05,
+      "loss": 0.7193,
+      "step": 6493
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477601551446145e-05,
+      "loss": 0.7604,
+      "step": 6494
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547619937933108e-05,
+      "loss": 0.7353,
+      "step": 6495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5474797053414096e-05,
+      "loss": 0.7653,
+      "step": 6496
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547339457373457e-05,
+      "loss": 0.7226,
+      "step": 6497
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547199194033189e-05,
+      "loss": 0.7296,
+      "step": 6498
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5470589153245462e-05,
+      "loss": 0.7401,
+      "step": 6499
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546918621251467e-05,
+      "loss": 0.7291,
+      "step": 6500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467783118178922e-05,
+      "loss": 0.7086,
+      "step": 6501
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466379870277623e-05,
+      "loss": 0.666,
+      "step": 6502
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5464976468850188e-05,
+      "loss": 0.649,
+      "step": 6503
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463572913936026e-05,
+      "loss": 0.7738,
+      "step": 6504
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462169205574565e-05,
+      "loss": 0.7357,
+      "step": 6505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546076534380522e-05,
+      "loss": 0.7266,
+      "step": 6506
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459361328667425e-05,
+      "loss": 0.7186,
+      "step": 6507
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457957160200608e-05,
+      "loss": 0.7844,
+      "step": 6508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456552838444215e-05,
+      "loss": 0.7356,
+      "step": 6509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5455148363437672e-05,
+      "loss": 0.7857,
+      "step": 6510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545374373522044e-05,
+      "loss": 0.6971,
+      "step": 6511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545233895383196e-05,
+      "loss": 0.7801,
+      "step": 6512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545093401931168e-05,
+      "loss": 0.6929,
+      "step": 6513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5449528931699072e-05,
+      "loss": 0.7321,
+      "step": 6514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5448123691033597e-05,
+      "loss": 0.7438,
+      "step": 6515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446718297354712e-05,
+      "loss": 0.7707,
+      "step": 6516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445312750701895e-05,
+      "loss": 0.6986,
+      "step": 6517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443907051114622e-05,
+      "loss": 0.7432,
+      "step": 6518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5442501198632373e-05,
+      "loss": 0.7042,
+      "step": 6519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441095193294623e-05,
+      "loss": 0.7315,
+      "step": 6520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439689035140873e-05,
+      "loss": 0.7015,
+      "step": 6521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543828272421061e-05,
+      "loss": 0.7902,
+      "step": 6522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543687626054333e-05,
+      "loss": 0.6599,
+      "step": 6523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5435469644178534e-05,
+      "loss": 0.7391,
+      "step": 6524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543406287515573e-05,
+      "loss": 0.6784,
+      "step": 6525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432655953514427e-05,
+      "loss": 0.751,
+      "step": 6526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5431248879294142e-05,
+      "loss": 0.7395,
+      "step": 6527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5429841652534387e-05,
+      "loss": 0.7729,
+      "step": 6528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542843427327469e-05,
+      "loss": 0.6518,
+      "step": 6529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5427026741554575e-05,
+      "loss": 0.7128,
+      "step": 6530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425619057413576e-05,
+      "loss": 0.7562,
+      "step": 6531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5424211220891224e-05,
+      "loss": 0.6883,
+      "step": 6532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5422803232027063e-05,
+      "loss": 0.7589,
+      "step": 6533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542139509086064e-05,
+      "loss": 0.7105,
+      "step": 6534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419986797431494e-05,
+      "loss": 0.7264,
+      "step": 6535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418578351779185e-05,
+      "loss": 0.7014,
+      "step": 6536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417169753943268e-05,
+      "loss": 0.8074,
+      "step": 6537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5415761003963307e-05,
+      "loss": 0.7696,
+      "step": 6538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414352101878858e-05,
+      "loss": 0.7476,
+      "step": 6539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54129430477295e-05,
+      "loss": 0.7416,
+      "step": 6540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541153384155481e-05,
+      "loss": 0.7296,
+      "step": 6541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5410124483394353e-05,
+      "loss": 0.7289,
+      "step": 6542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408714973287723e-05,
+      "loss": 0.6626,
+      "step": 6543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407305311274502e-05,
+      "loss": 0.8351,
+      "step": 6544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540589549739428e-05,
+      "loss": 0.6254,
+      "step": 6545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404485531686657e-05,
+      "loss": 0.7424,
+      "step": 6546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540307541419123e-05,
+      "loss": 0.7114,
+      "step": 6547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401665144947605e-05,
+      "loss": 0.6198,
+      "step": 6548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5400254723995384e-05,
+      "loss": 0.7823,
+      "step": 6549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5398844151374186e-05,
+      "loss": 0.8101,
+      "step": 6550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5397433427123624e-05,
+      "loss": 0.6387,
+      "step": 6551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539602255128332e-05,
+      "loss": 0.7557,
+      "step": 6552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53946115238929e-05,
+      "loss": 0.7329,
+      "step": 6553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393200344991993e-05,
+      "loss": 0.7636,
+      "step": 6554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391789014620236e-05,
+      "loss": 0.7795,
+      "step": 6555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5390377532817257e-05,
+      "loss": 0.6962,
+      "step": 6556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5388965899622708e-05,
+      "loss": 0.7112,
+      "step": 6557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538755411507623e-05,
+      "loss": 0.6992,
+      "step": 6558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386142179217477e-05,
+      "loss": 0.7259,
+      "step": 6559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5384730092086106e-05,
+      "loss": 0.7607,
+      "step": 6560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383317853721765e-05,
+      "loss": 0.7283,
+      "step": 6561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5381905464164134e-05,
+      "loss": 0.3307,
+      "step": 6562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380492923452867e-05,
+      "loss": 0.714,
+      "step": 6563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5379080231627646e-05,
+      "loss": 0.7614,
+      "step": 6564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377667388728136e-05,
+      "loss": 0.7297,
+      "step": 6565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5376254394794028e-05,
+      "loss": 0.6759,
+      "step": 6566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374841249865e-05,
+      "loss": 0.7543,
+      "step": 6567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373427953980745e-05,
+      "loss": 0.7903,
+      "step": 6568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372014507180953e-05,
+      "loss": 0.6627,
+      "step": 6569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370600909505325e-05,
+      "loss": 0.7803,
+      "step": 6570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369187160993556e-05,
+      "loss": 0.7733,
+      "step": 6571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367773261685362e-05,
+      "loss": 0.7012,
+      "step": 6572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5366359211620442e-05,
+      "loss": 0.7093,
+      "step": 6573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536494501083852e-05,
+      "loss": 0.7591,
+      "step": 6574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363530659379308e-05,
+      "loss": 0.7787,
+      "step": 6575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362116157282533e-05,
+      "loss": 0.6906,
+      "step": 6576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360701504587918e-05,
+      "loss": 0.696,
+      "step": 6577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53592867013352e-05,
+      "loss": 0.7516,
+      "step": 6578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357871747564108e-05,
+      "loss": 0.774,
+      "step": 6579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5356456643314385e-05,
+      "loss": 0.6784,
+      "step": 6580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355041388625774e-05,
+      "loss": 0.7345,
+      "step": 6581
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5353625983538022e-05,
+      "loss": 0.7001,
+      "step": 6582
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5352210428090888e-05,
+      "loss": 0.7939,
+      "step": 6583
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535079472232412e-05,
+      "loss": 0.7405,
+      "step": 6584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349378866277484e-05,
+      "loss": 0.7737,
+      "step": 6585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347962859990744e-05,
+      "loss": 0.7465,
+      "step": 6586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346546703503667e-05,
+      "loss": 0.7574,
+      "step": 6587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534513039685603e-05,
+      "loss": 0.7819,
+      "step": 6588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534371394008761e-05,
+      "loss": 0.7847,
+      "step": 6589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5342297333238187e-05,
+      "loss": 0.7312,
+      "step": 6590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340880576347546e-05,
+      "loss": 0.7137,
+      "step": 6591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533946366945548e-05,
+      "loss": 0.7538,
+      "step": 6592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5338046612601787e-05,
+      "loss": 0.7086,
+      "step": 6593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336629405826263e-05,
+      "loss": 0.7428,
+      "step": 6594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5335212049168705e-05,
+      "loss": 0.7036,
+      "step": 6595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333794542668926e-05,
+      "loss": 0.6732,
+      "step": 6596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533237688636674e-05,
+      "loss": 0.7448,
+      "step": 6597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5330959080301953e-05,
+      "loss": 0.745,
+      "step": 6598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5329541124514394e-05,
+      "loss": 0.743,
+      "step": 6599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328123019043885e-05,
+      "loss": 0.7606,
+      "step": 6600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532670476393025e-05,
+      "loss": 0.6834,
+      "step": 6601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325286359213326e-05,
+      "loss": 0.7042,
+      "step": 6602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5323867804932954e-05,
+      "loss": 0.7432,
+      "step": 6603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322449101128963e-05,
+      "loss": 0.8234,
+      "step": 6604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321030247841207e-05,
+      "loss": 0.7019,
+      "step": 6605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531961124510953e-05,
+      "loss": 0.7526,
+      "step": 6606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531819209297379e-05,
+      "loss": 0.6992,
+      "step": 6607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5316772791473843e-05,
+      "loss": 0.6898,
+      "step": 6608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315353340649547e-05,
+      "loss": 0.6907,
+      "step": 6609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313933740540773e-05,
+      "loss": 0.7148,
+      "step": 6610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531251399118739e-05,
+      "loss": 0.7519,
+      "step": 6611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531109409262927e-05,
+      "loss": 0.6926,
+      "step": 6612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5309674044906298e-05,
+      "loss": 0.7635,
+      "step": 6613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530825384805835e-05,
+      "loss": 0.7141,
+      "step": 6614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306833502125313e-05,
+      "loss": 0.7283,
+      "step": 6615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5305413007147084e-05,
+      "loss": 0.6822,
+      "step": 6616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303992363163553e-05,
+      "loss": 0.7637,
+      "step": 6617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530257157021462e-05,
+      "loss": 0.7281,
+      "step": 6618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301150628340192e-05,
+      "loss": 0.7926,
+      "step": 6619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299729537580174e-05,
+      "loss": 0.7069,
+      "step": 6620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529830829797448e-05,
+      "loss": 0.6901,
+      "step": 6621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529688690956302e-05,
+      "loss": 0.7021,
+      "step": 6622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295465372385727e-05,
+      "loss": 0.7137,
+      "step": 6623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294043686482515e-05,
+      "loss": 0.7105,
+      "step": 6624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292621851893318e-05,
+      "loss": 0.7101,
+      "step": 6625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291199868658065e-05,
+      "loss": 0.668,
+      "step": 6626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52897777368167e-05,
+      "loss": 0.7086,
+      "step": 6627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5288355456409153e-05,
+      "loss": 0.7922,
+      "step": 6628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528693302747538e-05,
+      "loss": 0.7437,
+      "step": 6629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5285510450055326e-05,
+      "loss": 0.6539,
+      "step": 6630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284087724188944e-05,
+      "loss": 0.7051,
+      "step": 6631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282664849916193e-05,
+      "loss": 0.7704,
+      "step": 6632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528124182727704e-05,
+      "loss": 0.7416,
+      "step": 6633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5279818656311443e-05,
+      "loss": 0.7009,
+      "step": 6634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278395337059376e-05,
+      "loss": 0.6611,
+      "step": 6635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527697186956082e-05,
+      "loss": 0.7775,
+      "step": 6636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527554825385574e-05,
+      "loss": 0.7008,
+      "step": 6637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527412448998413e-05,
+      "loss": 0.6972,
+      "step": 6638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272700577985976e-05,
+      "loss": 0.6811,
+      "step": 6639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527127651790127e-05,
+      "loss": 0.6844,
+      "step": 6640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5269852309769998e-05,
+      "loss": 0.631,
+      "step": 6641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526842795363217e-05,
+      "loss": 0.6854,
+      "step": 6642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5267003449527788e-05,
+      "loss": 0.7219,
+      "step": 6643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5265578797496856e-05,
+      "loss": 0.7276,
+      "step": 6644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526415399757939e-05,
+      "loss": 0.381,
+      "step": 6645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52627290498154e-05,
+      "loss": 0.7766,
+      "step": 6646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5261303954244917e-05,
+      "loss": 0.7036,
+      "step": 6647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525987871090796e-05,
+      "loss": 0.7551,
+      "step": 6648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258453319844555e-05,
+      "loss": 0.7394,
+      "step": 6649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257027781094736e-05,
+      "loss": 0.8019,
+      "step": 6650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525560209469854e-05,
+      "loss": 0.7611,
+      "step": 6651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254176260696012e-05,
+      "loss": 0.6413,
+      "step": 6652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252750279127196e-05,
+      "loss": 0.6603,
+      "step": 6653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5251324150032139e-05,
+      "loss": 0.7681,
+      "step": 6654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5249897873450895e-05,
+      "loss": 0.7897,
+      "step": 6655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524847144942352e-05,
+      "loss": 0.7948,
+      "step": 6656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524704487799008e-05,
+      "loss": 0.7119,
+      "step": 6657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5245618159190643e-05,
+      "loss": 0.7234,
+      "step": 6658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524419129306527e-05,
+      "loss": 0.7949,
+      "step": 6659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242764279654045e-05,
+      "loss": 0.658,
+      "step": 6660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241337118997037e-05,
+      "loss": 0.7754,
+      "step": 6661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239909811134337e-05,
+      "loss": 0.7844,
+      "step": 6662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238482356106028e-05,
+      "loss": 0.7282,
+      "step": 6663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237054753952203e-05,
+      "loss": 0.7623,
+      "step": 6664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5235627004712953e-05,
+      "loss": 0.708,
+      "step": 6665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234199108428379e-05,
+      "loss": 0.7906,
+      "step": 6666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5232771065138584e-05,
+      "loss": 0.7452,
+      "step": 6667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231342874883676e-05,
+      "loss": 0.7615,
+      "step": 6668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5229914537703767e-05,
+      "loss": 0.3622,
+      "step": 6669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5228486053638971e-05,
+      "loss": 0.683,
+      "step": 6670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522705742272941e-05,
+      "loss": 0.7368,
+      "step": 6671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5225628645015205e-05,
+      "loss": 0.8181,
+      "step": 6672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5224199720536486e-05,
+      "loss": 0.7595,
+      "step": 6673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222770649333383e-05,
+      "loss": 0.7477,
+      "step": 6674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221341431446037e-05,
+      "loss": 0.6808,
+      "step": 6675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219912066914582e-05,
+      "loss": 0.8393,
+      "step": 6676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218482555779167e-05,
+      "loss": 0.7642,
+      "step": 6677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217052898079937e-05,
+      "loss": 0.7366,
+      "step": 6678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215623093857051e-05,
+      "loss": 0.6474,
+      "step": 6679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214193143150655e-05,
+      "loss": 0.7451,
+      "step": 6680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212763046000923e-05,
+      "loss": 0.6579,
+      "step": 6681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211332802448006e-05,
+      "loss": 0.7184,
+      "step": 6682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209902412532086e-05,
+      "loss": 0.673,
+      "step": 6683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5208471876293334e-05,
+      "loss": 0.703,
+      "step": 6684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207041193771916e-05,
+      "loss": 0.6483,
+      "step": 6685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205610365008028e-05,
+      "loss": 0.658,
+      "step": 6686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5204179390041845e-05,
+      "loss": 0.6741,
+      "step": 6687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202748268913563e-05,
+      "loss": 0.712,
+      "step": 6688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201317001663374e-05,
+      "loss": 0.8064,
+      "step": 6689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5199885588331476e-05,
+      "loss": 0.752,
+      "step": 6690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198454028958067e-05,
+      "loss": 0.7817,
+      "step": 6691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519702232358336e-05,
+      "loss": 0.815,
+      "step": 6692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195590472247558e-05,
+      "loss": 0.6749,
+      "step": 6693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194158474990885e-05,
+      "loss": 0.7208,
+      "step": 6694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5192726331853552e-05,
+      "loss": 0.7659,
+      "step": 6695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191294042875781e-05,
+      "loss": 0.7156,
+      "step": 6696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189861608097802e-05,
+      "loss": 0.6687,
+      "step": 6697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188429027559841e-05,
+      "loss": 0.7103,
+      "step": 6698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518699630130214e-05,
+      "loss": 0.3782,
+      "step": 6699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185563429364934e-05,
+      "loss": 0.7569,
+      "step": 6700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184130411788462e-05,
+      "loss": 0.7104,
+      "step": 6701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182697248612979e-05,
+      "loss": 0.6731,
+      "step": 6702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181263939878728e-05,
+      "loss": 0.7438,
+      "step": 6703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179830485625973e-05,
+      "loss": 0.7795,
+      "step": 6704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5178396885894966e-05,
+      "loss": 0.7802,
+      "step": 6705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5176963140725974e-05,
+      "loss": 0.7621,
+      "step": 6706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175529250159265e-05,
+      "loss": 0.7443,
+      "step": 6707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174095214235106e-05,
+      "loss": 0.6637,
+      "step": 6708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5172661032993774e-05,
+      "loss": 0.6622,
+      "step": 6709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171226706475555e-05,
+      "loss": 0.7001,
+      "step": 6710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169792234720726e-05,
+      "loss": 0.7475,
+      "step": 6711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5168357617769576e-05,
+      "loss": 0.6995,
+      "step": 6712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166922855662401e-05,
+      "loss": 0.7493,
+      "step": 6713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5165487948439495e-05,
+      "loss": 0.7448,
+      "step": 6714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164052896141153e-05,
+      "loss": 0.8232,
+      "step": 6715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162617698807686e-05,
+      "loss": 0.8255,
+      "step": 6716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51611823564794e-05,
+      "loss": 0.7114,
+      "step": 6717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159746869196607e-05,
+      "loss": 0.7149,
+      "step": 6718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158311236999622e-05,
+      "loss": 0.6615,
+      "step": 6719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515687545992877e-05,
+      "loss": 0.7678,
+      "step": 6720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155439538024368e-05,
+      "loss": 0.7125,
+      "step": 6721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5154003471326753e-05,
+      "loss": 0.7622,
+      "step": 6722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152567259876252e-05,
+      "loss": 0.7229,
+      "step": 6723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151130903713206e-05,
+      "loss": 0.7235,
+      "step": 6724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514969440287795e-05,
+      "loss": 0.7,
+      "step": 6725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148257757410833e-05,
+      "loss": 0.7935,
+      "step": 6726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146820967352206e-05,
+      "loss": 0.8077,
+      "step": 6727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145384032742416e-05,
+      "loss": 0.6589,
+      "step": 6728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143946953621825e-05,
+      "loss": 0.6881,
+      "step": 6729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142509730030793e-05,
+      "loss": 0.7345,
+      "step": 6730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514107236200968e-05,
+      "loss": 0.7057,
+      "step": 6731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513963484959886e-05,
+      "loss": 0.7025,
+      "step": 6732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138197192838714e-05,
+      "loss": 0.7492,
+      "step": 6733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136759391769604e-05,
+      "loss": 0.715,
+      "step": 6734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135321446431923e-05,
+      "loss": 0.7135,
+      "step": 6735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513388335686605e-05,
+      "loss": 0.7913,
+      "step": 6736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5132445123112379e-05,
+      "loss": 0.7621,
+      "step": 6737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131006745211299e-05,
+      "loss": 0.6659,
+      "step": 6738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129568223203211e-05,
+      "loss": 0.7599,
+      "step": 6739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128129557128514e-05,
+      "loss": 0.7493,
+      "step": 6740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126690747027619e-05,
+      "loss": 0.7042,
+      "step": 6741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125251792940927e-05,
+      "loss": 0.7275,
+      "step": 6742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123812694908861e-05,
+      "loss": 0.7501,
+      "step": 6743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122373452971833e-05,
+      "loss": 0.755,
+      "step": 6744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120934067170266e-05,
+      "loss": 0.6692,
+      "step": 6745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119494537544586e-05,
+      "loss": 0.7369,
+      "step": 6746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118054864135223e-05,
+      "loss": 0.7911,
+      "step": 6747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116615046982612e-05,
+      "loss": 0.7375,
+      "step": 6748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115175086127193e-05,
+      "loss": 0.725,
+      "step": 6749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5113734981609405e-05,
+      "loss": 0.7664,
+      "step": 6750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511229473346969e-05,
+      "loss": 0.7529,
+      "step": 6751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110854341748504e-05,
+      "loss": 0.756,
+      "step": 6752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5109413806486302e-05,
+      "loss": 0.6716,
+      "step": 6753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107973127723537e-05,
+      "loss": 0.6868,
+      "step": 6754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106532305500677e-05,
+      "loss": 0.6638,
+      "step": 6755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105091339858185e-05,
+      "loss": 0.8753,
+      "step": 6756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103650230836528e-05,
+      "loss": 0.7207,
+      "step": 6757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102208978476185e-05,
+      "loss": 0.6801,
+      "step": 6758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100767582817639e-05,
+      "loss": 0.6896,
+      "step": 6759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099326043901361e-05,
+      "loss": 0.7263,
+      "step": 6760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5097884361767846e-05,
+      "loss": 0.7294,
+      "step": 6761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096442536457578e-05,
+      "loss": 0.7873,
+      "step": 6762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095000568011057e-05,
+      "loss": 0.7008,
+      "step": 6763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093558456468778e-05,
+      "loss": 0.6749,
+      "step": 6764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5092116201871246e-05,
+      "loss": 0.7343,
+      "step": 6765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5090673804258965e-05,
+      "loss": 0.6505,
+      "step": 6766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089231263672447e-05,
+      "loss": 0.8213,
+      "step": 6767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087788580152207e-05,
+      "loss": 0.6935,
+      "step": 6768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086345753738763e-05,
+      "loss": 0.7712,
+      "step": 6769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5084902784472638e-05,
+      "loss": 0.7555,
+      "step": 6770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083459672394354e-05,
+      "loss": 0.7576,
+      "step": 6771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082016417544447e-05,
+      "loss": 0.6554,
+      "step": 6772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080573019963449e-05,
+      "loss": 0.7928,
+      "step": 6773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50791294796919e-05,
+      "loss": 0.7332,
+      "step": 6774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077685796770348e-05,
+      "loss": 0.7361,
+      "step": 6775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076241971239326e-05,
+      "loss": 0.7021,
+      "step": 6776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074798003139393e-05,
+      "loss": 0.728,
+      "step": 6777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073353892511105e-05,
+      "loss": 0.7385,
+      "step": 6778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071909639395021e-05,
+      "loss": 0.6985,
+      "step": 6779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070465243831698e-05,
+      "loss": 0.6751,
+      "step": 6780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5069020705861705e-05,
+      "loss": 0.6939,
+      "step": 6781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5067576025525615e-05,
+      "loss": 0.7599,
+      "step": 6782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066131202863999e-05,
+      "loss": 0.6854,
+      "step": 6783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064686237917438e-05,
+      "loss": 0.7646,
+      "step": 6784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5063241130726515e-05,
+      "loss": 0.7002,
+      "step": 6785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061795881331813e-05,
+      "loss": 0.6909,
+      "step": 6786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060350489773927e-05,
+      "loss": 0.714,
+      "step": 6787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5058904956093452e-05,
+      "loss": 0.7336,
+      "step": 6788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057459280330984e-05,
+      "loss": 0.7119,
+      "step": 6789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056013462527123e-05,
+      "loss": 0.7518,
+      "step": 6790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054567502722482e-05,
+      "loss": 0.7115,
+      "step": 6791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5053121400957666e-05,
+      "loss": 0.6595,
+      "step": 6792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051675157273292e-05,
+      "loss": 0.7242,
+      "step": 6793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050228771709975e-05,
+      "loss": 0.7616,
+      "step": 6794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048782244308346e-05,
+      "loss": 0.7047,
+      "step": 6795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047335575109023e-05,
+      "loss": 0.6842,
+      "step": 6796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045888764152637e-05,
+      "loss": 0.708,
+      "step": 6797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504444181147983e-05,
+      "loss": 0.7426,
+      "step": 6798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042994717131234e-05,
+      "loss": 0.7186,
+      "step": 6799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5041547481147491e-05,
+      "loss": 0.7581,
+      "step": 6800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040100103569248e-05,
+      "loss": 0.7374,
+      "step": 6801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5038652584437161e-05,
+      "loss": 0.6708,
+      "step": 6802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037204923791878e-05,
+      "loss": 0.6618,
+      "step": 6803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5035757121674057e-05,
+      "loss": 0.7747,
+      "step": 6804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034309178124367e-05,
+      "loss": 0.7848,
+      "step": 6805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5032861093183466e-05,
+      "loss": 0.6346,
+      "step": 6806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5031412866892025e-05,
+      "loss": 0.7107,
+      "step": 6807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029964499290728e-05,
+      "loss": 0.7378,
+      "step": 6808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5028515990420243e-05,
+      "loss": 0.7054,
+      "step": 6809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027067340321254e-05,
+      "loss": 0.7469,
+      "step": 6810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502561854903445e-05,
+      "loss": 0.6928,
+      "step": 6811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024169616600518e-05,
+      "loss": 0.7835,
+      "step": 6812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022720543060154e-05,
+      "loss": 0.7936,
+      "step": 6813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021271328454056e-05,
+      "loss": 0.7337,
+      "step": 6814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019821972822927e-05,
+      "loss": 0.752,
+      "step": 6815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018372476207473e-05,
+      "loss": 0.7495,
+      "step": 6816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016922838648395e-05,
+      "loss": 0.7693,
+      "step": 6817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5015473060186422e-05,
+      "loss": 0.6882,
+      "step": 6818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014023140862262e-05,
+      "loss": 0.7615,
+      "step": 6819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012573080716636e-05,
+      "loss": 0.6645,
+      "step": 6820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011122879790274e-05,
+      "loss": 0.69,
+      "step": 6821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009672538123902e-05,
+      "loss": 0.7074,
+      "step": 6822
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008222055758258e-05,
+      "loss": 0.736,
+      "step": 6823
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500677143273408e-05,
+      "loss": 0.7577,
+      "step": 6824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5005320669092101e-05,
+      "loss": 0.7237,
+      "step": 6825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003869764873073e-05,
+      "loss": 0.7609,
+      "step": 6826
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002418720117743e-05,
+      "loss": 0.7339,
+      "step": 6827
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000967534866873e-05,
+      "loss": 0.6699,
+      "step": 6828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499951620916121e-05,
+      "loss": 0.7066,
+      "step": 6829
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998064743041516e-05,
+      "loss": 0.7275,
+      "step": 6830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996613136548559e-05,
+      "loss": 0.768,
+      "step": 6831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499516138972311e-05,
+      "loss": 0.7584,
+      "step": 6832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993709502605937e-05,
+      "loss": 0.7558,
+      "step": 6833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992257475237824e-05,
+      "loss": 0.6758,
+      "step": 6834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990805307659543e-05,
+      "loss": 0.7361,
+      "step": 6835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989352999911885e-05,
+      "loss": 0.6988,
+      "step": 6836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987900552035638e-05,
+      "loss": 0.7794,
+      "step": 6837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4986447964071596e-05,
+      "loss": 0.7883,
+      "step": 6838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4984995236060552e-05,
+      "loss": 0.7687,
+      "step": 6839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498354236804331e-05,
+      "loss": 0.7099,
+      "step": 6840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.498208936006067e-05,
+      "loss": 0.7458,
+      "step": 6841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980636212153446e-05,
+      "loss": 0.7137,
+      "step": 6842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4979182924362445e-05,
+      "loss": 0.7523,
+      "step": 6843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977729496728492e-05,
+      "loss": 0.6976,
+      "step": 6844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976275929292397e-05,
+      "loss": 0.727,
+      "step": 6845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974822222094991e-05,
+      "loss": 0.7864,
+      "step": 6846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49733683751771e-05,
+      "loss": 0.7059,
+      "step": 6847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4971914388579554e-05,
+      "loss": 0.6723,
+      "step": 6848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497046026234319e-05,
+      "loss": 0.7182,
+      "step": 6849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4969005996508857e-05,
+      "loss": 0.7535,
+      "step": 6850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967551591117383e-05,
+      "loss": 0.7709,
+      "step": 6851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966097046209624e-05,
+      "loss": 0.7365,
+      "step": 6852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964642361826431e-05,
+      "loss": 0.7223,
+      "step": 6853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963187538008664e-05,
+      "loss": 0.7091,
+      "step": 6854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961732574797175e-05,
+      "loss": 0.7123,
+      "step": 6855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.496027747223283e-05,
+      "loss": 0.7231,
+      "step": 6856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958822230356499e-05,
+      "loss": 0.6881,
+      "step": 6857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957366849209047e-05,
+      "loss": 0.6884,
+      "step": 6858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4955911328831357e-05,
+      "loss": 0.6927,
+      "step": 6859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954455669264305e-05,
+      "loss": 0.7482,
+      "step": 6860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952999870548772e-05,
+      "loss": 0.7356,
+      "step": 6861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951543932725643e-05,
+      "loss": 0.7881,
+      "step": 6862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950087855835816e-05,
+      "loss": 0.7231,
+      "step": 6863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4948631639920179e-05,
+      "loss": 0.6882,
+      "step": 6864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947175285019636e-05,
+      "loss": 0.7433,
+      "step": 6865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945718791175084e-05,
+      "loss": 0.7083,
+      "step": 6866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4944262158427431e-05,
+      "loss": 0.7102,
+      "step": 6867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942805386817591e-05,
+      "loss": 0.7629,
+      "step": 6868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941348476386476e-05,
+      "loss": 0.7283,
+      "step": 6869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939891427175003e-05,
+      "loss": 0.6317,
+      "step": 6870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938434239224095e-05,
+      "loss": 0.7269,
+      "step": 6871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936976912574675e-05,
+      "loss": 0.7213,
+      "step": 6872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935519447267679e-05,
+      "loss": 0.7608,
+      "step": 6873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934061843344034e-05,
+      "loss": 0.6877,
+      "step": 6874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493260410084468e-05,
+      "loss": 0.6901,
+      "step": 6875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931146219810562e-05,
+      "loss": 0.72,
+      "step": 6876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929688200282622e-05,
+      "loss": 0.7809,
+      "step": 6877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4928230042301807e-05,
+      "loss": 0.7198,
+      "step": 6878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926771745909072e-05,
+      "loss": 0.7208,
+      "step": 6879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492531331114538e-05,
+      "loss": 0.7487,
+      "step": 6880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4923854738051685e-05,
+      "loss": 0.7244,
+      "step": 6881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922396026668948e-05,
+      "loss": 0.7961,
+      "step": 6882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4920937177038149e-05,
+      "loss": 0.6961,
+      "step": 6883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4919478189200252e-05,
+      "loss": 0.7898,
+      "step": 6884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918019063196236e-05,
+      "loss": 0.7434,
+      "step": 6885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916559799067082e-05,
+      "loss": 0.7339,
+      "step": 6886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915100396853774e-05,
+      "loss": 0.7148,
+      "step": 6887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49136408565973e-05,
+      "loss": 0.7075,
+      "step": 6888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491218117833865e-05,
+      "loss": 0.7207,
+      "step": 6889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910721362118825e-05,
+      "loss": 0.7416,
+      "step": 6890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490926140797882e-05,
+      "loss": 0.6359,
+      "step": 6891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907801315959633e-05,
+      "loss": 0.7043,
+      "step": 6892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4906341086102285e-05,
+      "loss": 0.6888,
+      "step": 6893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904880718447781e-05,
+      "loss": 0.7166,
+      "step": 6894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903420213037131e-05,
+      "loss": 0.7596,
+      "step": 6895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901959569911362e-05,
+      "loss": 0.7978,
+      "step": 6896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900498789111493e-05,
+      "loss": 0.7345,
+      "step": 6897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899037870678552e-05,
+      "loss": 0.7035,
+      "step": 6898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897576814653572e-05,
+      "loss": 0.7402,
+      "step": 6899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489611562107758e-05,
+      "loss": 0.7556,
+      "step": 6900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894654289991622e-05,
+      "loss": 0.7283,
+      "step": 6901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893192821436735e-05,
+      "loss": 0.7126,
+      "step": 6902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891731215453967e-05,
+      "loss": 0.764,
+      "step": 6903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489026947208437e-05,
+      "loss": 0.7557,
+      "step": 6904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888807591368994e-05,
+      "loss": 0.8218,
+      "step": 6905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887345573348903e-05,
+      "loss": 0.7662,
+      "step": 6906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488588341806515e-05,
+      "loss": 0.6489,
+      "step": 6907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884421125558806e-05,
+      "loss": 0.7308,
+      "step": 6908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488295869587094e-05,
+      "loss": 0.7413,
+      "step": 6909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881496129042626e-05,
+      "loss": 0.7862,
+      "step": 6910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880033425114937e-05,
+      "loss": 0.6979,
+      "step": 6911
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878570584128955e-05,
+      "loss": 0.7985,
+      "step": 6912
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877107606125767e-05,
+      "loss": 0.7653,
+      "step": 6913
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487564449114646e-05,
+      "loss": 0.6978,
+      "step": 6914
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874181239232127e-05,
+      "loss": 0.7173,
+      "step": 6915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872717850423861e-05,
+      "loss": 0.6971,
+      "step": 6916
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871254324762764e-05,
+      "loss": 0.715,
+      "step": 6917
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869790662289944e-05,
+      "loss": 0.7204,
+      "step": 6918
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868326863046507e-05,
+      "loss": 0.7589,
+      "step": 6919
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486686292707356e-05,
+      "loss": 0.6823,
+      "step": 6920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865398854412222e-05,
+      "loss": 0.7137,
+      "step": 6921
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863934645103608e-05,
+      "loss": 0.7199,
+      "step": 6922
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4862470299188849e-05,
+      "loss": 0.6588,
+      "step": 6923
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861005816709064e-05,
+      "loss": 0.6115,
+      "step": 6924
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859541197705392e-05,
+      "loss": 0.7694,
+      "step": 6925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858076442218958e-05,
+      "loss": 0.7232,
+      "step": 6926
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856611550290907e-05,
+      "loss": 0.7485,
+      "step": 6927
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4855146521962382e-05,
+      "loss": 0.6496,
+      "step": 6928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853681357274522e-05,
+      "loss": 0.6946,
+      "step": 6929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4852216056268485e-05,
+      "loss": 0.7833,
+      "step": 6930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4850750618985422e-05,
+      "loss": 0.7826,
+      "step": 6931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849285045466487e-05,
+      "loss": 0.8047,
+      "step": 6932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847819335752844e-05,
+      "loss": 0.8298,
+      "step": 6933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4846353489885661e-05,
+      "loss": 0.7206,
+      "step": 6934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844887507906106e-05,
+      "loss": 0.6653,
+      "step": 6935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4843421389855345e-05,
+      "loss": 0.7632,
+      "step": 6936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4841955135774562e-05,
+      "loss": 0.6352,
+      "step": 6937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840488745704937e-05,
+      "loss": 0.6519,
+      "step": 6938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839022219687653e-05,
+      "loss": 0.8014,
+      "step": 6939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837555557763898e-05,
+      "loss": 0.6569,
+      "step": 6940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836088759974863e-05,
+      "loss": 0.7623,
+      "step": 6941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834621826361746e-05,
+      "loss": 0.7233,
+      "step": 6942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833154756965745e-05,
+      "loss": 0.7195,
+      "step": 6943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831687551828062e-05,
+      "loss": 0.684,
+      "step": 6944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830220210989909e-05,
+      "loss": 0.7756,
+      "step": 6945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828752734492494e-05,
+      "loss": 0.7033,
+      "step": 6946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4827285122377028e-05,
+      "loss": 0.7693,
+      "step": 6947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825817374684737e-05,
+      "loss": 0.6972,
+      "step": 6948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482434949145684e-05,
+      "loss": 0.7832,
+      "step": 6949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4822881472734563e-05,
+      "loss": 0.7346,
+      "step": 6950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821413318559134e-05,
+      "loss": 0.7344,
+      "step": 6951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4819945028971795e-05,
+      "loss": 0.7478,
+      "step": 6952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818476604013772e-05,
+      "loss": 0.3496,
+      "step": 6953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817008043726318e-05,
+      "loss": 0.7952,
+      "step": 6954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481553934815067e-05,
+      "loss": 0.7277,
+      "step": 6955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814070517328079e-05,
+      "loss": 0.8205,
+      "step": 6956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4812601551299795e-05,
+      "loss": 0.7654,
+      "step": 6957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811132450107085e-05,
+      "loss": 0.7927,
+      "step": 6958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809663213791201e-05,
+      "loss": 0.7097,
+      "step": 6959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808193842393408e-05,
+      "loss": 0.3748,
+      "step": 6960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806724335954975e-05,
+      "loss": 0.7333,
+      "step": 6961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805254694517175e-05,
+      "loss": 0.7284,
+      "step": 6962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803784918121282e-05,
+      "loss": 0.703,
+      "step": 6963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4802315006808578e-05,
+      "loss": 0.7066,
+      "step": 6964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800844960620344e-05,
+      "loss": 0.7704,
+      "step": 6965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799374779597866e-05,
+      "loss": 0.7213,
+      "step": 6966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797904463782435e-05,
+      "loss": 0.8313,
+      "step": 6967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796434013215348e-05,
+      "loss": 0.7416,
+      "step": 6968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794963427937903e-05,
+      "loss": 0.7575,
+      "step": 6969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793492707991401e-05,
+      "loss": 0.7513,
+      "step": 6970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4792021853417149e-05,
+      "loss": 0.729,
+      "step": 6971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790550864256453e-05,
+      "loss": 0.7274,
+      "step": 6972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4789079740550631e-05,
+      "loss": 0.771,
+      "step": 6973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787608482341003e-05,
+      "loss": 0.6904,
+      "step": 6974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786137089668882e-05,
+      "loss": 0.7154,
+      "step": 6975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4784665562575596e-05,
+      "loss": 0.6868,
+      "step": 6976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783193901102472e-05,
+      "loss": 0.7493,
+      "step": 6977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4781722105290849e-05,
+      "loss": 0.8024,
+      "step": 6978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780250175182058e-05,
+      "loss": 0.6942,
+      "step": 6979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778778110817437e-05,
+      "loss": 0.7413,
+      "step": 6980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777305912238338e-05,
+      "loss": 0.6593,
+      "step": 6981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775833579486097e-05,
+      "loss": 0.7228,
+      "step": 6982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774361112602072e-05,
+      "loss": 0.7006,
+      "step": 6983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772888511627622e-05,
+      "loss": 0.7269,
+      "step": 6984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771415776604094e-05,
+      "loss": 0.7504,
+      "step": 6985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769942907572863e-05,
+      "loss": 0.6933,
+      "step": 6986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768469904575286e-05,
+      "loss": 0.7269,
+      "step": 6987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766996767652736e-05,
+      "loss": 0.6975,
+      "step": 6988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4765523496846592e-05,
+      "loss": 0.7101,
+      "step": 6989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764050092198225e-05,
+      "loss": 0.7041,
+      "step": 6990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476257655374902e-05,
+      "loss": 0.7512,
+      "step": 6991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761102881540357e-05,
+      "loss": 0.7678,
+      "step": 6992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4759629075613631e-05,
+      "loss": 0.6953,
+      "step": 6993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4758155136010232e-05,
+      "loss": 0.7017,
+      "step": 6994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756681062771557e-05,
+      "loss": 0.7559,
+      "step": 6995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755206855939004e-05,
+      "loss": 0.7482,
+      "step": 6996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753732515553978e-05,
+      "loss": 0.6917,
+      "step": 6997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.475225804165789e-05,
+      "loss": 0.7141,
+      "step": 6998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750783434292147e-05,
+      "loss": 0.7653,
+      "step": 6999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4749308693498166e-05,
+      "loss": 0.6964,
+      "step": 7000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747833819317363e-05,
+      "loss": 0.6898,
+      "step": 7001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4746358811791164e-05,
+      "loss": 0.7527,
+      "step": 7002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744883670960996e-05,
+      "loss": 0.712,
+      "step": 7003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743408396868286e-05,
+      "loss": 0.7325,
+      "step": 7004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741932989554468e-05,
+      "loss": 0.7202,
+      "step": 7005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740457449060985e-05,
+      "loss": 0.7424,
+      "step": 7006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738981775429272e-05,
+      "loss": 0.7197,
+      "step": 7007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4737505968700774e-05,
+      "loss": 0.6584,
+      "step": 7008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736030028916945e-05,
+      "loss": 0.6725,
+      "step": 7009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734553956119235e-05,
+      "loss": 0.6978,
+      "step": 7010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733077750349098e-05,
+      "loss": 0.6707,
+      "step": 7011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4731601411647996e-05,
+      "loss": 0.7354,
+      "step": 7012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4730124940057396e-05,
+      "loss": 0.7108,
+      "step": 7013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4728648335618758e-05,
+      "loss": 0.7122,
+      "step": 7014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472717159837356e-05,
+      "loss": 0.734,
+      "step": 7015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725694728363273e-05,
+      "loss": 0.6948,
+      "step": 7016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4724217725629377e-05,
+      "loss": 0.7041,
+      "step": 7017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722740590213355e-05,
+      "loss": 0.7538,
+      "step": 7018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472126332215669e-05,
+      "loss": 0.7145,
+      "step": 7019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471978592150088e-05,
+      "loss": 0.7336,
+      "step": 7020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471830838828741e-05,
+      "loss": 0.7173,
+      "step": 7021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716830722557776e-05,
+      "loss": 0.7151,
+      "step": 7022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471535292435349e-05,
+      "loss": 0.6956,
+      "step": 7023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713874993716047e-05,
+      "loss": 0.7604,
+      "step": 7024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712396930686957e-05,
+      "loss": 0.7368,
+      "step": 7025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710918735307735e-05,
+      "loss": 0.7139,
+      "step": 7026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709440407619894e-05,
+      "loss": 0.7331,
+      "step": 7027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707961947664958e-05,
+      "loss": 0.6954,
+      "step": 7028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4706483355484447e-05,
+      "loss": 0.7761,
+      "step": 7029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705004631119885e-05,
+      "loss": 0.8064,
+      "step": 7030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703525774612808e-05,
+      "loss": 0.7308,
+      "step": 7031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4702046786004749e-05,
+      "loss": 0.7781,
+      "step": 7032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700567665337247e-05,
+      "loss": 0.6686,
+      "step": 7033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699088412651841e-05,
+      "loss": 0.7477,
+      "step": 7034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697609027990078e-05,
+      "loss": 0.7418,
+      "step": 7035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469612951139351e-05,
+      "loss": 0.662,
+      "step": 7036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694649862903686e-05,
+      "loss": 0.7331,
+      "step": 7037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693170082562163e-05,
+      "loss": 0.7521,
+      "step": 7038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4691690170410504e-05,
+      "loss": 0.7327,
+      "step": 7039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690210126490272e-05,
+      "loss": 0.7591,
+      "step": 7040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688729950843035e-05,
+      "loss": 0.6693,
+      "step": 7041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4687249643510362e-05,
+      "loss": 0.684,
+      "step": 7042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4685769204533831e-05,
+      "loss": 0.3394,
+      "step": 7043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684288633955021e-05,
+      "loss": 0.6872,
+      "step": 7044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4682807931815513e-05,
+      "loss": 0.741,
+      "step": 7045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4681327098156897e-05,
+      "loss": 0.6689,
+      "step": 7046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679846133020755e-05,
+      "loss": 0.7292,
+      "step": 7047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467836503644869e-05,
+      "loss": 0.701,
+      "step": 7048
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4676883808482294e-05,
+      "loss": 0.6525,
+      "step": 7049
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467540244916317e-05,
+      "loss": 0.7009,
+      "step": 7050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4673920958532919e-05,
+      "loss": 0.7398,
+      "step": 7051
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672439336633155e-05,
+      "loss": 0.7416,
+      "step": 7052
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670957583505488e-05,
+      "loss": 0.7445,
+      "step": 7053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4669475699191531e-05,
+      "loss": 0.7225,
+      "step": 7054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4667993683732908e-05,
+      "loss": 0.7444,
+      "step": 7055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666511537171237e-05,
+      "loss": 0.7229,
+      "step": 7056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466502925954815e-05,
+      "loss": 0.7592,
+      "step": 7057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663546850905276e-05,
+      "loss": 0.7077,
+      "step": 7058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662064311284247e-05,
+      "loss": 0.7139,
+      "step": 7059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660581640726704e-05,
+      "loss": 0.6701,
+      "step": 7060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659098839274286e-05,
+      "loss": 0.6635,
+      "step": 7061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657615906968639e-05,
+      "loss": 0.8149,
+      "step": 7062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656132843851415e-05,
+      "loss": 0.7383,
+      "step": 7063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654649649964258e-05,
+      "loss": 0.726,
+      "step": 7064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653166325348837e-05,
+      "loss": 0.7343,
+      "step": 7065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651682870046803e-05,
+      "loss": 0.7204,
+      "step": 7066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650199284099817e-05,
+      "loss": 0.7321,
+      "step": 7067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648715567549558e-05,
+      "loss": 0.7286,
+      "step": 7068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647231720437687e-05,
+      "loss": 0.7175,
+      "step": 7069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645747742805883e-05,
+      "loss": 0.8383,
+      "step": 7070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4644263634695821e-05,
+      "loss": 0.7784,
+      "step": 7071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642779396149185e-05,
+      "loss": 0.72,
+      "step": 7072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464129502720766e-05,
+      "loss": 0.7487,
+      "step": 7073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639810527912936e-05,
+      "loss": 0.7757,
+      "step": 7074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463832589830671e-05,
+      "loss": 0.662,
+      "step": 7075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636841138430668e-05,
+      "loss": 0.7724,
+      "step": 7076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4635356248326518e-05,
+      "loss": 0.7128,
+      "step": 7077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633871228035965e-05,
+      "loss": 0.7243,
+      "step": 7078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632386077600712e-05,
+      "loss": 0.6925,
+      "step": 7079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630900797062473e-05,
+      "loss": 0.752,
+      "step": 7080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629415386462963e-05,
+      "loss": 0.7419,
+      "step": 7081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627929845843896e-05,
+      "loss": 0.6464,
+      "step": 7082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626444175247002e-05,
+      "loss": 0.6847,
+      "step": 7083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624958374714e-05,
+      "loss": 0.6901,
+      "step": 7084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4623472444286626e-05,
+      "loss": 0.6769,
+      "step": 7085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621986384006605e-05,
+      "loss": 0.7296,
+      "step": 7086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462050019391568e-05,
+      "loss": 0.6887,
+      "step": 7087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461901387405559e-05,
+      "loss": 0.7113,
+      "step": 7088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617527424468077e-05,
+      "loss": 0.7603,
+      "step": 7089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461604084519489e-05,
+      "loss": 0.7583,
+      "step": 7090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614554136277784e-05,
+      "loss": 0.7629,
+      "step": 7091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4613067297758505e-05,
+      "loss": 0.7882,
+      "step": 7092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611580329678822e-05,
+      "loss": 0.7333,
+      "step": 7093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610093232080494e-05,
+      "loss": 0.8083,
+      "step": 7094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608606005005281e-05,
+      "loss": 0.7677,
+      "step": 7095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460711864849496e-05,
+      "loss": 0.657,
+      "step": 7096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4605631162591299e-05,
+      "loss": 0.6661,
+      "step": 7097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604143547336077e-05,
+      "loss": 0.766,
+      "step": 7098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602655802771075e-05,
+      "loss": 0.6734,
+      "step": 7099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4601167928938079e-05,
+      "loss": 0.789,
+      "step": 7100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4599679925878873e-05,
+      "loss": 0.6877,
+      "step": 7101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598191793635247e-05,
+      "loss": 0.7584,
+      "step": 7102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596703532249e-05,
+      "loss": 0.7147,
+      "step": 7103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595215141761934e-05,
+      "loss": 0.7263,
+      "step": 7104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593726622215841e-05,
+      "loss": 0.6704,
+      "step": 7105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4592237973652532e-05,
+      "loss": 0.7066,
+      "step": 7106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4590749196113821e-05,
+      "loss": 0.7659,
+      "step": 7107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589260289641512e-05,
+      "loss": 0.7646,
+      "step": 7108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587771254277429e-05,
+      "loss": 0.7202,
+      "step": 7109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458628209006339e-05,
+      "loss": 0.6845,
+      "step": 7110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584792797041218e-05,
+      "loss": 0.6688,
+      "step": 7111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583303375252743e-05,
+      "loss": 0.7858,
+      "step": 7112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581813824739793e-05,
+      "loss": 0.8067,
+      "step": 7113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4580324145544205e-05,
+      "loss": 0.7137,
+      "step": 7114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578834337707818e-05,
+      "loss": 0.7075,
+      "step": 7115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457734440127247e-05,
+      "loss": 0.7781,
+      "step": 7116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575854336280011e-05,
+      "loss": 0.6793,
+      "step": 7117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4574364142772291e-05,
+      "loss": 0.7606,
+      "step": 7118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572873820791159e-05,
+      "loss": 0.7856,
+      "step": 7119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571383370378474e-05,
+      "loss": 0.6678,
+      "step": 7120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569892791576093e-05,
+      "loss": 0.7535,
+      "step": 7121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568402084425881e-05,
+      "loss": 0.7131,
+      "step": 7122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566911248969708e-05,
+      "loss": 0.6685,
+      "step": 7123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565420285249445e-05,
+      "loss": 0.7255,
+      "step": 7124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4563929193306963e-05,
+      "loss": 0.7387,
+      "step": 7125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456243797318414e-05,
+      "loss": 0.6575,
+      "step": 7126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560946624922862e-05,
+      "loss": 0.7671,
+      "step": 7127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559455148565009e-05,
+      "loss": 0.7205,
+      "step": 7128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557963544152474e-05,
+      "loss": 0.7452,
+      "step": 7129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556471811727151e-05,
+      "loss": 0.6856,
+      "step": 7130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455497995133093e-05,
+      "loss": 0.7682,
+      "step": 7131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553487963005712e-05,
+      "loss": 0.7281,
+      "step": 7132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551995846793405e-05,
+      "loss": 0.7874,
+      "step": 7133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4550503602735913e-05,
+      "loss": 0.7675,
+      "step": 7134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549011230875145e-05,
+      "loss": 0.6658,
+      "step": 7135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547518731253017e-05,
+      "loss": 0.7036,
+      "step": 7136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546026103911444e-05,
+      "loss": 0.6979,
+      "step": 7137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454453334889235e-05,
+      "loss": 0.7014,
+      "step": 7138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543040466237662e-05,
+      "loss": 0.7308,
+      "step": 7139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4541547455989306e-05,
+      "loss": 0.7324,
+      "step": 7140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540054318189212e-05,
+      "loss": 0.6998,
+      "step": 7141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538561052879315e-05,
+      "loss": 0.7272,
+      "step": 7142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4537067660101558e-05,
+      "loss": 0.7081,
+      "step": 7143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535574139897882e-05,
+      "loss": 0.7105,
+      "step": 7144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453408049231023e-05,
+      "loss": 0.802,
+      "step": 7145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4532586717380559e-05,
+      "loss": 0.7635,
+      "step": 7146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531092815150815e-05,
+      "loss": 0.6523,
+      "step": 7147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529598785662962e-05,
+      "loss": 0.7214,
+      "step": 7148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528104628958955e-05,
+      "loss": 0.6701,
+      "step": 7149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526610345080763e-05,
+      "loss": 0.6923,
+      "step": 7150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525115934070348e-05,
+      "loss": 0.7961,
+      "step": 7151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4523621395969685e-05,
+      "loss": 0.7292,
+      "step": 7152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522126730820749e-05,
+      "loss": 0.7342,
+      "step": 7153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4520631938665516e-05,
+      "loss": 0.6811,
+      "step": 7154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519137019545974e-05,
+      "loss": 0.7594,
+      "step": 7155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517641973504103e-05,
+      "loss": 0.7467,
+      "step": 7156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451614680058189e-05,
+      "loss": 0.6339,
+      "step": 7157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514651500821333e-05,
+      "loss": 0.7574,
+      "step": 7158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513156074264427e-05,
+      "loss": 0.7035,
+      "step": 7159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451166052095317e-05,
+      "loss": 0.7145,
+      "step": 7160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510164840929569e-05,
+      "loss": 0.7828,
+      "step": 7161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508669034235627e-05,
+      "loss": 0.8104,
+      "step": 7162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507173100913354e-05,
+      "loss": 0.719,
+      "step": 7163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505677041004771e-05,
+      "loss": 0.7415,
+      "step": 7164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4504180854551889e-05,
+      "loss": 0.6829,
+      "step": 7165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450268454159673e-05,
+      "loss": 0.6924,
+      "step": 7166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4501188102181318e-05,
+      "loss": 0.7382,
+      "step": 7167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499691536347684e-05,
+      "loss": 0.7447,
+      "step": 7168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498194844137863e-05,
+      "loss": 0.6753,
+      "step": 7169
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496698025593882e-05,
+      "loss": 0.6877,
+      "step": 7170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495201080757786e-05,
+      "loss": 0.7295,
+      "step": 7171
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493704009671614e-05,
+      "loss": 0.7447,
+      "step": 7172
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4492206812377414e-05,
+      "loss": 0.7087,
+      "step": 7173
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490709488917239e-05,
+      "loss": 0.7026,
+      "step": 7174
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489212039333133e-05,
+      "loss": 0.7135,
+      "step": 7175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487714463667164e-05,
+      "loss": 0.7332,
+      "step": 7176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.448621676196138e-05,
+      "loss": 0.7643,
+      "step": 7177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484718934257854e-05,
+      "loss": 0.7534,
+      "step": 7178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4483220980598649e-05,
+      "loss": 0.7503,
+      "step": 7179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481722901025838e-05,
+      "loss": 0.7704,
+      "step": 7180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4480224695581496e-05,
+      "loss": 0.7104,
+      "step": 7181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478726364307695e-05,
+      "loss": 0.7106,
+      "step": 7182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477227907246525e-05,
+      "loss": 0.7279,
+      "step": 7183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475729324440063e-05,
+      "loss": 0.6572,
+      "step": 7184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4474230615930404e-05,
+      "loss": 0.6765,
+      "step": 7185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472731781759639e-05,
+      "loss": 0.7936,
+      "step": 7186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471232821969856e-05,
+      "loss": 0.7354,
+      "step": 7187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469733736603161e-05,
+      "loss": 0.7365,
+      "step": 7188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446823452570166e-05,
+      "loss": 0.729,
+      "step": 7189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466735189307453e-05,
+      "loss": 0.6981,
+      "step": 7190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465235727462648e-05,
+      "loss": 0.7726,
+      "step": 7191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4463736140209364e-05,
+      "loss": 0.6708,
+      "step": 7192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462236427589713e-05,
+      "loss": 0.7617,
+      "step": 7193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460736589645819e-05,
+      "loss": 0.8124,
+      "step": 7194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4459236626419803e-05,
+      "loss": 0.7587,
+      "step": 7195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457736537953793e-05,
+      "loss": 0.7564,
+      "step": 7196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456236324289919e-05,
+      "loss": 0.7436,
+      "step": 7197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454735985470316e-05,
+      "loss": 0.6934,
+      "step": 7198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453235521537125e-05,
+      "loss": 0.7143,
+      "step": 7199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4451734932532484e-05,
+      "loss": 0.7621,
+      "step": 7200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450234218498536e-05,
+      "loss": 0.7083,
+      "step": 7201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448733379477433e-05,
+      "loss": 0.7429,
+      "step": 7202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447232415511323e-05,
+      "loss": 0.7662,
+      "step": 7203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4445731326642367e-05,
+      "loss": 0.707,
+      "step": 7204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444230112912721e-05,
+      "loss": 0.6396,
+      "step": 7205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442728774364545e-05,
+      "loss": 0.6865,
+      "step": 7206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441227311040004e-05,
+      "loss": 0.3583,
+      "step": 7207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4439725722981274e-05,
+      "loss": 0.7769,
+      "step": 7208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438224010230526e-05,
+      "loss": 0.3618,
+      "step": 7209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4436722172829931e-05,
+      "loss": 0.7472,
+      "step": 7210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435220210821672e-05,
+      "loss": 0.7385,
+      "step": 7211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433718124247936e-05,
+      "loss": 0.7061,
+      "step": 7212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432215913150902e-05,
+      "loss": 0.7069,
+      "step": 7213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430713577572767e-05,
+      "loss": 0.6835,
+      "step": 7214
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4429211117555726e-05,
+      "loss": 0.7275,
+      "step": 7215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442770853314197e-05,
+      "loss": 0.7391,
+      "step": 7216
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4426205824373708e-05,
+      "loss": 0.7624,
+      "step": 7217
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424702991293134e-05,
+      "loss": 0.7051,
+      "step": 7218
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423200033942461e-05,
+      "loss": 0.7535,
+      "step": 7219
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4421696952363903e-05,
+      "loss": 0.7421,
+      "step": 7220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420193746599675e-05,
+      "loss": 0.7334,
+      "step": 7221
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418690416691989e-05,
+      "loss": 0.6935,
+      "step": 7222
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4417186962683072e-05,
+      "loss": 0.7622,
+      "step": 7223
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415683384615147e-05,
+      "loss": 0.3728,
+      "step": 7224
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414179682530447e-05,
+      "loss": 0.702,
+      "step": 7225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4412675856471198e-05,
+      "loss": 0.6808,
+      "step": 7226
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441117190647964e-05,
+      "loss": 0.7346,
+      "step": 7227
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409667832598012e-05,
+      "loss": 0.798,
+      "step": 7228
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4408163634868552e-05,
+      "loss": 0.6917,
+      "step": 7229
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406659313333511e-05,
+      "loss": 0.7465,
+      "step": 7230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440515486803514e-05,
+      "loss": 0.7388,
+      "step": 7231
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403650299015687e-05,
+      "loss": 0.6437,
+      "step": 7232
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4402145606317412e-05,
+      "loss": 0.7102,
+      "step": 7233
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400640789982578e-05,
+      "loss": 0.7558,
+      "step": 7234
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399135850053441e-05,
+      "loss": 0.3413,
+      "step": 7235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397630786572272e-05,
+      "loss": 0.6719,
+      "step": 7236
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396125599581343e-05,
+      "loss": 0.7032,
+      "step": 7237
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394620289122925e-05,
+      "loss": 0.7937,
+      "step": 7238
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393114855239298e-05,
+      "loss": 0.7114,
+      "step": 7239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4391609297972743e-05,
+      "loss": 0.7029,
+      "step": 7240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439010361736554e-05,
+      "loss": 0.7072,
+      "step": 7241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4388597813459981e-05,
+      "loss": 0.811,
+      "step": 7242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387091886298356e-05,
+      "loss": 0.6873,
+      "step": 7243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4385585835922962e-05,
+      "loss": 0.6674,
+      "step": 7244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4384079662376093e-05,
+      "loss": 0.6819,
+      "step": 7245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382573365700054e-05,
+      "loss": 0.697,
+      "step": 7246
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381066945937148e-05,
+      "loss": 0.8138,
+      "step": 7247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379560403129684e-05,
+      "loss": 0.7341,
+      "step": 7248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378053737319975e-05,
+      "loss": 0.7247,
+      "step": 7249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376546948550337e-05,
+      "loss": 0.7468,
+      "step": 7250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375040036863089e-05,
+      "loss": 0.7011,
+      "step": 7251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373533002300547e-05,
+      "loss": 0.7122,
+      "step": 7252
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372025844905049e-05,
+      "loss": 0.7509,
+      "step": 7253
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4370518564718916e-05,
+      "loss": 0.7457,
+      "step": 7254
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369011161784481e-05,
+      "loss": 0.7212,
+      "step": 7255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4367503636144082e-05,
+      "loss": 0.7193,
+      "step": 7256
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365995987840058e-05,
+      "loss": 0.7181,
+      "step": 7257
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364488216914753e-05,
+      "loss": 0.8226,
+      "step": 7258
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362980323410511e-05,
+      "loss": 0.7626,
+      "step": 7259
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361472307369688e-05,
+      "loss": 0.5939,
+      "step": 7260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435996416883463e-05,
+      "loss": 0.7626,
+      "step": 7261
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358455907847695e-05,
+      "loss": 0.749,
+      "step": 7262
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4356947524451247e-05,
+      "loss": 0.7622,
+      "step": 7263
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435543901868765e-05,
+      "loss": 0.7466,
+      "step": 7264
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435393039059927e-05,
+      "loss": 0.6954,
+      "step": 7265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352421640228475e-05,
+      "loss": 0.7076,
+      "step": 7266
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4350912767617642e-05,
+      "loss": 0.7398,
+      "step": 7267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349403772809146e-05,
+      "loss": 0.6709,
+      "step": 7268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4347894655845369e-05,
+      "loss": 0.7437,
+      "step": 7269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346385416768699e-05,
+      "loss": 0.7017,
+      "step": 7270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344876055621518e-05,
+      "loss": 0.7502,
+      "step": 7271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434336657244622e-05,
+      "loss": 0.7912,
+      "step": 7272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341856967285197e-05,
+      "loss": 0.7432,
+      "step": 7273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340347240180852e-05,
+      "loss": 0.8073,
+      "step": 7274
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.7383,
+      "step": 7275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337327420311795e-05,
+      "loss": 0.7593,
+      "step": 7276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4335817327631897e-05,
+      "loss": 0.7275,
+      "step": 7277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334307113178303e-05,
+      "loss": 0.764,
+      "step": 7278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332796776993423e-05,
+      "loss": 0.7782,
+      "step": 7279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4331286319119681e-05,
+      "loss": 0.6856,
+      "step": 7280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329775739599496e-05,
+      "loss": 0.6216,
+      "step": 7281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328265038475295e-05,
+      "loss": 0.6993,
+      "step": 7282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326754215789504e-05,
+      "loss": 0.6439,
+      "step": 7283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4325243271584557e-05,
+      "loss": 0.7384,
+      "step": 7284
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432373220590289e-05,
+      "loss": 0.6559,
+      "step": 7285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4322221018786944e-05,
+      "loss": 0.7052,
+      "step": 7286
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320709710279156e-05,
+      "loss": 0.7,
+      "step": 7287
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4319198280421973e-05,
+      "loss": 0.7557,
+      "step": 7288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431768672925785e-05,
+      "loss": 0.7182,
+      "step": 7289
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316175056829238e-05,
+      "loss": 0.7672,
+      "step": 7290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314663263178587e-05,
+      "loss": 0.6818,
+      "step": 7291
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4313151348348364e-05,
+      "loss": 0.7185,
+      "step": 7292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4311639312381025e-05,
+      "loss": 0.6953,
+      "step": 7293
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310127155319042e-05,
+      "loss": 0.7012,
+      "step": 7294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308614877204883e-05,
+      "loss": 0.7303,
+      "step": 7295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307102478081023e-05,
+      "loss": 0.68,
+      "step": 7296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305589957989932e-05,
+      "loss": 0.8036,
+      "step": 7297
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4304077316974096e-05,
+      "loss": 0.6816,
+      "step": 7298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302564555075997e-05,
+      "loss": 0.7526,
+      "step": 7299
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4301051672338121e-05,
+      "loss": 0.7709,
+      "step": 7300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429953866880296e-05,
+      "loss": 0.8173,
+      "step": 7301
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298025544513003e-05,
+      "loss": 0.6974,
+      "step": 7302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296512299510752e-05,
+      "loss": 0.8162,
+      "step": 7303
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294998933838704e-05,
+      "loss": 0.7289,
+      "step": 7304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4293485447539363e-05,
+      "loss": 0.6994,
+      "step": 7305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429197184065524e-05,
+      "loss": 0.748,
+      "step": 7306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290458113228838e-05,
+      "loss": 0.704,
+      "step": 7307
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288944265302676e-05,
+      "loss": 0.6975,
+      "step": 7308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4287430296919273e-05,
+      "loss": 0.7097,
+      "step": 7309
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285916208121144e-05,
+      "loss": 0.6577,
+      "step": 7310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284401998950816e-05,
+      "loss": 0.6144,
+      "step": 7311
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4282887669450817e-05,
+      "loss": 0.6859,
+      "step": 7312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428137321966368e-05,
+      "loss": 0.7615,
+      "step": 7313
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427985864963193e-05,
+      "loss": 0.6868,
+      "step": 7314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4278343959398113e-05,
+      "loss": 0.813,
+      "step": 7315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276829149004769e-05,
+      "loss": 0.7875,
+      "step": 7316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275314218494438e-05,
+      "loss": 0.7361,
+      "step": 7317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273799167909672e-05,
+      "loss": 0.7028,
+      "step": 7318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272283997293022e-05,
+      "loss": 0.6917,
+      "step": 7319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427076870668704e-05,
+      "loss": 0.6648,
+      "step": 7320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426925329613428e-05,
+      "loss": 0.7177,
+      "step": 7321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426773776567731e-05,
+      "loss": 0.7562,
+      "step": 7322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266222115358692e-05,
+      "loss": 0.7615,
+      "step": 7323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264706345220996e-05,
+      "loss": 0.7584,
+      "step": 7324
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263190455306788e-05,
+      "loss": 0.6934,
+      "step": 7325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4261674445658647e-05,
+      "loss": 0.6983,
+      "step": 7326
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4260158316319146e-05,
+      "loss": 0.7812,
+      "step": 7327
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258642067330872e-05,
+      "loss": 0.7068,
+      "step": 7328
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257125698736409e-05,
+      "loss": 0.6706,
+      "step": 7329
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255609210578343e-05,
+      "loss": 0.7608,
+      "step": 7330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254092602899264e-05,
+      "loss": 0.6106,
+      "step": 7331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4252575875741768e-05,
+      "loss": 0.691,
+      "step": 7332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251059029148455e-05,
+      "loss": 0.709,
+      "step": 7333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249542063161925e-05,
+      "loss": 0.627,
+      "step": 7334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248024977824784e-05,
+      "loss": 0.6541,
+      "step": 7335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4246507773179637e-05,
+      "loss": 0.819,
+      "step": 7336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244990449269097e-05,
+      "loss": 0.7435,
+      "step": 7337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424347300613578e-05,
+      "loss": 0.7137,
+      "step": 7338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4241955443822302e-05,
+      "loss": 0.7667,
+      "step": 7339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240437762371287e-05,
+      "loss": 0.7267,
+      "step": 7340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4238919961825357e-05,
+      "loss": 0.6794,
+      "step": 7341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237402042227144e-05,
+      "loss": 0.7297,
+      "step": 7342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235884003619279e-05,
+      "loss": 0.7733,
+      "step": 7343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234365846044392e-05,
+      "loss": 0.7322,
+      "step": 7344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232847569545127e-05,
+      "loss": 0.6795,
+      "step": 7345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231329174164124e-05,
+      "loss": 0.685,
+      "step": 7346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229810659944026e-05,
+      "loss": 0.7772,
+      "step": 7347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4228292026927481e-05,
+      "loss": 0.6718,
+      "step": 7348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226773275157142e-05,
+      "loss": 0.694,
+      "step": 7349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225254404675666e-05,
+      "loss": 0.6774,
+      "step": 7350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4223735415525709e-05,
+      "loss": 0.7358,
+      "step": 7351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222216307749932e-05,
+      "loss": 0.7853,
+      "step": 7352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220697081391002e-05,
+      "loss": 0.6381,
+      "step": 7353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219177736491585e-05,
+      "loss": 0.758,
+      "step": 7354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4217658273094355e-05,
+      "loss": 0.7263,
+      "step": 7355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216138691241985e-05,
+      "loss": 0.784,
+      "step": 7356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4214618990977154e-05,
+      "loss": 0.7185,
+      "step": 7357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213099172342543e-05,
+      "loss": 0.7775,
+      "step": 7358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211579235380838e-05,
+      "loss": 0.7843,
+      "step": 7359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210059180134725e-05,
+      "loss": 0.7402,
+      "step": 7360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208539006646903e-05,
+      "loss": 0.6809,
+      "step": 7361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207018714960052e-05,
+      "loss": 0.7624,
+      "step": 7362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4205498305116883e-05,
+      "loss": 0.7952,
+      "step": 7363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203977777160094e-05,
+      "loss": 0.7336,
+      "step": 7364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4202457131132392e-05,
+      "loss": 0.6842,
+      "step": 7365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4200936367076481e-05,
+      "loss": 0.6883,
+      "step": 7366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199415485035073e-05,
+      "loss": 0.7169,
+      "step": 7367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197894485050885e-05,
+      "loss": 0.7264,
+      "step": 7368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196373367166632e-05,
+      "loss": 0.7686,
+      "step": 7369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194852131425038e-05,
+      "loss": 0.6964,
+      "step": 7370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419333077786883e-05,
+      "loss": 0.7058,
+      "step": 7371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191809306540729e-05,
+      "loss": 0.7866,
+      "step": 7372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190287717483473e-05,
+      "loss": 0.7473,
+      "step": 7373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188766010739794e-05,
+      "loss": 0.7525,
+      "step": 7374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418724418635243e-05,
+      "loss": 0.7614,
+      "step": 7375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418572224436412e-05,
+      "loss": 0.6804,
+      "step": 7376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4184200184817614e-05,
+      "loss": 0.7466,
+      "step": 7377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4182678007755653e-05,
+      "loss": 0.7597,
+      "step": 7378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181155713220994e-05,
+      "loss": 0.7401,
+      "step": 7379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179633301256388e-05,
+      "loss": 0.6708,
+      "step": 7380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178110771904596e-05,
+      "loss": 0.7693,
+      "step": 7381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176588125208373e-05,
+      "loss": 0.6778,
+      "step": 7382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4175065361210489e-05,
+      "loss": 0.7326,
+      "step": 7383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173542479953712e-05,
+      "loss": 0.7586,
+      "step": 7384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417201948148081e-05,
+      "loss": 0.7098,
+      "step": 7385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4170496365834557e-05,
+      "loss": 0.7204,
+      "step": 7386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168973133057731e-05,
+      "loss": 0.7362,
+      "step": 7387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4167449783193114e-05,
+      "loss": 0.7804,
+      "step": 7388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4165926316283489e-05,
+      "loss": 0.6711,
+      "step": 7389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4164402732371642e-05,
+      "loss": 0.6639,
+      "step": 7390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416287903150037e-05,
+      "loss": 0.7173,
+      "step": 7391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416135521371246e-05,
+      "loss": 0.6617,
+      "step": 7392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415983127905071e-05,
+      "loss": 0.7241,
+      "step": 7393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158307227557924e-05,
+      "loss": 0.7067,
+      "step": 7394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156783059276902e-05,
+      "loss": 0.7025,
+      "step": 7395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4155258774250454e-05,
+      "loss": 0.7767,
+      "step": 7396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4153734372521388e-05,
+      "loss": 0.732,
+      "step": 7397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415220985413252e-05,
+      "loss": 0.7309,
+      "step": 7398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150685219126665e-05,
+      "loss": 0.7392,
+      "step": 7399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149160467546646e-05,
+      "loss": 0.7872,
+      "step": 7400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4147635599435283e-05,
+      "loss": 0.7996,
+      "step": 7401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146110614835405e-05,
+      "loss": 0.651,
+      "step": 7402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144585513789837e-05,
+      "loss": 0.6938,
+      "step": 7403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414306029634142e-05,
+      "loss": 0.7562,
+      "step": 7404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141534962532986e-05,
+      "loss": 0.7498,
+      "step": 7405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4140009512407375e-05,
+      "loss": 0.7416,
+      "step": 7406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4138483946007431e-05,
+      "loss": 0.73,
+      "step": 7407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136958263376002e-05,
+      "loss": 0.7743,
+      "step": 7408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135432464555934e-05,
+      "loss": 0.7222,
+      "step": 7409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133906549590082e-05,
+      "loss": 0.7768,
+      "step": 7410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41323805185213e-05,
+      "loss": 0.7748,
+      "step": 7411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4130854371392451e-05,
+      "loss": 0.7385,
+      "step": 7412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129328108246396e-05,
+      "loss": 0.6959,
+      "step": 7413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127801729125998e-05,
+      "loss": 0.7699,
+      "step": 7414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412627523407413e-05,
+      "loss": 0.6994,
+      "step": 7415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124748623133662e-05,
+      "loss": 0.6802,
+      "step": 7416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123221896347473e-05,
+      "loss": 0.7234,
+      "step": 7417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121695053758437e-05,
+      "loss": 0.7218,
+      "step": 7418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412016809540944e-05,
+      "loss": 0.7042,
+      "step": 7419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4118641021343371e-05,
+      "loss": 0.7927,
+      "step": 7420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117113831603108e-05,
+      "loss": 0.7032,
+      "step": 7421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411558652623155e-05,
+      "loss": 0.8765,
+      "step": 7422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4114059105271591e-05,
+      "loss": 0.6974,
+      "step": 7423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411253156876613e-05,
+      "loss": 0.6936,
+      "step": 7424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111003916758068e-05,
+      "loss": 0.8439,
+      "step": 7425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4109476149290313e-05,
+      "loss": 0.6763,
+      "step": 7426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107948266405767e-05,
+      "loss": 0.6822,
+      "step": 7427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106420268147343e-05,
+      "loss": 0.7639,
+      "step": 7428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104892154557959e-05,
+      "loss": 0.6168,
+      "step": 7429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103363925680534e-05,
+      "loss": 0.7985,
+      "step": 7430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101835581557983e-05,
+      "loss": 0.7272,
+      "step": 7431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100307122233234e-05,
+      "loss": 0.741,
+      "step": 7432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4098778547749215e-05,
+      "loss": 0.6319,
+      "step": 7433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097249858148854e-05,
+      "loss": 0.7457,
+      "step": 7434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4095721053475086e-05,
+      "loss": 0.8003,
+      "step": 7435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094192133770854e-05,
+      "loss": 0.7207,
+      "step": 7436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092663099079089e-05,
+      "loss": 0.7513,
+      "step": 7437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409113394944274e-05,
+      "loss": 0.7695,
+      "step": 7438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089604684904758e-05,
+      "loss": 0.7337,
+      "step": 7439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4088075305508087e-05,
+      "loss": 0.7072,
+      "step": 7440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408654581129568e-05,
+      "loss": 0.7069,
+      "step": 7441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085016202310496e-05,
+      "loss": 0.7245,
+      "step": 7442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083486478595494e-05,
+      "loss": 0.7704,
+      "step": 7443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081956640193639e-05,
+      "loss": 0.6783,
+      "step": 7444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4080426687147895e-05,
+      "loss": 0.7472,
+      "step": 7445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078896619501237e-05,
+      "loss": 0.7901,
+      "step": 7446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407736643729663e-05,
+      "loss": 0.7199,
+      "step": 7447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075836140577049e-05,
+      "loss": 0.7767,
+      "step": 7448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4074305729385482e-05,
+      "loss": 0.7841,
+      "step": 7449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072775203764906e-05,
+      "loss": 0.7961,
+      "step": 7450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4071244563758308e-05,
+      "loss": 0.6774,
+      "step": 7451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4069713809408675e-05,
+      "loss": 0.7392,
+      "step": 7452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068182940759e-05,
+      "loss": 0.7463,
+      "step": 7453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406665195785228e-05,
+      "loss": 0.7338,
+      "step": 7454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065120860731511e-05,
+      "loss": 0.7643,
+      "step": 7455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40635896494397e-05,
+      "loss": 0.7281,
+      "step": 7456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4062058324019842e-05,
+      "loss": 0.7095,
+      "step": 7457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060526884514951e-05,
+      "loss": 0.7345,
+      "step": 7458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058995330968041e-05,
+      "loss": 0.726,
+      "step": 7459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4057463663422121e-05,
+      "loss": 0.738,
+      "step": 7460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055931881920214e-05,
+      "loss": 0.7097,
+      "step": 7461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054399986505334e-05,
+      "loss": 0.6658,
+      "step": 7462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405286797722051e-05,
+      "loss": 0.6484,
+      "step": 7463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405133585410877e-05,
+      "loss": 0.7024,
+      "step": 7464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049803617213143e-05,
+      "loss": 0.8049,
+      "step": 7465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404827126657666e-05,
+      "loss": 0.6816,
+      "step": 7466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4046738802242363e-05,
+      "loss": 0.7351,
+      "step": 7467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045206224253284e-05,
+      "loss": 0.7191,
+      "step": 7468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404367353265248e-05,
+      "loss": 0.7456,
+      "step": 7469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4042140727482985e-05,
+      "loss": 0.7128,
+      "step": 7470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4040607808787851e-05,
+      "loss": 0.6822,
+      "step": 7471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039074776610135e-05,
+      "loss": 0.7228,
+      "step": 7472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403754163099289e-05,
+      "loss": 0.7265,
+      "step": 7473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036008371979175e-05,
+      "loss": 0.7259,
+      "step": 7474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034474999612055e-05,
+      "loss": 0.7068,
+      "step": 7475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032941513934593e-05,
+      "loss": 0.6804,
+      "step": 7476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031407914989857e-05,
+      "loss": 0.6976,
+      "step": 7477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029874202820921e-05,
+      "loss": 0.7933,
+      "step": 7478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402834037747086e-05,
+      "loss": 0.7138,
+      "step": 7479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4026806438982752e-05,
+      "loss": 0.787,
+      "step": 7480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4025272387399676e-05,
+      "loss": 0.7345,
+      "step": 7481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023738222764722e-05,
+      "loss": 0.764,
+      "step": 7482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022203945120971e-05,
+      "loss": 0.7517,
+      "step": 7483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020669554511521e-05,
+      "loss": 0.6986,
+      "step": 7484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019135050979464e-05,
+      "loss": 0.6914,
+      "step": 7485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4017600434567893e-05,
+      "loss": 0.6614,
+      "step": 7486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016065705319914e-05,
+      "loss": 0.7163,
+      "step": 7487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014530863278627e-05,
+      "loss": 0.759,
+      "step": 7488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401299590848714e-05,
+      "loss": 0.7435,
+      "step": 7489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011460840988564e-05,
+      "loss": 0.7208,
+      "step": 7490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009925660826014e-05,
+      "loss": 0.7736,
+      "step": 7491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008390368042597e-05,
+      "loss": 0.7167,
+      "step": 7492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4006854962681443e-05,
+      "loss": 0.6976,
+      "step": 7493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400531944478567e-05,
+      "loss": 0.7254,
+      "step": 7494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400378381439841e-05,
+      "loss": 0.7648,
+      "step": 7495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400224807156278e-05,
+      "loss": 0.7191,
+      "step": 7496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000712216321924e-05,
+      "loss": 0.8012,
+      "step": 7497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399917624871897e-05,
+      "loss": 0.6259,
+      "step": 7498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3997640168797055e-05,
+      "loss": 0.6514,
+      "step": 7499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3996103976599326e-05,
+      "loss": 0.7305,
+      "step": 7500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399456767216893e-05,
+      "loss": 0.731,
+      "step": 7501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993031255549006e-05,
+      "loss": 0.7817,
+      "step": 7502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991494726782712e-05,
+      "loss": 0.7589,
+      "step": 7503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3989958085913202e-05,
+      "loss": 0.6793,
+      "step": 7504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988421332983631e-05,
+      "loss": 0.8239,
+      "step": 7505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3986884468037158e-05,
+      "loss": 0.6926,
+      "step": 7506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398534749111695e-05,
+      "loss": 0.7058,
+      "step": 7507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983810402266171e-05,
+      "loss": 0.7098,
+      "step": 7508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982273201527994e-05,
+      "loss": 0.7328,
+      "step": 7509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398073588894559e-05,
+      "loss": 0.8264,
+      "step": 7510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979198464562136e-05,
+      "loss": 0.7606,
+      "step": 7511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977660928420812e-05,
+      "loss": 0.7462,
+      "step": 7512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3976123280564794e-05,
+      "loss": 0.7139,
+      "step": 7513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397458552103728e-05,
+      "loss": 0.7782,
+      "step": 7514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397304764988145e-05,
+      "loss": 0.7355,
+      "step": 7515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39715096671405e-05,
+      "loss": 0.7058,
+      "step": 7516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3969971572857619e-05,
+      "loss": 0.7375,
+      "step": 7517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396843336707601e-05,
+      "loss": 0.6519,
+      "step": 7518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966895049838876e-05,
+      "loss": 0.7814,
+      "step": 7519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3965356621189416e-05,
+      "loss": 0.675,
+      "step": 7520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3963818081170843e-05,
+      "loss": 0.6917,
+      "step": 7521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962279429826367e-05,
+      "loss": 0.6602,
+      "step": 7522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960740667199195e-05,
+      "loss": 0.704,
+      "step": 7523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959201793332554e-05,
+      "loss": 0.7551,
+      "step": 7524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957662808269655e-05,
+      "loss": 0.6959,
+      "step": 7525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3956123712053727e-05,
+      "loss": 0.6997,
+      "step": 7526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954584504727993e-05,
+      "loss": 0.787,
+      "step": 7527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953045186335685e-05,
+      "loss": 0.771,
+      "step": 7528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3951505756920036e-05,
+      "loss": 0.6585,
+      "step": 7529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3949966216524278e-05,
+      "loss": 0.7391,
+      "step": 7530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3948426565191655e-05,
+      "loss": 0.7466,
+      "step": 7531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3946886802965405e-05,
+      "loss": 0.7218,
+      "step": 7532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945346929888773e-05,
+      "loss": 0.7178,
+      "step": 7533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394380694600501e-05,
+      "loss": 0.7239,
+      "step": 7534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942266851357365e-05,
+      "loss": 0.7713,
+      "step": 7535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394072664598909e-05,
+      "loss": 0.7236,
+      "step": 7536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3939186329943448e-05,
+      "loss": 0.808,
+      "step": 7537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937645903263697e-05,
+      "loss": 0.6818,
+      "step": 7538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936105365993098e-05,
+      "loss": 0.6911,
+      "step": 7539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3934564718174923e-05,
+      "loss": 0.7292,
+      "step": 7540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393302395985244e-05,
+      "loss": 0.7277,
+      "step": 7541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931483091068917e-05,
+      "loss": 0.7058,
+      "step": 7542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929942111867636e-05,
+      "loss": 0.7105,
+      "step": 7543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3928401022291877e-05,
+      "loss": 0.7177,
+      "step": 7544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3926859822384919e-05,
+      "loss": 0.6268,
+      "step": 7545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925318512190047e-05,
+      "loss": 0.7156,
+      "step": 7546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923777091750552e-05,
+      "loss": 0.7511,
+      "step": 7547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922235561109725e-05,
+      "loss": 0.7205,
+      "step": 7548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3920693920310856e-05,
+      "loss": 0.6807,
+      "step": 7549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919152169397253e-05,
+      "loss": 0.7377,
+      "step": 7550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917610308412205e-05,
+      "loss": 0.7453,
+      "step": 7551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3916068337399026e-05,
+      "loss": 0.7107,
+      "step": 7552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914526256401018e-05,
+      "loss": 0.6871,
+      "step": 7553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391298406546149e-05,
+      "loss": 0.7657,
+      "step": 7554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911441764623756e-05,
+      "loss": 0.7159,
+      "step": 7555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909899353931136e-05,
+      "loss": 0.7058,
+      "step": 7556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908356833426945e-05,
+      "loss": 0.739,
+      "step": 7557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390681420315451e-05,
+      "loss": 0.7087,
+      "step": 7558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905271463157153e-05,
+      "loss": 0.6699,
+      "step": 7559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903728613478205e-05,
+      "loss": 0.7622,
+      "step": 7560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902185654160995e-05,
+      "loss": 0.7136,
+      "step": 7561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390064258524886e-05,
+      "loss": 0.707,
+      "step": 7562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899099406785135e-05,
+      "loss": 0.6902,
+      "step": 7563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897556118813164e-05,
+      "loss": 0.7936,
+      "step": 7564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389601272137629e-05,
+      "loss": 0.7211,
+      "step": 7565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894469214517862e-05,
+      "loss": 0.6974,
+      "step": 7566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892925598281228e-05,
+      "loss": 0.7244,
+      "step": 7567
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389138187270974e-05,
+      "loss": 0.7458,
+      "step": 7568
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3889838037846754e-05,
+      "loss": 0.7263,
+      "step": 7569
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888294093735636e-05,
+      "loss": 0.7257,
+      "step": 7570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3886750040419744e-05,
+      "loss": 0.7946,
+      "step": 7571
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388520587794244e-05,
+      "loss": 0.7255,
+      "step": 7572
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38836616063471e-05,
+      "loss": 0.7529,
+      "step": 7573
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882117225677087e-05,
+      "loss": 0.6826,
+      "step": 7574
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880572735975783e-05,
+      "loss": 0.701,
+      "step": 7575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3879028137286565e-05,
+      "loss": 0.7561,
+      "step": 7576
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877483429652811e-05,
+      "loss": 0.7731,
+      "step": 7577
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875938613117903e-05,
+      "loss": 0.7044,
+      "step": 7578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874393687725238e-05,
+      "loss": 0.6602,
+      "step": 7579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872848653518194e-05,
+      "loss": 0.7247,
+      "step": 7580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387130351054017e-05,
+      "loss": 0.755,
+      "step": 7581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386975825883456e-05,
+      "loss": 0.694,
+      "step": 7582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868212898444766e-05,
+      "loss": 0.7336,
+      "step": 7583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866667429414188e-05,
+      "loss": 0.7031,
+      "step": 7584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3865121851786235e-05,
+      "loss": 0.7405,
+      "step": 7585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863576165604311e-05,
+      "loss": 0.692,
+      "step": 7586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386203037091183e-05,
+      "loss": 0.7252,
+      "step": 7587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38604844677522e-05,
+      "loss": 0.3533,
+      "step": 7588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385893845616885e-05,
+      "loss": 0.7506,
+      "step": 7589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385739233620519e-05,
+      "loss": 0.7446,
+      "step": 7590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855846107904652e-05,
+      "loss": 0.7149,
+      "step": 7591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854299771310656e-05,
+      "loss": 0.7899,
+      "step": 7592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3852753326466634e-05,
+      "loss": 0.7662,
+      "step": 7593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385120677341602e-05,
+      "loss": 0.7184,
+      "step": 7594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849660112202248e-05,
+      "loss": 0.7858,
+      "step": 7595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384811334286876e-05,
+      "loss": 0.6953,
+      "step": 7596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846566465458993e-05,
+      "loss": 0.686,
+      "step": 7597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3845019480016393e-05,
+      "loss": 0.7313,
+      "step": 7598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843472386584411e-05,
+      "loss": 0.7506,
+      "step": 7599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841925185206497e-05,
+      "loss": 0.668,
+      "step": 7600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840377875926103e-05,
+      "loss": 0.7345,
+      "step": 7601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838830458786686e-05,
+      "loss": 0.6669,
+      "step": 7602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3837282933831708e-05,
+      "loss": 0.658,
+      "step": 7603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835735301104628e-05,
+      "loss": 0.7078,
+      "step": 7604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834187560648922e-05,
+      "loss": 0.7275,
+      "step": 7605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832639712508047e-05,
+      "loss": 0.6499,
+      "step": 7606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831091756725483e-05,
+      "loss": 0.7276,
+      "step": 7607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829543693344699e-05,
+      "loss": 0.7399,
+      "step": 7608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382799552240918e-05,
+      "loss": 0.7206,
+      "step": 7609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826447243962401e-05,
+      "loss": 0.3604,
+      "step": 7610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824898858047853e-05,
+      "loss": 0.6787,
+      "step": 7611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823350364709016e-05,
+      "loss": 0.6981,
+      "step": 7612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821801763989385e-05,
+      "loss": 0.7657,
+      "step": 7613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382025305593245e-05,
+      "loss": 0.7642,
+      "step": 7614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3818704240581712e-05,
+      "loss": 0.3593,
+      "step": 7615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3817155317980667e-05,
+      "loss": 0.7433,
+      "step": 7616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3815606288172816e-05,
+      "loss": 0.7234,
+      "step": 7617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3814057151201666e-05,
+      "loss": 0.6959,
+      "step": 7618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3812507907110725e-05,
+      "loss": 0.7121,
+      "step": 7619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810958555943506e-05,
+      "loss": 0.7445,
+      "step": 7620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809409097743523e-05,
+      "loss": 0.7343,
+      "step": 7621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807859532554291e-05,
+      "loss": 0.6335,
+      "step": 7622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806309860419329e-05,
+      "loss": 0.7255,
+      "step": 7623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3804760081382164e-05,
+      "loss": 0.7609,
+      "step": 7624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3803210195486325e-05,
+      "loss": 0.7454,
+      "step": 7625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801660202775335e-05,
+      "loss": 0.7676,
+      "step": 7626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3800110103292731e-05,
+      "loss": 0.7582,
+      "step": 7627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3798559897082044e-05,
+      "loss": 0.7288,
+      "step": 7628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797009584186818e-05,
+      "loss": 0.7817,
+      "step": 7629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795459164650589e-05,
+      "loss": 0.7366,
+      "step": 7630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793908638516904e-05,
+      "loss": 0.6962,
+      "step": 7631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.379235800582931e-05,
+      "loss": 0.7319,
+      "step": 7632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3790807266631355e-05,
+      "loss": 0.7426,
+      "step": 7633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3789256420966594e-05,
+      "loss": 0.707,
+      "step": 7634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378770546887859e-05,
+      "loss": 0.7645,
+      "step": 7635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378615441041089e-05,
+      "loss": 0.7692,
+      "step": 7636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3784603245607064e-05,
+      "loss": 0.8207,
+      "step": 7637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783051974510677e-05,
+      "loss": 0.7219,
+      "step": 7638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781500597165294e-05,
+      "loss": 0.7158,
+      "step": 7639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377994911361449e-05,
+      "loss": 0.7306,
+      "step": 7640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778397523901838e-05,
+      "loss": 0.6962,
+      "step": 7641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776845828070913e-05,
+      "loss": 0.7006,
+      "step": 7642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3775294026165295e-05,
+      "loss": 0.7104,
+      "step": 7643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773742118228572e-05,
+      "loss": 0.7073,
+      "step": 7644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3772190104304328e-05,
+      "loss": 0.6944,
+      "step": 7645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770637984436148e-05,
+      "loss": 0.6747,
+      "step": 7646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769085758667628e-05,
+      "loss": 0.6774,
+      "step": 7647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3767533427042364e-05,
+      "loss": 0.726,
+      "step": 7648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765980989603951e-05,
+      "loss": 0.773,
+      "step": 7649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3764428446395992e-05,
+      "loss": 0.7302,
+      "step": 7650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3762875797462092e-05,
+      "loss": 0.7382,
+      "step": 7651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761323042845855e-05,
+      "loss": 0.6758,
+      "step": 7652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375977018259089e-05,
+      "loss": 0.7467,
+      "step": 7653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3758217216740816e-05,
+      "loss": 0.6853,
+      "step": 7654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756664145339244e-05,
+      "loss": 0.6965,
+      "step": 7655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755110968429793e-05,
+      "loss": 0.7576,
+      "step": 7656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753557686056085e-05,
+      "loss": 0.6814,
+      "step": 7657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752004298261747e-05,
+      "loss": 0.754,
+      "step": 7658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3750450805090405e-05,
+      "loss": 0.7427,
+      "step": 7659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374889720658569e-05,
+      "loss": 0.7131,
+      "step": 7660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747343502791235e-05,
+      "loss": 0.7895,
+      "step": 7661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745789693750678e-05,
+      "loss": 0.7623,
+      "step": 7662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3744235779507654e-05,
+      "loss": 0.7409,
+      "step": 7663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742681760105814e-05,
+      "loss": 0.7296,
+      "step": 7664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741127635588795e-05,
+      "loss": 0.7184,
+      "step": 7665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373957340600025e-05,
+      "loss": 0.7398,
+      "step": 7666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373801907138383e-05,
+      "loss": 0.7198,
+      "step": 7667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736464631783188e-05,
+      "loss": 0.7716,
+      "step": 7668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734910087241981e-05,
+      "loss": 0.7449,
+      "step": 7669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733355437803873e-05,
+      "loss": 0.6704,
+      "step": 7670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3731800683512521e-05,
+      "loss": 0.719,
+      "step": 7671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730245824411597e-05,
+      "loss": 0.6802,
+      "step": 7672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3728690860544761e-05,
+      "loss": 0.7194,
+      "step": 7673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727135791955699e-05,
+      "loss": 0.7548,
+      "step": 7674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3725580618688075e-05,
+      "loss": 0.7843,
+      "step": 7675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724025340785569e-05,
+      "loss": 0.7534,
+      "step": 7676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722469958291865e-05,
+      "loss": 0.3471,
+      "step": 7677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720914471250644e-05,
+      "loss": 0.7794,
+      "step": 7678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719358879705591e-05,
+      "loss": 0.6613,
+      "step": 7679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37178031837004e-05,
+      "loss": 0.7199,
+      "step": 7680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3716247383278762e-05,
+      "loss": 0.685,
+      "step": 7681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371469147848437e-05,
+      "loss": 0.7523,
+      "step": 7682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3713135469360926e-05,
+      "loss": 0.7731,
+      "step": 7683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711579355952127e-05,
+      "loss": 0.748,
+      "step": 7684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371002313830168e-05,
+      "loss": 0.7518,
+      "step": 7685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708466816453294e-05,
+      "loss": 0.7941,
+      "step": 7686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3706910390450679e-05,
+      "loss": 0.6972,
+      "step": 7687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370535386033754e-05,
+      "loss": 0.7008,
+      "step": 7688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703797226157603e-05,
+      "loss": 0.7866,
+      "step": 7689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702240487954583e-05,
+      "loss": 0.7737,
+      "step": 7690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37006836457722e-05,
+      "loss": 0.6465,
+      "step": 7691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699126699654183e-05,
+      "loss": 0.7656,
+      "step": 7692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3697569649644256e-05,
+      "loss": 0.6848,
+      "step": 7693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369601249578615e-05,
+      "loss": 0.7758,
+      "step": 7694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36944552381236e-05,
+      "loss": 0.7156,
+      "step": 7695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692897876700345e-05,
+      "loss": 0.77,
+      "step": 7696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3691340411560117e-05,
+      "loss": 0.7022,
+      "step": 7697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689782842746662e-05,
+      "loss": 0.6624,
+      "step": 7698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688225170303727e-05,
+      "loss": 0.6818,
+      "step": 7699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3686667394275057e-05,
+      "loss": 0.7618,
+      "step": 7700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3685109514704404e-05,
+      "loss": 0.7043,
+      "step": 7701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683551531635523e-05,
+      "loss": 0.6932,
+      "step": 7702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368199344511217e-05,
+      "loss": 0.7459,
+      "step": 7703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680435255178104e-05,
+      "loss": 0.678,
+      "step": 7704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678876961877088e-05,
+      "loss": 0.7193,
+      "step": 7705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.367731856525289e-05,
+      "loss": 0.7353,
+      "step": 7706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3675760065349276e-05,
+      "loss": 0.6724,
+      "step": 7707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674201462210012e-05,
+      "loss": 0.7525,
+      "step": 7708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672642755878882e-05,
+      "loss": 0.7686,
+      "step": 7709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3671083946399658e-05,
+      "loss": 0.7235,
+      "step": 7710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366952503381612e-05,
+      "loss": 0.7835,
+      "step": 7711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667966018172052e-05,
+      "loss": 0.6734,
+      "step": 7712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366640689951124e-05,
+      "loss": 0.7329,
+      "step": 7713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3664847677877471e-05,
+      "loss": 0.7286,
+      "step": 7714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663288353314536e-05,
+      "loss": 0.6743,
+      "step": 7715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661728925866234e-05,
+      "loss": 0.7319,
+      "step": 7716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3660169395576357e-05,
+      "loss": 0.6406,
+      "step": 7717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658609762488706e-05,
+      "loss": 0.7176,
+      "step": 7718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657050026647089e-05,
+      "loss": 0.759,
+      "step": 7719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655490188095307e-05,
+      "loss": 0.7484,
+      "step": 7720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365393024687717e-05,
+      "loss": 0.7595,
+      "step": 7721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365237020303649e-05,
+      "loss": 0.7517,
+      "step": 7722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650810056617084e-05,
+      "loss": 0.7761,
+      "step": 7723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649249807662765e-05,
+      "loss": 0.6465,
+      "step": 7724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647689456217356e-05,
+      "loss": 0.7001,
+      "step": 7725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3646129002324684e-05,
+      "loss": 0.752,
+      "step": 7726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644568446028567e-05,
+      "loss": 0.7444,
+      "step": 7727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3643007787372838e-05,
+      "loss": 0.6944,
+      "step": 7728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.364144702640133e-05,
+      "loss": 0.7587,
+      "step": 7729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639886163157878e-05,
+      "loss": 0.7433,
+      "step": 7730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638325197686317e-05,
+      "loss": 0.7281,
+      "step": 7731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363676413003049e-05,
+      "loss": 0.6963,
+      "step": 7732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635202960234242e-05,
+      "loss": 0.6645,
+      "step": 7733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633641688341413e-05,
+      "loss": 0.7285,
+      "step": 7734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363208031439586e-05,
+      "loss": 0.7352,
+      "step": 7735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363051883844143e-05,
+      "loss": 0.6219,
+      "step": 7736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628957260521976e-05,
+      "loss": 0.6887,
+      "step": 7737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627395580681357e-05,
+      "loss": 0.7362,
+      "step": 7738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362583379896344e-05,
+      "loss": 0.75,
+      "step": 7739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624271915412084e-05,
+      "loss": 0.7157,
+      "step": 7740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622709930071152e-05,
+      "loss": 0.6808,
+      "step": 7741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621147842984519e-05,
+      "loss": 0.754,
+      "step": 7742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361958565419605e-05,
+      "loss": 0.7227,
+      "step": 7743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618023363749626e-05,
+      "loss": 0.7431,
+      "step": 7744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616460971689126e-05,
+      "loss": 0.7515,
+      "step": 7745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614898478058423e-05,
+      "loss": 0.7478,
+      "step": 7746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3613335882901406e-05,
+      "loss": 0.7734,
+      "step": 7747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3611773186261959e-05,
+      "loss": 0.6727,
+      "step": 7748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610210388183976e-05,
+      "loss": 0.7888,
+      "step": 7749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3608647488711343e-05,
+      "loss": 0.7436,
+      "step": 7750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607084487887959e-05,
+      "loss": 0.7535,
+      "step": 7751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605521385757718e-05,
+      "loss": 0.7431,
+      "step": 7752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360395818236452e-05,
+      "loss": 0.767,
+      "step": 7753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602394877752275e-05,
+      "loss": 0.7385,
+      "step": 7754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3600831471964883e-05,
+      "loss": 0.7676,
+      "step": 7755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599267965046257e-05,
+      "loss": 0.3416,
+      "step": 7756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597704357040304e-05,
+      "loss": 0.6863,
+      "step": 7757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596140647990943e-05,
+      "loss": 0.6835,
+      "step": 7758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594576837942091e-05,
+      "loss": 0.6843,
+      "step": 7759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3593012926937668e-05,
+      "loss": 0.6718,
+      "step": 7760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591448915021595e-05,
+      "loss": 0.6891,
+      "step": 7761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589884802237803e-05,
+      "loss": 0.7247,
+      "step": 7762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3588320588630215e-05,
+      "loss": 0.6868,
+      "step": 7763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3586756274242768e-05,
+      "loss": 0.7486,
+      "step": 7764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585191859119393e-05,
+      "loss": 0.708,
+      "step": 7765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358362734330403e-05,
+      "loss": 0.7534,
+      "step": 7766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3582062726840616e-05,
+      "loss": 0.6304,
+      "step": 7767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580498009773098e-05,
+      "loss": 0.7298,
+      "step": 7768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578933192145421e-05,
+      "loss": 0.718,
+      "step": 7769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357736827400153e-05,
+      "loss": 0.8193,
+      "step": 7770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575803255385383e-05,
+      "loss": 0.6682,
+      "step": 7771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574238136340927e-05,
+      "loss": 0.7296,
+      "step": 7772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572672916912122e-05,
+      "loss": 0.7209,
+      "step": 7773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571107597142932e-05,
+      "loss": 0.7036,
+      "step": 7774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569542177077316e-05,
+      "loss": 0.7764,
+      "step": 7775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567976656759241e-05,
+      "loss": 0.7352,
+      "step": 7776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566411036232672e-05,
+      "loss": 0.7145,
+      "step": 7777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3564845315541586e-05,
+      "loss": 0.8278,
+      "step": 7778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356327949472995e-05,
+      "loss": 0.6648,
+      "step": 7779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561713573841748e-05,
+      "loss": 0.7319,
+      "step": 7780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3560147552920959e-05,
+      "loss": 0.7448,
+      "step": 7781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558581432011559e-05,
+      "loss": 0.7386,
+      "step": 7782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557015211157538e-05,
+      "loss": 0.723,
+      "step": 7783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555448890402886e-05,
+      "loss": 0.7222,
+      "step": 7784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3553882469791591e-05,
+      "loss": 0.7343,
+      "step": 7785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552315949367648e-05,
+      "loss": 0.6502,
+      "step": 7786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355074932917505e-05,
+      "loss": 0.7181,
+      "step": 7787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549182609257804e-05,
+      "loss": 0.7435,
+      "step": 7788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547615789659902e-05,
+      "loss": 0.6289,
+      "step": 7789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 0.7192,
+      "step": 7790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3544481851598174e-05,
+      "loss": 0.7504,
+      "step": 7791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542914733222365e-05,
+      "loss": 0.7097,
+      "step": 7792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541347515341939e-05,
+      "loss": 0.6927,
+      "step": 7793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539780198000918e-05,
+      "loss": 0.6509,
+      "step": 7794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3538212781243317e-05,
+      "loss": 0.8414,
+      "step": 7795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536645265113157e-05,
+      "loss": 0.7707,
+      "step": 7796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353507764965447e-05,
+      "loss": 0.7185,
+      "step": 7797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353350993491127e-05,
+      "loss": 0.7995,
+      "step": 7798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3531942120927597e-05,
+      "loss": 0.6884,
+      "step": 7799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530374207747484e-05,
+      "loss": 0.799,
+      "step": 7800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3528806195414965e-05,
+      "loss": 0.7311,
+      "step": 7801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3527238083974074e-05,
+      "loss": 0.6697,
+      "step": 7802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525669873468856e-05,
+      "loss": 0.6937,
+      "step": 7803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524101563943356e-05,
+      "loss": 0.7565,
+      "step": 7804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522533155441622e-05,
+      "loss": 0.7451,
+      "step": 7805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520964648007699e-05,
+      "loss": 0.7526,
+      "step": 7806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519396041685644e-05,
+      "loss": 0.7659,
+      "step": 7807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517827336519505e-05,
+      "loss": 0.7469,
+      "step": 7808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3516258532553348e-05,
+      "loss": 0.7731,
+      "step": 7809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514689629831232e-05,
+      "loss": 0.7095,
+      "step": 7810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513120628397215e-05,
+      "loss": 0.7457,
+      "step": 7811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351155152829537e-05,
+      "loss": 0.7606,
+      "step": 7812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350998232956976e-05,
+      "loss": 0.6558,
+      "step": 7813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3508413032264462e-05,
+      "loss": 0.6556,
+      "step": 7814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506843636423548e-05,
+      "loss": 0.7337,
+      "step": 7815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505274142091093e-05,
+      "loss": 0.6766,
+      "step": 7816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3503704549311185e-05,
+      "loss": 0.6759,
+      "step": 7817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502134858127895e-05,
+      "loss": 0.6953,
+      "step": 7818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3500565068585318e-05,
+      "loss": 0.8268,
+      "step": 7819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498995180727543e-05,
+      "loss": 0.7139,
+      "step": 7820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497425194598654e-05,
+      "loss": 0.6915,
+      "step": 7821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495855110242748e-05,
+      "loss": 0.7739,
+      "step": 7822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3494284927703925e-05,
+      "loss": 0.7507,
+      "step": 7823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349271464702628e-05,
+      "loss": 0.7005,
+      "step": 7824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3491144268253915e-05,
+      "loss": 0.7554,
+      "step": 7825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3489573791430938e-05,
+      "loss": 0.696,
+      "step": 7826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3488003216601458e-05,
+      "loss": 0.6957,
+      "step": 7827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3486432543809578e-05,
+      "loss": 0.7464,
+      "step": 7828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484861773099416e-05,
+      "loss": 0.7396,
+      "step": 7829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3483290904515091e-05,
+      "loss": 0.6993,
+      "step": 7830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3481719938100719e-05,
+      "loss": 0.6272,
+      "step": 7831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480148873900419e-05,
+      "loss": 0.7373,
+      "step": 7832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478577711958316e-05,
+      "loss": 0.7078,
+      "step": 7833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477006452318538e-05,
+      "loss": 0.657,
+      "step": 7834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475435095025215e-05,
+      "loss": 0.6812,
+      "step": 7835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347386364012248e-05,
+      "loss": 0.7725,
+      "step": 7836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3472292087654465e-05,
+      "loss": 0.7077,
+      "step": 7837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347072043766531e-05,
+      "loss": 0.6851,
+      "step": 7838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469148690199157e-05,
+      "loss": 0.7278,
+      "step": 7839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346757684530015e-05,
+      "loss": 0.7621,
+      "step": 7840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346600490301243e-05,
+      "loss": 0.7715,
+      "step": 7841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346443286338015e-05,
+      "loss": 0.7088,
+      "step": 7842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346286072644746e-05,
+      "loss": 0.7144,
+      "step": 7843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461288492258513e-05,
+      "loss": 0.7151,
+      "step": 7844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345971616085747e-05,
+      "loss": 0.7537,
+      "step": 7845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458143732288489e-05,
+      "loss": 0.7467,
+      "step": 7846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345657120659573e-05,
+      "loss": 0.6877,
+      "step": 7847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454998583823358e-05,
+      "loss": 0.7188,
+      "step": 7848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453425864015549e-05,
+      "loss": 0.7593,
+      "step": 7849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3451853047216463e-05,
+      "loss": 0.7589,
+      "step": 7850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450280133470279e-05,
+      "loss": 0.6954,
+      "step": 7851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448707122821171e-05,
+      "loss": 0.7944,
+      "step": 7852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3447134015313324e-05,
+      "loss": 0.7238,
+      "step": 7853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344556081099091e-05,
+      "loss": 0.7191,
+      "step": 7854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3443987509898118e-05,
+      "loss": 0.3619,
+      "step": 7855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442414112079139e-05,
+      "loss": 0.6736,
+      "step": 7856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440840617578155e-05,
+      "loss": 0.6999,
+      "step": 7857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343926702643936e-05,
+      "loss": 0.7633,
+      "step": 7858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437693338706957e-05,
+      "loss": 0.6938,
+      "step": 7859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3436119554425133e-05,
+      "loss": 0.6772,
+      "step": 7860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3434545673638093e-05,
+      "loss": 0.701,
+      "step": 7861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432971696390044e-05,
+      "loss": 0.6736,
+      "step": 7862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3431397622725187e-05,
+      "loss": 0.6786,
+      "step": 7863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3429823452687732e-05,
+      "loss": 0.6954,
+      "step": 7864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3428249186321892e-05,
+      "loss": 0.7401,
+      "step": 7865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426674823671882e-05,
+      "loss": 0.6968,
+      "step": 7866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425100364781911e-05,
+      "loss": 0.7476,
+      "step": 7867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3423525809696205e-05,
+      "loss": 0.6819,
+      "step": 7868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342195115845899e-05,
+      "loss": 0.7615,
+      "step": 7869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420376411114484e-05,
+      "loss": 0.3429,
+      "step": 7870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418801567706916e-05,
+      "loss": 0.6877,
+      "step": 7871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341722662828052e-05,
+      "loss": 0.7442,
+      "step": 7872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415651592879522e-05,
+      "loss": 0.7891,
+      "step": 7873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414076461548162e-05,
+      "loss": 0.7014,
+      "step": 7874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3412501234330684e-05,
+      "loss": 0.7501,
+      "step": 7875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341092591127132e-05,
+      "loss": 0.731,
+      "step": 7876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3409350492414315e-05,
+      "loss": 0.705,
+      "step": 7877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340777497780392e-05,
+      "loss": 0.7226,
+      "step": 7878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340619936748438e-05,
+      "loss": 0.6754,
+      "step": 7879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340462366149995e-05,
+      "loss": 0.6619,
+      "step": 7880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403047859894882e-05,
+      "loss": 0.7586,
+      "step": 7881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3401471962713437e-05,
+      "loss": 0.7402,
+      "step": 7882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3399895969999868e-05,
+      "loss": 0.6843,
+      "step": 7883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398319881798443e-05,
+      "loss": 0.7501,
+      "step": 7884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339674369815343e-05,
+      "loss": 0.7581,
+      "step": 7885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339516741910909e-05,
+      "loss": 0.674,
+      "step": 7886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393591044709694e-05,
+      "loss": 0.7341,
+      "step": 7887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339201457499952e-05,
+      "loss": 0.6675,
+      "step": 7888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390438010022843e-05,
+      "loss": 0.6956,
+      "step": 7889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388861349823941e-05,
+      "loss": 0.7611,
+      "step": 7890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3387284594447095e-05,
+      "loss": 0.6754,
+      "step": 7891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385707743936588e-05,
+      "loss": 0.6884,
+      "step": 7892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3384130798336705e-05,
+      "loss": 0.7438,
+      "step": 7893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382553757691742e-05,
+      "loss": 0.7322,
+      "step": 7894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380976622045984e-05,
+      "loss": 0.6928,
+      "step": 7895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337939939144373e-05,
+      "loss": 0.7231,
+      "step": 7896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377822065929276e-05,
+      "loss": 0.7499,
+      "step": 7897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337624464554692e-05,
+      "loss": 0.7189,
+      "step": 7898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374667130340968e-05,
+      "loss": 0.6603,
+      "step": 7899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373089520355725e-05,
+      "loss": 0.7778,
+      "step": 7900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3371511815635497e-05,
+      "loss": 0.767,
+      "step": 7901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369934016224598e-05,
+      "loss": 0.662,
+      "step": 7902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3368356122167332e-05,
+      "loss": 0.7402,
+      "step": 7903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366778133508027e-05,
+      "loss": 0.7437,
+      "step": 7904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365200050290998e-05,
+      "loss": 0.6322,
+      "step": 7905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363621872560563e-05,
+      "loss": 0.6866,
+      "step": 7906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362043600361047e-05,
+      "loss": 0.7056,
+      "step": 7907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336046523373678e-05,
+      "loss": 0.7711,
+      "step": 7908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358886772732085e-05,
+      "loss": 0.6941,
+      "step": 7909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3357308217391298e-05,
+      "loss": 0.838,
+      "step": 7910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355729567758757e-05,
+      "loss": 0.7562,
+      "step": 7911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354150823878792e-05,
+      "loss": 0.7933,
+      "step": 7912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352571985795744e-05,
+      "loss": 0.632,
+      "step": 7913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335099305355396e-05,
+      "loss": 0.7727,
+      "step": 7914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349414027197787e-05,
+      "loss": 0.7594,
+      "step": 7915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347834906771565e-05,
+      "loss": 0.7848,
+      "step": 7916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346255692319645e-05,
+      "loss": 0.6759,
+      "step": 7917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3344676383886386e-05,
+      "loss": 0.7183,
+      "step": 7918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343096981516141e-05,
+      "loss": 0.6975,
+      "step": 7919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341517485253269e-05,
+      "loss": 0.701,
+      "step": 7920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339937895142127e-05,
+      "loss": 0.7083,
+      "step": 7921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3338358211227085e-05,
+      "loss": 0.6935,
+      "step": 7922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336778433552502e-05,
+      "loss": 0.8273,
+      "step": 7923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335198562162757e-05,
+      "loss": 0.6978,
+      "step": 7924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333361859710221e-05,
+      "loss": 0.6988,
+      "step": 7925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332038538415242e-05,
+      "loss": 0.7444,
+      "step": 7926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330458386146228e-05,
+      "loss": 0.7246,
+      "step": 7927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3328878140339548e-05,
+      "loss": 0.7756,
+      "step": 7928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327297801039582e-05,
+      "loss": 0.7266,
+      "step": 7929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3325717368290718e-05,
+      "loss": 0.8455,
+      "step": 7930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324136842137344e-05,
+      "loss": 0.7368,
+      "step": 7931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3322556222623843e-05,
+      "loss": 0.6861,
+      "step": 7932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3320975509794616e-05,
+      "loss": 0.6732,
+      "step": 7933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319394703694051e-05,
+      "loss": 0.7297,
+      "step": 7934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317813804366551e-05,
+      "loss": 0.7341,
+      "step": 7935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316232811856513e-05,
+      "loss": 0.8069,
+      "step": 7936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314651726208345e-05,
+      "loss": 0.7599,
+      "step": 7937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313070547466445e-05,
+      "loss": 0.7836,
+      "step": 7938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311489275675224e-05,
+      "loss": 0.7199,
+      "step": 7939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3309907910879096e-05,
+      "loss": 0.6967,
+      "step": 7940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308326453122476e-05,
+      "loss": 0.7634,
+      "step": 7941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306744902449772e-05,
+      "loss": 0.7098,
+      "step": 7942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3305163258905408e-05,
+      "loss": 0.6967,
+      "step": 7943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3303581522533806e-05,
+      "loss": 0.7613,
+      "step": 7944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301999693379388e-05,
+      "loss": 0.7863,
+      "step": 7945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330041777148658e-05,
+      "loss": 0.7208,
+      "step": 7946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298835756899815e-05,
+      "loss": 0.6819,
+      "step": 7947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3297253649663517e-05,
+      "loss": 0.6374,
+      "step": 7948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295671449822126e-05,
+      "loss": 0.6729,
+      "step": 7949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294089157420082e-05,
+      "loss": 0.723,
+      "step": 7950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3292506772501819e-05,
+      "loss": 0.622,
+      "step": 7951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290924295111779e-05,
+      "loss": 0.6905,
+      "step": 7952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328934172529441e-05,
+      "loss": 0.6854,
+      "step": 7953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287759063094154e-05,
+      "loss": 0.7626,
+      "step": 7954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286176308555467e-05,
+      "loss": 0.6993,
+      "step": 7955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3284593461722798e-05,
+      "loss": 0.719,
+      "step": 7956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283010522640606e-05,
+      "loss": 0.6919,
+      "step": 7957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.328142749135334e-05,
+      "loss": 0.6892,
+      "step": 7958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279844367905472e-05,
+      "loss": 0.7569,
+      "step": 7959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3278261152341459e-05,
+      "loss": 0.7708,
+      "step": 7960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3276677844705761e-05,
+      "loss": 0.7666,
+      "step": 7961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275094445042855e-05,
+      "loss": 0.6654,
+      "step": 7962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273510953397205e-05,
+      "loss": 0.7135,
+      "step": 7963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271927369813292e-05,
+      "loss": 0.7016,
+      "step": 7964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270343694335584e-05,
+      "loss": 0.7382,
+      "step": 7965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268759927008562e-05,
+      "loss": 0.72,
+      "step": 7966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267176067876707e-05,
+      "loss": 0.7982,
+      "step": 7967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32655921169845e-05,
+      "loss": 0.3344,
+      "step": 7968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3264008074376436e-05,
+      "loss": 0.7324,
+      "step": 7969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262423940096994e-05,
+      "loss": 0.6933,
+      "step": 7970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326083971419067e-05,
+      "loss": 0.7986,
+      "step": 7971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3259255396701956e-05,
+      "loss": 0.6918,
+      "step": 7972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3257670987675346e-05,
+      "loss": 0.7243,
+      "step": 7973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256086487155345e-05,
+      "loss": 0.5941,
+      "step": 7974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325450189518645e-05,
+      "loss": 0.7738,
+      "step": 7975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325291721181317e-05,
+      "loss": 0.6876,
+      "step": 7976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251332437080003e-05,
+      "loss": 0.7075,
+      "step": 7977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3249747571031464e-05,
+      "loss": 0.7655,
+      "step": 7978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248162613712066e-05,
+      "loss": 0.7407,
+      "step": 7979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246577565166324e-05,
+      "loss": 0.6759,
+      "step": 7980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244992425438748e-05,
+      "loss": 0.723,
+      "step": 7981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243407194573863e-05,
+      "loss": 0.6885,
+      "step": 7982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.324182187261619e-05,
+      "loss": 0.6565,
+      "step": 7983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3240236459610251e-05,
+      "loss": 0.7114,
+      "step": 7984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3238650955600581e-05,
+      "loss": 0.7339,
+      "step": 7985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237065360631702e-05,
+      "loss": 0.6998,
+      "step": 7986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323547967474815e-05,
+      "loss": 0.7437,
+      "step": 7987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233893897994457e-05,
+      "loss": 0.7281,
+      "step": 7988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3232308030415161e-05,
+      "loss": 0.7221,
+      "step": 7989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230722072054803e-05,
+      "loss": 0.6555,
+      "step": 7990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3229136022957927e-05,
+      "loss": 0.7471,
+      "step": 7991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227549883169073e-05,
+      "loss": 0.7378,
+      "step": 7992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225963652732794e-05,
+      "loss": 0.762,
+      "step": 7993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3224377331693639e-05,
+      "loss": 0.7627,
+      "step": 7994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3222790920096155e-05,
+      "loss": 0.8118,
+      "step": 7995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3221204417984907e-05,
+      "loss": 0.667,
+      "step": 7996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219617825404447e-05,
+      "loss": 0.6687,
+      "step": 7997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218031142399335e-05,
+      "loss": 0.7508,
+      "step": 7998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3216444369014135e-05,
+      "loss": 0.6449,
+      "step": 7999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321485750529341e-05,
+      "loss": 0.7239,
+      "step": 8000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321327055128173e-05,
+      "loss": 0.6909,
+      "step": 8001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3211683507023669e-05,
+      "loss": 0.6532,
+      "step": 8002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321009637256379e-05,
+      "loss": 0.7552,
+      "step": 8003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208509147946678e-05,
+      "loss": 0.7216,
+      "step": 8004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3206921833216908e-05,
+      "loss": 0.7531,
+      "step": 8005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3205334428419063e-05,
+      "loss": 0.6669,
+      "step": 8006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320374693359772e-05,
+      "loss": 0.8213,
+      "step": 8007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320215934879747e-05,
+      "loss": 0.7469,
+      "step": 8008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200571674062896e-05,
+      "loss": 0.6773,
+      "step": 8009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198983909438596e-05,
+      "loss": 0.7855,
+      "step": 8010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197396054969156e-05,
+      "loss": 0.7029,
+      "step": 8011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319580811069918e-05,
+      "loss": 0.6997,
+      "step": 8012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194220076673254e-05,
+      "loss": 0.7049,
+      "step": 8013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319263195293599e-05,
+      "loss": 0.7119,
+      "step": 8014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191043739531988e-05,
+      "loss": 0.7508,
+      "step": 8015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318945543650585e-05,
+      "loss": 0.6761,
+      "step": 8016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318786704390219e-05,
+      "loss": 0.7924,
+      "step": 8017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3186278561765616e-05,
+      "loss": 0.7738,
+      "step": 8018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3184689990140738e-05,
+      "loss": 0.7341,
+      "step": 8019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3183101329072177e-05,
+      "loss": 0.7308,
+      "step": 8020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3181512578604552e-05,
+      "loss": 0.7701,
+      "step": 8021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317992373878248e-05,
+      "loss": 0.6551,
+      "step": 8022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178334809650583e-05,
+      "loss": 0.6997,
+      "step": 8023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176745791253493e-05,
+      "loss": 0.7663,
+      "step": 8024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175156683635836e-05,
+      "loss": 0.6995,
+      "step": 8025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173567486842243e-05,
+      "loss": 0.7603,
+      "step": 8026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3171978200917343e-05,
+      "loss": 0.6875,
+      "step": 8027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170388825905778e-05,
+      "loss": 0.7803,
+      "step": 8028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3168799361852184e-05,
+      "loss": 0.715,
+      "step": 8029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167209808801202e-05,
+      "loss": 0.6722,
+      "step": 8030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165620166797475e-05,
+      "loss": 0.7276,
+      "step": 8031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316403043588565e-05,
+      "loss": 0.798,
+      "step": 8032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162440616110372e-05,
+      "loss": 0.7148,
+      "step": 8033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3160850707516297e-05,
+      "loss": 0.7257,
+      "step": 8034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159260710148076e-05,
+      "loss": 0.7156,
+      "step": 8035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157670624050364e-05,
+      "loss": 0.7536,
+      "step": 8036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3156080449267821e-05,
+      "loss": 0.7856,
+      "step": 8037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154490185845109e-05,
+      "loss": 0.7099,
+      "step": 8038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152899833826887e-05,
+      "loss": 0.7477,
+      "step": 8039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151309393257824e-05,
+      "loss": 0.6816,
+      "step": 8040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149718864182591e-05,
+      "loss": 0.6791,
+      "step": 8041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314812824664585e-05,
+      "loss": 0.7866,
+      "step": 8042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146537540692284e-05,
+      "loss": 0.7678,
+      "step": 8043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144946746366562e-05,
+      "loss": 0.7019,
+      "step": 8044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3143355863713368e-05,
+      "loss": 0.769,
+      "step": 8045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314176489277738e-05,
+      "loss": 0.783,
+      "step": 8046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140173833603278e-05,
+      "loss": 0.7118,
+      "step": 8047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138582686235752e-05,
+      "loss": 0.6863,
+      "step": 8048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136991450719493e-05,
+      "loss": 0.784,
+      "step": 8049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3135400127099182e-05,
+      "loss": 0.7032,
+      "step": 8050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3133808715419524e-05,
+      "loss": 0.6896,
+      "step": 8051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132217215725205e-05,
+      "loss": 0.696,
+      "step": 8052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130625628060927e-05,
+      "loss": 0.6801,
+      "step": 8053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129033952471388e-05,
+      "loss": 0.6769,
+      "step": 8054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3127442189001296e-05,
+      "loss": 0.7266,
+      "step": 8055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125850337695354e-05,
+      "loss": 0.3409,
+      "step": 8056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3124258398598268e-05,
+      "loss": 0.7365,
+      "step": 8057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312266637175475e-05,
+      "loss": 0.679,
+      "step": 8058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121074257209514e-05,
+      "loss": 0.7448,
+      "step": 8059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3119482055007276e-05,
+      "loss": 0.7112,
+      "step": 8060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117889765192753e-05,
+      "loss": 0.7101,
+      "step": 8061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3116297387810662e-05,
+      "loss": 0.753,
+      "step": 8062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3114704922905728e-05,
+      "loss": 0.6235,
+      "step": 8063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113112370522678e-05,
+      "loss": 0.7301,
+      "step": 8064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111519730706237e-05,
+      "loss": 0.3543,
+      "step": 8065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109927003501138e-05,
+      "loss": 0.6578,
+      "step": 8066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310833418895211e-05,
+      "loss": 0.7428,
+      "step": 8067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3106741287103891e-05,
+      "loss": 0.7762,
+      "step": 8068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105148298001216e-05,
+      "loss": 0.7542,
+      "step": 8069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3103555221688829e-05,
+      "loss": 0.6992,
+      "step": 8070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310196205821147e-05,
+      "loss": 0.7047,
+      "step": 8071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100368807613882e-05,
+      "loss": 0.7121,
+      "step": 8072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3098775469940814e-05,
+      "loss": 0.7353,
+      "step": 8073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3097182045237014e-05,
+      "loss": 0.7229,
+      "step": 8074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095588533547239e-05,
+      "loss": 0.6823,
+      "step": 8075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093994934916236e-05,
+      "loss": 0.724,
+      "step": 8076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3092401249388773e-05,
+      "loss": 0.6687,
+      "step": 8077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090807477009597e-05,
+      "loss": 0.7885,
+      "step": 8078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3089213617823478e-05,
+      "loss": 0.6507,
+      "step": 8079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087619671875175e-05,
+      "loss": 0.7243,
+      "step": 8080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086025639209465e-05,
+      "loss": 0.7187,
+      "step": 8081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084431519871106e-05,
+      "loss": 0.6656,
+      "step": 8082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082837313904873e-05,
+      "loss": 0.6523,
+      "step": 8083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3081243021355542e-05,
+      "loss": 0.727,
+      "step": 8084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307964864226789e-05,
+      "loss": 0.7049,
+      "step": 8085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078054176686692e-05,
+      "loss": 0.6677,
+      "step": 8086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076459624656734e-05,
+      "loss": 0.6878,
+      "step": 8087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074864986222797e-05,
+      "loss": 0.6508,
+      "step": 8088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073270261429667e-05,
+      "loss": 0.6842,
+      "step": 8089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071675450322136e-05,
+      "loss": 0.709,
+      "step": 8090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070080552944993e-05,
+      "loss": 0.7324,
+      "step": 8091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306848556934303e-05,
+      "loss": 0.6887,
+      "step": 8092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3066890499561043e-05,
+      "loss": 0.7577,
+      "step": 8093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065295343643833e-05,
+      "loss": 0.6656,
+      "step": 8094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063700101636202e-05,
+      "loss": 0.7607,
+      "step": 8095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3062104773582943e-05,
+      "loss": 0.684,
+      "step": 8096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060509359528878e-05,
+      "loss": 0.7874,
+      "step": 8097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3058913859518801e-05,
+      "loss": 0.7015,
+      "step": 8098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057318273597531e-05,
+      "loss": 0.73,
+      "step": 8099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055722601809877e-05,
+      "loss": 0.7087,
+      "step": 8100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054126844200652e-05,
+      "loss": 0.7245,
+      "step": 8101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052531000814679e-05,
+      "loss": 0.7184,
+      "step": 8102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050935071696775e-05,
+      "loss": 0.7104,
+      "step": 8103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049339056891763e-05,
+      "loss": 0.6672,
+      "step": 8104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304774295644447e-05,
+      "loss": 0.6863,
+      "step": 8105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304614677039972e-05,
+      "loss": 0.7312,
+      "step": 8106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3044550498802346e-05,
+      "loss": 0.7022,
+      "step": 8107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042954141697175e-05,
+      "loss": 0.7065,
+      "step": 8108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3041357699129044e-05,
+      "loss": 0.651,
+      "step": 8109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3039761171142795e-05,
+      "loss": 0.688,
+      "step": 8110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038164557783262e-05,
+      "loss": 0.7472,
+      "step": 8111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036567859095286e-05,
+      "loss": 0.7839,
+      "step": 8112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034971075123712e-05,
+      "loss": 0.7205,
+      "step": 8113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303337420591339e-05,
+      "loss": 0.6911,
+      "step": 8114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031777251509166e-05,
+      "loss": 0.7407,
+      "step": 8115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030180211955892e-05,
+      "loss": 0.6547,
+      "step": 8116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028583087298418e-05,
+      "loss": 0.67,
+      "step": 8117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3026985877581609e-05,
+      "loss": 0.6959,
+      "step": 8118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025388582850311e-05,
+      "loss": 0.8306,
+      "step": 8119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023791203149396e-05,
+      "loss": 0.6903,
+      "step": 8120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022193738523723e-05,
+      "loss": 0.8013,
+      "step": 8121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3020596189018156e-05,
+      "loss": 0.694,
+      "step": 8122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3018998554677564e-05,
+      "loss": 0.6682,
+      "step": 8123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017400835546821e-05,
+      "loss": 0.7388,
+      "step": 8124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3015803031670796e-05,
+      "loss": 0.616,
+      "step": 8125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014205143094363e-05,
+      "loss": 0.7495,
+      "step": 8126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012607169862404e-05,
+      "loss": 0.7322,
+      "step": 8127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011009112019795e-05,
+      "loss": 0.744,
+      "step": 8128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009410969611419e-05,
+      "loss": 0.6558,
+      "step": 8129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300781274268216e-05,
+      "loss": 0.7662,
+      "step": 8130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006214431276909e-05,
+      "loss": 0.7268,
+      "step": 8131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3004616035440551e-05,
+      "loss": 0.8339,
+      "step": 8132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300301755521798e-05,
+      "loss": 0.7669,
+      "step": 8133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3001418990654092e-05,
+      "loss": 0.6638,
+      "step": 8134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999820341793778e-05,
+      "loss": 0.7649,
+      "step": 8135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998221608681944e-05,
+      "loss": 0.7194,
+      "step": 8136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996622791363484e-05,
+      "loss": 0.7256,
+      "step": 8137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2995023889883307e-05,
+      "loss": 0.7483,
+      "step": 8138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993424904286315e-05,
+      "loss": 0.7307,
+      "step": 8139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299182583461742e-05,
+      "loss": 0.7123,
+      "step": 8140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990226680921529e-05,
+      "loss": 0.7186,
+      "step": 8141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988627443243563e-05,
+      "loss": 0.7129,
+      "step": 8142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987028121628425e-05,
+      "loss": 0.6937,
+      "step": 8143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298542871612104e-05,
+      "loss": 0.6993,
+      "step": 8144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2983829226766328e-05,
+      "loss": 0.7261,
+      "step": 8145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982229653609215e-05,
+      "loss": 0.6919,
+      "step": 8146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980629996694618e-05,
+      "loss": 0.8132,
+      "step": 8147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979030256067468e-05,
+      "loss": 0.7136,
+      "step": 8148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977430431772694e-05,
+      "loss": 0.6888,
+      "step": 8149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297583052385523e-05,
+      "loss": 0.6479,
+      "step": 8150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974230532360008e-05,
+      "loss": 0.7562,
+      "step": 8151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972630457331969e-05,
+      "loss": 0.7342,
+      "step": 8152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971030298816043e-05,
+      "loss": 0.7247,
+      "step": 8153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969430056857177e-05,
+      "loss": 0.7819,
+      "step": 8154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2967829731500319e-05,
+      "loss": 0.7038,
+      "step": 8155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966229322790406e-05,
+      "loss": 0.7613,
+      "step": 8156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964628830772393e-05,
+      "loss": 0.72,
+      "step": 8157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2963028255491227e-05,
+      "loss": 0.7316,
+      "step": 8158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296142759699186e-05,
+      "loss": 0.724,
+      "step": 8159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295982685531925e-05,
+      "loss": 0.7099,
+      "step": 8160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958226030518355e-05,
+      "loss": 0.642,
+      "step": 8161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2956625122634137e-05,
+      "loss": 0.7286,
+      "step": 8162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2955024131711553e-05,
+      "loss": 0.7003,
+      "step": 8163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953423057795567e-05,
+      "loss": 0.7291,
+      "step": 8164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951821900931154e-05,
+      "loss": 0.3445,
+      "step": 8165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950220661163278e-05,
+      "loss": 0.6737,
+      "step": 8166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294861933853691e-05,
+      "loss": 0.703,
+      "step": 8167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2947017933097024e-05,
+      "loss": 0.6906,
+      "step": 8168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2945416444888597e-05,
+      "loss": 0.7067,
+      "step": 8169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943814873956609e-05,
+      "loss": 0.7147,
+      "step": 8170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294221322034604e-05,
+      "loss": 0.813,
+      "step": 8171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940611484101875e-05,
+      "loss": 0.7721,
+      "step": 8172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939009665269098e-05,
+      "loss": 0.7039,
+      "step": 8173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2937407763892691e-05,
+      "loss": 0.6796,
+      "step": 8174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935805780017656e-05,
+      "loss": 0.8174,
+      "step": 8175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934203713688977e-05,
+      "loss": 0.7355,
+      "step": 8176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2932601564951655e-05,
+      "loss": 0.7971,
+      "step": 8177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293099933385068e-05,
+      "loss": 0.7125,
+      "step": 8178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929397020431055e-05,
+      "loss": 0.7655,
+      "step": 8179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927794624737784e-05,
+      "loss": 0.721,
+      "step": 8180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292619214681587e-05,
+      "loss": 0.7565,
+      "step": 8181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924589586710317e-05,
+      "loss": 0.6887,
+      "step": 8182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922986944466134e-05,
+      "loss": 0.7324,
+      "step": 8183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921384220128333e-05,
+      "loss": 0.7585,
+      "step": 8184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291978141374193e-05,
+      "loss": 0.7474,
+      "step": 8185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2918178525351937e-05,
+      "loss": 0.7313,
+      "step": 8186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2916575555003373e-05,
+      "loss": 0.697,
+      "step": 8187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2914972502741258e-05,
+      "loss": 0.6617,
+      "step": 8188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2913369368610618e-05,
+      "loss": 0.8015,
+      "step": 8189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911766152656472e-05,
+      "loss": 0.66,
+      "step": 8190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2910162854923854e-05,
+      "loss": 0.7277,
+      "step": 8191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908559475457784e-05,
+      "loss": 0.7061,
+      "step": 8192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906956014303303e-05,
+      "loss": 0.7269,
+      "step": 8193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290535247150544e-05,
+      "loss": 0.7549,
+      "step": 8194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903748847109232e-05,
+      "loss": 0.3308,
+      "step": 8195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290214514115972e-05,
+      "loss": 0.6836,
+      "step": 8196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900541353701943e-05,
+      "loss": 0.6694,
+      "step": 8197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898937484780944e-05,
+      "loss": 0.71,
+      "step": 8198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289733353444177e-05,
+      "loss": 0.6678,
+      "step": 8199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2895729502729465e-05,
+      "loss": 0.7141,
+      "step": 8200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2894125389689087e-05,
+      "loss": 0.735,
+      "step": 8201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892521195365679e-05,
+      "loss": 0.6575,
+      "step": 8202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.28909169198043e-05,
+      "loss": 0.7443,
+      "step": 8203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2889312563050007e-05,
+      "loss": 0.7913,
+      "step": 8204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288770812514786e-05,
+      "loss": 0.7739,
+      "step": 8205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2886103606142916e-05,
+      "loss": 0.6483,
+      "step": 8206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2884499006080249e-05,
+      "loss": 0.7323,
+      "step": 8207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288289432500491e-05,
+      "loss": 0.6677,
+      "step": 8208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881289562961981e-05,
+      "loss": 0.6924,
+      "step": 8209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879684719996526e-05,
+      "loss": 0.697,
+      "step": 8210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287807979615362e-05,
+      "loss": 0.6893,
+      "step": 8211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876474791478335e-05,
+      "loss": 0.7372,
+      "step": 8212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287486970601575e-05,
+      "loss": 0.7568,
+      "step": 8213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2873264539810945e-05,
+      "loss": 0.8123,
+      "step": 8214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871659292909004e-05,
+      "loss": 0.6514,
+      "step": 8215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870053965355007e-05,
+      "loss": 0.7499,
+      "step": 8216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868448557194046e-05,
+      "loss": 0.7108,
+      "step": 8217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2866843068471204e-05,
+      "loss": 0.7481,
+      "step": 8218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865237499231573e-05,
+      "loss": 0.726,
+      "step": 8219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2863631849520252e-05,
+      "loss": 0.7234,
+      "step": 8220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862026119382332e-05,
+      "loss": 0.6899,
+      "step": 8221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860420308862906e-05,
+      "loss": 0.6882,
+      "step": 8222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2858814418007085e-05,
+      "loss": 0.6987,
+      "step": 8223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285720844685996e-05,
+      "loss": 0.7155,
+      "step": 8224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855602395466642e-05,
+      "loss": 0.7175,
+      "step": 8225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853996263872239e-05,
+      "loss": 0.6249,
+      "step": 8226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852390052121857e-05,
+      "loss": 0.7064,
+      "step": 8227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2850783760260608e-05,
+      "loss": 0.6648,
+      "step": 8228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28491773883336e-05,
+      "loss": 0.7127,
+      "step": 8229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847570936385962e-05,
+      "loss": 0.7303,
+      "step": 8230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2845964404462801e-05,
+      "loss": 0.6866,
+      "step": 8231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844357792609242e-05,
+      "loss": 0.714,
+      "step": 8232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2842751100870405e-05,
+      "loss": 0.733,
+      "step": 8233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841144329291416e-05,
+      "loss": 0.7036,
+      "step": 8234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839537477917401e-05,
+      "loss": 0.7142,
+      "step": 8235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837930546793491e-05,
+      "loss": 0.7583,
+      "step": 8236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283632353596482e-05,
+      "loss": 0.7107,
+      "step": 8237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834716445476514e-05,
+      "loss": 0.7436,
+      "step": 8238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833109275373712e-05,
+      "loss": 0.7057,
+      "step": 8239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283150202570156e-05,
+      "loss": 0.3373,
+      "step": 8240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2829894696505187e-05,
+      "loss": 0.7172,
+      "step": 8241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828287287829744e-05,
+      "loss": 0.7038,
+      "step": 8242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282667979972037e-05,
+      "loss": 0.602,
+      "step": 8243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825072232222216e-05,
+      "loss": 0.6921,
+      "step": 8244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282346458538043e-05,
+      "loss": 0.7178,
+      "step": 8245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2821856859240166e-05,
+      "loss": 0.6965,
+      "step": 8246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2820249053846574e-05,
+      "loss": 0.7346,
+      "step": 8247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281864116924481e-05,
+      "loss": 0.7113,
+      "step": 8248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817033205480033e-05,
+      "loss": 0.7026,
+      "step": 8249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281542516259741e-05,
+      "loss": 0.691,
+      "step": 8250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813817040642093e-05,
+      "loss": 0.799,
+      "step": 8251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2812208839659257e-05,
+      "loss": 0.6276,
+      "step": 8252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281060055969406e-05,
+      "loss": 0.6985,
+      "step": 8253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808992200791676e-05,
+      "loss": 0.6841,
+      "step": 8254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280738376299728e-05,
+      "loss": 0.7387,
+      "step": 8255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280577524635604e-05,
+      "loss": 0.6978,
+      "step": 8256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804166650913135e-05,
+      "loss": 0.6992,
+      "step": 8257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802557976713741e-05,
+      "loss": 0.7267,
+      "step": 8258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2800949223803039e-05,
+      "loss": 0.7262,
+      "step": 8259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799340392226213e-05,
+      "loss": 0.7211,
+      "step": 8260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279773148202845e-05,
+      "loss": 0.7045,
+      "step": 8261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796122493254932e-05,
+      "loss": 0.6832,
+      "step": 8262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2794513425950851e-05,
+      "loss": 0.7724,
+      "step": 8263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27929042801614e-05,
+      "loss": 0.7026,
+      "step": 8264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279129505593177e-05,
+      "loss": 0.7053,
+      "step": 8265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789685753307159e-05,
+      "loss": 0.7298,
+      "step": 8266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788076372332761e-05,
+      "loss": 0.761,
+      "step": 8267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278646691305378e-05,
+      "loss": 0.7115,
+      "step": 8268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278485737551542e-05,
+      "loss": 0.7391,
+      "step": 8269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783247759762882e-05,
+      "loss": 0.6938,
+      "step": 8270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2781638065841376e-05,
+      "loss": 0.7325,
+      "step": 8271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780028293796108e-05,
+      "loss": 0.7491,
+      "step": 8272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778418443672289e-05,
+      "loss": 0.6728,
+      "step": 8273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776808515515136e-05,
+      "loss": 0.7312,
+      "step": 8274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2775198509369863e-05,
+      "loss": 0.6804,
+      "step": 8275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2773588425281692e-05,
+      "loss": 0.7104,
+      "step": 8276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771978263295835e-05,
+      "loss": 0.7075,
+      "step": 8277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2770368023457515e-05,
+      "loss": 0.676,
+      "step": 8278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2768757705811965e-05,
+      "loss": 0.7101,
+      "step": 8279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767147310404404e-05,
+      "loss": 0.6952,
+      "step": 8280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765536837280061e-05,
+      "loss": 0.7111,
+      "step": 8281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763926286484174e-05,
+      "loss": 0.3834,
+      "step": 8282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2762315658061968e-05,
+      "loss": 0.7021,
+      "step": 8283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2760704952058677e-05,
+      "loss": 0.7566,
+      "step": 8284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759094168519548e-05,
+      "loss": 0.3564,
+      "step": 8285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757483307489817e-05,
+      "loss": 0.6983,
+      "step": 8286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275587236901472e-05,
+      "loss": 0.6925,
+      "step": 8287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275426135313951e-05,
+      "loss": 0.7181,
+      "step": 8288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2752650259909425e-05,
+      "loss": 0.6601,
+      "step": 8289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751039089369722e-05,
+      "loss": 0.8091,
+      "step": 8290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274942784156564e-05,
+      "loss": 0.7362,
+      "step": 8291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2747816516542445e-05,
+      "loss": 0.7728,
+      "step": 8292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746205114345384e-05,
+      "loss": 0.8177,
+      "step": 8293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274459363501971e-05,
+      "loss": 0.7657,
+      "step": 8294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2742982078610696e-05,
+      "loss": 0.7672,
+      "step": 8295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274137044516359e-05,
+      "loss": 0.7715,
+      "step": 8296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739758734723663e-05,
+      "loss": 0.6955,
+      "step": 8297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738146947336177e-05,
+      "loss": 0.7244,
+      "step": 8298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27365350830464e-05,
+      "loss": 0.6947,
+      "step": 8299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2734923141899605e-05,
+      "loss": 0.6677,
+      "step": 8300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733311123941063e-05,
+      "loss": 0.6767,
+      "step": 8301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273169902921605e-05,
+      "loss": 0.8474,
+      "step": 8302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730086857769837e-05,
+      "loss": 0.6901,
+      "step": 8303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2728474609647706e-05,
+      "loss": 0.6763,
+      "step": 8304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2726862284894939e-05,
+      "loss": 0.7337,
+      "step": 8305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725249883556819e-05,
+      "loss": 0.6454,
+      "step": 8306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723637405678628e-05,
+      "loss": 0.693,
+      "step": 8307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722024851305656e-05,
+      "loss": 0.6932,
+      "step": 8308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720412220483193e-05,
+      "loss": 0.7639,
+      "step": 8309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718799513256528e-05,
+      "loss": 0.7313,
+      "step": 8310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717186729670958e-05,
+      "loss": 0.6978,
+      "step": 8311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715573869771777e-05,
+      "loss": 0.7316,
+      "step": 8312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2713960933604283e-05,
+      "loss": 0.7322,
+      "step": 8313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2712347921213772e-05,
+      "loss": 0.6776,
+      "step": 8314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2710734832645557e-05,
+      "loss": 0.7254,
+      "step": 8315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709121667944933e-05,
+      "loss": 0.7165,
+      "step": 8316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270750842715721e-05,
+      "loss": 0.7447,
+      "step": 8317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2705895110327693e-05,
+      "loss": 0.712,
+      "step": 8318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704281717501699e-05,
+      "loss": 0.6945,
+      "step": 8319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702668248724537e-05,
+      "loss": 0.725,
+      "step": 8320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701054704041524e-05,
+      "loss": 0.7224,
+      "step": 8321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699441083497977e-05,
+      "loss": 0.6895,
+      "step": 8322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697827387139215e-05,
+      "loss": 0.7353,
+      "step": 8323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696213615010553e-05,
+      "loss": 0.6905,
+      "step": 8324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269459976715733e-05,
+      "loss": 0.7275,
+      "step": 8325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692985843624857e-05,
+      "loss": 0.7236,
+      "step": 8326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269137184445847e-05,
+      "loss": 0.7044,
+      "step": 8327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689757769703495e-05,
+      "loss": 0.6853,
+      "step": 8328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688143619405266e-05,
+      "loss": 0.7075,
+      "step": 8329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2686529393609116e-05,
+      "loss": 0.8085,
+      "step": 8330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684915092360387e-05,
+      "loss": 0.7317,
+      "step": 8331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268330071570441e-05,
+      "loss": 0.7436,
+      "step": 8332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681686263686528e-05,
+      "loss": 0.7759,
+      "step": 8333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680071736352082e-05,
+      "loss": 0.7455,
+      "step": 8334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2678457133746422e-05,
+      "loss": 0.7557,
+      "step": 8335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676842455914891e-05,
+      "loss": 0.6581,
+      "step": 8336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267522770290284e-05,
+      "loss": 0.7447,
+      "step": 8337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673612874755619e-05,
+      "loss": 0.6924,
+      "step": 8338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267199797151858e-05,
+      "loss": 0.7309,
+      "step": 8339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267038299323708e-05,
+      "loss": 0.7082,
+      "step": 8340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266876793995648e-05,
+      "loss": 0.7284,
+      "step": 8341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667152811722132e-05,
+      "loss": 0.775,
+      "step": 8342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665537608579404e-05,
+      "loss": 0.6986,
+      "step": 8343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2663922330573655e-05,
+      "loss": 0.7184,
+      "step": 8344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662306977750255e-05,
+      "loss": 0.6835,
+      "step": 8345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660691550154571e-05,
+      "loss": 0.6826,
+      "step": 8346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2659076047831972e-05,
+      "loss": 0.6976,
+      "step": 8347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2657460470827834e-05,
+      "loss": 0.6842,
+      "step": 8348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655844819187522e-05,
+      "loss": 0.6994,
+      "step": 8349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2654229092956422e-05,
+      "loss": 0.6524,
+      "step": 8350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265261329217991e-05,
+      "loss": 0.7463,
+      "step": 8351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2650997416903367e-05,
+      "loss": 0.7726,
+      "step": 8352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649381467172173e-05,
+      "loss": 0.7333,
+      "step": 8353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2647765443031715e-05,
+      "loss": 0.6872,
+      "step": 8354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2646149344527376e-05,
+      "loss": 0.7339,
+      "step": 8355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2644533171704552e-05,
+      "loss": 0.7075,
+      "step": 8356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264291692460863e-05,
+      "loss": 0.7252,
+      "step": 8357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641300603285e-05,
+      "loss": 0.7294,
+      "step": 8358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639684207779062e-05,
+      "loss": 0.7355,
+      "step": 8359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2638067738136215e-05,
+      "loss": 0.6366,
+      "step": 8360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263645119440185e-05,
+      "loss": 0.6718,
+      "step": 8361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2634834576621378e-05,
+      "loss": 0.6499,
+      "step": 8362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2633217884840196e-05,
+      "loss": 0.7601,
+      "step": 8363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631601119103714e-05,
+      "loss": 0.714,
+      "step": 8364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2629984279457334e-05,
+      "loss": 0.7437,
+      "step": 8365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2628367365946472e-05,
+      "loss": 0.7167,
+      "step": 8366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2626750378616539e-05,
+      "loss": 0.6668,
+      "step": 8367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2625133317512945e-05,
+      "loss": 0.7287,
+      "step": 8368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2623516182681106e-05,
+      "loss": 0.704,
+      "step": 8369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621898974166446e-05,
+      "loss": 0.6554,
+      "step": 8370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620281692014383e-05,
+      "loss": 0.7252,
+      "step": 8371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618664336270335e-05,
+      "loss": 0.7027,
+      "step": 8372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2617046906979726e-05,
+      "loss": 0.751,
+      "step": 8373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2615429404187989e-05,
+      "loss": 0.7038,
+      "step": 8374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613811827940548e-05,
+      "loss": 0.7852,
+      "step": 8375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612194178282835e-05,
+      "loss": 0.7812,
+      "step": 8376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610576455260284e-05,
+      "loss": 0.7,
+      "step": 8377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2608958658918321e-05,
+      "loss": 0.7653,
+      "step": 8378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607340789302392e-05,
+      "loss": 0.7419,
+      "step": 8379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605722846457934e-05,
+      "loss": 0.7273,
+      "step": 8380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2604104830430388e-05,
+      "loss": 0.682,
+      "step": 8381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602486741265192e-05,
+      "loss": 0.7461,
+      "step": 8382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600868579007795e-05,
+      "loss": 0.6292,
+      "step": 8383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2599250343703644e-05,
+      "loss": 0.6833,
+      "step": 8384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597632035398185e-05,
+      "loss": 0.7377,
+      "step": 8385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596013654136874e-05,
+      "loss": 0.6703,
+      "step": 8386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2594395199965162e-05,
+      "loss": 0.7842,
+      "step": 8387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.25927766729285e-05,
+      "loss": 0.6905,
+      "step": 8388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591158073072347e-05,
+      "loss": 0.7297,
+      "step": 8389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589539400442168e-05,
+      "loss": 0.6794,
+      "step": 8390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.258792065508342e-05,
+      "loss": 0.6959,
+      "step": 8391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2586301837041563e-05,
+      "loss": 0.7638,
+      "step": 8392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584682946362067e-05,
+      "loss": 0.7741,
+      "step": 8393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583063983090398e-05,
+      "loss": 0.666,
+      "step": 8394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2581444947272027e-05,
+      "loss": 0.7322,
+      "step": 8395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2579825838952427e-05,
+      "loss": 0.6648,
+      "step": 8396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2578206658177064e-05,
+      "loss": 0.7138,
+      "step": 8397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2576587404991418e-05,
+      "loss": 0.7203,
+      "step": 8398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2574968079440969e-05,
+      "loss": 0.7759,
+      "step": 8399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573348681571188e-05,
+      "loss": 0.8317,
+      "step": 8400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257172921142757e-05,
+      "loss": 0.7662,
+      "step": 8401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570109669055589e-05,
+      "loss": 0.8279,
+      "step": 8402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568490054500733e-05,
+      "loss": 0.7146,
+      "step": 8403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2566870367808488e-05,
+      "loss": 0.7127,
+      "step": 8404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.256525060902435e-05,
+      "loss": 0.7574,
+      "step": 8405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2563630778193805e-05,
+      "loss": 0.7296,
+      "step": 8406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2562010875362347e-05,
+      "loss": 0.7379,
+      "step": 8407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2560390900575472e-05,
+      "loss": 0.7186,
+      "step": 8408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558770853878681e-05,
+      "loss": 0.6876,
+      "step": 8409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2557150735317471e-05,
+      "loss": 0.6777,
+      "step": 8410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2555530544937345e-05,
+      "loss": 0.6488,
+      "step": 8411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553910282783807e-05,
+      "loss": 0.747,
+      "step": 8412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552289948902362e-05,
+      "loss": 0.7022,
+      "step": 8413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550669543338519e-05,
+      "loss": 0.6888,
+      "step": 8414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549049066137785e-05,
+      "loss": 0.6785,
+      "step": 8415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547428517345677e-05,
+      "loss": 0.7515,
+      "step": 8416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545807897007705e-05,
+      "loss": 0.7422,
+      "step": 8417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544187205169385e-05,
+      "loss": 0.6865,
+      "step": 8418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542566441876237e-05,
+      "loss": 0.6748,
+      "step": 8419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2540945607173783e-05,
+      "loss": 0.6765,
+      "step": 8420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539324701107539e-05,
+      "loss": 0.6398,
+      "step": 8421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537703723723034e-05,
+      "loss": 0.6781,
+      "step": 8422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253608267506579e-05,
+      "loss": 0.3292,
+      "step": 8423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534461555181337e-05,
+      "loss": 0.7158,
+      "step": 8424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2532840364115206e-05,
+      "loss": 0.7426,
+      "step": 8425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253121910191293e-05,
+      "loss": 0.7471,
+      "step": 8426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529597768620039e-05,
+      "loss": 0.7053,
+      "step": 8427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2527976364282073e-05,
+      "loss": 0.6689,
+      "step": 8428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526354888944564e-05,
+      "loss": 0.7009,
+      "step": 8429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252473334265306e-05,
+      "loss": 0.6891,
+      "step": 8430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523111725453098e-05,
+      "loss": 0.7103,
+      "step": 8431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521490037390224e-05,
+      "loss": 0.6761,
+      "step": 8432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2519868278509984e-05,
+      "loss": 0.83,
+      "step": 8433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251824644885792e-05,
+      "loss": 0.6944,
+      "step": 8434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251662454847959e-05,
+      "loss": 0.7398,
+      "step": 8435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515002577420542e-05,
+      "loss": 0.7027,
+      "step": 8436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513380535726331e-05,
+      "loss": 0.7116,
+      "step": 8437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2511758423442515e-05,
+      "loss": 0.6635,
+      "step": 8438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510136240614648e-05,
+      "loss": 0.6463,
+      "step": 8439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508513987288289e-05,
+      "loss": 0.7354,
+      "step": 8440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506891663509005e-05,
+      "loss": 0.7383,
+      "step": 8441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505269269322358e-05,
+      "loss": 0.6231,
+      "step": 8442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250364680477391e-05,
+      "loss": 0.6885,
+      "step": 8443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250202426990923e-05,
+      "loss": 0.7277,
+      "step": 8444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2500401664773894e-05,
+      "loss": 0.7908,
+      "step": 8445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498778989413464e-05,
+      "loss": 0.7252,
+      "step": 8446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497156243873522e-05,
+      "loss": 0.7607,
+      "step": 8447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249553342819964e-05,
+      "loss": 0.7049,
+      "step": 8448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2493910542437395e-05,
+      "loss": 0.678,
+      "step": 8449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492287586632368e-05,
+      "loss": 0.7172,
+      "step": 8450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2490664560830139e-05,
+      "loss": 0.667,
+      "step": 8451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489041465076297e-05,
+      "loss": 0.7236,
+      "step": 8452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2487418299416419e-05,
+      "loss": 0.618,
+      "step": 8453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2485795063896096e-05,
+      "loss": 0.7309,
+      "step": 8454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248417175856092e-05,
+      "loss": 0.6993,
+      "step": 8455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482548383456482e-05,
+      "loss": 0.7805,
+      "step": 8456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248092493862837e-05,
+      "loss": 0.7288,
+      "step": 8457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2479301424122186e-05,
+      "loss": 0.7884,
+      "step": 8458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2477677839983522e-05,
+      "loss": 0.6498,
+      "step": 8459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247605418625798e-05,
+      "loss": 0.7357,
+      "step": 8460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474430462991163e-05,
+      "loss": 0.7278,
+      "step": 8461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2472806670228667e-05,
+      "loss": 0.7026,
+      "step": 8462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471182808016107e-05,
+      "loss": 0.6661,
+      "step": 8463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246955887639908e-05,
+      "loss": 0.6534,
+      "step": 8464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2467934875423204e-05,
+      "loss": 0.7056,
+      "step": 8465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466310805134083e-05,
+      "loss": 0.725,
+      "step": 8466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2464686665577334e-05,
+      "loss": 0.7395,
+      "step": 8467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463062456798572e-05,
+      "loss": 0.7234,
+      "step": 8468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2461438178843409e-05,
+      "loss": 0.6186,
+      "step": 8469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459813831757468e-05,
+      "loss": 0.7709,
+      "step": 8470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245818941558637e-05,
+      "loss": 0.7161,
+      "step": 8471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456564930375734e-05,
+      "loss": 0.7219,
+      "step": 8472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454940376171187e-05,
+      "loss": 0.6815,
+      "step": 8473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453315753018356e-05,
+      "loss": 0.6559,
+      "step": 8474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451691060962867e-05,
+      "loss": 0.7022,
+      "step": 8475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2450066300050354e-05,
+      "loss": 0.7258,
+      "step": 8476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448441470326445e-05,
+      "loss": 0.6254,
+      "step": 8477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446816571836779e-05,
+      "loss": 0.723,
+      "step": 8478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445191604626984e-05,
+      "loss": 0.6969,
+      "step": 8479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244356656874271e-05,
+      "loss": 0.6879,
+      "step": 8480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2441941464229587e-05,
+      "loss": 0.663,
+      "step": 8481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440316291133261e-05,
+      "loss": 0.6502,
+      "step": 8482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438691049499374e-05,
+      "loss": 0.6973,
+      "step": 8483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437065739373573e-05,
+      "loss": 0.772,
+      "step": 8484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435440360801504e-05,
+      "loss": 0.6874,
+      "step": 8485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433814913828821e-05,
+      "loss": 0.6993,
+      "step": 8486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432189398501176e-05,
+      "loss": 0.6705,
+      "step": 8487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2430563814864214e-05,
+      "loss": 0.7386,
+      "step": 8488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428938162963596e-05,
+      "loss": 0.726,
+      "step": 8489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427312442844976e-05,
+      "loss": 0.7002,
+      "step": 8490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425686654554023e-05,
+      "loss": 0.7213,
+      "step": 8491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2424060798136389e-05,
+      "loss": 0.7242,
+      "step": 8492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422434873637737e-05,
+      "loss": 0.6932,
+      "step": 8493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420808881103735e-05,
+      "loss": 0.6675,
+      "step": 8494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419182820580048e-05,
+      "loss": 0.7677,
+      "step": 8495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417556692112347e-05,
+      "loss": 0.7568,
+      "step": 8496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2415930495746301e-05,
+      "loss": 0.7168,
+      "step": 8497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414304231527586e-05,
+      "loss": 0.6667,
+      "step": 8498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412677899501865e-05,
+      "loss": 0.7586,
+      "step": 8499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411051499714832e-05,
+      "loss": 0.7151,
+      "step": 8500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2409425032212152e-05,
+      "loss": 0.6678,
+      "step": 8501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407798497039509e-05,
+      "loss": 0.7384,
+      "step": 8502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406171894242587e-05,
+      "loss": 0.7582,
+      "step": 8503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2404545223867067e-05,
+      "loss": 0.7385,
+      "step": 8504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402918485958635e-05,
+      "loss": 0.7519,
+      "step": 8505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401291680562983e-05,
+      "loss": 0.76,
+      "step": 8506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2399664807725798e-05,
+      "loss": 0.7048,
+      "step": 8507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398037867492769e-05,
+      "loss": 0.6905,
+      "step": 8508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239641085990959e-05,
+      "loss": 0.7213,
+      "step": 8509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2394783785021964e-05,
+      "loss": 0.6971,
+      "step": 8510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393156642875579e-05,
+      "loss": 0.7292,
+      "step": 8511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2391529433516139e-05,
+      "loss": 0.6688,
+      "step": 8512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389902156989344e-05,
+      "loss": 0.7346,
+      "step": 8513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388274813340892e-05,
+      "loss": 0.6473,
+      "step": 8514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386647402616494e-05,
+      "loss": 0.6705,
+      "step": 8515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2385019924861856e-05,
+      "loss": 0.7326,
+      "step": 8516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383392380122687e-05,
+      "loss": 0.6765,
+      "step": 8517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2381764768444696e-05,
+      "loss": 0.6633,
+      "step": 8518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380137089873592e-05,
+      "loss": 0.6988,
+      "step": 8519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378509344455094e-05,
+      "loss": 0.633,
+      "step": 8520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376881532234919e-05,
+      "loss": 0.721,
+      "step": 8521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2375253653258778e-05,
+      "loss": 0.7357,
+      "step": 8522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2373625707572401e-05,
+      "loss": 0.7355,
+      "step": 8523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371997695221498e-05,
+      "loss": 0.6993,
+      "step": 8524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370369616251803e-05,
+      "loss": 0.784,
+      "step": 8525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368741470709036e-05,
+      "loss": 0.6923,
+      "step": 8526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367113258638925e-05,
+      "loss": 0.8069,
+      "step": 8527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365484980087198e-05,
+      "loss": 0.6273,
+      "step": 8528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2363856635099591e-05,
+      "loss": 0.7463,
+      "step": 8529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2362228223721833e-05,
+      "loss": 0.7104,
+      "step": 8530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2360599745999657e-05,
+      "loss": 0.5939,
+      "step": 8531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2358971201978801e-05,
+      "loss": 0.7388,
+      "step": 8532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235734259170501e-05,
+      "loss": 0.738,
+      "step": 8533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2355713915224014e-05,
+      "loss": 0.6917,
+      "step": 8534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354085172581562e-05,
+      "loss": 0.7218,
+      "step": 8535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2352456363823397e-05,
+      "loss": 0.7034,
+      "step": 8536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2350827488995262e-05,
+      "loss": 0.6851,
+      "step": 8537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2349198548142911e-05,
+      "loss": 0.6851,
+      "step": 8538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347569541312086e-05,
+      "loss": 0.6796,
+      "step": 8539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2345940468548542e-05,
+      "loss": 0.6873,
+      "step": 8540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344311329898033e-05,
+      "loss": 0.7203,
+      "step": 8541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342682125406315e-05,
+      "loss": 0.7419,
+      "step": 8542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2341052855119144e-05,
+      "loss": 0.7627,
+      "step": 8543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339423519082277e-05,
+      "loss": 0.7773,
+      "step": 8544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337794117341476e-05,
+      "loss": 0.7163,
+      "step": 8545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2336164649942508e-05,
+      "loss": 0.671,
+      "step": 8546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233453511693113e-05,
+      "loss": 0.7008,
+      "step": 8547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233290551835311e-05,
+      "loss": 0.7305,
+      "step": 8548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331275854254222e-05,
+      "loss": 0.7271,
+      "step": 8549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232964612468023e-05,
+      "loss": 0.6298,
+      "step": 8550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2328016329676909e-05,
+      "loss": 0.7242,
+      "step": 8551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232638646929003e-05,
+      "loss": 0.6806,
+      "step": 8552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232475654356537e-05,
+      "loss": 0.778,
+      "step": 8553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2323126552548702e-05,
+      "loss": 0.7188,
+      "step": 8554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321496496285814e-05,
+      "loss": 0.7238,
+      "step": 8555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2319866374822485e-05,
+      "loss": 0.7845,
+      "step": 8556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231823618820449e-05,
+      "loss": 0.6839,
+      "step": 8557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316605936477621e-05,
+      "loss": 0.7294,
+      "step": 8558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231497561968766e-05,
+      "loss": 0.7201,
+      "step": 8559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23133452378804e-05,
+      "loss": 0.6449,
+      "step": 8560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311714791101628e-05,
+      "loss": 0.7232,
+      "step": 8561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310084279397137e-05,
+      "loss": 0.6808,
+      "step": 8562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2308453702812719e-05,
+      "loss": 0.6178,
+      "step": 8563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2306823061394171e-05,
+      "loss": 0.6329,
+      "step": 8564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305192355187292e-05,
+      "loss": 0.7498,
+      "step": 8565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2303561584237877e-05,
+      "loss": 0.7118,
+      "step": 8566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230193074859173e-05,
+      "loss": 0.7141,
+      "step": 8567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300299848294657e-05,
+      "loss": 0.7338,
+      "step": 8568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298668883392455e-05,
+      "loss": 0.7199,
+      "step": 8569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2297037853930935e-05,
+      "loss": 0.71,
+      "step": 8570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2295406759955908e-05,
+      "loss": 0.7259,
+      "step": 8571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229377560151318e-05,
+      "loss": 0.7314,
+      "step": 8572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292144378648565e-05,
+      "loss": 0.7404,
+      "step": 8573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2290513091407871e-05,
+      "loss": 0.7282,
+      "step": 8574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288881739836925e-05,
+      "loss": 0.7746,
+      "step": 8575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287250323981534e-05,
+      "loss": 0.678,
+      "step": 8576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2285618843887522e-05,
+      "loss": 0.7679,
+      "step": 8577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283987299600711e-05,
+      "loss": 0.698,
+      "step": 8578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2282355691166914e-05,
+      "loss": 0.7208,
+      "step": 8579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280724018631968e-05,
+      "loss": 0.7106,
+      "step": 8580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2279092282041696e-05,
+      "loss": 0.7042,
+      "step": 8581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2277460481441926e-05,
+      "loss": 0.7181,
+      "step": 8582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275828616878485e-05,
+      "loss": 0.7607,
+      "step": 8583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274196688397203e-05,
+      "loss": 0.7157,
+      "step": 8584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227256469604392e-05,
+      "loss": 0.6365,
+      "step": 8585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270932639864471e-05,
+      "loss": 0.7845,
+      "step": 8586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269300519904687e-05,
+      "loss": 0.6935,
+      "step": 8587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2267668336210411e-05,
+      "loss": 0.7176,
+      "step": 8588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2266036088827484e-05,
+      "loss": 0.7389,
+      "step": 8589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2264403777801748e-05,
+      "loss": 0.7727,
+      "step": 8590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262771403179048e-05,
+      "loss": 0.697,
+      "step": 8591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2261138965005229e-05,
+      "loss": 0.727,
+      "step": 8592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2259506463326138e-05,
+      "loss": 0.7527,
+      "step": 8593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257873898187625e-05,
+      "loss": 0.6893,
+      "step": 8594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256241269635543e-05,
+      "loss": 0.6572,
+      "step": 8595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254608577715745e-05,
+      "loss": 0.6377,
+      "step": 8596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252975822474085e-05,
+      "loss": 0.6415,
+      "step": 8597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251343003956421e-05,
+      "loss": 0.6632,
+      "step": 8598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2249710122208609e-05,
+      "loss": 0.7676,
+      "step": 8599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2248077177276512e-05,
+      "loss": 0.673,
+      "step": 8600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246444169205994e-05,
+      "loss": 0.7491,
+      "step": 8601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244811098042912e-05,
+      "loss": 0.7548,
+      "step": 8602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243177963833138e-05,
+      "loss": 0.7402,
+      "step": 8603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241544766622536e-05,
+      "loss": 0.7217,
+      "step": 8604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239911506456975e-05,
+      "loss": 0.6791,
+      "step": 8605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223827818338233e-05,
+      "loss": 0.6674,
+      "step": 8606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223664479744447e-05,
+      "loss": 0.6692,
+      "step": 8607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2235011348689274e-05,
+      "loss": 0.3461,
+      "step": 8608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223337783716261e-05,
+      "loss": 0.7318,
+      "step": 8609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223174426291036e-05,
+      "loss": 0.7135,
+      "step": 8610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230110625978411e-05,
+      "loss": 0.6869,
+      "step": 8611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228476926412636e-05,
+      "loss": 0.7498,
+      "step": 8612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2226843164258917e-05,
+      "loss": 0.6761,
+      "step": 8613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.6683,
+      "step": 8614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2223575452371206e-05,
+      "loss": 0.7086,
+      "step": 8615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221941502728985e-05,
+      "loss": 0.7374,
+      "step": 8616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2220307490682373e-05,
+      "loss": 0.7773,
+      "step": 8617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218673416277267e-05,
+      "loss": 0.6691,
+      "step": 8618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217039279559553e-05,
+      "loss": 0.6751,
+      "step": 8619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215405080575133e-05,
+      "loss": 0.6929,
+      "step": 8620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213770819369903e-05,
+      "loss": 0.678,
+      "step": 8621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221213649598976e-05,
+      "loss": 0.7209,
+      "step": 8622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221050211048061e-05,
+      "loss": 0.6808,
+      "step": 8623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208867662888346e-05,
+      "loss": 0.6281,
+      "step": 8624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220723315325888e-05,
+      "loss": 0.7438,
+      "step": 8625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2205598581638116e-05,
+      "loss": 0.7468,
+      "step": 8626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203963948071964e-05,
+      "loss": 0.7174,
+      "step": 8627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2202329252606329e-05,
+      "loss": 0.7356,
+      "step": 8628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200694495287122e-05,
+      "loss": 0.7294,
+      "step": 8629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199059676160263e-05,
+      "loss": 0.8217,
+      "step": 8630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197424795271662e-05,
+      "loss": 0.7214,
+      "step": 8631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195789852667235e-05,
+      "loss": 0.7395,
+      "step": 8632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194154848392901e-05,
+      "loss": 0.6198,
+      "step": 8633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219251978249458e-05,
+      "loss": 0.7897,
+      "step": 8634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190884655018193e-05,
+      "loss": 0.697,
+      "step": 8635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189249466009664e-05,
+      "loss": 0.7586,
+      "step": 8636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187614215514923e-05,
+      "loss": 0.6562,
+      "step": 8637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185978903579889e-05,
+      "loss": 0.7562,
+      "step": 8638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184343530250491e-05,
+      "loss": 0.7243,
+      "step": 8639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182708095572667e-05,
+      "loss": 0.7305,
+      "step": 8640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.218107259959234e-05,
+      "loss": 0.6879,
+      "step": 8641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2179437042355449e-05,
+      "loss": 0.7226,
+      "step": 8642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177801423907928e-05,
+      "loss": 0.7159,
+      "step": 8643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2176165744295716e-05,
+      "loss": 0.6481,
+      "step": 8644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2174530003564749e-05,
+      "loss": 0.6744,
+      "step": 8645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217289420176097e-05,
+      "loss": 0.7671,
+      "step": 8646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217125833893032e-05,
+      "loss": 0.73,
+      "step": 8647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2169622415118745e-05,
+      "loss": 0.7559,
+      "step": 8648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167986430372187e-05,
+      "loss": 0.6661,
+      "step": 8649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166350384736596e-05,
+      "loss": 0.6858,
+      "step": 8650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164714278257922e-05,
+      "loss": 0.6229,
+      "step": 8651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163078110982114e-05,
+      "loss": 0.6579,
+      "step": 8652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2161441882955127e-05,
+      "loss": 0.725,
+      "step": 8653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159805594222912e-05,
+      "loss": 0.6444,
+      "step": 8654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2158169244831428e-05,
+      "loss": 0.7447,
+      "step": 8655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215653283482663e-05,
+      "loss": 0.7277,
+      "step": 8656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2154896364254483e-05,
+      "loss": 0.6896,
+      "step": 8657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215325983316094e-05,
+      "loss": 0.7633,
+      "step": 8658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215162324159197e-05,
+      "loss": 0.7679,
+      "step": 8659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2149986589593535e-05,
+      "loss": 0.7414,
+      "step": 8660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21483498772116e-05,
+      "loss": 0.7182,
+      "step": 8661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146713104492137e-05,
+      "loss": 0.7008,
+      "step": 8662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145076271481113e-05,
+      "loss": 0.6846,
+      "step": 8663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143439378224497e-05,
+      "loss": 0.6994,
+      "step": 8664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141802424768265e-05,
+      "loss": 0.7181,
+      "step": 8665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140165411158395e-05,
+      "loss": 0.7216,
+      "step": 8666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2138528337440857e-05,
+      "loss": 0.7441,
+      "step": 8667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136891203661635e-05,
+      "loss": 0.7311,
+      "step": 8668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2135254009866702e-05,
+      "loss": 0.7743,
+      "step": 8669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133616756102042e-05,
+      "loss": 0.7191,
+      "step": 8670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131979442413639e-05,
+      "loss": 0.7101,
+      "step": 8671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213034206884748e-05,
+      "loss": 0.7181,
+      "step": 8672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212870463544955e-05,
+      "loss": 0.716,
+      "step": 8673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127067142265834e-05,
+      "loss": 0.7235,
+      "step": 8674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125429589342326e-05,
+      "loss": 0.6826,
+      "step": 8675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123791976725019e-05,
+      "loss": 0.6527,
+      "step": 8676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2122154304459899e-05,
+      "loss": 0.7065,
+      "step": 8677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120516572592964e-05,
+      "loss": 0.7297,
+      "step": 8678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118878781170213e-05,
+      "loss": 0.6832,
+      "step": 8679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117240930237643e-05,
+      "loss": 0.6348,
+      "step": 8680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115603019841255e-05,
+      "loss": 0.6988,
+      "step": 8681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113965050027047e-05,
+      "loss": 0.7301,
+      "step": 8682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112327020841029e-05,
+      "loss": 0.7557,
+      "step": 8683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110688932329195e-05,
+      "loss": 0.6609,
+      "step": 8684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210905078453756e-05,
+      "loss": 0.7068,
+      "step": 8685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107412577512134e-05,
+      "loss": 0.7469,
+      "step": 8686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105774311298919e-05,
+      "loss": 0.7177,
+      "step": 8687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104135985943932e-05,
+      "loss": 0.7278,
+      "step": 8688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102497601493186e-05,
+      "loss": 0.7225,
+      "step": 8689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100859157992693e-05,
+      "loss": 0.349,
+      "step": 8690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099220655488471e-05,
+      "loss": 0.7179,
+      "step": 8691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097582094026544e-05,
+      "loss": 0.6851,
+      "step": 8692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209594347365292e-05,
+      "loss": 0.782,
+      "step": 8693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094304794413627e-05,
+      "loss": 0.7168,
+      "step": 8694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209266605635469e-05,
+      "loss": 0.7581,
+      "step": 8695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091027259522132e-05,
+      "loss": 0.6362,
+      "step": 8696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208938840396198e-05,
+      "loss": 0.6616,
+      "step": 8697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087749489720259e-05,
+      "loss": 0.7045,
+      "step": 8698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086110516843004e-05,
+      "loss": 0.7115,
+      "step": 8699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2084471485376242e-05,
+      "loss": 0.7579,
+      "step": 8700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208283239536601e-05,
+      "loss": 0.748,
+      "step": 8701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2081193246858338e-05,
+      "loss": 0.7337,
+      "step": 8702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079554039899266e-05,
+      "loss": 0.6849,
+      "step": 8703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207791477453483e-05,
+      "loss": 0.3431,
+      "step": 8704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207627545081107e-05,
+      "loss": 0.7358,
+      "step": 8705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2074636068774028e-05,
+      "loss": 0.807,
+      "step": 8706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072996628469746e-05,
+      "loss": 0.6832,
+      "step": 8707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207135712994427e-05,
+      "loss": 0.7836,
+      "step": 8708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069717573243644e-05,
+      "loss": 0.6759,
+      "step": 8709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2068077958413917e-05,
+      "loss": 0.6662,
+      "step": 8710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066438285501141e-05,
+      "loss": 0.7285,
+      "step": 8711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064798554551364e-05,
+      "loss": 0.6555,
+      "step": 8712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2063158765610638e-05,
+      "loss": 0.6847,
+      "step": 8713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2061518918725019e-05,
+      "loss": 0.6834,
+      "step": 8714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205987901394056e-05,
+      "loss": 0.6441,
+      "step": 8715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058239051303325e-05,
+      "loss": 0.671,
+      "step": 8716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2056599030859367e-05,
+      "loss": 0.6402,
+      "step": 8717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2054958952654751e-05,
+      "loss": 0.7333,
+      "step": 8718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053318816735535e-05,
+      "loss": 0.7145,
+      "step": 8719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051678623147787e-05,
+      "loss": 0.6792,
+      "step": 8720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050038371937572e-05,
+      "loss": 0.7249,
+      "step": 8721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048398063150959e-05,
+      "loss": 0.6922,
+      "step": 8722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2046757696834011e-05,
+      "loss": 0.7227,
+      "step": 8723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045117273032803e-05,
+      "loss": 0.7314,
+      "step": 8724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2043476791793404e-05,
+      "loss": 0.7424,
+      "step": 8725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2041836253161894e-05,
+      "loss": 0.7703,
+      "step": 8726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040195657184344e-05,
+      "loss": 0.7048,
+      "step": 8727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038555003906832e-05,
+      "loss": 0.6648,
+      "step": 8728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036914293375433e-05,
+      "loss": 0.6596,
+      "step": 8729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203527352563623e-05,
+      "loss": 0.7206,
+      "step": 8730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203363270073531e-05,
+      "loss": 0.7452,
+      "step": 8731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031991818718746e-05,
+      "loss": 0.6871,
+      "step": 8732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030350879632632e-05,
+      "loss": 0.6871,
+      "step": 8733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202870988352305e-05,
+      "loss": 0.6863,
+      "step": 8734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202706883043609e-05,
+      "loss": 0.7877,
+      "step": 8735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025427720417841e-05,
+      "loss": 0.6791,
+      "step": 8736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023786553514396e-05,
+      "loss": 0.7383,
+      "step": 8737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2022145329771849e-05,
+      "loss": 0.6783,
+      "step": 8738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020504049236286e-05,
+      "loss": 0.7588,
+      "step": 8739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018862711953812e-05,
+      "loss": 0.7026,
+      "step": 8740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2017221317970524e-05,
+      "loss": 0.6335,
+      "step": 8741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015579867332519e-05,
+      "loss": 0.652,
+      "step": 8742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2013938360085899e-05,
+      "loss": 0.7448,
+      "step": 8743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012296796276765e-05,
+      "loss": 0.7712,
+      "step": 8744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010655175951225e-05,
+      "loss": 0.7024,
+      "step": 8745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200901349915538e-05,
+      "loss": 0.675,
+      "step": 8746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007371765935343e-05,
+      "loss": 0.7723,
+      "step": 8747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005729976337218e-05,
+      "loss": 0.6664,
+      "step": 8748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2004088130407114e-05,
+      "loss": 0.667,
+      "step": 8749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200244622819115e-05,
+      "loss": 0.7274,
+      "step": 8750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000804269735436e-05,
+      "loss": 0.6304,
+      "step": 8751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999162255086088e-05,
+      "loss": 0.7322,
+      "step": 8752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997520184289217e-05,
+      "loss": 0.6344,
+      "step": 8753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199587805739095e-05,
+      "loss": 0.7475,
+      "step": 8754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994235874437403e-05,
+      "loss": 0.6943,
+      "step": 8755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992593635474698e-05,
+      "loss": 0.7095,
+      "step": 8756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1990951340548961e-05,
+      "loss": 0.7206,
+      "step": 8757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989308989706312e-05,
+      "loss": 0.7708,
+      "step": 8758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987666582992876e-05,
+      "loss": 0.7421,
+      "step": 8759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986024120454786e-05,
+      "loss": 0.7375,
+      "step": 8760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984381602138173e-05,
+      "loss": 0.7332,
+      "step": 8761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198273902808916e-05,
+      "loss": 0.7239,
+      "step": 8762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981096398353886e-05,
+      "loss": 0.6799,
+      "step": 8763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979453712978481e-05,
+      "loss": 0.7181,
+      "step": 8764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977810972009086e-05,
+      "loss": 0.7101,
+      "step": 8765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976168175491831e-05,
+      "loss": 0.6749,
+      "step": 8766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1974525323472865e-05,
+      "loss": 0.6852,
+      "step": 8767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972882415998317e-05,
+      "loss": 0.657,
+      "step": 8768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971239453114336e-05,
+      "loss": 0.6889,
+      "step": 8769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969596434867063e-05,
+      "loss": 0.6938,
+      "step": 8770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1967953361302642e-05,
+      "loss": 0.6528,
+      "step": 8771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966310232467225e-05,
+      "loss": 0.7072,
+      "step": 8772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1964667048406953e-05,
+      "loss": 0.7347,
+      "step": 8773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196302380916798e-05,
+      "loss": 0.6517,
+      "step": 8774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961380514796455e-05,
+      "loss": 0.7563,
+      "step": 8775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959737165338534e-05,
+      "loss": 0.6707,
+      "step": 8776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195809376084037e-05,
+      "loss": 0.7136,
+      "step": 8777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1956450301348117e-05,
+      "loss": 0.7091,
+      "step": 8778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954806786907934e-05,
+      "loss": 0.7217,
+      "step": 8779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1953163217565977e-05,
+      "loss": 0.7485,
+      "step": 8780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1951519593368412e-05,
+      "loss": 0.6671,
+      "step": 8781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1949875914361398e-05,
+      "loss": 0.7193,
+      "step": 8782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948232180591097e-05,
+      "loss": 0.7259,
+      "step": 8783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946588392103678e-05,
+      "loss": 0.7495,
+      "step": 8784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944944548945302e-05,
+      "loss": 0.648,
+      "step": 8785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1943300651162145e-05,
+      "loss": 0.6963,
+      "step": 8786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941656698800373e-05,
+      "loss": 0.6878,
+      "step": 8787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1940012691906153e-05,
+      "loss": 0.6761,
+      "step": 8788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938368630525661e-05,
+      "loss": 0.6616,
+      "step": 8789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1936724514705074e-05,
+      "loss": 0.7136,
+      "step": 8790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935080344490567e-05,
+      "loss": 0.6514,
+      "step": 8791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1933436119928312e-05,
+      "loss": 0.7186,
+      "step": 8792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1931791841064498e-05,
+      "loss": 0.7568,
+      "step": 8793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.193014750794529e-05,
+      "loss": 0.6872,
+      "step": 8794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1928503120616886e-05,
+      "loss": 0.7288,
+      "step": 8795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926858679125461e-05,
+      "loss": 0.7241,
+      "step": 8796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925214183517207e-05,
+      "loss": 0.6637,
+      "step": 8797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923569633838299e-05,
+      "loss": 0.713,
+      "step": 8798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921925030134934e-05,
+      "loss": 0.7544,
+      "step": 8799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19202803724533e-05,
+      "loss": 0.6225,
+      "step": 8800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918635660839583e-05,
+      "loss": 0.7304,
+      "step": 8801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916990895339982e-05,
+      "loss": 0.7567,
+      "step": 8802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915346076000692e-05,
+      "loss": 0.7301,
+      "step": 8803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913701202867898e-05,
+      "loss": 0.6442,
+      "step": 8804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1912056275987809e-05,
+      "loss": 0.7044,
+      "step": 8805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191041129540662e-05,
+      "loss": 0.6861,
+      "step": 8806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908766261170528e-05,
+      "loss": 0.7443,
+      "step": 8807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907121173325737e-05,
+      "loss": 0.6661,
+      "step": 8808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190547603191845e-05,
+      "loss": 0.7129,
+      "step": 8809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1903830836994871e-05,
+      "loss": 0.7382,
+      "step": 8810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190218558860121e-05,
+      "loss": 0.6751,
+      "step": 8811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900540286783669e-05,
+      "loss": 0.8221,
+      "step": 8812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1898894931588463e-05,
+      "loss": 0.7663,
+      "step": 8813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897249523061794e-05,
+      "loss": 0.6947,
+      "step": 8814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1895604061249882e-05,
+      "loss": 0.6891,
+      "step": 8815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893958546198939e-05,
+      "loss": 0.747,
+      "step": 8816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892312977955178e-05,
+      "loss": 0.7007,
+      "step": 8817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890667356564815e-05,
+      "loss": 0.7926,
+      "step": 8818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1889021682074074e-05,
+      "loss": 0.6715,
+      "step": 8819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887375954529167e-05,
+      "loss": 0.654,
+      "step": 8820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1885730173976323e-05,
+      "loss": 0.6422,
+      "step": 8821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884084340461757e-05,
+      "loss": 0.702,
+      "step": 8822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18824384540317e-05,
+      "loss": 0.7531,
+      "step": 8823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188079251473237e-05,
+      "loss": 0.732,
+      "step": 8824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187914652261e-05,
+      "loss": 0.6413,
+      "step": 8825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877500477710817e-05,
+      "loss": 0.6786,
+      "step": 8826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187585438008105e-05,
+      "loss": 0.7722,
+      "step": 8827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187420822976693e-05,
+      "loss": 0.6978,
+      "step": 8828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872562026814693e-05,
+      "loss": 0.7289,
+      "step": 8829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870915771270573e-05,
+      "loss": 0.7275,
+      "step": 8830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18692694631808e-05,
+      "loss": 0.6706,
+      "step": 8831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867623102591624e-05,
+      "loss": 0.6746,
+      "step": 8832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186597668954927e-05,
+      "loss": 0.7346,
+      "step": 8833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864330224099984e-05,
+      "loss": 0.6754,
+      "step": 8834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186268370629001e-05,
+      "loss": 0.7096,
+      "step": 8835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186103713616559e-05,
+      "loss": 0.725,
+      "step": 8836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859390513772969e-05,
+      "loss": 0.7503,
+      "step": 8837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857743839158391e-05,
+      "loss": 0.6223,
+      "step": 8838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856097112368104e-05,
+      "loss": 0.6589,
+      "step": 8839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185445033344836e-05,
+      "loss": 0.7055,
+      "step": 8840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852803502445407e-05,
+      "loss": 0.7196,
+      "step": 8841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851156619405501e-05,
+      "loss": 0.7067,
+      "step": 8842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849509684374891e-05,
+      "loss": 0.6534,
+      "step": 8843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847862697399835e-05,
+      "loss": 0.7106,
+      "step": 8844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846215658526586e-05,
+      "loss": 0.7308,
+      "step": 8845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1844568567801405e-05,
+      "loss": 0.7064,
+      "step": 8846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184292142527055e-05,
+      "loss": 0.7213,
+      "step": 8847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1841274230980284e-05,
+      "loss": 0.755,
+      "step": 8848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183962698497687e-05,
+      "loss": 0.6485,
+      "step": 8849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1837979687306564e-05,
+      "loss": 0.6843,
+      "step": 8850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183633233801564e-05,
+      "loss": 0.6967,
+      "step": 8851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834684937150363e-05,
+      "loss": 0.6197,
+      "step": 8852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1833037484756998e-05,
+      "loss": 0.7361,
+      "step": 8853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831389980881815e-05,
+      "loss": 0.741,
+      "step": 8854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1829742425571087e-05,
+      "loss": 0.7275,
+      "step": 8855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1828094818871085e-05,
+      "loss": 0.7726,
+      "step": 8856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826447160828085e-05,
+      "loss": 0.674,
+      "step": 8857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824799451488362e-05,
+      "loss": 0.6249,
+      "step": 8858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823151690898187e-05,
+      "loss": 0.6976,
+      "step": 8859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821503879103846e-05,
+      "loss": 0.6847,
+      "step": 8860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1819856016151617e-05,
+      "loss": 0.6634,
+      "step": 8861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1818208102087779e-05,
+      "loss": 0.6676,
+      "step": 8862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816560136958612e-05,
+      "loss": 0.7374,
+      "step": 8863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1814912120810407e-05,
+      "loss": 0.7853,
+      "step": 8864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813264053689442e-05,
+      "loss": 0.7115,
+      "step": 8865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811615935642009e-05,
+      "loss": 0.7309,
+      "step": 8866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809967766714395e-05,
+      "loss": 0.3223,
+      "step": 8867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180831954695289e-05,
+      "loss": 0.7339,
+      "step": 8868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180667127640378e-05,
+      "loss": 0.735,
+      "step": 8869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805022955113365e-05,
+      "loss": 0.7059,
+      "step": 8870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803374583127939e-05,
+      "loss": 0.7465,
+      "step": 8871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1801726160493789e-05,
+      "loss": 0.6746,
+      "step": 8872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800077687257218e-05,
+      "loss": 0.7404,
+      "step": 8873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1798429163464524e-05,
+      "loss": 0.688,
+      "step": 8874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1796780589162007e-05,
+      "loss": 0.6446,
+      "step": 8875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795131964395967e-05,
+      "loss": 0.7399,
+      "step": 8876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1793483289212707e-05,
+      "loss": 0.6778,
+      "step": 8877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791834563658531e-05,
+      "loss": 0.7408,
+      "step": 8878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1790185787779742e-05,
+      "loss": 0.6819,
+      "step": 8879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1788536961622645e-05,
+      "loss": 0.3504,
+      "step": 8880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786888085233557e-05,
+      "loss": 0.685,
+      "step": 8881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1785239158658782e-05,
+      "loss": 0.7037,
+      "step": 8882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783590181944628e-05,
+      "loss": 0.7196,
+      "step": 8883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781941155137413e-05,
+      "loss": 0.8111,
+      "step": 8884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1780292078283447e-05,
+      "loss": 0.7269,
+      "step": 8885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778642951429047e-05,
+      "loss": 0.7063,
+      "step": 8886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177699377462053e-05,
+      "loss": 0.7186,
+      "step": 8887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775344547904213e-05,
+      "loss": 0.7321,
+      "step": 8888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773695271326413e-05,
+      "loss": 0.6934,
+      "step": 8889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1772045944933454e-05,
+      "loss": 0.6538,
+      "step": 8890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177039656877166e-05,
+      "loss": 0.6774,
+      "step": 8891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1768747142887351e-05,
+      "loss": 0.6625,
+      "step": 8892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767097667326852e-05,
+      "loss": 0.7434,
+      "step": 8893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765448142136491e-05,
+      "loss": 0.6944,
+      "step": 8894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763798567362594e-05,
+      "loss": 0.7158,
+      "step": 8895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762148943051493e-05,
+      "loss": 0.7402,
+      "step": 8896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760499269249518e-05,
+      "loss": 0.7055,
+      "step": 8897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758849546002997e-05,
+      "loss": 0.6581,
+      "step": 8898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1757199773358268e-05,
+      "loss": 0.6658,
+      "step": 8899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755549951361662e-05,
+      "loss": 0.3715,
+      "step": 8900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753900080059519e-05,
+      "loss": 0.6678,
+      "step": 8901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752250159498175e-05,
+      "loss": 0.7528,
+      "step": 8902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1750600189723968e-05,
+      "loss": 0.6857,
+      "step": 8903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1748950170783237e-05,
+      "loss": 0.6644,
+      "step": 8904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747300102722328e-05,
+      "loss": 0.7246,
+      "step": 8905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174564998558758e-05,
+      "loss": 0.7173,
+      "step": 8906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174399981942534e-05,
+      "loss": 0.7608,
+      "step": 8907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1742349604281951e-05,
+      "loss": 0.7049,
+      "step": 8908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1740699340203763e-05,
+      "loss": 0.7153,
+      "step": 8909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1739049027237122e-05,
+      "loss": 0.7457,
+      "step": 8910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737398665428379e-05,
+      "loss": 0.672,
+      "step": 8911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735748254823887e-05,
+      "loss": 0.7244,
+      "step": 8912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1734097795469995e-05,
+      "loss": 0.7611,
+      "step": 8913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173244728741306e-05,
+      "loss": 0.7433,
+      "step": 8914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730796730699437e-05,
+      "loss": 0.7768,
+      "step": 8915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172914612537548e-05,
+      "loss": 0.7471,
+      "step": 8916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727495471487555e-05,
+      "loss": 0.6797,
+      "step": 8917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172584476908201e-05,
+      "loss": 0.7039,
+      "step": 8918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1724194018205215e-05,
+      "loss": 0.6587,
+      "step": 8919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722543218903528e-05,
+      "loss": 0.7927,
+      "step": 8920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720892371223314e-05,
+      "loss": 0.705,
+      "step": 8921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719241475210937e-05,
+      "loss": 0.7181,
+      "step": 8922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717590530912764e-05,
+      "loss": 0.6752,
+      "step": 8923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715939538375159e-05,
+      "loss": 0.6485,
+      "step": 8924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1714288497644497e-05,
+      "loss": 0.7341,
+      "step": 8925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1712637408767145e-05,
+      "loss": 0.7208,
+      "step": 8926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710986271789478e-05,
+      "loss": 0.7073,
+      "step": 8927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709335086757862e-05,
+      "loss": 0.3571,
+      "step": 8928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707683853718676e-05,
+      "loss": 0.7659,
+      "step": 8929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706032572718299e-05,
+      "loss": 0.6898,
+      "step": 8930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1704381243803102e-05,
+      "loss": 0.6783,
+      "step": 8931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702729867019467e-05,
+      "loss": 0.6321,
+      "step": 8932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701078442413774e-05,
+      "loss": 0.7532,
+      "step": 8933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699426970032399e-05,
+      "loss": 0.7719,
+      "step": 8934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169777544992173e-05,
+      "loss": 0.7805,
+      "step": 8935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696123882128152e-05,
+      "loss": 0.7852,
+      "step": 8936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694472266698045e-05,
+      "loss": 0.7221,
+      "step": 8937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1692820603677797e-05,
+      "loss": 0.7159,
+      "step": 8938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691168893113798e-05,
+      "loss": 0.802,
+      "step": 8939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689517135052434e-05,
+      "loss": 0.7326,
+      "step": 8940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16878653295401e-05,
+      "loss": 0.6601,
+      "step": 8941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1686213476623184e-05,
+      "loss": 0.7459,
+      "step": 8942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1684561576348082e-05,
+      "loss": 0.7106,
+      "step": 8943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682909628761184e-05,
+      "loss": 0.6882,
+      "step": 8944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681257633908887e-05,
+      "loss": 0.321,
+      "step": 8945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679605591837593e-05,
+      "loss": 0.7045,
+      "step": 8946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1677953502593695e-05,
+      "loss": 0.3193,
+      "step": 8947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1676301366223597e-05,
+      "loss": 0.7303,
+      "step": 8948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674649182773697e-05,
+      "loss": 0.7375,
+      "step": 8949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1672996952290398e-05,
+      "loss": 0.7276,
+      "step": 8950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671344674820104e-05,
+      "loss": 0.7916,
+      "step": 8951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1669692350409223e-05,
+      "loss": 0.7321,
+      "step": 8952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1668039979104158e-05,
+      "loss": 0.6762,
+      "step": 8953
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666387560951315e-05,
+      "loss": 0.6787,
+      "step": 8954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664735095997106e-05,
+      "loss": 0.6777,
+      "step": 8955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1663082584287944e-05,
+      "loss": 0.7464,
+      "step": 8956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1661430025870233e-05,
+      "loss": 0.7849,
+      "step": 8957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659777420790393e-05,
+      "loss": 0.7106,
+      "step": 8958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1658124769094834e-05,
+      "loss": 0.7399,
+      "step": 8959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1656472070829977e-05,
+      "loss": 0.6924,
+      "step": 8960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165481932604223e-05,
+      "loss": 0.6867,
+      "step": 8961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165316653477802e-05,
+      "loss": 0.6885,
+      "step": 8962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651513697083762e-05,
+      "loss": 0.834,
+      "step": 8963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164986081300588e-05,
+      "loss": 0.7209,
+      "step": 8964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648207882590788e-05,
+      "loss": 0.7119,
+      "step": 8965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646554905884922e-05,
+      "loss": 0.3525,
+      "step": 8966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1644901882934697e-05,
+      "loss": 0.7134,
+      "step": 8967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164324881378654e-05,
+      "loss": 0.6936,
+      "step": 8968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164159569848688e-05,
+      "loss": 0.721,
+      "step": 8969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163994253708215e-05,
+      "loss": 0.6729,
+      "step": 8970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1638289329618773e-05,
+      "loss": 0.7027,
+      "step": 8971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1636636076143186e-05,
+      "loss": 0.7388,
+      "step": 8972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634982776701815e-05,
+      "loss": 0.7092,
+      "step": 8973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633329431341098e-05,
+      "loss": 0.6527,
+      "step": 8974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1631676040107468e-05,
+      "loss": 0.6882,
+      "step": 8975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630022603047366e-05,
+      "loss": 0.7737,
+      "step": 8976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628369120207225e-05,
+      "loss": 0.732,
+      "step": 8977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1626715591633484e-05,
+      "loss": 0.7598,
+      "step": 8978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1625062017372586e-05,
+      "loss": 0.6828,
+      "step": 8979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162340839747097e-05,
+      "loss": 0.7305,
+      "step": 8980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1621754731975079e-05,
+      "loss": 0.7435,
+      "step": 8981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620101020931362e-05,
+      "loss": 0.6531,
+      "step": 8982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1618447264386256e-05,
+      "loss": 0.7296,
+      "step": 8983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616793462386214e-05,
+      "loss": 0.7677,
+      "step": 8984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.161513961497768e-05,
+      "loss": 0.7103,
+      "step": 8985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1613485722207105e-05,
+      "loss": 0.7614,
+      "step": 8986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.161183178412094e-05,
+      "loss": 0.7023,
+      "step": 8987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1610177800765636e-05,
+      "loss": 0.7392,
+      "step": 8988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608523772187646e-05,
+      "loss": 0.6819,
+      "step": 8989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606869698433423e-05,
+      "loss": 0.7281,
+      "step": 8990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605215579549426e-05,
+      "loss": 0.6671,
+      "step": 8991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160356141558211e-05,
+      "loss": 0.7065,
+      "step": 8992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160190720657793e-05,
+      "loss": 0.7363,
+      "step": 8993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160025295258335e-05,
+      "loss": 0.7316,
+      "step": 8994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598598653644826e-05,
+      "loss": 0.7459,
+      "step": 8995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596944309808824e-05,
+      "loss": 0.7328,
+      "step": 8996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595289921121803e-05,
+      "loss": 0.6709,
+      "step": 8997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593635487630235e-05,
+      "loss": 0.7136,
+      "step": 8998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1591981009380578e-05,
+      "loss": 0.6478,
+      "step": 8999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15903264864193e-05,
+      "loss": 0.6911,
+      "step": 9000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158867191879287e-05,
+      "loss": 0.7579,
+      "step": 9001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158701730654776e-05,
+      "loss": 0.7751,
+      "step": 9002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1585362649730441e-05,
+      "loss": 0.733,
+      "step": 9003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1583707948387376e-05,
+      "loss": 0.6763,
+      "step": 9004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582053202565051e-05,
+      "loss": 0.7149,
+      "step": 9005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158039841230993e-05,
+      "loss": 0.7327,
+      "step": 9006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1578743577668493e-05,
+      "loss": 0.6777,
+      "step": 9007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157708869868722e-05,
+      "loss": 0.7167,
+      "step": 9008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575433775412584e-05,
+      "loss": 0.7047,
+      "step": 9009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573778807891062e-05,
+      "loss": 0.7466,
+      "step": 9010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1572123796169146e-05,
+      "loss": 0.7254,
+      "step": 9011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570468740293306e-05,
+      "loss": 0.7466,
+      "step": 9012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156881364031003e-05,
+      "loss": 0.678,
+      "step": 9013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567158496265803e-05,
+      "loss": 0.6752,
+      "step": 9014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565503308207108e-05,
+      "loss": 0.7208,
+      "step": 9015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1563848076180435e-05,
+      "loss": 0.7636,
+      "step": 9016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156219280023227e-05,
+      "loss": 0.6573,
+      "step": 9017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560537480409107e-05,
+      "loss": 0.7744,
+      "step": 9018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1558882116757428e-05,
+      "loss": 0.7716,
+      "step": 9019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1557226709323728e-05,
+      "loss": 0.6294,
+      "step": 9020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555571258154503e-05,
+      "loss": 0.7509,
+      "step": 9021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553915763296247e-05,
+      "loss": 0.7183,
+      "step": 9022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1552260224795451e-05,
+      "loss": 0.6464,
+      "step": 9023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550604642698616e-05,
+      "loss": 0.6515,
+      "step": 9024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548949017052236e-05,
+      "loss": 0.7309,
+      "step": 9025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1547293347902813e-05,
+      "loss": 0.6358,
+      "step": 9026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545637635296846e-05,
+      "loss": 0.7439,
+      "step": 9027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543981879280843e-05,
+      "loss": 0.7209,
+      "step": 9028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1542326079901296e-05,
+      "loss": 0.6982,
+      "step": 9029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1540670237204715e-05,
+      "loss": 0.7097,
+      "step": 9030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1539014351237605e-05,
+      "loss": 0.7593,
+      "step": 9031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537358422046472e-05,
+      "loss": 0.767,
+      "step": 9032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153570244967782e-05,
+      "loss": 0.6807,
+      "step": 9033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534046434178166e-05,
+      "loss": 0.6924,
+      "step": 9034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532390375594012e-05,
+      "loss": 0.7295,
+      "step": 9035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1530734273971873e-05,
+      "loss": 0.7562,
+      "step": 9036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529078129358263e-05,
+      "loss": 0.7301,
+      "step": 9037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527421941799693e-05,
+      "loss": 0.7304,
+      "step": 9038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1525765711342675e-05,
+      "loss": 0.7069,
+      "step": 9039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152410943803373e-05,
+      "loss": 0.6521,
+      "step": 9040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1522453121919376e-05,
+      "loss": 0.7231,
+      "step": 9041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152079676304613e-05,
+      "loss": 0.7172,
+      "step": 9042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151914036146051e-05,
+      "loss": 0.7422,
+      "step": 9043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1517483917209039e-05,
+      "loss": 0.6882,
+      "step": 9044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1515827430338236e-05,
+      "loss": 0.7097,
+      "step": 9045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514170900894627e-05,
+      "loss": 0.6798,
+      "step": 9046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151251432892474e-05,
+      "loss": 0.7589,
+      "step": 9047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510857714475092e-05,
+      "loss": 0.7417,
+      "step": 9048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1509201057592216e-05,
+      "loss": 0.7736,
+      "step": 9049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150754435832264e-05,
+      "loss": 0.6658,
+      "step": 9050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150588761671289e-05,
+      "loss": 0.7874,
+      "step": 9051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15042308328095e-05,
+      "loss": 0.7011,
+      "step": 9052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502574006658998e-05,
+      "loss": 0.7344,
+      "step": 9053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1500917138307924e-05,
+      "loss": 0.6276,
+      "step": 9054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1499260227802798e-05,
+      "loss": 0.6966,
+      "step": 9055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149760327519017e-05,
+      "loss": 0.7271,
+      "step": 9056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149594628051657e-05,
+      "loss": 0.673,
+      "step": 9057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1494289243828537e-05,
+      "loss": 0.7486,
+      "step": 9058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1492632165172605e-05,
+      "loss": 0.6606,
+      "step": 9059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490975044595321e-05,
+      "loss": 0.7531,
+      "step": 9060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148931788214322e-05,
+      "loss": 0.7333,
+      "step": 9061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1487660677862848e-05,
+      "loss": 0.7817,
+      "step": 9062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486003431800752e-05,
+      "loss": 0.7277,
+      "step": 9063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1484346144003467e-05,
+      "loss": 0.6923,
+      "step": 9064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1482688814517543e-05,
+      "loss": 0.6368,
+      "step": 9065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148103144338953e-05,
+      "loss": 0.7218,
+      "step": 9066
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479374030665975e-05,
+      "loss": 0.79,
+      "step": 9067
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477716576393426e-05,
+      "loss": 0.7191,
+      "step": 9068
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147605908061843e-05,
+      "loss": 0.7885,
+      "step": 9069
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474401543387547e-05,
+      "loss": 0.6693,
+      "step": 9070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1472743964747324e-05,
+      "loss": 0.6645,
+      "step": 9071
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471086344744315e-05,
+      "loss": 0.6663,
+      "step": 9072
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1469428683425078e-05,
+      "loss": 0.7173,
+      "step": 9073
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467770980836169e-05,
+      "loss": 0.7546,
+      "step": 9074
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146611323702414e-05,
+      "loss": 0.6795,
+      "step": 9075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464455452035557e-05,
+      "loss": 0.7444,
+      "step": 9076
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1462797625916973e-05,
+      "loss": 0.7075,
+      "step": 9077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461139758714954e-05,
+      "loss": 0.6636,
+      "step": 9078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459481850476062e-05,
+      "loss": 0.7167,
+      "step": 9079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457823901246856e-05,
+      "loss": 0.6818,
+      "step": 9080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456165911073905e-05,
+      "loss": 0.6895,
+      "step": 9081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454507880003768e-05,
+      "loss": 0.6981,
+      "step": 9082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452849808083024e-05,
+      "loss": 0.7214,
+      "step": 9083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1451191695358229e-05,
+      "loss": 0.7083,
+      "step": 9084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449533541875953e-05,
+      "loss": 0.7305,
+      "step": 9085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447875347682774e-05,
+      "loss": 0.6992,
+      "step": 9086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1446217112825255e-05,
+      "loss": 0.6805,
+      "step": 9087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1444558837349972e-05,
+      "loss": 0.7199,
+      "step": 9088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14429005213035e-05,
+      "loss": 0.6726,
+      "step": 9089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441242164732409e-05,
+      "loss": 0.696,
+      "step": 9090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.143958376768328e-05,
+      "loss": 0.6664,
+      "step": 9091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1437925330202689e-05,
+      "loss": 0.6966,
+      "step": 9092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436266852337213e-05,
+      "loss": 0.7117,
+      "step": 9093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434608334133428e-05,
+      "loss": 0.7634,
+      "step": 9094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432949775637913e-05,
+      "loss": 0.7089,
+      "step": 9095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431291176897263e-05,
+      "loss": 0.73,
+      "step": 9096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429632537958047e-05,
+      "loss": 0.7872,
+      "step": 9097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1427973858866854e-05,
+      "loss": 0.6111,
+      "step": 9098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1426315139670268e-05,
+      "loss": 0.6693,
+      "step": 9099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424656380414874e-05,
+      "loss": 0.6929,
+      "step": 9100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142299758114726e-05,
+      "loss": 0.7477,
+      "step": 9101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421338741914018e-05,
+      "loss": 0.722,
+      "step": 9102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141967986276173e-05,
+      "loss": 0.7205,
+      "step": 9103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141802094373699e-05,
+      "loss": 0.7677,
+      "step": 9104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141636198488639e-05,
+      "loss": 0.72,
+      "step": 9105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1414702986256525e-05,
+      "loss": 0.7297,
+      "step": 9106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413043947893985e-05,
+      "loss": 0.679,
+      "step": 9107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411384869845367e-05,
+      "loss": 0.6644,
+      "step": 9108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1409725752157267e-05,
+      "loss": 0.717,
+      "step": 9109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1408066594876279e-05,
+      "loss": 0.7252,
+      "step": 9110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1406407398049004e-05,
+      "loss": 0.722,
+      "step": 9111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404748161722046e-05,
+      "loss": 0.677,
+      "step": 9112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1403088885941997e-05,
+      "loss": 0.7886,
+      "step": 9113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1401429570755464e-05,
+      "loss": 0.7362,
+      "step": 9114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399770216209048e-05,
+      "loss": 0.7437,
+      "step": 9115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1398110822349353e-05,
+      "loss": 0.7155,
+      "step": 9116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396451389222982e-05,
+      "loss": 0.6752,
+      "step": 9117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1394791916876545e-05,
+      "loss": 0.7575,
+      "step": 9118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393132405356648e-05,
+      "loss": 0.6875,
+      "step": 9119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391472854709896e-05,
+      "loss": 0.6503,
+      "step": 9120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389813264982903e-05,
+      "loss": 0.6941,
+      "step": 9121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138815363622228e-05,
+      "loss": 0.6707,
+      "step": 9122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138649396847463e-05,
+      "loss": 0.6928,
+      "step": 9123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384834261786575e-05,
+      "loss": 0.7418,
+      "step": 9124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383174516204723e-05,
+      "loss": 0.7681,
+      "step": 9125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381514731775691e-05,
+      "loss": 0.7061,
+      "step": 9126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379854908546094e-05,
+      "loss": 0.7226,
+      "step": 9127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378195046562557e-05,
+      "loss": 0.695,
+      "step": 9128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376535145871685e-05,
+      "loss": 0.7198,
+      "step": 9129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374875206520102e-05,
+      "loss": 0.7375,
+      "step": 9130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137321522855443e-05,
+      "loss": 0.7592,
+      "step": 9131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371555212021295e-05,
+      "loss": 0.6943,
+      "step": 9132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136989515696731e-05,
+      "loss": 0.697,
+      "step": 9133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1368235063439103e-05,
+      "loss": 0.6349,
+      "step": 9134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1366574931483299e-05,
+      "loss": 0.6821,
+      "step": 9135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1364914761146522e-05,
+      "loss": 0.7212,
+      "step": 9136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13632545524754e-05,
+      "loss": 0.7258,
+      "step": 9137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361594305516563e-05,
+      "loss": 0.7072,
+      "step": 9138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359934020316637e-05,
+      "loss": 0.7291,
+      "step": 9139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358273696922248e-05,
+      "loss": 0.6747,
+      "step": 9140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356613335380037e-05,
+      "loss": 0.7131,
+      "step": 9141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354952935736627e-05,
+      "loss": 0.357,
+      "step": 9142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1353292498038658e-05,
+      "loss": 0.6308,
+      "step": 9143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135163202233276e-05,
+      "loss": 0.6464,
+      "step": 9144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349971508665567e-05,
+      "loss": 0.7192,
+      "step": 9145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1348310957083718e-05,
+      "loss": 0.7127,
+      "step": 9146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1346650367633854e-05,
+      "loss": 0.7078,
+      "step": 9147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1344989740362609e-05,
+      "loss": 0.7281,
+      "step": 9148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343329075316623e-05,
+      "loss": 0.6966,
+      "step": 9149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1341668372542533e-05,
+      "loss": 0.6571,
+      "step": 9150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134000763208699e-05,
+      "loss": 0.7259,
+      "step": 9151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1338346853996629e-05,
+      "loss": 0.757,
+      "step": 9152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1336686038318096e-05,
+      "loss": 0.7186,
+      "step": 9153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335025185098038e-05,
+      "loss": 0.7398,
+      "step": 9154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1333364294383098e-05,
+      "loss": 0.3628,
+      "step": 9155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1331703366219924e-05,
+      "loss": 0.722,
+      "step": 9156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330042400655166e-05,
+      "loss": 0.6473,
+      "step": 9157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132838139773547e-05,
+      "loss": 0.7524,
+      "step": 9158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1326720357507488e-05,
+      "loss": 0.7388,
+      "step": 9159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1325059280017866e-05,
+      "loss": 0.7315,
+      "step": 9160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323398165313269e-05,
+      "loss": 0.7329,
+      "step": 9161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1321737013440338e-05,
+      "loss": 0.7557,
+      "step": 9162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320075824445731e-05,
+      "loss": 0.7632,
+      "step": 9163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1318414598376104e-05,
+      "loss": 0.7159,
+      "step": 9164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316753335278112e-05,
+      "loss": 0.6726,
+      "step": 9165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315092035198416e-05,
+      "loss": 0.6799,
+      "step": 9166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313430698183669e-05,
+      "loss": 0.6452,
+      "step": 9167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311769324280539e-05,
+      "loss": 0.7256,
+      "step": 9168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1310107913535677e-05,
+      "loss": 0.6979,
+      "step": 9169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1308446465995748e-05,
+      "loss": 0.7498,
+      "step": 9170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130678498170742e-05,
+      "loss": 0.6985,
+      "step": 9171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305123460717348e-05,
+      "loss": 0.7503,
+      "step": 9172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303461903072201e-05,
+      "loss": 0.6968,
+      "step": 9173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1301800308818647e-05,
+      "loss": 0.6675,
+      "step": 9174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300138678003348e-05,
+      "loss": 0.6478,
+      "step": 9175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298477010672972e-05,
+      "loss": 0.7116,
+      "step": 9176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1296815306874197e-05,
+      "loss": 0.6985,
+      "step": 9177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1295153566653678e-05,
+      "loss": 0.6591,
+      "step": 9178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293491790058098e-05,
+      "loss": 0.657,
+      "step": 9179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1291829977134118e-05,
+      "loss": 0.7125,
+      "step": 9180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290168127928424e-05,
+      "loss": 0.6793,
+      "step": 9181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1288506242487681e-05,
+      "loss": 0.7053,
+      "step": 9182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1286844320858566e-05,
+      "loss": 0.7011,
+      "step": 9183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1285182363087756e-05,
+      "loss": 0.6389,
+      "step": 9184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1283520369221924e-05,
+      "loss": 0.6799,
+      "step": 9185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281858339307753e-05,
+      "loss": 0.6795,
+      "step": 9186
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280196273391918e-05,
+      "loss": 0.6442,
+      "step": 9187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1278534171521102e-05,
+      "loss": 0.6684,
+      "step": 9188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276872033741987e-05,
+      "loss": 0.7278,
+      "step": 9189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127520986010125e-05,
+      "loss": 0.6881,
+      "step": 9190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1273547650645578e-05,
+      "loss": 0.677,
+      "step": 9191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271885405421651e-05,
+      "loss": 0.7256,
+      "step": 9192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1270223124476161e-05,
+      "loss": 0.6255,
+      "step": 9193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268560807855791e-05,
+      "loss": 0.7744,
+      "step": 9194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266898455607223e-05,
+      "loss": 0.6746,
+      "step": 9195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1265236067777149e-05,
+      "loss": 0.788,
+      "step": 9196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263573644412263e-05,
+      "loss": 0.7578,
+      "step": 9197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261911185559245e-05,
+      "loss": 0.6857,
+      "step": 9198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260248691264792e-05,
+      "loss": 0.6675,
+      "step": 9199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1258586161575595e-05,
+      "loss": 0.6859,
+      "step": 9200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256923596538349e-05,
+      "loss": 0.7088,
+      "step": 9201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1255260996199743e-05,
+      "loss": 0.3496,
+      "step": 9202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125359836060648e-05,
+      "loss": 0.6734,
+      "step": 9203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1251935689805249e-05,
+      "loss": 0.6827,
+      "step": 9204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1250272983842746e-05,
+      "loss": 0.6473,
+      "step": 9205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248610242765674e-05,
+      "loss": 0.7102,
+      "step": 9206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1246947466620733e-05,
+      "loss": 0.7488,
+      "step": 9207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245284655454617e-05,
+      "loss": 0.6876,
+      "step": 9208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243621809314029e-05,
+      "loss": 0.6395,
+      "step": 9209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241958928245674e-05,
+      "loss": 0.736,
+      "step": 9210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1240296012296252e-05,
+      "loss": 0.7144,
+      "step": 9211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1238633061512467e-05,
+      "loss": 0.7325,
+      "step": 9212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236970075941027e-05,
+      "loss": 0.7423,
+      "step": 9213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1235307055628633e-05,
+      "loss": 0.653,
+      "step": 9214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123364400062199e-05,
+      "loss": 0.6185,
+      "step": 9215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1231980910967816e-05,
+      "loss": 0.6736,
+      "step": 9216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1230317786712812e-05,
+      "loss": 0.6777,
+      "step": 9217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228654627903688e-05,
+      "loss": 0.6545,
+      "step": 9218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226991434587155e-05,
+      "loss": 0.7789,
+      "step": 9219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225328206809929e-05,
+      "loss": 0.6495,
+      "step": 9220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223664944618717e-05,
+      "loss": 0.3379,
+      "step": 9221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222001648060234e-05,
+      "loss": 0.6773,
+      "step": 9222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220338317181198e-05,
+      "loss": 0.7222,
+      "step": 9223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1218674952028321e-05,
+      "loss": 0.673,
+      "step": 9224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217011552648316e-05,
+      "loss": 0.7422,
+      "step": 9225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215348119087912e-05,
+      "loss": 0.6758,
+      "step": 9226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1213684651393815e-05,
+      "loss": 0.6551,
+      "step": 9227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121202114961275e-05,
+      "loss": 0.6956,
+      "step": 9228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121035761379144e-05,
+      "loss": 0.7239,
+      "step": 9229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12086940439766e-05,
+      "loss": 0.7535,
+      "step": 9230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207030440214957e-05,
+      "loss": 0.7009,
+      "step": 9231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 0.7382,
+      "step": 9232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120370313103815e-05,
+      "loss": 0.7442,
+      "step": 9233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202039425716436e-05,
+      "loss": 0.6823,
+      "step": 9234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200375686634813e-05,
+      "loss": 0.7333,
+      "step": 9235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198711913840015e-05,
+      "loss": 0.6599,
+      "step": 9236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1197048107378762e-05,
+      "loss": 0.6811,
+      "step": 9237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195384267297788e-05,
+      "loss": 0.7625,
+      "step": 9238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193720393643826e-05,
+      "loss": 0.7386,
+      "step": 9239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192056486463594e-05,
+      "loss": 0.6275,
+      "step": 9240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1190392545803837e-05,
+      "loss": 0.6713,
+      "step": 9241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188728571711283e-05,
+      "loss": 0.7281,
+      "step": 9242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1187064564232661e-05,
+      "loss": 0.7389,
+      "step": 9243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185400523414714e-05,
+      "loss": 0.6892,
+      "step": 9244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118373644930417e-05,
+      "loss": 0.6817,
+      "step": 9245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118207234194777e-05,
+      "loss": 0.7248,
+      "step": 9246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180408201392248e-05,
+      "loss": 0.6582,
+      "step": 9247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1178744027684346e-05,
+      "loss": 0.713,
+      "step": 9248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177079820870803e-05,
+      "loss": 0.7323,
+      "step": 9249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175415580998353e-05,
+      "loss": 0.7179,
+      "step": 9250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173751308113741e-05,
+      "loss": 0.6651,
+      "step": 9251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1172087002263715e-05,
+      "loss": 0.7389,
+      "step": 9252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170422663495008e-05,
+      "loss": 0.7116,
+      "step": 9253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116875829185437e-05,
+      "loss": 0.7241,
+      "step": 9254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167093887388542e-05,
+      "loss": 0.7095,
+      "step": 9255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165429450144271e-05,
+      "loss": 0.644,
+      "step": 9256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163764980168306e-05,
+      "loss": 0.7281,
+      "step": 9257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162100477507392e-05,
+      "loss": 0.6812,
+      "step": 9258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160435942208281e-05,
+      "loss": 0.7551,
+      "step": 9259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158771374317715e-05,
+      "loss": 0.6721,
+      "step": 9260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157106773882448e-05,
+      "loss": 0.709,
+      "step": 9261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155442140949237e-05,
+      "loss": 0.7321,
+      "step": 9262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153777475564826e-05,
+      "loss": 0.6909,
+      "step": 9263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115211277777597e-05,
+      "loss": 0.7397,
+      "step": 9264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1150448047629425e-05,
+      "loss": 0.7633,
+      "step": 9265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1148783285171945e-05,
+      "loss": 0.6768,
+      "step": 9266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1147118490450284e-05,
+      "loss": 0.6587,
+      "step": 9267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145453663511203e-05,
+      "loss": 0.7325,
+      "step": 9268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143788804401454e-05,
+      "loss": 0.7221,
+      "step": 9269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142123913167797e-05,
+      "loss": 0.7123,
+      "step": 9270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140458989856993e-05,
+      "loss": 0.7455,
+      "step": 9271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138794034515803e-05,
+      "loss": 0.6946,
+      "step": 9272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1137129047190987e-05,
+      "loss": 0.7279,
+      "step": 9273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1135464027929306e-05,
+      "loss": 0.7289,
+      "step": 9274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133798976777523e-05,
+      "loss": 0.7435,
+      "step": 9275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132133893782401e-05,
+      "loss": 0.7371,
+      "step": 9276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130468778990708e-05,
+      "loss": 0.7063,
+      "step": 9277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112880363244921e-05,
+      "loss": 0.6604,
+      "step": 9278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127138454204668e-05,
+      "loss": 0.6404,
+      "step": 9279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1125473244303853e-05,
+      "loss": 0.7479,
+      "step": 9280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123808002793537e-05,
+      "loss": 0.7379,
+      "step": 9281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122142729720481e-05,
+      "loss": 0.7083,
+      "step": 9282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120477425131462e-05,
+      "loss": 0.731,
+      "step": 9283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1118812089073245e-05,
+      "loss": 0.7172,
+      "step": 9284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1117146721592609e-05,
+      "loss": 0.7296,
+      "step": 9285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111548132273632e-05,
+      "loss": 0.6796,
+      "step": 9286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113815892551154e-05,
+      "loss": 0.7061,
+      "step": 9287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111215043108389e-05,
+      "loss": 0.7511,
+      "step": 9288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110484938381296e-05,
+      "loss": 0.6984,
+      "step": 9289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108819414490151e-05,
+      "loss": 0.7374,
+      "step": 9290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107153859457234e-05,
+      "loss": 0.7125,
+      "step": 9291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110548827332932e-05,
+      "loss": 0.7241,
+      "step": 9292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103822656153192e-05,
+      "loss": 0.7198,
+      "step": 9293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102157007975628e-05,
+      "loss": 0.7038,
+      "step": 9294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1100491328843401e-05,
+      "loss": 0.7496,
+      "step": 9295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098825618803304e-05,
+      "loss": 0.7138,
+      "step": 9296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097159877902115e-05,
+      "loss": 0.6606,
+      "step": 9297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095494106186618e-05,
+      "loss": 0.7119,
+      "step": 9298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1093828303703592e-05,
+      "loss": 0.756,
+      "step": 9299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092162470499826e-05,
+      "loss": 0.6653,
+      "step": 9300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090496606622109e-05,
+      "loss": 0.7597,
+      "step": 9301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088830712117223e-05,
+      "loss": 0.7541,
+      "step": 9302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1087164787031957e-05,
+      "loss": 0.7432,
+      "step": 9303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10854988314131e-05,
+      "loss": 0.7227,
+      "step": 9304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083832845307436e-05,
+      "loss": 0.7499,
+      "step": 9305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108216682876176e-05,
+      "loss": 0.6634,
+      "step": 9306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1080500781822867e-05,
+      "loss": 0.6304,
+      "step": 9307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1078834704537543e-05,
+      "loss": 0.6626,
+      "step": 9308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077168596952579e-05,
+      "loss": 0.715,
+      "step": 9309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1075502459114773e-05,
+      "loss": 0.7427,
+      "step": 9310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073836291070918e-05,
+      "loss": 0.6492,
+      "step": 9311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1072170092867808e-05,
+      "loss": 0.6807,
+      "step": 9312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107050386455224e-05,
+      "loss": 0.7578,
+      "step": 9313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1068837606171015e-05,
+      "loss": 0.7633,
+      "step": 9314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1067171317770919e-05,
+      "loss": 0.7285,
+      "step": 9315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065504999398762e-05,
+      "loss": 0.6827,
+      "step": 9316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1063838651101341e-05,
+      "loss": 0.645,
+      "step": 9317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1062172272925453e-05,
+      "loss": 0.6452,
+      "step": 9318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060505864917903e-05,
+      "loss": 0.7,
+      "step": 9319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105883942712549e-05,
+      "loss": 0.7136,
+      "step": 9320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1057172959595015e-05,
+      "loss": 0.6998,
+      "step": 9321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055506462373286e-05,
+      "loss": 0.7808,
+      "step": 9322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1053839935507106e-05,
+      "loss": 0.6146,
+      "step": 9323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1052173379043283e-05,
+      "loss": 0.7114,
+      "step": 9324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050506793028615e-05,
+      "loss": 0.6746,
+      "step": 9325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048840177509916e-05,
+      "loss": 0.8111,
+      "step": 9326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1047173532533994e-05,
+      "loss": 0.7502,
+      "step": 9327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045506858147654e-05,
+      "loss": 0.7389,
+      "step": 9328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043840154397706e-05,
+      "loss": 0.7174,
+      "step": 9329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042173421330961e-05,
+      "loss": 0.6718,
+      "step": 9330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1040506658994232e-05,
+      "loss": 0.7285,
+      "step": 9331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038839867434327e-05,
+      "loss": 0.7033,
+      "step": 9332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037173046698063e-05,
+      "loss": 0.7438,
+      "step": 9333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103550619683225e-05,
+      "loss": 0.658,
+      "step": 9334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10338393178837e-05,
+      "loss": 0.7037,
+      "step": 9335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1032172409899237e-05,
+      "loss": 0.6888,
+      "step": 9336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030505472925672e-05,
+      "loss": 0.6855,
+      "step": 9337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102883850700982e-05,
+      "loss": 0.7049,
+      "step": 9338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10271715121985e-05,
+      "loss": 0.6716,
+      "step": 9339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102550448853853e-05,
+      "loss": 0.6373,
+      "step": 9340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023837436076733e-05,
+      "loss": 0.6937,
+      "step": 9341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1022170354859922e-05,
+      "loss": 0.7748,
+      "step": 9342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020503244934924e-05,
+      "loss": 0.6602,
+      "step": 9343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018836106348558e-05,
+      "loss": 0.7325,
+      "step": 9344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017168939147644e-05,
+      "loss": 0.712,
+      "step": 9345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101550174337901e-05,
+      "loss": 0.7112,
+      "step": 9346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1013834519089482e-05,
+      "loss": 0.7175,
+      "step": 9347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1012167266325877e-05,
+      "loss": 0.6925,
+      "step": 9348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010499985135026e-05,
+      "loss": 0.7584,
+      "step": 9349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1008832675563754e-05,
+      "loss": 0.7538,
+      "step": 9350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.100716533765889e-05,
+      "loss": 0.746,
+      "step": 9351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005497971467258e-05,
+      "loss": 0.7454,
+      "step": 9352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003830577035693e-05,
+      "loss": 0.3446,
+      "step": 9353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002163154411016e-05,
+      "loss": 0.7007,
+      "step": 9354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1000495703640063e-05,
+      "loss": 0.7436,
+      "step": 9355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099882822476967e-05,
+      "loss": 0.7336,
+      "step": 9356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997160717846658e-05,
+      "loss": 0.7233,
+      "step": 9357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0995493182917869e-05,
+      "loss": 0.6545,
+      "step": 9358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099382562003013e-05,
+      "loss": 0.7848,
+      "step": 9359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099215802923028e-05,
+      "loss": 0.705,
+      "step": 9360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990490410565154e-05,
+      "loss": 0.7401,
+      "step": 9361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988822764081585e-05,
+      "loss": 0.7,
+      "step": 9362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987155089826413e-05,
+      "loss": 0.6717,
+      "step": 9363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0985487387846474e-05,
+      "loss": 0.6597,
+      "step": 9364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983819658188603e-05,
+      "loss": 0.7254,
+      "step": 9365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982151900899644e-05,
+      "loss": 0.6666,
+      "step": 9366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0980484116026435e-05,
+      "loss": 0.6949,
+      "step": 9367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0978816303615815e-05,
+      "loss": 0.7466,
+      "step": 9368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977148463714631e-05,
+      "loss": 0.7455,
+      "step": 9369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975480596369717e-05,
+      "loss": 0.7015,
+      "step": 9370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0973812701627922e-05,
+      "loss": 0.7062,
+      "step": 9371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972144779536085e-05,
+      "loss": 0.7125,
+      "step": 9372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097047683014106e-05,
+      "loss": 0.7262,
+      "step": 9373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968808853489679e-05,
+      "loss": 0.6658,
+      "step": 9374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0967140849628796e-05,
+      "loss": 0.7024,
+      "step": 9375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0965472818605256e-05,
+      "loss": 0.62,
+      "step": 9376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963804760465908e-05,
+      "loss": 0.655,
+      "step": 9377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0962136675257598e-05,
+      "loss": 0.717,
+      "step": 9378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0960468563027178e-05,
+      "loss": 0.7349,
+      "step": 9379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958800423821491e-05,
+      "loss": 0.3176,
+      "step": 9380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957132257687397e-05,
+      "loss": 0.713,
+      "step": 9381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955464064671743e-05,
+      "loss": 0.8124,
+      "step": 9382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095379584482138e-05,
+      "loss": 0.726,
+      "step": 9383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095212759818316e-05,
+      "loss": 0.7679,
+      "step": 9384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950459324803939e-05,
+      "loss": 0.7003,
+      "step": 9385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0948791024730572e-05,
+      "loss": 0.7534,
+      "step": 9386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0947122698009912e-05,
+      "loss": 0.7476,
+      "step": 9387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0945454344688812e-05,
+      "loss": 0.7496,
+      "step": 9388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943785964814138e-05,
+      "loss": 0.7101,
+      "step": 9389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942117558432737e-05,
+      "loss": 0.6931,
+      "step": 9390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940449125591473e-05,
+      "loss": 0.7817,
+      "step": 9391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938780666337204e-05,
+      "loss": 0.6865,
+      "step": 9392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0937112180716787e-05,
+      "loss": 0.6785,
+      "step": 9393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0935443668777086e-05,
+      "loss": 0.656,
+      "step": 9394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933775130564959e-05,
+      "loss": 0.6892,
+      "step": 9395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0932106566127267e-05,
+      "loss": 0.7288,
+      "step": 9396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0930437975510877e-05,
+      "loss": 0.7736,
+      "step": 9397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0928769358762648e-05,
+      "loss": 0.7313,
+      "step": 9398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0927100715929448e-05,
+      "loss": 0.6331,
+      "step": 9399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925432047058135e-05,
+      "loss": 0.6203,
+      "step": 9400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923763352195582e-05,
+      "loss": 0.7286,
+      "step": 9401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0922094631388656e-05,
+      "loss": 0.6185,
+      "step": 9402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920425884684214e-05,
+      "loss": 0.7375,
+      "step": 9403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091875711212913e-05,
+      "loss": 0.7014,
+      "step": 9404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0917088313770272e-05,
+      "loss": 0.7282,
+      "step": 9405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0915419489654511e-05,
+      "loss": 0.6793,
+      "step": 9406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091375063982871e-05,
+      "loss": 0.6629,
+      "step": 9407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091208176433975e-05,
+      "loss": 0.6708,
+      "step": 9408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910412863234495e-05,
+      "loss": 0.7173,
+      "step": 9409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908743936559814e-05,
+      "loss": 0.7375,
+      "step": 9410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0907074984362587e-05,
+      "loss": 0.7107,
+      "step": 9411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905406006689687e-05,
+      "loss": 0.7133,
+      "step": 9412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903737003587982e-05,
+      "loss": 0.7211,
+      "step": 9413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902067975104354e-05,
+      "loss": 0.6984,
+      "step": 9414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090039892128567e-05,
+      "loss": 0.7428,
+      "step": 9415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0898729842178815e-05,
+      "loss": 0.6718,
+      "step": 9416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897060737830661e-05,
+      "loss": 0.6122,
+      "step": 9417
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895391608288088e-05,
+      "loss": 0.6684,
+      "step": 9418
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0893722453597972e-05,
+      "loss": 0.7485,
+      "step": 9419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892053273807193e-05,
+      "loss": 0.7342,
+      "step": 9420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890384068962632e-05,
+      "loss": 0.716,
+      "step": 9421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0888714839111168e-05,
+      "loss": 0.7671,
+      "step": 9422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088704558429968e-05,
+      "loss": 0.7062,
+      "step": 9423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885376304575055e-05,
+      "loss": 0.6587,
+      "step": 9424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0883706999984174e-05,
+      "loss": 0.6872,
+      "step": 9425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882037670573919e-05,
+      "loss": 0.6776,
+      "step": 9426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0880368316391174e-05,
+      "loss": 0.7246,
+      "step": 9427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878698937482828e-05,
+      "loss": 0.6747,
+      "step": 9428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877029533895758e-05,
+      "loss": 0.6707,
+      "step": 9429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875360105676855e-05,
+      "loss": 0.6439,
+      "step": 9430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873690652873007e-05,
+      "loss": 0.7111,
+      "step": 9431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08720211755311e-05,
+      "loss": 0.6577,
+      "step": 9432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087035167369802e-05,
+      "loss": 0.6176,
+      "step": 9433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0868682147420662e-05,
+      "loss": 0.7434,
+      "step": 9434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0867012596745905e-05,
+      "loss": 0.7043,
+      "step": 9435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0865343021720649e-05,
+      "loss": 0.7282,
+      "step": 9436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086367342239178e-05,
+      "loss": 0.7166,
+      "step": 9437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862003798806195e-05,
+      "loss": 0.6225,
+      "step": 9438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086033415101078e-05,
+      "loss": 0.3318,
+      "step": 9439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0858664479052429e-05,
+      "loss": 0.6542,
+      "step": 9440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085699478297804e-05,
+      "loss": 0.6887,
+      "step": 9441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08553250628345e-05,
+      "loss": 0.6246,
+      "step": 9442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0853655318668712e-05,
+      "loss": 0.7079,
+      "step": 9443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851985550527568e-05,
+      "loss": 0.6741,
+      "step": 9444
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085031575845796e-05,
+      "loss": 0.6798,
+      "step": 9445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084864594250679e-05,
+      "loss": 0.6747,
+      "step": 9446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846976102720958e-05,
+      "loss": 0.6956,
+      "step": 9447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845306239147355e-05,
+      "loss": 0.6899,
+      "step": 9448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843636351832888e-05,
+      "loss": 0.7321,
+      "step": 9449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084196644082445e-05,
+      "loss": 0.7221,
+      "step": 9450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840296506168943e-05,
+      "loss": 0.6962,
+      "step": 9451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083862654791327e-05,
+      "loss": 0.689,
+      "step": 9452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083695656610433e-05,
+      "loss": 0.673,
+      "step": 9453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835286560789031e-05,
+      "loss": 0.6892,
+      "step": 9454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833616532014268e-05,
+      "loss": 0.7138,
+      "step": 9455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083194647982695e-05,
+      "loss": 0.7244,
+      "step": 9456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830276404273982e-05,
+      "loss": 0.6914,
+      "step": 9457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0828606305402266e-05,
+      "loss": 0.6828,
+      "step": 9458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826936183258705e-05,
+      "loss": 0.6876,
+      "step": 9459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825266037890211e-05,
+      "loss": 0.7462,
+      "step": 9460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823595869343689e-05,
+      "loss": 0.7203,
+      "step": 9461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821925677666045e-05,
+      "loss": 0.6154,
+      "step": 9462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820255462904187e-05,
+      "loss": 0.7144,
+      "step": 9463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818585225105029e-05,
+      "loss": 0.6984,
+      "step": 9464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816914964315472e-05,
+      "loss": 0.679,
+      "step": 9465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815244680582435e-05,
+      "loss": 0.7149,
+      "step": 9466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813574373952825e-05,
+      "loss": 0.6846,
+      "step": 9467
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811904044473551e-05,
+      "loss": 0.6325,
+      "step": 9468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0810233692191525e-05,
+      "loss": 0.6601,
+      "step": 9469
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0808563317153664e-05,
+      "loss": 0.6386,
+      "step": 9470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0806892919406879e-05,
+      "loss": 0.7587,
+      "step": 9471
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805222498998085e-05,
+      "loss": 0.6789,
+      "step": 9472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0803552055974195e-05,
+      "loss": 0.6734,
+      "step": 9473
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801881590382126e-05,
+      "loss": 0.7385,
+      "step": 9474
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080021110226879e-05,
+      "loss": 0.6161,
+      "step": 9475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798540591681109e-05,
+      "loss": 0.7249,
+      "step": 9476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796870058665998e-05,
+      "loss": 0.6955,
+      "step": 9477
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0795199503270375e-05,
+      "loss": 0.6413,
+      "step": 9478
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793528925541157e-05,
+      "loss": 0.3541,
+      "step": 9479
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791858325525262e-05,
+      "loss": 0.7263,
+      "step": 9480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790187703269613e-05,
+      "loss": 0.7297,
+      "step": 9481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0788517058821129e-05,
+      "loss": 0.7486,
+      "step": 9482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786846392226733e-05,
+      "loss": 0.7097,
+      "step": 9483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785175703533341e-05,
+      "loss": 0.707,
+      "step": 9484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783504992787882e-05,
+      "loss": 0.6706,
+      "step": 9485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0781834260037272e-05,
+      "loss": 0.6973,
+      "step": 9486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780163505328438e-05,
+      "loss": 0.6635,
+      "step": 9487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0778492728708306e-05,
+      "loss": 0.7254,
+      "step": 9488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776821930223796e-05,
+      "loss": 0.6724,
+      "step": 9489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0775151109921835e-05,
+      "loss": 0.6555,
+      "step": 9490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773480267849353e-05,
+      "loss": 0.7001,
+      "step": 9491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771809404053271e-05,
+      "loss": 0.6847,
+      "step": 9492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770138518580519e-05,
+      "loss": 0.7222,
+      "step": 9493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768467611478025e-05,
+      "loss": 0.7758,
+      "step": 9494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766796682792712e-05,
+      "loss": 0.7291,
+      "step": 9495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0765125732571515e-05,
+      "loss": 0.6377,
+      "step": 9496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076345476086136e-05,
+      "loss": 0.7906,
+      "step": 9497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761783767709182e-05,
+      "loss": 0.7755,
+      "step": 9498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760112753161907e-05,
+      "loss": 0.7332,
+      "step": 9499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0758441717266465e-05,
+      "loss": 0.7004,
+      "step": 9500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756770660069794e-05,
+      "loss": 0.7068,
+      "step": 9501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075509958161882e-05,
+      "loss": 0.8073,
+      "step": 9502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753428481960483e-05,
+      "loss": 0.7545,
+      "step": 9503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751757361141711e-05,
+      "loss": 0.7181,
+      "step": 9504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0750086219209439e-05,
+      "loss": 0.6848,
+      "step": 9505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748415056210604e-05,
+      "loss": 0.6995,
+      "step": 9506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074674387219214e-05,
+      "loss": 0.7399,
+      "step": 9507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0745072667200981e-05,
+      "loss": 0.7051,
+      "step": 9508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0743401441284074e-05,
+      "loss": 0.6424,
+      "step": 9509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074173019448834e-05,
+      "loss": 0.3434,
+      "step": 9510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074005892686073e-05,
+      "loss": 0.7102,
+      "step": 9511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0738387638448178e-05,
+      "loss": 0.684,
+      "step": 9512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736716329297623e-05,
+      "loss": 0.7303,
+      "step": 9513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735044999456003e-05,
+      "loss": 0.7353,
+      "step": 9514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073337364897026e-05,
+      "loss": 0.7034,
+      "step": 9515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731702277887335e-05,
+      "loss": 0.7142,
+      "step": 9516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0730030886254169e-05,
+      "loss": 0.6998,
+      "step": 9517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07283594741177e-05,
+      "loss": 0.7167,
+      "step": 9518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072668804152488e-05,
+      "loss": 0.7315,
+      "step": 9519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0725016588522639e-05,
+      "loss": 0.7762,
+      "step": 9520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0723345115157933e-05,
+      "loss": 0.751,
+      "step": 9521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07216736214777e-05,
+      "loss": 0.6897,
+      "step": 9522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720002107528886e-05,
+      "loss": 0.6858,
+      "step": 9523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718330573358436e-05,
+      "loss": 0.7746,
+      "step": 9524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0716659019013294e-05,
+      "loss": 0.7638,
+      "step": 9525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071498744454041e-05,
+      "loss": 0.6711,
+      "step": 9526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071331584998673e-05,
+      "loss": 0.7059,
+      "step": 9527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0711644235399199e-05,
+      "loss": 0.709,
+      "step": 9528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0709972600824772e-05,
+      "loss": 0.7474,
+      "step": 9529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0708300946310386e-05,
+      "loss": 0.7383,
+      "step": 9530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0706629271903002e-05,
+      "loss": 0.7327,
+      "step": 9531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704957577649566e-05,
+      "loss": 0.6916,
+      "step": 9532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703285863597026e-05,
+      "loss": 0.7343,
+      "step": 9533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701614129792335e-05,
+      "loss": 0.7671,
+      "step": 9534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0699942376282443e-05,
+      "loss": 0.7185,
+      "step": 9535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698270603114303e-05,
+      "loss": 0.6859,
+      "step": 9536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696598810334868e-05,
+      "loss": 0.7405,
+      "step": 9537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069492699799109e-05,
+      "loss": 0.7874,
+      "step": 9538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069325516612993e-05,
+      "loss": 0.7521,
+      "step": 9539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069158331479833e-05,
+      "loss": 0.7346,
+      "step": 9540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689911444043249e-05,
+      "loss": 0.6834,
+      "step": 9541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068823955391165e-05,
+      "loss": 0.6702,
+      "step": 9542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0686567644450478e-05,
+      "loss": 0.6797,
+      "step": 9543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0684895715706697e-05,
+      "loss": 0.6416,
+      "step": 9544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0683223767727259e-05,
+      "loss": 0.7085,
+      "step": 9545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0681551800559126e-05,
+      "loss": 0.7131,
+      "step": 9546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679879814249254e-05,
+      "loss": 0.6867,
+      "step": 9547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678207808844605e-05,
+      "loss": 0.3471,
+      "step": 9548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676535784392131e-05,
+      "loss": 0.7241,
+      "step": 9549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674863740938797e-05,
+      "loss": 0.7832,
+      "step": 9550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673191678531558e-05,
+      "loss": 0.6997,
+      "step": 9551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0671519597217384e-05,
+      "loss": 0.7119,
+      "step": 9552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0669847497043227e-05,
+      "loss": 0.6615,
+      "step": 9553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0668175378056053e-05,
+      "loss": 0.7132,
+      "step": 9554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0666503240302824e-05,
+      "loss": 0.6957,
+      "step": 9555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0664831083830503e-05,
+      "loss": 0.7151,
+      "step": 9556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663158908686053e-05,
+      "loss": 0.7023,
+      "step": 9557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066148671491644e-05,
+      "loss": 0.7039,
+      "step": 9558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659814502568626e-05,
+      "loss": 0.7845,
+      "step": 9559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658142271689576e-05,
+      "loss": 0.704,
+      "step": 9560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656470022326254e-05,
+      "loss": 0.719,
+      "step": 9561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065479775452563e-05,
+      "loss": 0.635,
+      "step": 9562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0653125468334667e-05,
+      "loss": 0.7707,
+      "step": 9563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651453163800334e-05,
+      "loss": 0.7159,
+      "step": 9564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0649780840969597e-05,
+      "loss": 0.6602,
+      "step": 9565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0648108499889427e-05,
+      "loss": 0.6761,
+      "step": 9566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064643614060679e-05,
+      "loss": 0.7792,
+      "step": 9567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0644763763168658e-05,
+      "loss": 0.6951,
+      "step": 9568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643091367621996e-05,
+      "loss": 0.6701,
+      "step": 9569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641418954013776e-05,
+      "loss": 0.6678,
+      "step": 9570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063974652239097e-05,
+      "loss": 0.7128,
+      "step": 9571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0638074072800548e-05,
+      "loss": 0.7171,
+      "step": 9572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0636401605289481e-05,
+      "loss": 0.7519,
+      "step": 9573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634729119904745e-05,
+      "loss": 0.6365,
+      "step": 9574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0633056616693304e-05,
+      "loss": 0.6697,
+      "step": 9575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063138409570214e-05,
+      "loss": 0.6176,
+      "step": 9576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629711556978223e-05,
+      "loss": 0.7021,
+      "step": 9577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062803900056853e-05,
+      "loss": 0.7487,
+      "step": 9578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626366426520031e-05,
+      "loss": 0.6891,
+      "step": 9579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0624693834879703e-05,
+      "loss": 0.6227,
+      "step": 9580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0623021225694521e-05,
+      "loss": 0.7366,
+      "step": 9581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621348599011462e-05,
+      "loss": 0.6405,
+      "step": 9582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0619675954877505e-05,
+      "loss": 0.6906,
+      "step": 9583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618003293339623e-05,
+      "loss": 0.7008,
+      "step": 9584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616330614444792e-05,
+      "loss": 0.7107,
+      "step": 9585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0614657918239996e-05,
+      "loss": 0.7051,
+      "step": 9586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0612985204772213e-05,
+      "loss": 0.7072,
+      "step": 9587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611312474088417e-05,
+      "loss": 0.6612,
+      "step": 9588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0609639726235592e-05,
+      "loss": 0.7851,
+      "step": 9589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607966961260713e-05,
+      "loss": 0.6577,
+      "step": 9590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606294179210768e-05,
+      "loss": 0.7142,
+      "step": 9591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0604621380132731e-05,
+      "loss": 0.7157,
+      "step": 9592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602948564073588e-05,
+      "loss": 0.6899,
+      "step": 9593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060127573108032e-05,
+      "loss": 0.69,
+      "step": 9594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0599602881199906e-05,
+      "loss": 0.6948,
+      "step": 9595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059793001447933e-05,
+      "loss": 0.683,
+      "step": 9596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059625713096558e-05,
+      "loss": 0.7521,
+      "step": 9597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594584230705635e-05,
+      "loss": 0.7404,
+      "step": 9598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059291131374648e-05,
+      "loss": 0.6614,
+      "step": 9599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591238380135102e-05,
+      "loss": 0.701,
+      "step": 9600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589565429918484e-05,
+      "loss": 0.6924,
+      "step": 9601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0587892463143611e-05,
+      "loss": 0.7404,
+      "step": 9602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586219479857474e-05,
+      "loss": 0.6771,
+      "step": 9603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0584546480107056e-05,
+      "loss": 0.6761,
+      "step": 9604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582873463939343e-05,
+      "loss": 0.7235,
+      "step": 9605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0581200431401322e-05,
+      "loss": 0.7403,
+      "step": 9606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579527382539984e-05,
+      "loss": 0.7074,
+      "step": 9607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577854317402317e-05,
+      "loss": 0.6893,
+      "step": 9608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0576181236035307e-05,
+      "loss": 0.3386,
+      "step": 9609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574508138485949e-05,
+      "loss": 0.7397,
+      "step": 9610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0572835024801227e-05,
+      "loss": 0.6994,
+      "step": 9611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571161895028134e-05,
+      "loss": 0.679,
+      "step": 9612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569488749213664e-05,
+      "loss": 0.6851,
+      "step": 9613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05678155874048e-05,
+      "loss": 0.6315,
+      "step": 9614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0566142409648539e-05,
+      "loss": 0.6969,
+      "step": 9615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0564469215991869e-05,
+      "loss": 0.6301,
+      "step": 9616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056279600648179e-05,
+      "loss": 0.6683,
+      "step": 9617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0561122781165292e-05,
+      "loss": 0.6679,
+      "step": 9618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0559449540089364e-05,
+      "loss": 0.7672,
+      "step": 9619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557776283301003e-05,
+      "loss": 0.6759,
+      "step": 9620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0556103010847203e-05,
+      "loss": 0.7256,
+      "step": 9621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0554429722774962e-05,
+      "loss": 0.6066,
+      "step": 9622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552756419131272e-05,
+      "loss": 0.6603,
+      "step": 9623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0551083099963125e-05,
+      "loss": 0.6638,
+      "step": 9624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549409765317521e-05,
+      "loss": 0.7301,
+      "step": 9625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547736415241459e-05,
+      "loss": 0.701,
+      "step": 9626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0546063049781932e-05,
+      "loss": 0.6518,
+      "step": 9627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054438966898594e-05,
+      "loss": 0.6695,
+      "step": 9628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542716272900477e-05,
+      "loss": 0.7132,
+      "step": 9629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541042861572543e-05,
+      "loss": 0.7363,
+      "step": 9630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053936943504914e-05,
+      "loss": 0.7195,
+      "step": 9631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537695993377261e-05,
+      "loss": 0.7458,
+      "step": 9632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536022536603914e-05,
+      "loss": 0.6802,
+      "step": 9633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053434906477609e-05,
+      "loss": 0.6485,
+      "step": 9634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0532675577940792e-05,
+      "loss": 0.6568,
+      "step": 9635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0531002076145023e-05,
+      "loss": 0.7174,
+      "step": 9636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529328559435782e-05,
+      "loss": 0.6961,
+      "step": 9637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527655027860073e-05,
+      "loss": 0.6838,
+      "step": 9638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0525981481464898e-05,
+      "loss": 0.6833,
+      "step": 9639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0524307920297253e-05,
+      "loss": 0.692,
+      "step": 9640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522634344404148e-05,
+      "loss": 0.7011,
+      "step": 9641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0520960753832583e-05,
+      "loss": 0.7003,
+      "step": 9642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519287148629566e-05,
+      "loss": 0.7151,
+      "step": 9643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517613528842096e-05,
+      "loss": 0.6575,
+      "step": 9644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0515939894517178e-05,
+      "loss": 0.678,
+      "step": 9645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514266245701819e-05,
+      "loss": 0.7141,
+      "step": 9646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512592582443024e-05,
+      "loss": 0.6837,
+      "step": 9647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510918904787796e-05,
+      "loss": 0.627,
+      "step": 9648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509245212783147e-05,
+      "loss": 0.7197,
+      "step": 9649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507571506476075e-05,
+      "loss": 0.7704,
+      "step": 9650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505897785913595e-05,
+      "loss": 0.7027,
+      "step": 9651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050422405114271e-05,
+      "loss": 0.7535,
+      "step": 9652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050255030221043e-05,
+      "loss": 0.6869,
+      "step": 9653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050087653916376e-05,
+      "loss": 0.7095,
+      "step": 9654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499202762049711e-05,
+      "loss": 0.7029,
+      "step": 9655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0497528970915293e-05,
+      "loss": 0.7393,
+      "step": 9656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049585516580751e-05,
+      "loss": 0.6918,
+      "step": 9657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494181346773376e-05,
+      "loss": 0.7092,
+      "step": 9658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492507513859904e-05,
+      "loss": 0.6472,
+      "step": 9659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490833667114098e-05,
+      "loss": 0.7717,
+      "step": 9660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0489159806582969e-05,
+      "loss": 0.7605,
+      "step": 9661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487485932313535e-05,
+      "loss": 0.685,
+      "step": 9662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0485812044352802e-05,
+      "loss": 0.6701,
+      "step": 9663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484138142747783e-05,
+      "loss": 0.6937,
+      "step": 9664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482464227545493e-05,
+      "loss": 0.6849,
+      "step": 9665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048079029879294e-05,
+      "loss": 0.7617,
+      "step": 9666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047911635653714e-05,
+      "loss": 0.663,
+      "step": 9667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0477442400825109e-05,
+      "loss": 0.8069,
+      "step": 9668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0475768431703858e-05,
+      "loss": 0.7691,
+      "step": 9669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04740944492204e-05,
+      "loss": 0.7418,
+      "step": 9670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047242045342175e-05,
+      "loss": 0.7482,
+      "step": 9671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047074644435493e-05,
+      "loss": 0.7164,
+      "step": 9672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469072422066944e-05,
+      "loss": 0.7076,
+      "step": 9673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0467398386604816e-05,
+      "loss": 0.7374,
+      "step": 9674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046572433801556e-05,
+      "loss": 0.7347,
+      "step": 9675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0464050276346192e-05,
+      "loss": 0.8,
+      "step": 9676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046237620164373e-05,
+      "loss": 0.7092,
+      "step": 9677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046070211395519e-05,
+      "loss": 0.6656,
+      "step": 9678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045902801332759e-05,
+      "loss": 0.7629,
+      "step": 9679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457353899807947e-05,
+      "loss": 0.7111,
+      "step": 9680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455679773443279e-05,
+      "loss": 0.6829,
+      "step": 9681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0454005634280611e-05,
+      "loss": 0.7338,
+      "step": 9682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0452331482366954e-05,
+      "loss": 0.7331,
+      "step": 9683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450657317749328e-05,
+      "loss": 0.746,
+      "step": 9684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044898314047476e-05,
+      "loss": 0.7309,
+      "step": 9685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0447308950590265e-05,
+      "loss": 0.7598,
+      "step": 9686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044563474814286e-05,
+      "loss": 0.701,
+      "step": 9687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0443960533179577e-05,
+      "loss": 0.687,
+      "step": 9688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442286305747427e-05,
+      "loss": 0.7105,
+      "step": 9689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440612065893431e-05,
+      "loss": 0.7825,
+      "step": 9690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438937813664618e-05,
+      "loss": 0.7121,
+      "step": 9691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437263549108008e-05,
+      "loss": 0.7362,
+      "step": 9692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0435589272270622e-05,
+      "loss": 0.7291,
+      "step": 9693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043391498319948e-05,
+      "loss": 0.6634,
+      "step": 9694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043224068194161e-05,
+      "loss": 0.6704,
+      "step": 9695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430566368544035e-05,
+      "loss": 0.7161,
+      "step": 9696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042889204305378e-05,
+      "loss": 0.6909,
+      "step": 9697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427217705517868e-05,
+      "loss": 0.6794,
+      "step": 9698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042554335598332e-05,
+      "loss": 0.6742,
+      "step": 9699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423868994497166e-05,
+      "loss": 0.7051,
+      "step": 9700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422194621106428e-05,
+      "loss": 0.6256,
+      "step": 9701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420520235858134e-05,
+      "loss": 0.7343,
+      "step": 9702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041884583879931e-05,
+      "loss": 0.6836,
+      "step": 9703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041717142997698e-05,
+      "loss": 0.6972,
+      "step": 9704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0415497009438172e-05,
+      "loss": 0.7131,
+      "step": 9705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413822577229913e-05,
+      "loss": 0.6628,
+      "step": 9706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0412148133399232e-05,
+      "loss": 0.7092,
+      "step": 9707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0410473677993156e-05,
+      "loss": 0.7316,
+      "step": 9708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408799211058711e-05,
+      "loss": 0.6611,
+      "step": 9709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407124732642924e-05,
+      "loss": 0.713,
+      "step": 9710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405450242792827e-05,
+      "loss": 0.6479,
+      "step": 9711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403775741555448e-05,
+      "loss": 0.7205,
+      "step": 9712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402101228977816e-05,
+      "loss": 0.6377,
+      "step": 9713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400426705106964e-05,
+      "loss": 0.6835,
+      "step": 9714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0398752169989914e-05,
+      "loss": 0.7489,
+      "step": 9715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397077623673701e-05,
+      "loss": 0.7623,
+      "step": 9716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0395403066205356e-05,
+      "loss": 0.7505,
+      "step": 9717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393728497631909e-05,
+      "loss": 0.7004,
+      "step": 9718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0392053918000393e-05,
+      "loss": 0.6949,
+      "step": 9719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390379327357836e-05,
+      "loss": 0.6035,
+      "step": 9720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038870472575127e-05,
+      "loss": 0.7041,
+      "step": 9721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038703011322773e-05,
+      "loss": 0.6631,
+      "step": 9722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0385355489834246e-05,
+      "loss": 0.7213,
+      "step": 9723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383680855617855e-05,
+      "loss": 0.7163,
+      "step": 9724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382006210625582e-05,
+      "loss": 0.7382,
+      "step": 9725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380331554904463e-05,
+      "loss": 0.7402,
+      "step": 9726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037865688850154e-05,
+      "loss": 0.7263,
+      "step": 9727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0376982211463835e-05,
+      "loss": 0.6851,
+      "step": 9728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037530752383839e-05,
+      "loss": 0.78,
+      "step": 9729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0373632825672236e-05,
+      "loss": 0.7093,
+      "step": 9730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371958117012406e-05,
+      "loss": 0.716,
+      "step": 9731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0370283397905941e-05,
+      "loss": 0.6672,
+      "step": 9732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0368608668399871e-05,
+      "loss": 0.7172,
+      "step": 9733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366933928541236e-05,
+      "loss": 0.6275,
+      "step": 9734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0365259178377066e-05,
+      "loss": 0.6883,
+      "step": 9735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363584417954403e-05,
+      "loss": 0.6608,
+      "step": 9736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0361909647320282e-05,
+      "loss": 0.7078,
+      "step": 9737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036023486652174e-05,
+      "loss": 0.7427,
+      "step": 9738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358560075605812e-05,
+      "loss": 0.644,
+      "step": 9739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0356885274619536e-05,
+      "loss": 0.7126,
+      "step": 9740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355210463609951e-05,
+      "loss": 0.6892,
+      "step": 9741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353535642624093e-05,
+      "loss": 0.722,
+      "step": 9742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0351860811709002e-05,
+      "loss": 0.6663,
+      "step": 9743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035018597091172e-05,
+      "loss": 0.643,
+      "step": 9744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034851112027928e-05,
+      "loss": 0.7032,
+      "step": 9745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346836259858716e-05,
+      "loss": 0.632,
+      "step": 9746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345161389697083e-05,
+      "loss": 0.6723,
+      "step": 9747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0343486509841408e-05,
+      "loss": 0.6691,
+      "step": 9748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341811620338736e-05,
+      "loss": 0.6826,
+      "step": 9749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0340136721236103e-05,
+      "loss": 0.7413,
+      "step": 9750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0338461812580554e-05,
+      "loss": 0.6562,
+      "step": 9751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336786894419129e-05,
+      "loss": 0.7278,
+      "step": 9752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335111966798868e-05,
+      "loss": 0.7221,
+      "step": 9753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0333437029766811e-05,
+      "loss": 0.7734,
+      "step": 9754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033176208337e-05,
+      "loss": 0.7144,
+      "step": 9755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0330087127655476e-05,
+      "loss": 0.7086,
+      "step": 9756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0328412162670284e-05,
+      "loss": 0.6665,
+      "step": 9757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0326737188461465e-05,
+      "loss": 0.7262,
+      "step": 9758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032506220507606e-05,
+      "loss": 0.6709,
+      "step": 9759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0323387212561113e-05,
+      "loss": 0.7434,
+      "step": 9760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321712210963668e-05,
+      "loss": 0.694,
+      "step": 9761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0320037200330765e-05,
+      "loss": 0.7313,
+      "step": 9762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0318362180709455e-05,
+      "loss": 0.6952,
+      "step": 9763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0316687152146774e-05,
+      "loss": 0.773,
+      "step": 9764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315012114689767e-05,
+      "loss": 0.674,
+      "step": 9765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031333706838548e-05,
+      "loss": 0.733,
+      "step": 9766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031166201328096e-05,
+      "loss": 0.6888,
+      "step": 9767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309986949423245e-05,
+      "loss": 0.6891,
+      "step": 9768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308311876859387e-05,
+      "loss": 0.7376,
+      "step": 9769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0306636795636429e-05,
+      "loss": 0.6654,
+      "step": 9770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0304961705801415e-05,
+      "loss": 0.7098,
+      "step": 9771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0303286607401392e-05,
+      "loss": 0.7032,
+      "step": 9772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301611500483408e-05,
+      "loss": 0.6931,
+      "step": 9773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299936385094505e-05,
+      "loss": 0.6949,
+      "step": 9774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298261261281732e-05,
+      "loss": 0.6833,
+      "step": 9775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0296586129092137e-05,
+      "loss": 0.7159,
+      "step": 9776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294910988572764e-05,
+      "loss": 0.7438,
+      "step": 9777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029323583977066e-05,
+      "loss": 0.6846,
+      "step": 9778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291560682732878e-05,
+      "loss": 0.6975,
+      "step": 9779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028988551750646e-05,
+      "loss": 0.6855,
+      "step": 9780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0288210344138451e-05,
+      "loss": 0.7412,
+      "step": 9781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286535162675908e-05,
+      "loss": 0.6787,
+      "step": 9782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284859973165877e-05,
+      "loss": 0.6951,
+      "step": 9783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283184775655401e-05,
+      "loss": 0.7416,
+      "step": 9784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281509570191534e-05,
+      "loss": 0.7031,
+      "step": 9785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0279834356821324e-05,
+      "loss": 0.7228,
+      "step": 9786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278159135591818e-05,
+      "loss": 0.7207,
+      "step": 9787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0276483906550067e-05,
+      "loss": 0.6573,
+      "step": 9788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274808669743126e-05,
+      "loss": 0.7259,
+      "step": 9789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273133425218034e-05,
+      "loss": 0.7001,
+      "step": 9790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0271458173021846e-05,
+      "loss": 0.7332,
+      "step": 9791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269782913201618e-05,
+      "loss": 0.7695,
+      "step": 9792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268107645804392e-05,
+      "loss": 0.7459,
+      "step": 9793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0266432370877225e-05,
+      "loss": 0.7238,
+      "step": 9794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264757088467163e-05,
+      "loss": 0.596,
+      "step": 9795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263081798621262e-05,
+      "loss": 0.7354,
+      "step": 9796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261406501386568e-05,
+      "loss": 0.7076,
+      "step": 9797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0259731196810139e-05,
+      "loss": 0.7066,
+      "step": 9798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0258055884939023e-05,
+      "loss": 0.7216,
+      "step": 9799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256380565820271e-05,
+      "loss": 0.7145,
+      "step": 9800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0254705239500935e-05,
+      "loss": 0.735,
+      "step": 9801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253029906028073e-05,
+      "loss": 0.6929,
+      "step": 9802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251354565448732e-05,
+      "loss": 0.6365,
+      "step": 9803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0249679217809964e-05,
+      "loss": 0.6814,
+      "step": 9804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0248003863158829e-05,
+      "loss": 0.6391,
+      "step": 9805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0246328501542373e-05,
+      "loss": 0.6244,
+      "step": 9806
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244653133007651e-05,
+      "loss": 0.7116,
+      "step": 9807
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024297775760172e-05,
+      "loss": 0.7288,
+      "step": 9808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241302375371633e-05,
+      "loss": 0.7255,
+      "step": 9809
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0239626986364441e-05,
+      "loss": 0.6485,
+      "step": 9810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237951590627197e-05,
+      "loss": 0.6905,
+      "step": 9811
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236276188206962e-05,
+      "loss": 0.7176,
+      "step": 9812
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0234600779150785e-05,
+      "loss": 0.7114,
+      "step": 9813
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023292536350572e-05,
+      "loss": 0.6967,
+      "step": 9814
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0231249941318828e-05,
+      "loss": 0.6863,
+      "step": 9815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229574512637159e-05,
+      "loss": 0.6688,
+      "step": 9816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022789907750777e-05,
+      "loss": 0.6971,
+      "step": 9817
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0226223635977719e-05,
+      "loss": 0.6914,
+      "step": 9818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224548188094054e-05,
+      "loss": 0.6877,
+      "step": 9819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0222872733903838e-05,
+      "loss": 0.6996,
+      "step": 9820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0221197273454123e-05,
+      "loss": 0.6831,
+      "step": 9821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219521806791967e-05,
+      "loss": 0.6984,
+      "step": 9822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021784633396443e-05,
+      "loss": 0.6625,
+      "step": 9823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021617085501856e-05,
+      "loss": 0.7407,
+      "step": 9824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021449537000142e-05,
+      "loss": 0.625,
+      "step": 9825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0212819878960065e-05,
+      "loss": 0.6425,
+      "step": 9826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0211144381941554e-05,
+      "loss": 0.664,
+      "step": 9827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0209468878992942e-05,
+      "loss": 0.6794,
+      "step": 9828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0207793370161287e-05,
+      "loss": 0.6884,
+      "step": 9829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0206117855493646e-05,
+      "loss": 0.6181,
+      "step": 9830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204442335037077e-05,
+      "loss": 0.6862,
+      "step": 9831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202766808838638e-05,
+      "loss": 0.6666,
+      "step": 9832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0201091276945388e-05,
+      "loss": 0.7681,
+      "step": 9833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199415739404381e-05,
+      "loss": 0.7302,
+      "step": 9834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197740196262684e-05,
+      "loss": 0.7427,
+      "step": 9835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196064647567346e-05,
+      "loss": 0.7047,
+      "step": 9836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194389093365428e-05,
+      "loss": 0.7207,
+      "step": 9837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0192713533703997e-05,
+      "loss": 0.7443,
+      "step": 9838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191037968630103e-05,
+      "loss": 0.7515,
+      "step": 9839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189362398190806e-05,
+      "loss": 0.3497,
+      "step": 9840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0187686822433166e-05,
+      "loss": 0.6703,
+      "step": 9841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0186011241404248e-05,
+      "loss": 0.6899,
+      "step": 9842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184335655151103e-05,
+      "loss": 0.6973,
+      "step": 9843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0182660063720797e-05,
+      "loss": 0.6639,
+      "step": 9844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180984467160386e-05,
+      "loss": 0.7447,
+      "step": 9845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017930886551693e-05,
+      "loss": 0.7098,
+      "step": 9846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177633258837494e-05,
+      "loss": 0.6818,
+      "step": 9847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175957647169135e-05,
+      "loss": 0.7531,
+      "step": 9848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017428203055891e-05,
+      "loss": 0.7198,
+      "step": 9849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172606409053887e-05,
+      "loss": 0.6524,
+      "step": 9850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017093078270112e-05,
+      "loss": 0.7989,
+      "step": 9851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0169255151547672e-05,
+      "loss": 0.6827,
+      "step": 9852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0167579515640607e-05,
+      "loss": 0.7393,
+      "step": 9853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0165903875026983e-05,
+      "loss": 0.7071,
+      "step": 9854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164228229753863e-05,
+      "loss": 0.7322,
+      "step": 9855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0162552579868304e-05,
+      "loss": 0.6705,
+      "step": 9856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160876925417373e-05,
+      "loss": 0.7296,
+      "step": 9857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0159201266448129e-05,
+      "loss": 0.6892,
+      "step": 9858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157525603007631e-05,
+      "loss": 0.7087,
+      "step": 9859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155849935142947e-05,
+      "loss": 0.6831,
+      "step": 9860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154174262901138e-05,
+      "loss": 0.6649,
+      "step": 9861
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015249858632926e-05,
+      "loss": 0.7283,
+      "step": 9862
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150822905474381e-05,
+      "loss": 0.6895,
+      "step": 9863
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149147220383566e-05,
+      "loss": 0.7074,
+      "step": 9864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0147471531103867e-05,
+      "loss": 0.7175,
+      "step": 9865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0145795837682356e-05,
+      "loss": 0.7747,
+      "step": 9866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144120140166092e-05,
+      "loss": 0.7406,
+      "step": 9867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142444438602139e-05,
+      "loss": 0.6978,
+      "step": 9868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0140768733037558e-05,
+      "loss": 0.7865,
+      "step": 9869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139093023519415e-05,
+      "loss": 0.3464,
+      "step": 9870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137417310094771e-05,
+      "loss": 0.6479,
+      "step": 9871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135741592810688e-05,
+      "loss": 0.6954,
+      "step": 9872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134065871714233e-05,
+      "loss": 0.7513,
+      "step": 9873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0132390146852471e-05,
+      "loss": 0.5961,
+      "step": 9874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013071441827246e-05,
+      "loss": 0.6989,
+      "step": 9875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129038686021262e-05,
+      "loss": 0.7394,
+      "step": 9876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012736295014595e-05,
+      "loss": 0.6712,
+      "step": 9877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0125687210693582e-05,
+      "loss": 0.7177,
+      "step": 9878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0124011467711224e-05,
+      "loss": 0.7049,
+      "step": 9879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122335721245938e-05,
+      "loss": 0.7222,
+      "step": 9880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0120659971344786e-05,
+      "loss": 0.7714,
+      "step": 9881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011898421805484e-05,
+      "loss": 0.6981,
+      "step": 9882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0117308461423159e-05,
+      "loss": 0.6592,
+      "step": 9883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011563270149681e-05,
+      "loss": 0.7025,
+      "step": 9884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113956938322853e-05,
+      "loss": 0.7111,
+      "step": 9885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112281171948353e-05,
+      "loss": 0.6656,
+      "step": 9886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110605402420384e-05,
+      "loss": 0.7008,
+      "step": 9887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108929629786001e-05,
+      "loss": 0.6777,
+      "step": 9888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107253854092274e-05,
+      "loss": 0.6654,
+      "step": 9889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105578075386266e-05,
+      "loss": 0.3557,
+      "step": 9890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010390229371504e-05,
+      "loss": 0.6631,
+      "step": 9891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0102226509125666e-05,
+      "loss": 0.7097,
+      "step": 9892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0100550721665208e-05,
+      "loss": 0.6832,
+      "step": 9893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098874931380728e-05,
+      "loss": 0.6868,
+      "step": 9894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0097199138319296e-05,
+      "loss": 0.7233,
+      "step": 9895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009552334252797e-05,
+      "loss": 0.6876,
+      "step": 9896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009384754405383e-05,
+      "loss": 0.7023,
+      "step": 9897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0092171742943927e-05,
+      "loss": 0.6197,
+      "step": 9898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090495939245332e-05,
+      "loss": 0.6334,
+      "step": 9899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0088820133005115e-05,
+      "loss": 0.6642,
+      "step": 9900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087144324270332e-05,
+      "loss": 0.6587,
+      "step": 9901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085468513088059e-05,
+      "loss": 0.7031,
+      "step": 9902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0083792699505359e-05,
+      "loss": 0.7573,
+      "step": 9903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0082116883569295e-05,
+      "loss": 0.6788,
+      "step": 9904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080441065326936e-05,
+      "loss": 0.6272,
+      "step": 9905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078765244825348e-05,
+      "loss": 0.7306,
+      "step": 9906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0077089422111597e-05,
+      "loss": 0.6268,
+      "step": 9907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0075413597232749e-05,
+      "loss": 0.6779,
+      "step": 9908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007373777023587e-05,
+      "loss": 0.7428,
+      "step": 9909
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007206194116803e-05,
+      "loss": 0.7281,
+      "step": 9910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070386110076286e-05,
+      "loss": 0.7422,
+      "step": 9911
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068710277007715e-05,
+      "loss": 0.3382,
+      "step": 9912
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067034442009385e-05,
+      "loss": 0.7648,
+      "step": 9913
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065358605128352e-05,
+      "loss": 0.7286,
+      "step": 9914
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063682766411688e-05,
+      "loss": 0.6956,
+      "step": 9915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0062006925906462e-05,
+      "loss": 0.7164,
+      "step": 9916
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006033108365974e-05,
+      "loss": 0.7529,
+      "step": 9917
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0058655239718587e-05,
+      "loss": 0.7323,
+      "step": 9918
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056979394130071e-05,
+      "loss": 0.6543,
+      "step": 9919
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005530354694126e-05,
+      "loss": 0.7136,
+      "step": 9920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053627698199215e-05,
+      "loss": 0.7374,
+      "step": 9921
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051951847951011e-05,
+      "loss": 0.7013,
+      "step": 9922
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050275996243714e-05,
+      "loss": 0.6822,
+      "step": 9923
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004860014312439e-05,
+      "loss": 0.7275,
+      "step": 9924
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046924288640103e-05,
+      "loss": 0.7158,
+      "step": 9925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045248432837922e-05,
+      "loss": 0.6682,
+      "step": 9926
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043572575764916e-05,
+      "loss": 0.6585,
+      "step": 9927
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041896717468152e-05,
+      "loss": 0.6933,
+      "step": 9928
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040220857994698e-05,
+      "loss": 0.7491,
+      "step": 9929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003854499739162e-05,
+      "loss": 0.6606,
+      "step": 9930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036869135705982e-05,
+      "loss": 0.6332,
+      "step": 9931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035193272984857e-05,
+      "loss": 0.609,
+      "step": 9932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033517409275313e-05,
+      "loss": 0.6485,
+      "step": 9933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0031841544624413e-05,
+      "loss": 0.6998,
+      "step": 9934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030165679079225e-05,
+      "loss": 0.6842,
+      "step": 9935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028489812686819e-05,
+      "loss": 0.7052,
+      "step": 9936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026813945494262e-05,
+      "loss": 0.6675,
+      "step": 9937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025138077548621e-05,
+      "loss": 0.713,
+      "step": 9938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023462208896967e-05,
+      "loss": 0.6547,
+      "step": 9939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0021786339586364e-05,
+      "loss": 0.6686,
+      "step": 9940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020110469663875e-05,
+      "loss": 0.7213,
+      "step": 9941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018434599176579e-05,
+      "loss": 0.6886,
+      "step": 9942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0016758728171533e-05,
+      "loss": 0.781,
+      "step": 9943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0015082856695812e-05,
+      "loss": 0.7636,
+      "step": 9944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001340698479648e-05,
+      "loss": 0.5851,
+      "step": 9945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0011731112520606e-05,
+      "loss": 0.7535,
+      "step": 9946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0010055239915258e-05,
+      "loss": 0.7089,
+      "step": 9947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0008379367027502e-05,
+      "loss": 0.735,
+      "step": 9948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000670349390441e-05,
+      "loss": 0.7611,
+      "step": 9949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0005027620593046e-05,
+      "loss": 0.6472,
+      "step": 9950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0003351747140476e-05,
+      "loss": 0.6854,
+      "step": 9951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0001675873593772e-05,
+      "loss": 0.6817,
+      "step": 9952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.7327,
+      "step": 9953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99832412640623e-06,
+      "loss": 0.7543,
+      "step": 9954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996648252859526e-06,
+      "loss": 0.7448,
+      "step": 9955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994972379406957e-06,
+      "loss": 0.329,
+      "step": 9956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993296506095595e-06,
+      "loss": 0.6933,
+      "step": 9957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9916206329725e-06,
+      "loss": 0.6832,
+      "step": 9958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989944760084745e-06,
+      "loss": 0.6789,
+      "step": 9959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.988268887479397e-06,
+      "loss": 0.6858,
+      "step": 9960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986593015203523e-06,
+      "loss": 0.8147,
+      "step": 9961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98491714330419e-06,
+      "loss": 0.7753,
+      "step": 9962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983241271828472e-06,
+      "loss": 0.717,
+      "step": 9963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981565400823428e-06,
+      "loss": 0.7404,
+      "step": 9964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979889530336128e-06,
+      "loss": 0.3556,
+      "step": 9965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97821366041364e-06,
+      "loss": 0.7258,
+      "step": 9966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976537791103034e-06,
+      "loss": 0.7224,
+      "step": 9967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.974861922451378e-06,
+      "loss": 0.6475,
+      "step": 9968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973186054505738e-06,
+      "loss": 0.7062,
+      "step": 9969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971510187313184e-06,
+      "loss": 0.7359,
+      "step": 9970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969834320920777e-06,
+      "loss": 0.707,
+      "step": 9971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96815845537559e-06,
+      "loss": 0.6881,
+      "step": 9972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966482590724692e-06,
+      "loss": 0.7384,
+      "step": 9973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964806727015144e-06,
+      "loss": 0.6779,
+      "step": 9974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96313086429402e-06,
+      "loss": 0.7168,
+      "step": 9975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961455002608385e-06,
+      "loss": 0.671,
+      "step": 9976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959779142005307e-06,
+      "loss": 0.7101,
+      "step": 9977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958103282531851e-06,
+      "loss": 0.7052,
+      "step": 9978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956427424235087e-06,
+      "loss": 0.7153,
+      "step": 9979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954751567162082e-06,
+      "loss": 0.6722,
+      "step": 9980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9530757113599e-06,
+      "loss": 0.6587,
+      "step": 9981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951399856875612e-06,
+      "loss": 0.736,
+      "step": 9982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949724003756291e-06,
+      "loss": 0.6824,
+      "step": 9983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948048152048992e-06,
+      "loss": 0.668,
+      "step": 9984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.946372301800787e-06,
+      "loss": 0.7026,
+      "step": 9985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.944696453058744e-06,
+      "loss": 0.6487,
+      "step": 9986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94302060586993e-06,
+      "loss": 0.6956,
+      "step": 9987
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.941344760281415e-06,
+      "loss": 0.6987,
+      "step": 9988
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.939668916340266e-06,
+      "loss": 0.6942,
+      "step": 9989
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.937993074093541e-06,
+      "loss": 0.7049,
+      "step": 9990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936317233588314e-06,
+      "loss": 0.6545,
+      "step": 9991
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934641394871651e-06,
+      "loss": 0.6423,
+      "step": 9992
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932965557990618e-06,
+      "loss": 0.7011,
+      "step": 9993
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931289722992285e-06,
+      "loss": 0.6859,
+      "step": 9994
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.929613889923714e-06,
+      "loss": 0.6418,
+      "step": 9995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.927938058831977e-06,
+      "loss": 0.7071,
+      "step": 9996
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.926262229764132e-06,
+      "loss": 0.7363,
+      "step": 9997
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.924586402767255e-06,
+      "loss": 0.7277,
+      "step": 9998
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.922910577888406e-06,
+      "loss": 0.7186,
+      "step": 9999
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.921234755174654e-06,
+      "loss": 0.6233,
+      "step": 10000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.919558934673065e-06,
+      "loss": 0.7562,
+      "step": 10001
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.917883116430708e-06,
+      "loss": 0.7812,
+      "step": 10002
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916207300494646e-06,
+      "loss": 0.7147,
+      "step": 10003
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914531486911944e-06,
+      "loss": 0.6909,
+      "step": 10004
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91285567572967e-06,
+      "loss": 0.6831,
+      "step": 10005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.911179866994888e-06,
+      "loss": 0.6985,
+      "step": 10006
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90950406075467e-06,
+      "loss": 0.7567,
+      "step": 10007
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907828257056075e-06,
+      "loss": 0.6514,
+      "step": 10008
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906152455946176e-06,
+      "loss": 0.7367,
+      "step": 10009
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904476657472031e-06,
+      "loss": 0.7422,
+      "step": 10010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.902800861680708e-06,
+      "loss": 0.7384,
+      "step": 10011
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901125068619273e-06,
+      "loss": 0.688,
+      "step": 10012
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899449278334796e-06,
+      "loss": 0.6955,
+      "step": 10013
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897773490874334e-06,
+      "loss": 0.3473,
+      "step": 10014
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896097706284962e-06,
+      "loss": 0.7505,
+      "step": 10015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894421924613739e-06,
+      "loss": 0.7352,
+      "step": 10016
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89274614590773e-06,
+      "loss": 0.7072,
+      "step": 10017
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891070370214e-06,
+      "loss": 0.6998,
+      "step": 10018
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889394597579619e-06,
+      "loss": 0.6625,
+      "step": 10019
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887718828051646e-06,
+      "loss": 0.6556,
+      "step": 10020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88604306167715e-06,
+      "loss": 0.7026,
+      "step": 10021
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884367298503197e-06,
+      "loss": 0.748,
+      "step": 10022
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.882691538576845e-06,
+      "loss": 0.7917,
+      "step": 10023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881015781945164e-06,
+      "loss": 0.7159,
+      "step": 10024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879340028655216e-06,
+      "loss": 0.6988,
+      "step": 10025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877664278754067e-06,
+      "loss": 0.6931,
+      "step": 10026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875988532288778e-06,
+      "loss": 0.648,
+      "step": 10027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.874312789306423e-06,
+      "loss": 0.6674,
+      "step": 10028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872637049854054e-06,
+      "loss": 0.5803,
+      "step": 10029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87096131397874e-06,
+      "loss": 0.7465,
+      "step": 10030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869285581727543e-06,
+      "loss": 0.6982,
+      "step": 10031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.867609853147532e-06,
+      "loss": 0.6638,
+      "step": 10032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865934128285767e-06,
+      "loss": 0.6984,
+      "step": 10033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864258407189312e-06,
+      "loss": 0.6886,
+      "step": 10034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862582689905232e-06,
+      "loss": 0.7094,
+      "step": 10035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.860906976480588e-06,
+      "loss": 0.6554,
+      "step": 10036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859231266962444e-06,
+      "loss": 0.6431,
+      "step": 10037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857555561397864e-06,
+      "loss": 0.7165,
+      "step": 10038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85587985983391e-06,
+      "loss": 0.6649,
+      "step": 10039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.854204162317646e-06,
+      "loss": 0.7739,
+      "step": 10040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.852528468896136e-06,
+      "loss": 0.6965,
+      "step": 10041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850852779616439e-06,
+      "loss": 0.6922,
+      "step": 10042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84917709452562e-06,
+      "loss": 0.6426,
+      "step": 10043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847501413670742e-06,
+      "loss": 0.7564,
+      "step": 10044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845825737098867e-06,
+      "loss": 0.7491,
+      "step": 10045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844150064857055e-06,
+      "loss": 0.7196,
+      "step": 10046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842474396992367e-06,
+      "loss": 0.741,
+      "step": 10047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.840798733551878e-06,
+      "loss": 0.6833,
+      "step": 10048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839123074582633e-06,
+      "loss": 0.7168,
+      "step": 10049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8374474201317e-06,
+      "loss": 0.6754,
+      "step": 10050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835771770246142e-06,
+      "loss": 0.6818,
+      "step": 10051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834096124973018e-06,
+      "loss": 0.634,
+      "step": 10052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.832420484359395e-06,
+      "loss": 0.7488,
+      "step": 10053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830744848452331e-06,
+      "loss": 0.7178,
+      "step": 10054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829069217298882e-06,
+      "loss": 0.6038,
+      "step": 10055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827393590946116e-06,
+      "loss": 0.698,
+      "step": 10056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825717969441092e-06,
+      "loss": 0.6848,
+      "step": 10057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824042352830869e-06,
+      "loss": 0.7143,
+      "step": 10058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822366741162507e-06,
+      "loss": 0.3217,
+      "step": 10059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82069113448307e-06,
+      "loss": 0.7424,
+      "step": 10060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819015532839618e-06,
+      "loss": 0.6837,
+      "step": 10061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817339936279208e-06,
+      "loss": 0.7106,
+      "step": 10062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8156643448489e-06,
+      "loss": 0.7149,
+      "step": 10063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813988758595757e-06,
+      "loss": 0.6619,
+      "step": 10064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.812313177566835e-06,
+      "loss": 0.7286,
+      "step": 10065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810637601809196e-06,
+      "loss": 0.7297,
+      "step": 10066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808962031369902e-06,
+      "loss": 0.7229,
+      "step": 10067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807286466296008e-06,
+      "loss": 0.6459,
+      "step": 10068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805610906634574e-06,
+      "loss": 0.6675,
+      "step": 10069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803935352432657e-06,
+      "loss": 0.6916,
+      "step": 10070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802259803737318e-06,
+      "loss": 0.7311,
+      "step": 10071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800584260595617e-06,
+      "loss": 0.7623,
+      "step": 10072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798908723054614e-06,
+      "loss": 0.6524,
+      "step": 10073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797233191161367e-06,
+      "loss": 0.6896,
+      "step": 10074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.795557664962928e-06,
+      "loss": 0.6891,
+      "step": 10075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.793882144506357e-06,
+      "loss": 0.6554,
+      "step": 10076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792206629838714e-06,
+      "loss": 0.7302,
+      "step": 10077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79053112100706e-06,
+      "loss": 0.7036,
+      "step": 10078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788855618058446e-06,
+      "loss": 0.6272,
+      "step": 10079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787180121039938e-06,
+      "loss": 0.6707,
+      "step": 10080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785504629998582e-06,
+      "loss": 0.6985,
+      "step": 10081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783829144981443e-06,
+      "loss": 0.6638,
+      "step": 10082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.782153666035575e-06,
+      "loss": 0.6617,
+      "step": 10083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780478193208034e-06,
+      "loss": 0.6937,
+      "step": 10084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778802726545879e-06,
+      "loss": 0.6961,
+      "step": 10085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.777127266096165e-06,
+      "loss": 0.7261,
+      "step": 10086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775451811905951e-06,
+      "loss": 0.7224,
+      "step": 10087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773776364022288e-06,
+      "loss": 0.6853,
+      "step": 10088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.772100922492232e-06,
+      "loss": 0.7473,
+      "step": 10089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770425487362843e-06,
+      "loss": 0.6689,
+      "step": 10090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.768750058681174e-06,
+      "loss": 0.7002,
+      "step": 10091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767074636494278e-06,
+      "loss": 0.7182,
+      "step": 10092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76539922084922e-06,
+      "loss": 0.6666,
+      "step": 10093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763723811793042e-06,
+      "loss": 0.8016,
+      "step": 10094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762048409372806e-06,
+      "loss": 0.6294,
+      "step": 10095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.760373013635562e-06,
+      "loss": 0.6465,
+      "step": 10096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.758697624628369e-06,
+      "loss": 0.6913,
+      "step": 10097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75702224239828e-06,
+      "loss": 0.7957,
+      "step": 10098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.755346866992347e-06,
+      "loss": 0.7062,
+      "step": 10099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.753671498457629e-06,
+      "loss": 0.6596,
+      "step": 10100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751996136841174e-06,
+      "loss": 0.6384,
+      "step": 10101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750320782190038e-06,
+      "loss": 0.7211,
+      "step": 10102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748645434551271e-06,
+      "loss": 0.7312,
+      "step": 10103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74697009397193e-06,
+      "loss": 0.7267,
+      "step": 10104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.745294760499067e-06,
+      "loss": 0.6609,
+      "step": 10105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743619434179734e-06,
+      "loss": 0.6651,
+      "step": 10106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741944115060982e-06,
+      "loss": 0.7365,
+      "step": 10107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740268803189864e-06,
+      "loss": 0.6591,
+      "step": 10108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738593498613435e-06,
+      "loss": 0.7041,
+      "step": 10109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736918201378742e-06,
+      "loss": 0.7005,
+      "step": 10110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735242911532839e-06,
+      "loss": 0.6504,
+      "step": 10111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733567629122777e-06,
+      "loss": 0.6884,
+      "step": 10112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.731892354195613e-06,
+      "loss": 0.6705,
+      "step": 10113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.730217086798387e-06,
+      "loss": 0.6811,
+      "step": 10114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728541826978157e-06,
+      "loss": 0.698,
+      "step": 10115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.726866574781969e-06,
+      "loss": 0.6841,
+      "step": 10116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725191330256877e-06,
+      "loss": 0.702,
+      "step": 10117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.723516093449933e-06,
+      "loss": 0.6978,
+      "step": 10118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721840864408187e-06,
+      "loss": 0.7706,
+      "step": 10119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72016564317868e-06,
+      "loss": 0.7086,
+      "step": 10120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718490429808468e-06,
+      "loss": 0.6474,
+      "step": 10121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7168152243446e-06,
+      "loss": 0.7527,
+      "step": 10122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715140026834126e-06,
+      "loss": 0.6921,
+      "step": 10123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713464837324092e-06,
+      "loss": 0.679,
+      "step": 10124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.711789655861549e-06,
+      "loss": 0.6652,
+      "step": 10125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710114482493545e-06,
+      "loss": 0.6982,
+      "step": 10126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708439317267126e-06,
+      "loss": 0.7073,
+      "step": 10127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706764160229343e-06,
+      "loss": 0.7073,
+      "step": 10128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70508901142724e-06,
+      "loss": 0.7873,
+      "step": 10129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703413870907867e-06,
+      "loss": 0.7319,
+      "step": 10130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70173873871827e-06,
+      "loss": 0.6752,
+      "step": 10131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700063614905499e-06,
+      "loss": 0.6933,
+      "step": 10132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698388499516597e-06,
+      "loss": 0.6395,
+      "step": 10133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.696713392598611e-06,
+      "loss": 0.7338,
+      "step": 10134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695038294198588e-06,
+      "loss": 0.6658,
+      "step": 10135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.693363204363575e-06,
+      "loss": 0.5771,
+      "step": 10136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691688123140613e-06,
+      "loss": 0.7296,
+      "step": 10137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690013050576755e-06,
+      "loss": 0.6694,
+      "step": 10138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688337986719048e-06,
+      "loss": 0.7256,
+      "step": 10139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.686662931614526e-06,
+      "loss": 0.6604,
+      "step": 10140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684987885310236e-06,
+      "loss": 0.7574,
+      "step": 10141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68331284785323e-06,
+      "loss": 0.6569,
+      "step": 10142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.681637819290548e-06,
+      "loss": 0.6735,
+      "step": 10143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.679962799669233e-06,
+      "loss": 0.6729,
+      "step": 10144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678287789036333e-06,
+      "loss": 0.677,
+      "step": 10145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.67661278743889e-06,
+      "loss": 0.7038,
+      "step": 10146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.674937794923942e-06,
+      "loss": 0.6215,
+      "step": 10147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.673262811538537e-06,
+      "loss": 0.616,
+      "step": 10148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671587837329717e-06,
+      "loss": 0.6709,
+      "step": 10149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669912872344526e-06,
+      "loss": 0.7181,
+      "step": 10150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668237916630004e-06,
+      "loss": 0.7084,
+      "step": 10151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666562970233196e-06,
+      "loss": 0.71,
+      "step": 10152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664888033201137e-06,
+      "loss": 0.7259,
+      "step": 10153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663213105580875e-06,
+      "loss": 0.6967,
+      "step": 10154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.661538187419448e-06,
+      "loss": 0.7635,
+      "step": 10155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.659863278763899e-06,
+      "loss": 0.7373,
+      "step": 10156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658188379661265e-06,
+      "loss": 0.6944,
+      "step": 10157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656513490158592e-06,
+      "loss": 0.7222,
+      "step": 10158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.654838610302922e-06,
+      "loss": 0.7675,
+      "step": 10159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.653163740141285e-06,
+      "loss": 0.6606,
+      "step": 10160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651488879720724e-06,
+      "loss": 0.6793,
+      "step": 10161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649814029088282e-06,
+      "loss": 0.6765,
+      "step": 10162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.648139188290998e-06,
+      "loss": 0.6834,
+      "step": 10163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646464357375906e-06,
+      "loss": 0.6705,
+      "step": 10164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644789536390052e-06,
+      "loss": 0.6923,
+      "step": 10165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.643114725380467e-06,
+      "loss": 0.7332,
+      "step": 10166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641439924394191e-06,
+      "loss": 0.7124,
+      "step": 10167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.639765133478262e-06,
+      "loss": 0.6644,
+      "step": 10168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63809035267972e-06,
+      "loss": 0.641,
+      "step": 10169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636415582045597e-06,
+      "loss": 0.7659,
+      "step": 10170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634740821622935e-06,
+      "loss": 0.6993,
+      "step": 10171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633066071458769e-06,
+      "loss": 0.6618,
+      "step": 10172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.631391331600132e-06,
+      "loss": 0.7158,
+      "step": 10173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629716602094062e-06,
+      "loss": 0.7039,
+      "step": 10174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628041882987597e-06,
+      "loss": 0.7396,
+      "step": 10175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626367174327769e-06,
+      "loss": 0.6695,
+      "step": 10176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.624692476161612e-06,
+      "loss": 0.3214,
+      "step": 10177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62301778853617e-06,
+      "loss": 0.6694,
+      "step": 10178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.621343111498465e-06,
+      "loss": 0.6653,
+      "step": 10179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619668445095539e-06,
+      "loss": 0.6952,
+      "step": 10180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61799378937442e-06,
+      "loss": 0.6802,
+      "step": 10181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61631914438215e-06,
+      "loss": 0.695,
+      "step": 10182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614644510165755e-06,
+      "loss": 0.6771,
+      "step": 10183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612969886772272e-06,
+      "loss": 0.7385,
+      "step": 10184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611295274248735e-06,
+      "loss": 0.6146,
+      "step": 10185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609620672642167e-06,
+      "loss": 0.6557,
+      "step": 10186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607946081999609e-06,
+      "loss": 0.6874,
+      "step": 10187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606271502368093e-06,
+      "loss": 0.7526,
+      "step": 10188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.604596933794646e-06,
+      "loss": 0.6983,
+      "step": 10189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6029223763263e-06,
+      "loss": 0.6935,
+      "step": 10190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60124783001009e-06,
+      "loss": 0.6494,
+      "step": 10191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599573294893042e-06,
+      "loss": 0.6479,
+      "step": 10192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597898771022185e-06,
+      "loss": 0.6562,
+      "step": 10193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596224258444554e-06,
+      "loss": 0.6314,
+      "step": 10194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594549757207175e-06,
+      "loss": 0.7016,
+      "step": 10195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592875267357077e-06,
+      "loss": 0.693,
+      "step": 10196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59120078894129e-06,
+      "loss": 0.6846,
+      "step": 10197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589526322006849e-06,
+      "loss": 0.6902,
+      "step": 10198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.587851866600771e-06,
+      "loss": 0.7542,
+      "step": 10199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586177422770089e-06,
+      "loss": 0.7236,
+      "step": 10200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.584502990561831e-06,
+      "loss": 0.7216,
+      "step": 10201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582828570023022e-06,
+      "loss": 0.6039,
+      "step": 10202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581154161200692e-06,
+      "loss": 0.6899,
+      "step": 10203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57947976414187e-06,
+      "loss": 0.6425,
+      "step": 10204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577805378893577e-06,
+      "loss": 0.6719,
+      "step": 10205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576131005502837e-06,
+      "loss": 0.6339,
+      "step": 10206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574456644016683e-06,
+      "loss": 0.6932,
+      "step": 10207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572782294482136e-06,
+      "loss": 0.6237,
+      "step": 10208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571107956946222e-06,
+      "loss": 0.7206,
+      "step": 10209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569433631455965e-06,
+      "loss": 0.6767,
+      "step": 10210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567759318058391e-06,
+      "loss": 0.7301,
+      "step": 10211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566085016800523e-06,
+      "loss": 0.608,
+      "step": 10212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.564410727729383e-06,
+      "loss": 0.7642,
+      "step": 10213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562736450891995e-06,
+      "loss": 0.7399,
+      "step": 10214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561062186335383e-06,
+      "loss": 0.6965,
+      "step": 10215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55938793410657e-06,
+      "loss": 0.6721,
+      "step": 10216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55771369425258e-06,
+      "loss": 0.7463,
+      "step": 10217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.556039466820428e-06,
+      "loss": 0.7139,
+      "step": 10218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554365251857141e-06,
+      "loss": 0.6885,
+      "step": 10219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552691049409739e-06,
+      "loss": 0.7316,
+      "step": 10220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551016859525243e-06,
+      "loss": 0.6955,
+      "step": 10221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54934268225067e-06,
+      "loss": 0.7074,
+      "step": 10222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547668517633048e-06,
+      "loss": 0.6852,
+      "step": 10223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545994365719396e-06,
+      "loss": 0.7209,
+      "step": 10224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.544320226556725e-06,
+      "loss": 0.7052,
+      "step": 10225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542646100192056e-06,
+      "loss": 0.6367,
+      "step": 10226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540971986672413e-06,
+      "loss": 0.6863,
+      "step": 10227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539297886044812e-06,
+      "loss": 0.735,
+      "step": 10228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537623798356272e-06,
+      "loss": 0.6988,
+      "step": 10229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.535949723653812e-06,
+      "loss": 0.7024,
+      "step": 10230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534275661984444e-06,
+      "loss": 0.3415,
+      "step": 10231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532601613395187e-06,
+      "loss": 0.731,
+      "step": 10232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530927577933058e-06,
+      "loss": 0.6699,
+      "step": 10233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529253555645074e-06,
+      "loss": 0.7189,
+      "step": 10234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.527579546578251e-06,
+      "loss": 0.7188,
+      "step": 10235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525905550779601e-06,
+      "loss": 0.7021,
+      "step": 10236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524231568296147e-06,
+      "loss": 0.7131,
+      "step": 10237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522557599174895e-06,
+      "loss": 0.6511,
+      "step": 10238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520883643462862e-06,
+      "loss": 0.6617,
+      "step": 10239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.519209701207062e-06,
+      "loss": 0.7094,
+      "step": 10240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51753577245451e-06,
+      "loss": 0.712,
+      "step": 10241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515861857252218e-06,
+      "loss": 0.6413,
+      "step": 10242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.514187955647203e-06,
+      "loss": 0.6904,
+      "step": 10243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51251406768647e-06,
+      "loss": 0.7163,
+      "step": 10244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.510840193417034e-06,
+      "loss": 0.7151,
+      "step": 10245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509166332885906e-06,
+      "loss": 0.6296,
+      "step": 10246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5074924861401e-06,
+      "loss": 0.7117,
+      "step": 10247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505818653226626e-06,
+      "loss": 0.6814,
+      "step": 10248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504144834192492e-06,
+      "loss": 0.7288,
+      "step": 10249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.502471029084714e-06,
+      "loss": 0.6578,
+      "step": 10250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500797237950292e-06,
+      "loss": 0.7115,
+      "step": 10251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499123460836242e-06,
+      "loss": 0.6052,
+      "step": 10252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497449697789573e-06,
+      "loss": 0.7053,
+      "step": 10253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495775948857292e-06,
+      "loss": 0.3325,
+      "step": 10254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494102214086407e-06,
+      "loss": 0.674,
+      "step": 10255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492428493523927e-06,
+      "loss": 0.6556,
+      "step": 10256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490754787216858e-06,
+      "loss": 0.7811,
+      "step": 10257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489081095212207e-06,
+      "loss": 0.6733,
+      "step": 10258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48740741755698e-06,
+      "loss": 0.6246,
+      "step": 10259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485733754298184e-06,
+      "loss": 0.3332,
+      "step": 10260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484060105482824e-06,
+      "loss": 0.7975,
+      "step": 10261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482386471157905e-06,
+      "loss": 0.7113,
+      "step": 10262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480712851370439e-06,
+      "loss": 0.7701,
+      "step": 10263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.479039246167419e-06,
+      "loss": 0.7083,
+      "step": 10264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477365655595855e-06,
+      "loss": 0.6935,
+      "step": 10265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47569207970275e-06,
+      "loss": 0.6555,
+      "step": 10266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474018518535105e-06,
+      "loss": 0.6829,
+      "step": 10267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472344972139928e-06,
+      "loss": 0.7048,
+      "step": 10268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.470671440564223e-06,
+      "loss": 0.7421,
+      "step": 10269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.468997923854982e-06,
+      "loss": 0.6866,
+      "step": 10270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46732442205921e-06,
+      "loss": 0.7177,
+      "step": 10271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465650935223913e-06,
+      "loss": 0.7281,
+      "step": 10272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463977463396089e-06,
+      "loss": 0.7275,
+      "step": 10273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462304006622737e-06,
+      "loss": 0.7044,
+      "step": 10274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460630564950861e-06,
+      "loss": 0.6218,
+      "step": 10275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458957138427458e-06,
+      "loss": 0.7273,
+      "step": 10276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457283727099525e-06,
+      "loss": 0.6131,
+      "step": 10277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.455610331014064e-06,
+      "loss": 0.7049,
+      "step": 10278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45393695021807e-06,
+      "loss": 0.694,
+      "step": 10279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452263584758543e-06,
+      "loss": 0.7294,
+      "step": 10280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.45059023468248e-06,
+      "loss": 0.6449,
+      "step": 10281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44891690003688e-06,
+      "loss": 0.646,
+      "step": 10282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447243580868735e-06,
+      "loss": 0.6573,
+      "step": 10283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445570277225042e-06,
+      "loss": 0.7054,
+      "step": 10284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.443896989152799e-06,
+      "loss": 0.7855,
+      "step": 10285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442223716699e-06,
+      "loss": 0.691,
+      "step": 10286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440550459910637e-06,
+      "loss": 0.637,
+      "step": 10287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43887721883471e-06,
+      "loss": 0.6869,
+      "step": 10288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437203993518214e-06,
+      "loss": 0.6083,
+      "step": 10289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435530784008133e-06,
+      "loss": 0.6635,
+      "step": 10290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433857590351465e-06,
+      "loss": 0.7432,
+      "step": 10291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432184412595204e-06,
+      "loss": 0.7049,
+      "step": 10292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430511250786341e-06,
+      "loss": 0.7298,
+      "step": 10293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428838104971867e-06,
+      "loss": 0.6403,
+      "step": 10294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427164975198776e-06,
+      "loss": 0.6921,
+      "step": 10295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425491861514055e-06,
+      "loss": 0.6982,
+      "step": 10296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.423818763964694e-06,
+      "loss": 0.7784,
+      "step": 10297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422145682597685e-06,
+      "loss": 0.698,
+      "step": 10298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420472617460018e-06,
+      "loss": 0.6949,
+      "step": 10299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41879956859868e-06,
+      "loss": 0.6488,
+      "step": 10300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41712653606066e-06,
+      "loss": 0.7101,
+      "step": 10301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41545351989295e-06,
+      "loss": 0.6819,
+      "step": 10302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41378052014253e-06,
+      "loss": 0.665,
+      "step": 10303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41210753685639e-06,
+      "loss": 0.6821,
+      "step": 10304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410434570081519e-06,
+      "loss": 0.72,
+      "step": 10305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408761619864901e-06,
+      "loss": 0.6499,
+      "step": 10306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40708868625352e-06,
+      "loss": 0.6282,
+      "step": 10307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40541576929437e-06,
+      "loss": 0.7315,
+      "step": 10308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.403742869034424e-06,
+      "loss": 0.6597,
+      "step": 10309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402069985520674e-06,
+      "loss": 0.6689,
+      "step": 10310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.400397118800097e-06,
+      "loss": 0.745,
+      "step": 10311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398724268919684e-06,
+      "loss": 0.7553,
+      "step": 10312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397051435926414e-06,
+      "loss": 0.6677,
+      "step": 10313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395378619867269e-06,
+      "loss": 0.697,
+      "step": 10314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393705820789237e-06,
+      "loss": 0.7316,
+      "step": 10315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392033038739288e-06,
+      "loss": 0.7649,
+      "step": 10316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.390360273764411e-06,
+      "loss": 0.7545,
+      "step": 10317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388687525911584e-06,
+      "loss": 0.7201,
+      "step": 10318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.387014795227789e-06,
+      "loss": 0.7363,
+      "step": 10319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385342081760004e-06,
+      "loss": 0.6937,
+      "step": 10320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38366938555521e-06,
+      "loss": 0.7209,
+      "step": 10321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38199670666038e-06,
+      "loss": 0.6846,
+      "step": 10322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.3803240451225e-06,
+      "loss": 0.6412,
+      "step": 10323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37865140098854e-06,
+      "loss": 0.7809,
+      "step": 10324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.376978774305482e-06,
+      "loss": 0.6905,
+      "step": 10325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.3753061651203e-06,
+      "loss": 0.7215,
+      "step": 10326
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373633573479972e-06,
+      "loss": 0.7296,
+      "step": 10327
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.371960999431476e-06,
+      "loss": 0.7385,
+      "step": 10328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37028844302178e-06,
+      "loss": 0.65,
+      "step": 10329
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368615904297861e-06,
+      "loss": 0.692,
+      "step": 10330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.366943383306697e-06,
+      "loss": 0.7195,
+      "step": 10331
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365270880095258e-06,
+      "loss": 0.7173,
+      "step": 10332
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.363598394710519e-06,
+      "loss": 0.6654,
+      "step": 10333
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.361925927199458e-06,
+      "loss": 0.6744,
+      "step": 10334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360253477609035e-06,
+      "loss": 0.7452,
+      "step": 10335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358581045986227e-06,
+      "loss": 0.7151,
+      "step": 10336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356908632378007e-06,
+      "loss": 0.6851,
+      "step": 10337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.355236236831345e-06,
+      "loss": 0.6743,
+      "step": 10338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.353563859393211e-06,
+      "loss": 0.7431,
+      "step": 10339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.351891500110575e-06,
+      "loss": 0.6735,
+      "step": 10340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350219159030405e-06,
+      "loss": 0.7081,
+      "step": 10341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.348546836199668e-06,
+      "loss": 0.6852,
+      "step": 10342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.346874531665336e-06,
+      "loss": 0.7015,
+      "step": 10343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345202245474373e-06,
+      "loss": 0.7834,
+      "step": 10344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343529977673748e-06,
+      "loss": 0.6942,
+      "step": 10345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.341857728310428e-06,
+      "loss": 0.6597,
+      "step": 10346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34018549743138e-06,
+      "loss": 0.6938,
+      "step": 10347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338513285083564e-06,
+      "loss": 0.6569,
+      "step": 10348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336841091313949e-06,
+      "loss": 0.6428,
+      "step": 10349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.3351689161695e-06,
+      "loss": 0.6827,
+      "step": 10350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33349675969718e-06,
+      "loss": 0.6766,
+      "step": 10351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331824621943948e-06,
+      "loss": 0.3274,
+      "step": 10352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330152502956774e-06,
+      "loss": 0.7178,
+      "step": 10353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.328480402782622e-06,
+      "loss": 0.7576,
+      "step": 10354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326808321468445e-06,
+      "loss": 0.7217,
+      "step": 10355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325136259061207e-06,
+      "loss": 0.6758,
+      "step": 10356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32346421560787e-06,
+      "loss": 0.6528,
+      "step": 10357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321792191155398e-06,
+      "loss": 0.6977,
+      "step": 10358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320120185750745e-06,
+      "loss": 0.7382,
+      "step": 10359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318448199440877e-06,
+      "loss": 0.7143,
+      "step": 10360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.316776232272743e-06,
+      "loss": 0.6929,
+      "step": 10361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.315104284293306e-06,
+      "loss": 0.7408,
+      "step": 10362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313432355549524e-06,
+      "loss": 0.6659,
+      "step": 10363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311760446088353e-06,
+      "loss": 0.729,
+      "step": 10364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310088555956751e-06,
+      "loss": 0.718,
+      "step": 10365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308416685201673e-06,
+      "loss": 0.6554,
+      "step": 10366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306744833870076e-06,
+      "loss": 0.7114,
+      "step": 10367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.305073002008911e-06,
+      "loss": 0.7136,
+      "step": 10368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303401189665135e-06,
+      "loss": 0.7099,
+      "step": 10369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301729396885699e-06,
+      "loss": 0.7158,
+      "step": 10370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300057623717559e-06,
+      "loss": 0.6504,
+      "step": 10371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298385870207667e-06,
+      "loss": 0.6885,
+      "step": 10372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29671413640298e-06,
+      "loss": 0.7379,
+      "step": 10373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29504242235044e-06,
+      "loss": 0.7045,
+      "step": 10374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293370728097002e-06,
+      "loss": 0.712,
+      "step": 10375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291699053689615e-06,
+      "loss": 0.7504,
+      "step": 10376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290027399175232e-06,
+      "loss": 0.7008,
+      "step": 10377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.288355764600801e-06,
+      "loss": 0.7205,
+      "step": 10378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286684150013272e-06,
+      "loss": 0.7201,
+      "step": 10379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285012555459595e-06,
+      "loss": 0.6811,
+      "step": 10380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28334098098671e-06,
+      "loss": 0.6624,
+      "step": 10381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28166942664157e-06,
+      "loss": 0.6811,
+      "step": 10382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279997892471118e-06,
+      "loss": 0.6668,
+      "step": 10383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.278326378522302e-06,
+      "loss": 0.6306,
+      "step": 10384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.276654884842068e-06,
+      "loss": 0.634,
+      "step": 10385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274983411477363e-06,
+      "loss": 0.7326,
+      "step": 10386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273311958475126e-06,
+      "loss": 0.6762,
+      "step": 10387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271640525882303e-06,
+      "loss": 0.7331,
+      "step": 10388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.269969113745836e-06,
+      "loss": 0.7036,
+      "step": 10389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26829772211267e-06,
+      "loss": 0.7609,
+      "step": 10390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266626351029743e-06,
+      "loss": 0.6658,
+      "step": 10391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264955000544e-06,
+      "loss": 0.6726,
+      "step": 10392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263283670702382e-06,
+      "loss": 0.6768,
+      "step": 10393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261612361551827e-06,
+      "loss": 0.7238,
+      "step": 10394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259941073139272e-06,
+      "loss": 0.7252,
+      "step": 10395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25826980551166e-06,
+      "loss": 0.6786,
+      "step": 10396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25659855871593e-06,
+      "loss": 0.6894,
+      "step": 10397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.254927332799017e-06,
+      "loss": 0.6359,
+      "step": 10398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253256127807866e-06,
+      "loss": 0.623,
+      "step": 10399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.251584943789403e-06,
+      "loss": 0.6711,
+      "step": 10400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249913780790564e-06,
+      "loss": 0.6646,
+      "step": 10401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248242638858294e-06,
+      "loss": 0.6767,
+      "step": 10402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24657151803952e-06,
+      "loss": 0.7726,
+      "step": 10403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.24490041838118e-06,
+      "loss": 0.7208,
+      "step": 10404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243229339930208e-06,
+      "loss": 0.724,
+      "step": 10405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.241558282733538e-06,
+      "loss": 0.618,
+      "step": 10406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239887246838097e-06,
+      "loss": 0.6531,
+      "step": 10407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238216232290821e-06,
+      "loss": 0.5953,
+      "step": 10408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236545239138643e-06,
+      "loss": 0.6385,
+      "step": 10409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.234874267428489e-06,
+      "loss": 0.6594,
+      "step": 10410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233203317207291e-06,
+      "loss": 0.6819,
+      "step": 10411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231532388521982e-06,
+      "loss": 0.7183,
+      "step": 10412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229861481419486e-06,
+      "loss": 0.6857,
+      "step": 10413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228190595946734e-06,
+      "loss": 0.7448,
+      "step": 10414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22651973215065e-06,
+      "loss": 0.6642,
+      "step": 10415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.224848890078167e-06,
+      "loss": 0.7074,
+      "step": 10416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223178069776206e-06,
+      "loss": 0.6316,
+      "step": 10417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221507271291696e-06,
+      "loss": 0.6866,
+      "step": 10418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219836494671567e-06,
+      "loss": 0.6959,
+      "step": 10419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218165739962733e-06,
+      "loss": 0.6899,
+      "step": 10420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216495007212122e-06,
+      "loss": 0.7656,
+      "step": 10421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21482429646666e-06,
+      "loss": 0.6671,
+      "step": 10422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21315360777327e-06,
+      "loss": 0.6692,
+      "step": 10423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.211482941178873e-06,
+      "loss": 0.6692,
+      "step": 10424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209812296730392e-06,
+      "loss": 0.6493,
+      "step": 10425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208141674474741e-06,
+      "loss": 0.7125,
+      "step": 10426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206471074458848e-06,
+      "loss": 0.7218,
+      "step": 10427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204800496729628e-06,
+      "loss": 0.6966,
+      "step": 10428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203129941334004e-06,
+      "loss": 0.6976,
+      "step": 10429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201459408318891e-06,
+      "loss": 0.7242,
+      "step": 10430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19978889773121e-06,
+      "loss": 0.6135,
+      "step": 10431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19811840961788e-06,
+      "loss": 0.6681,
+      "step": 10432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196447944025808e-06,
+      "loss": 0.7009,
+      "step": 10433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194777501001919e-06,
+      "loss": 0.6524,
+      "step": 10434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193107080593123e-06,
+      "loss": 0.7255,
+      "step": 10435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.191436682846337e-06,
+      "loss": 0.6449,
+      "step": 10436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.189766307808475e-06,
+      "loss": 0.6868,
+      "step": 10437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188095955526456e-06,
+      "loss": 0.7395,
+      "step": 10438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18642562604718e-06,
+      "loss": 0.7465,
+      "step": 10439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184755319417569e-06,
+      "loss": 0.7226,
+      "step": 10440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18308503568453e-06,
+      "loss": 0.627,
+      "step": 10441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181414774894973e-06,
+      "loss": 0.69,
+      "step": 10442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179744537095811e-06,
+      "loss": 0.6895,
+      "step": 10443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178074322333955e-06,
+      "loss": 0.6859,
+      "step": 10444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176404130656316e-06,
+      "loss": 0.7304,
+      "step": 10445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17473396210979e-06,
+      "loss": 0.6484,
+      "step": 10446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173063816741296e-06,
+      "loss": 0.6931,
+      "step": 10447
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171393694597738e-06,
+      "loss": 0.703,
+      "step": 10448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16972359572602e-06,
+      "loss": 0.623,
+      "step": 10449
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16805352017305e-06,
+      "loss": 0.6924,
+      "step": 10450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166383467985735e-06,
+      "loss": 0.7291,
+      "step": 10451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164713439210972e-06,
+      "loss": 0.6993,
+      "step": 10452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163043433895671e-06,
+      "loss": 0.6768,
+      "step": 10453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161373452086732e-06,
+      "loss": 0.7404,
+      "step": 10454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15970349383106e-06,
+      "loss": 0.6157,
+      "step": 10455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158033559175554e-06,
+      "loss": 0.6665,
+      "step": 10456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156363648167116e-06,
+      "loss": 0.6984,
+      "step": 10457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154693760852649e-06,
+      "loss": 0.6417,
+      "step": 10458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153023897279048e-06,
+      "loss": 0.713,
+      "step": 10459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.151354057493213e-06,
+      "loss": 0.6924,
+      "step": 10460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.149684241542044e-06,
+      "loss": 0.7466,
+      "step": 10461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148014449472436e-06,
+      "loss": 0.6512,
+      "step": 10462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14634468133129e-06,
+      "loss": 0.7526,
+      "step": 10463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144674937165503e-06,
+      "loss": 0.6963,
+      "step": 10464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.143005217021965e-06,
+      "loss": 0.6794,
+      "step": 10465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.141335520947573e-06,
+      "loss": 0.7274,
+      "step": 10466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139665848989222e-06,
+      "loss": 0.7075,
+      "step": 10467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137996201193807e-06,
+      "loss": 0.6622,
+      "step": 10468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136326577608218e-06,
+      "loss": 0.749,
+      "step": 10469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134656978279351e-06,
+      "loss": 0.7429,
+      "step": 10470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132987403254097e-06,
+      "loss": 0.7189,
+      "step": 10471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.131317852579344e-06,
+      "loss": 0.7209,
+      "step": 10472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129648326301983e-06,
+      "loss": 0.6653,
+      "step": 10473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127978824468904e-06,
+      "loss": 0.7275,
+      "step": 10474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126309347126996e-06,
+      "loss": 0.7173,
+      "step": 10475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.124639894323148e-06,
+      "loss": 0.6654,
+      "step": 10476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122970466104247e-06,
+      "loss": 0.6989,
+      "step": 10477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121301062517179e-06,
+      "loss": 0.7005,
+      "step": 10478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119631683608828e-06,
+      "loss": 0.7365,
+      "step": 10479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117962329426084e-06,
+      "loss": 0.6015,
+      "step": 10480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11629300001583e-06,
+      "loss": 0.6541,
+      "step": 10481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.114623695424945e-06,
+      "loss": 0.7045,
+      "step": 10482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112954415700321e-06,
+      "loss": 0.6515,
+      "step": 10483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.111285160888839e-06,
+      "loss": 0.7014,
+      "step": 10484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109615931037375e-06,
+      "loss": 0.6911,
+      "step": 10485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10794672619281e-06,
+      "loss": 0.6933,
+      "step": 10486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106277546402032e-06,
+      "loss": 0.7252,
+      "step": 10487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104608391711916e-06,
+      "loss": 0.6488,
+      "step": 10488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10293926216934e-06,
+      "loss": 0.7197,
+      "step": 10489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10127015782119e-06,
+      "loss": 0.6862,
+      "step": 10490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.099601078714333e-06,
+      "loss": 0.7258,
+      "step": 10491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09793202489565e-06,
+      "loss": 0.7402,
+      "step": 10492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09626299641202e-06,
+      "loss": 0.6672,
+      "step": 10493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094593993310316e-06,
+      "loss": 0.731,
+      "step": 10494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.092925015637413e-06,
+      "loss": 0.705,
+      "step": 10495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091256063440186e-06,
+      "loss": 0.7166,
+      "step": 10496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08958713676551e-06,
+      "loss": 0.7455,
+      "step": 10497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087918235660254e-06,
+      "loss": 0.7386,
+      "step": 10498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086249360171291e-06,
+      "loss": 0.6636,
+      "step": 10499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084580510345494e-06,
+      "loss": 0.7063,
+      "step": 10500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08291168622973e-06,
+      "loss": 0.7362,
+      "step": 10501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081242887870872e-06,
+      "loss": 0.669,
+      "step": 10502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079574115315786e-06,
+      "loss": 0.7476,
+      "step": 10503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.077905368611351e-06,
+      "loss": 0.7384,
+      "step": 10504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07623664780442e-06,
+      "loss": 0.8003,
+      "step": 10505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074567952941867e-06,
+      "loss": 0.7052,
+      "step": 10506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072899284070554e-06,
+      "loss": 0.3041,
+      "step": 10507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.071230641237352e-06,
+      "loss": 0.7128,
+      "step": 10508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069562024489123e-06,
+      "loss": 0.6369,
+      "step": 10509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067893433872738e-06,
+      "loss": 0.6797,
+      "step": 10510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.066224869435045e-06,
+      "loss": 0.6335,
+      "step": 10511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.064556331222918e-06,
+      "loss": 0.6195,
+      "step": 10512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062887819283217e-06,
+      "loss": 0.7321,
+      "step": 10513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061219333662798e-06,
+      "loss": 0.732,
+      "step": 10514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059550874408529e-06,
+      "loss": 0.7309,
+      "step": 10515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.057882441567265e-06,
+      "loss": 0.685,
+      "step": 10516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056214035185867e-06,
+      "loss": 0.6884,
+      "step": 10517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05454565531119e-06,
+      "loss": 0.7134,
+      "step": 10518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052877301990093e-06,
+      "loss": 0.7617,
+      "step": 10519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051208975269432e-06,
+      "loss": 0.712,
+      "step": 10520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049540675196064e-06,
+      "loss": 0.7274,
+      "step": 10521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.047872401816842e-06,
+      "loss": 0.697,
+      "step": 10522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046204155178626e-06,
+      "loss": 0.6897,
+      "step": 10523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044535935328262e-06,
+      "loss": 0.7325,
+      "step": 10524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042867742312605e-06,
+      "loss": 0.7209,
+      "step": 10525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04119957617851e-06,
+      "loss": 0.6582,
+      "step": 10526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.039531436972824e-06,
+      "loss": 0.6829,
+      "step": 10527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037863324742402e-06,
+      "loss": 0.6823,
+      "step": 10528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036195239534092e-06,
+      "loss": 0.7153,
+      "step": 10529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.034527181394747e-06,
+      "loss": 0.7033,
+      "step": 10530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032859150371205e-06,
+      "loss": 0.7186,
+      "step": 10531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.031191146510325e-06,
+      "loss": 0.6938,
+      "step": 10532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.029523169858944e-06,
+      "loss": 0.6843,
+      "step": 10533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027855220463915e-06,
+      "loss": 0.6962,
+      "step": 10534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02618729837208e-06,
+      "loss": 0.6776,
+      "step": 10535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024519403630286e-06,
+      "loss": 0.7349,
+      "step": 10536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022851536285374e-06,
+      "loss": 0.7066,
+      "step": 10537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021183696384186e-06,
+      "loss": 0.6733,
+      "step": 10538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.019515883973569e-06,
+      "loss": 0.6322,
+      "step": 10539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01784809910036e-06,
+      "loss": 0.6073,
+      "step": 10540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.0161803418114e-06,
+      "loss": 0.6345,
+      "step": 10541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01451261215353e-06,
+      "loss": 0.7165,
+      "step": 10542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012844910173592e-06,
+      "loss": 0.7316,
+      "step": 10543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011177235918418e-06,
+      "loss": 0.7911,
+      "step": 10544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00950958943485e-06,
+      "loss": 0.6983,
+      "step": 10545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.007841970769722e-06,
+      "loss": 0.6868,
+      "step": 10546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00617437996987e-06,
+      "loss": 0.6639,
+      "step": 10547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.004506817082133e-06,
+      "loss": 0.6016,
+      "step": 10548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002839282153346e-06,
+      "loss": 0.6858,
+      "step": 10549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.001171775230335e-06,
+      "loss": 0.6805,
+      "step": 10550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99950429635994e-06,
+      "loss": 0.7628,
+      "step": 10551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997836845588987e-06,
+      "loss": 0.7128,
+      "step": 10552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996169422964312e-06,
+      "loss": 0.7338,
+      "step": 10553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994502028532744e-06,
+      "loss": 0.6595,
+      "step": 10554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992834662341112e-06,
+      "loss": 0.721,
+      "step": 10555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991167324436249e-06,
+      "loss": 0.7006,
+      "step": 10556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989500014864975e-06,
+      "loss": 0.6745,
+      "step": 10557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987832733674125e-06,
+      "loss": 0.6854,
+      "step": 10558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98616548091052e-06,
+      "loss": 0.7042,
+      "step": 10559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98449825662099e-06,
+      "loss": 0.7111,
+      "step": 10560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982831060852356e-06,
+      "loss": 0.6282,
+      "step": 10561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981163893651447e-06,
+      "loss": 0.7526,
+      "step": 10562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97949675506508e-06,
+      "loss": 0.633,
+      "step": 10563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.977829645140081e-06,
+      "loss": 0.3494,
+      "step": 10564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976162563923272e-06,
+      "loss": 0.6923,
+      "step": 10565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.974495511461472e-06,
+      "loss": 0.7262,
+      "step": 10566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972828487801502e-06,
+      "loss": 0.7194,
+      "step": 10567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97116149299018e-06,
+      "loss": 0.7105,
+      "step": 10568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969494527074333e-06,
+      "loss": 0.3704,
+      "step": 10569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.967827590100766e-06,
+      "loss": 0.7015,
+      "step": 10570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966160682116301e-06,
+      "loss": 0.7279,
+      "step": 10571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964493803167751e-06,
+      "loss": 0.6703,
+      "step": 10572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962826953301939e-06,
+      "loss": 0.7173,
+      "step": 10573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961160132565675e-06,
+      "loss": 0.6679,
+      "step": 10574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.959493341005773e-06,
+      "loss": 0.7206,
+      "step": 10575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95782657866904e-06,
+      "loss": 0.6821,
+      "step": 10576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956159845602297e-06,
+      "loss": 0.651,
+      "step": 10577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954493141852348e-06,
+      "loss": 0.6199,
+      "step": 10578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952826467466008e-06,
+      "loss": 0.6893,
+      "step": 10579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951159822490084e-06,
+      "loss": 0.6384,
+      "step": 10580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949493206971386e-06,
+      "loss": 0.6786,
+      "step": 10581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947826620956722e-06,
+      "loss": 0.6597,
+      "step": 10582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.946160064492895e-06,
+      "loss": 0.6672,
+      "step": 10583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.944493537626716e-06,
+      "loss": 0.6682,
+      "step": 10584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.942827040404987e-06,
+      "loss": 0.6662,
+      "step": 10585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941160572874514e-06,
+      "loss": 0.7001,
+      "step": 10586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9394941350821e-06,
+      "loss": 0.6327,
+      "step": 10587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93782772707455e-06,
+      "loss": 0.6686,
+      "step": 10588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936161348898664e-06,
+      "loss": 0.6977,
+      "step": 10589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934495000601241e-06,
+      "loss": 0.6766,
+      "step": 10590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.932828682229083e-06,
+      "loss": 0.6772,
+      "step": 10591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93116239382899e-06,
+      "loss": 0.7037,
+      "step": 10592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92949613544776e-06,
+      "loss": 0.6443,
+      "step": 10593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.927829907132194e-06,
+      "loss": 0.7394,
+      "step": 10594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926163708929087e-06,
+      "loss": 0.7354,
+      "step": 10595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924497540885229e-06,
+      "loss": 0.694,
+      "step": 10596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922831403047423e-06,
+      "loss": 0.7132,
+      "step": 10597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92116529546246e-06,
+      "loss": 0.6887,
+      "step": 10598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919499218177136e-06,
+      "loss": 0.6813,
+      "step": 10599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.917833171238239e-06,
+      "loss": 0.594,
+      "step": 10600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916167154692568e-06,
+      "loss": 0.7003,
+      "step": 10601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.914501168586906e-06,
+      "loss": 0.6306,
+      "step": 10602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912835212968047e-06,
+      "loss": 0.6729,
+      "step": 10603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91116928788278e-06,
+      "loss": 0.7085,
+      "step": 10604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.909503393377893e-06,
+      "loss": 0.7017,
+      "step": 10605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.907837529500174e-06,
+      "loss": 0.7098,
+      "step": 10606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90617169629641e-06,
+      "loss": 0.6811,
+      "step": 10607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904505893813387e-06,
+      "loss": 0.666,
+      "step": 10608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902840122097888e-06,
+      "loss": 0.6868,
+      "step": 10609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.901174381196698e-06,
+      "loss": 0.6978,
+      "step": 10610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8995086711566e-06,
+      "loss": 0.6625,
+      "step": 10611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897842992024376e-06,
+      "loss": 0.7305,
+      "step": 10612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89617734384681e-06,
+      "loss": 0.7054,
+      "step": 10613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.894511726670683e-06,
+      "loss": 0.7061,
+      "step": 10614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892846140542771e-06,
+      "loss": 0.7272,
+      "step": 10615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891180585509854e-06,
+      "loss": 0.6891,
+      "step": 10616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889515061618707e-06,
+      "loss": 0.752,
+      "step": 10617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.887849568916112e-06,
+      "loss": 0.6677,
+      "step": 10618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886184107448846e-06,
+      "loss": 0.6576,
+      "step": 10619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88451867726368e-06,
+      "loss": 0.7204,
+      "step": 10620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882853278407394e-06,
+      "loss": 0.6729,
+      "step": 10621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881187910926757e-06,
+      "loss": 0.6724,
+      "step": 10622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.879522574868541e-06,
+      "loss": 0.6214,
+      "step": 10623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87785727027952e-06,
+      "loss": 0.6982,
+      "step": 10624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876191997206466e-06,
+      "loss": 0.7446,
+      "step": 10625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.874526755696147e-06,
+      "loss": 0.7121,
+      "step": 10626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872861545795335e-06,
+      "loss": 0.6547,
+      "step": 10627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871196367550794e-06,
+      "loss": 0.6736,
+      "step": 10628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869531221009294e-06,
+      "loss": 0.6657,
+      "step": 10629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8678661062176e-06,
+      "loss": 0.7775,
+      "step": 10630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86620102322248e-06,
+      "loss": 0.6834,
+      "step": 10631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864535972070698e-06,
+      "loss": 0.7021,
+      "step": 10632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.862870952809015e-06,
+      "loss": 0.6104,
+      "step": 10633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8612059654842e-06,
+      "loss": 0.6297,
+      "step": 10634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85954101014301e-06,
+      "loss": 0.6569,
+      "step": 10635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857876086832207e-06,
+      "loss": 0.6476,
+      "step": 10636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856211195598549e-06,
+      "loss": 0.7311,
+      "step": 10637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854546336488799e-06,
+      "loss": 0.7211,
+      "step": 10638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852881509549716e-06,
+      "loss": 0.6762,
+      "step": 10639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85121671482806e-06,
+      "loss": 0.7042,
+      "step": 10640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.849551952370578e-06,
+      "loss": 0.6933,
+      "step": 10641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847887222224032e-06,
+      "loss": 0.6502,
+      "step": 10642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.846222524435178e-06,
+      "loss": 0.6718,
+      "step": 10643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844557859050767e-06,
+      "loss": 0.3806,
+      "step": 10644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842893226117552e-06,
+      "loss": 0.6321,
+      "step": 10645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841228625682287e-06,
+      "loss": 0.6557,
+      "step": 10646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839564057791725e-06,
+      "loss": 0.6815,
+      "step": 10647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837899522492612e-06,
+      "loss": 0.5889,
+      "step": 10648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836235019831697e-06,
+      "loss": 0.7324,
+      "step": 10649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834570549855732e-06,
+      "loss": 0.7278,
+      "step": 10650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832906112611461e-06,
+      "loss": 0.7325,
+      "step": 10651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831241708145634e-06,
+      "loss": 0.6993,
+      "step": 10652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829577336504997e-06,
+      "loss": 0.6749,
+      "step": 10653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82791299773629e-06,
+      "loss": 0.733,
+      "step": 10654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82624869188626e-06,
+      "loss": 0.6685,
+      "step": 10655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.824584419001652e-06,
+      "loss": 0.6413,
+      "step": 10656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8229201791292e-06,
+      "loss": 0.7145,
+      "step": 10657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821255972315656e-06,
+      "loss": 0.6604,
+      "step": 10658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.819591798607753e-06,
+      "loss": 0.6908,
+      "step": 10659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817927658052237e-06,
+      "loss": 0.6715,
+      "step": 10660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816263550695832e-06,
+      "loss": 0.6604,
+      "step": 10661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.814599476585291e-06,
+      "loss": 0.6902,
+      "step": 10662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81293543576734e-06,
+      "loss": 0.6625,
+      "step": 10663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811271428288722e-06,
+      "loss": 0.7119,
+      "step": 10664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.809607454196165e-06,
+      "loss": 0.7031,
+      "step": 10665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80794351353641e-06,
+      "loss": 0.7098,
+      "step": 10666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80627960635618e-06,
+      "loss": 0.5976,
+      "step": 10667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804615732702213e-06,
+      "loss": 0.7781,
+      "step": 10668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80295189262124e-06,
+      "loss": 0.6139,
+      "step": 10669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.801288086159989e-06,
+      "loss": 0.7559,
+      "step": 10670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799624313365189e-06,
+      "loss": 0.7626,
+      "step": 10671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797960574283568e-06,
+      "loss": 0.7037,
+      "step": 10672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796296868961854e-06,
+      "loss": 0.731,
+      "step": 10673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 0.7606,
+      "step": 10674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.792969559785046e-06,
+      "loss": 0.6711,
+      "step": 10675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791305956023403e-06,
+      "loss": 0.6256,
+      "step": 10676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789642386208561e-06,
+      "loss": 0.6903,
+      "step": 10677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.787978850387248e-06,
+      "loss": 0.7319,
+      "step": 10678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78631534860619e-06,
+      "loss": 0.6675,
+      "step": 10679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784651880912093e-06,
+      "loss": 0.7257,
+      "step": 10680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.782988447351685e-06,
+      "loss": 0.7259,
+      "step": 10681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781325047971682e-06,
+      "loss": 0.6906,
+      "step": 10682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779661682818804e-06,
+      "loss": 0.7928,
+      "step": 10683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777998351939766e-06,
+      "loss": 0.7198,
+      "step": 10684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776335055381284e-06,
+      "loss": 0.7233,
+      "step": 10685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.774671793190075e-06,
+      "loss": 0.7252,
+      "step": 10686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773008565412847e-06,
+      "loss": 0.6604,
+      "step": 10687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771345372096315e-06,
+      "loss": 0.7057,
+      "step": 10688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769682213287191e-06,
+      "loss": 0.6592,
+      "step": 10689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768019089032185e-06,
+      "loss": 0.7115,
+      "step": 10690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76635599937801e-06,
+      "loss": 0.6782,
+      "step": 10691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.764692944371372e-06,
+      "loss": 0.6994,
+      "step": 10692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76302992405898e-06,
+      "loss": 0.7203,
+      "step": 10693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761366938487536e-06,
+      "loss": 0.7206,
+      "step": 10694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759703987703751e-06,
+      "loss": 0.6929,
+      "step": 10695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.758041071754328e-06,
+      "loss": 0.6709,
+      "step": 10696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.756378190685973e-06,
+      "loss": 0.7181,
+      "step": 10697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754715344545383e-06,
+      "loss": 0.6464,
+      "step": 10698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753052533379272e-06,
+      "loss": 0.6787,
+      "step": 10699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751389757234328e-06,
+      "loss": 0.7085,
+      "step": 10700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749727016157256e-06,
+      "loss": 0.7167,
+      "step": 10701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.748064310194755e-06,
+      "loss": 0.6721,
+      "step": 10702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.746401639393523e-06,
+      "loss": 0.6869,
+      "step": 10703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744739003800257e-06,
+      "loss": 0.6912,
+      "step": 10704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.743076403461656e-06,
+      "loss": 0.7225,
+      "step": 10705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.741413838424407e-06,
+      "loss": 0.6779,
+      "step": 10706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73975130873521e-06,
+      "loss": 0.7214,
+      "step": 10707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.738088814440758e-06,
+      "loss": 0.6708,
+      "step": 10708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73642635558774e-06,
+      "loss": 0.674,
+      "step": 10709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734763932222852e-06,
+      "loss": 0.7275,
+      "step": 10710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733101544392778e-06,
+      "loss": 0.6088,
+      "step": 10711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731439192144214e-06,
+      "loss": 0.6599,
+      "step": 10712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72977687552384e-06,
+      "loss": 0.6789,
+      "step": 10713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72811459457835e-06,
+      "loss": 0.6564,
+      "step": 10714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.726452349354425e-06,
+      "loss": 0.7252,
+      "step": 10715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724790139898753e-06,
+      "loss": 0.6388,
+      "step": 10716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723127966258015e-06,
+      "loss": 0.6451,
+      "step": 10717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721465828478901e-06,
+      "loss": 0.7133,
+      "step": 10718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719803726608085e-06,
+      "loss": 0.7356,
+      "step": 10719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718141660692252e-06,
+      "loss": 0.6224,
+      "step": 10720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71647963077808e-06,
+      "loss": 0.7102,
+      "step": 10721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.714817636912247e-06,
+      "loss": 0.5805,
+      "step": 10722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.713155679141436e-06,
+      "loss": 0.6999,
+      "step": 10723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711493757512319e-06,
+      "loss": 0.631,
+      "step": 10724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.70983187207158e-06,
+      "loss": 0.7314,
+      "step": 10725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.708170022865885e-06,
+      "loss": 0.7203,
+      "step": 10726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706508209941906e-06,
+      "loss": 0.7443,
+      "step": 10727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704846433346324e-06,
+      "loss": 0.6962,
+      "step": 10728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703184693125808e-06,
+      "loss": 0.6285,
+      "step": 10729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701522989327028e-06,
+      "loss": 0.7595,
+      "step": 10730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699861321996657e-06,
+      "loss": 0.6853,
+      "step": 10731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698199691181356e-06,
+      "loss": 0.7149,
+      "step": 10732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6965380969278e-06,
+      "loss": 0.7823,
+      "step": 10733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694876539282655e-06,
+      "loss": 0.7325,
+      "step": 10734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693215018292584e-06,
+      "loss": 0.7346,
+      "step": 10735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.691553534004254e-06,
+      "loss": 0.7478,
+      "step": 10736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689892086464325e-06,
+      "loss": 0.6247,
+      "step": 10737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688230675719466e-06,
+      "loss": 0.7082,
+      "step": 10738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686569301816333e-06,
+      "loss": 0.7389,
+      "step": 10739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.684907964801587e-06,
+      "loss": 0.752,
+      "step": 10740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68324666472189e-06,
+      "loss": 0.6907,
+      "step": 10741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.681585401623898e-06,
+      "loss": 0.6824,
+      "step": 10742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67992417555427e-06,
+      "loss": 0.6887,
+      "step": 10743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.678262986559667e-06,
+      "loss": 0.7056,
+      "step": 10744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676601834686736e-06,
+      "loss": 0.6367,
+      "step": 10745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.674940719982135e-06,
+      "loss": 0.6129,
+      "step": 10746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673279642492513e-06,
+      "loss": 0.7212,
+      "step": 10747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671618602264531e-06,
+      "loss": 0.6703,
+      "step": 10748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669957599344836e-06,
+      "loss": 0.7293,
+      "step": 10749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.668296633780075e-06,
+      "loss": 0.6833,
+      "step": 10750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666635705616904e-06,
+      "loss": 0.7194,
+      "step": 10751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.664974814901963e-06,
+      "loss": 0.7419,
+      "step": 10752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663313961681905e-06,
+      "loss": 0.7047,
+      "step": 10753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.661653146003373e-06,
+      "loss": 0.7194,
+      "step": 10754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659992367913013e-06,
+      "loss": 0.6782,
+      "step": 10755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658331627457468e-06,
+      "loss": 0.7258,
+      "step": 10756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656670924683382e-06,
+      "loss": 0.701,
+      "step": 10757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.655010259637396e-06,
+      "loss": 0.7017,
+      "step": 10758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65334963236615e-06,
+      "loss": 0.6784,
+      "step": 10759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.651689042916283e-06,
+      "loss": 0.6748,
+      "step": 10760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650028491334436e-06,
+      "loss": 0.6397,
+      "step": 10761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648367977667245e-06,
+      "loss": 0.6774,
+      "step": 10762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.646707501961344e-06,
+      "loss": 0.651,
+      "step": 10763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645047064263376e-06,
+      "loss": 0.6826,
+      "step": 10764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643386664619968e-06,
+      "loss": 0.6937,
+      "step": 10765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641726303077755e-06,
+      "loss": 0.7208,
+      "step": 10766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640065979683367e-06,
+      "loss": 0.7382,
+      "step": 10767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638405694483439e-06,
+      "loss": 0.6546,
+      "step": 10768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.636745447524599e-06,
+      "loss": 0.6998,
+      "step": 10769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.635085238853483e-06,
+      "loss": 0.7097,
+      "step": 10770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633425068516705e-06,
+      "loss": 0.7,
+      "step": 10771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6317649365609e-06,
+      "loss": 0.6585,
+      "step": 10772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.630104843032693e-06,
+      "loss": 0.6522,
+      "step": 10773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628444787978709e-06,
+      "loss": 0.6161,
+      "step": 10774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.626784771445568e-06,
+      "loss": 0.6955,
+      "step": 10775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6251247934799e-06,
+      "loss": 0.7364,
+      "step": 10776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62346485412832e-06,
+      "loss": 0.7831,
+      "step": 10777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62180495343745e-06,
+      "loss": 0.7651,
+      "step": 10778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620145091453907e-06,
+      "loss": 0.6007,
+      "step": 10779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618485268224312e-06,
+      "loss": 0.6061,
+      "step": 10780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61682548379528e-06,
+      "loss": 0.7374,
+      "step": 10781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61516573821343e-06,
+      "loss": 0.7111,
+      "step": 10782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613506031525375e-06,
+      "loss": 0.7081,
+      "step": 10783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611846363777727e-06,
+      "loss": 0.6617,
+      "step": 10784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610186735017102e-06,
+      "loss": 0.6146,
+      "step": 10785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608527145290106e-06,
+      "loss": 0.7326,
+      "step": 10786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606867594643353e-06,
+      "loss": 0.7349,
+      "step": 10787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.605208083123455e-06,
+      "loss": 0.6784,
+      "step": 10788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603548610777018e-06,
+      "loss": 0.6602,
+      "step": 10789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.601889177650652e-06,
+      "loss": 0.6906,
+      "step": 10790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600229783790957e-06,
+      "loss": 0.6727,
+      "step": 10791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59857042924454e-06,
+      "loss": 0.6941,
+      "step": 10792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.596911114058005e-06,
+      "loss": 0.6717,
+      "step": 10793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595251838277955e-06,
+      "loss": 0.6618,
+      "step": 10794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.593592601950996e-06,
+      "loss": 0.3375,
+      "step": 10795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591933405123724e-06,
+      "loss": 0.6611,
+      "step": 10796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.590274247842737e-06,
+      "loss": 0.7015,
+      "step": 10797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588615130154636e-06,
+      "loss": 0.6523,
+      "step": 10798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586956052106017e-06,
+      "loss": 0.6901,
+      "step": 10799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.585297013743476e-06,
+      "loss": 0.6758,
+      "step": 10800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.583638015113611e-06,
+      "loss": 0.6541,
+      "step": 10801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.581979056263011e-06,
+      "loss": 0.6639,
+      "step": 10802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580320137238275e-06,
+      "loss": 0.7159,
+      "step": 10803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578661258085987e-06,
+      "loss": 0.7294,
+      "step": 10804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577002418852742e-06,
+      "loss": 0.6748,
+      "step": 10805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.57534361958513e-06,
+      "loss": 0.6542,
+      "step": 10806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573684860329737e-06,
+      "loss": 0.5929,
+      "step": 10807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.572026141133148e-06,
+      "loss": 0.6685,
+      "step": 10808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570367462041958e-06,
+      "loss": 0.6566,
+      "step": 10809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.568708823102742e-06,
+      "loss": 0.7167,
+      "step": 10810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567050224362088e-06,
+      "loss": 0.691,
+      "step": 10811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565391665866576e-06,
+      "loss": 0.7397,
+      "step": 10812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563733147662792e-06,
+      "loss": 0.6998,
+      "step": 10813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.562074669797313e-06,
+      "loss": 0.6631,
+      "step": 10814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56041623231672e-06,
+      "loss": 0.6634,
+      "step": 10815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558757835267593e-06,
+      "loss": 0.6683,
+      "step": 10816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557099478696503e-06,
+      "loss": 0.666,
+      "step": 10817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55544116265003e-06,
+      "loss": 0.681,
+      "step": 10818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553782887174747e-06,
+      "loss": 0.7274,
+      "step": 10819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55212465231723e-06,
+      "loss": 0.6415,
+      "step": 10820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550466458124049e-06,
+      "loss": 0.7009,
+      "step": 10821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548808304641776e-06,
+      "loss": 0.7009,
+      "step": 10822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547150191916981e-06,
+      "loss": 0.7104,
+      "step": 10823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545492119996233e-06,
+      "loss": 0.6512,
+      "step": 10824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5438340889261e-06,
+      "loss": 0.7093,
+      "step": 10825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542176098753146e-06,
+      "loss": 0.6467,
+      "step": 10826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540518149523943e-06,
+      "loss": 0.6308,
+      "step": 10827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.538860241285046e-06,
+      "loss": 0.6305,
+      "step": 10828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.537202374083032e-06,
+      "loss": 0.6605,
+      "step": 10829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53554454796445e-06,
+      "loss": 0.6793,
+      "step": 10830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533886762975864e-06,
+      "loss": 0.6634,
+      "step": 10831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532229019163835e-06,
+      "loss": 0.6424,
+      "step": 10832
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.530571316574923e-06,
+      "loss": 0.6814,
+      "step": 10833
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528913655255685e-06,
+      "loss": 0.7369,
+      "step": 10834
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527256035252681e-06,
+      "loss": 0.6459,
+      "step": 10835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525598456612454e-06,
+      "loss": 0.643,
+      "step": 10836
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523940919381571e-06,
+      "loss": 0.7877,
+      "step": 10837
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522283423606577e-06,
+      "loss": 0.6501,
+      "step": 10838
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520625969334028e-06,
+      "loss": 0.6158,
+      "step": 10839
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518968556610471e-06,
+      "loss": 0.707,
+      "step": 10840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517311185482457e-06,
+      "loss": 0.7356,
+      "step": 10841
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515653855996538e-06,
+      "loss": 0.7489,
+      "step": 10842
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513996568199255e-06,
+      "loss": 0.6742,
+      "step": 10843
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.512339322137153e-06,
+      "loss": 0.6133,
+      "step": 10844
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510682117856783e-06,
+      "loss": 0.7152,
+      "step": 10845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509024955404682e-06,
+      "loss": 0.7197,
+      "step": 10846
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507367834827396e-06,
+      "loss": 0.6954,
+      "step": 10847
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.505710756171468e-06,
+      "loss": 0.7191,
+      "step": 10848
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504053719483433e-06,
+      "loss": 0.6997,
+      "step": 10849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502396724809834e-06,
+      "loss": 0.6842,
+      "step": 10850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500739772197203e-06,
+      "loss": 0.692,
+      "step": 10851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.499082861692081e-06,
+      "loss": 0.6851,
+      "step": 10852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497425993341002e-06,
+      "loss": 0.7385,
+      "step": 10853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.495769167190502e-06,
+      "loss": 0.6996,
+      "step": 10854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494112383287115e-06,
+      "loss": 0.7639,
+      "step": 10855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.492455641677365e-06,
+      "loss": 0.6879,
+      "step": 10856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.490798942407786e-06,
+      "loss": 0.7064,
+      "step": 10857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48914228552491e-06,
+      "loss": 0.7033,
+      "step": 10858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.487485671075263e-06,
+      "loss": 0.7563,
+      "step": 10859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485829099105373e-06,
+      "loss": 0.721,
+      "step": 10860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484172569661764e-06,
+      "loss": 0.6389,
+      "step": 10861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482516082790965e-06,
+      "loss": 0.7058,
+      "step": 10862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480859638539492e-06,
+      "loss": 0.7048,
+      "step": 10863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479203236953872e-06,
+      "loss": 0.7253,
+      "step": 10864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.477546878080625e-06,
+      "loss": 0.7257,
+      "step": 10865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47589056196627e-06,
+      "loss": 0.668,
+      "step": 10866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474234288657326e-06,
+      "loss": 0.7386,
+      "step": 10867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472578058200314e-06,
+      "loss": 0.7491,
+      "step": 10868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470921870641742e-06,
+      "loss": 0.7212,
+      "step": 10869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46926572602813e-06,
+      "loss": 0.6512,
+      "step": 10870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467609624405992e-06,
+      "loss": 0.7534,
+      "step": 10871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465953565821839e-06,
+      "loss": 0.6592,
+      "step": 10872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464297550322181e-06,
+      "loss": 0.6666,
+      "step": 10873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46264157795353e-06,
+      "loss": 0.7859,
+      "step": 10874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4609856487624e-06,
+      "loss": 0.7416,
+      "step": 10875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45932976279529e-06,
+      "loss": 0.7422,
+      "step": 10876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.457673920098705e-06,
+      "loss": 0.6859,
+      "step": 10877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45601812071916e-06,
+      "loss": 0.595,
+      "step": 10878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454362364703152e-06,
+      "loss": 0.6802,
+      "step": 10879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.452706652097187e-06,
+      "loss": 0.6661,
+      "step": 10880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451050982947767e-06,
+      "loss": 0.6836,
+      "step": 10881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44939535730139e-06,
+      "loss": 0.7133,
+      "step": 10882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.447739775204554e-06,
+      "loss": 0.6888,
+      "step": 10883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.446084236703758e-06,
+      "loss": 0.651,
+      "step": 10884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.444428741845499e-06,
+      "loss": 0.6839,
+      "step": 10885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442773290676274e-06,
+      "loss": 0.6775,
+      "step": 10886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441117883242575e-06,
+      "loss": 0.6794,
+      "step": 10887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439462519590898e-06,
+      "loss": 0.6768,
+      "step": 10888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437807199767731e-06,
+      "loss": 0.6393,
+      "step": 10889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436151923819567e-06,
+      "loss": 0.6508,
+      "step": 10890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434496691792893e-06,
+      "loss": 0.6892,
+      "step": 10891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432841503734199e-06,
+      "loss": 0.7113,
+      "step": 10892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43118635968997e-06,
+      "loss": 0.6801,
+      "step": 10893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4295312597067e-06,
+      "loss": 0.6848,
+      "step": 10894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42787620383086e-06,
+      "loss": 0.7163,
+      "step": 10895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426221192108939e-06,
+      "loss": 0.6858,
+      "step": 10896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424566224587417e-06,
+      "loss": 0.6806,
+      "step": 10897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422911301312781e-06,
+      "loss": 0.7027,
+      "step": 10898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421256422331507e-06,
+      "loss": 0.6118,
+      "step": 10899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419601587690069e-06,
+      "loss": 0.699,
+      "step": 10900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417946797434955e-06,
+      "loss": 0.727,
+      "step": 10901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.416292051612625e-06,
+      "loss": 0.7192,
+      "step": 10902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.414637350269564e-06,
+      "loss": 0.6907,
+      "step": 10903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412982693452241e-06,
+      "loss": 0.7241,
+      "step": 10904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41132808120713e-06,
+      "loss": 0.6353,
+      "step": 10905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.409673513580702e-06,
+      "loss": 0.7226,
+      "step": 10906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408018990619427e-06,
+      "loss": 0.6957,
+      "step": 10907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406364512369769e-06,
+      "loss": 0.605,
+      "step": 10908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404710078878198e-06,
+      "loss": 0.6535,
+      "step": 10909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40305569019118e-06,
+      "loss": 0.6249,
+      "step": 10910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401401346355177e-06,
+      "loss": 0.6811,
+      "step": 10911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399747047416654e-06,
+      "loss": 0.6336,
+      "step": 10912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398092793422072e-06,
+      "loss": 0.7036,
+      "step": 10913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396438584417897e-06,
+      "loss": 0.731,
+      "step": 10914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394784420450578e-06,
+      "loss": 0.6935,
+      "step": 10915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39313030156658e-06,
+      "loss": 0.7507,
+      "step": 10916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.391476227812357e-06,
+      "loss": 0.6279,
+      "step": 10917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.389822199234366e-06,
+      "loss": 0.6611,
+      "step": 10918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.38816821587906e-06,
+      "loss": 0.6491,
+      "step": 10919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3865142777929e-06,
+      "loss": 0.7424,
+      "step": 10920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.384860385022325e-06,
+      "loss": 0.7022,
+      "step": 10921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38320653761379e-06,
+      "loss": 0.7492,
+      "step": 10922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.381552735613745e-06,
+      "loss": 0.6888,
+      "step": 10923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379898979068642e-06,
+      "loss": 0.7255,
+      "step": 10924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37824526802492e-06,
+      "loss": 0.5955,
+      "step": 10925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37659160252903e-06,
+      "loss": 0.7235,
+      "step": 10926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374937982627418e-06,
+      "loss": 0.7085,
+      "step": 10927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.373284408366518e-06,
+      "loss": 0.6981,
+      "step": 10928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371630879792778e-06,
+      "loss": 0.7099,
+      "step": 10929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369977396952636e-06,
+      "loss": 0.6988,
+      "step": 10930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.368323959892532e-06,
+      "loss": 0.7354,
+      "step": 10931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366670568658905e-06,
+      "loss": 0.6038,
+      "step": 10932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36501722329819e-06,
+      "loss": 0.7127,
+      "step": 10933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36336392385682e-06,
+      "loss": 0.7485,
+      "step": 10934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36171067038123e-06,
+      "loss": 0.6996,
+      "step": 10935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.360057462917853e-06,
+      "loss": 0.6691,
+      "step": 10936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358404301513121e-06,
+      "loss": 0.623,
+      "step": 10937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356751186213462e-06,
+      "loss": 0.6521,
+      "step": 10938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355098117065305e-06,
+      "loss": 0.7315,
+      "step": 10939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353445094115084e-06,
+      "loss": 0.6251,
+      "step": 10940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351792117409214e-06,
+      "loss": 0.6633,
+      "step": 10941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.350139186994124e-06,
+      "loss": 0.7723,
+      "step": 10942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34848630291624e-06,
+      "loss": 0.7357,
+      "step": 10943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.346833465221982e-06,
+      "loss": 0.6656,
+      "step": 10944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34518067395777e-06,
+      "loss": 0.6288,
+      "step": 10945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.343527929170028e-06,
+      "loss": 0.6993,
+      "step": 10946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.341875230905168e-06,
+      "loss": 0.7177,
+      "step": 10947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34022257920961e-06,
+      "loss": 0.6045,
+      "step": 10948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338569974129769e-06,
+      "loss": 0.7128,
+      "step": 10949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33691741571206e-06,
+      "loss": 0.7002,
+      "step": 10950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335264904002895e-06,
+      "loss": 0.7208,
+      "step": 10951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333612439048687e-06,
+      "loss": 0.754,
+      "step": 10952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331960020895847e-06,
+      "loss": 0.6825,
+      "step": 10953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.330307649590782e-06,
+      "loss": 0.3343,
+      "step": 10954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328655325179898e-06,
+      "loss": 0.6795,
+      "step": 10955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327003047709605e-06,
+      "loss": 0.631,
+      "step": 10956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325350817226305e-06,
+      "loss": 0.7097,
+      "step": 10957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323698633776403e-06,
+      "loss": 0.67,
+      "step": 10958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322046497406309e-06,
+      "loss": 0.756,
+      "step": 10959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320394408162412e-06,
+      "loss": 0.689,
+      "step": 10960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318742366091116e-06,
+      "loss": 0.3345,
+      "step": 10961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31709037123882e-06,
+      "loss": 0.7185,
+      "step": 10962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.315438423651923e-06,
+      "loss": 0.6336,
+      "step": 10963
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313786523376818e-06,
+      "loss": 0.6528,
+      "step": 10964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312134670459901e-06,
+      "loss": 0.7035,
+      "step": 10965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31048286494757e-06,
+      "loss": 0.6975,
+      "step": 10966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308831106886206e-06,
+      "loss": 0.7903,
+      "step": 10967
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.307179396322206e-06,
+      "loss": 0.6986,
+      "step": 10968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305527733301958e-06,
+      "loss": 0.6956,
+      "step": 10969
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303876117871853e-06,
+      "loss": 0.6977,
+      "step": 10970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30222455007827e-06,
+      "loss": 0.7323,
+      "step": 10971
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.300573029967605e-06,
+      "loss": 0.7014,
+      "step": 10972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298921557586231e-06,
+      "loss": 0.6928,
+      "step": 10973
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297270132980538e-06,
+      "loss": 0.7008,
+      "step": 10974
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.295618756196902e-06,
+      "loss": 0.6367,
+      "step": 10975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293967427281705e-06,
+      "loss": 0.6587,
+      "step": 10976
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.292316146281325e-06,
+      "loss": 0.7005,
+      "step": 10977
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29066491324214e-06,
+      "loss": 0.7049,
+      "step": 10978
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289013728210529e-06,
+      "loss": 0.6329,
+      "step": 10979
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287362591232859e-06,
+      "loss": 0.6761,
+      "step": 10980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.285711502355506e-06,
+      "loss": 0.723,
+      "step": 10981
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284060461624845e-06,
+      "loss": 0.7457,
+      "step": 10982
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28240946908724e-06,
+      "loss": 0.6359,
+      "step": 10983
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280758524789065e-06,
+      "loss": 0.6524,
+      "step": 10984
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279107628776693e-06,
+      "loss": 0.7695,
+      "step": 10985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277456781096477e-06,
+      "loss": 0.6739,
+      "step": 10986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275805981794786e-06,
+      "loss": 0.7164,
+      "step": 10987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274155230917991e-06,
+      "loss": 0.6716,
+      "step": 10988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272504528512448e-06,
+      "loss": 0.7357,
+      "step": 10989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270853874624519e-06,
+      "loss": 0.7444,
+      "step": 10990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269203269300563e-06,
+      "loss": 0.674,
+      "step": 10991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267552712586942e-06,
+      "loss": 0.6341,
+      "step": 10992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.265902204530007e-06,
+      "loss": 0.7073,
+      "step": 10993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264251745176116e-06,
+      "loss": 0.7197,
+      "step": 10994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262601334571623e-06,
+      "loss": 0.6536,
+      "step": 10995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260950972762882e-06,
+      "loss": 0.7133,
+      "step": 10996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25930065979624e-06,
+      "loss": 0.7402,
+      "step": 10997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.257650395718054e-06,
+      "loss": 0.7101,
+      "step": 10998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256000180574665e-06,
+      "loss": 0.7051,
+      "step": 10999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254350014412423e-06,
+      "loss": 0.678,
+      "step": 11000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252699897277675e-06,
+      "loss": 0.7034,
+      "step": 11001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251049829216765e-06,
+      "loss": 0.6964,
+      "step": 11002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.249399810276034e-06,
+      "loss": 0.6671,
+      "step": 11003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247749840501825e-06,
+      "loss": 0.6851,
+      "step": 11004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246099919940484e-06,
+      "loss": 0.6632,
+      "step": 11005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.244450048638341e-06,
+      "loss": 0.7027,
+      "step": 11006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.242800226641734e-06,
+      "loss": 0.6589,
+      "step": 11007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.241150453997005e-06,
+      "loss": 0.7378,
+      "step": 11008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239500730750485e-06,
+      "loss": 0.6487,
+      "step": 11009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.237851056948507e-06,
+      "loss": 0.758,
+      "step": 11010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236201432637408e-06,
+      "loss": 0.6954,
+      "step": 11011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234551857863514e-06,
+      "loss": 0.7036,
+      "step": 11012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232902332673151e-06,
+      "loss": 0.6609,
+      "step": 11013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231252857112652e-06,
+      "loss": 0.6932,
+      "step": 11014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229603431228344e-06,
+      "loss": 0.6536,
+      "step": 11015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.227954055066548e-06,
+      "loss": 0.6915,
+      "step": 11016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226304728673588e-06,
+      "loss": 0.7252,
+      "step": 11017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224655452095792e-06,
+      "loss": 0.7178,
+      "step": 11018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.223006225379474e-06,
+      "loss": 0.7324,
+      "step": 11019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221357048570956e-06,
+      "loss": 0.7992,
+      "step": 11020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.219707921716555e-06,
+      "loss": 0.6791,
+      "step": 11021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218058844862592e-06,
+      "loss": 0.6336,
+      "step": 11022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216409818055372e-06,
+      "loss": 0.7145,
+      "step": 11023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.214760841341223e-06,
+      "loss": 0.7315,
+      "step": 11024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213111914766448e-06,
+      "loss": 0.7321,
+      "step": 11025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211463038377358e-06,
+      "loss": 0.6251,
+      "step": 11026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.209814212220262e-06,
+      "loss": 0.7477,
+      "step": 11027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208165436341474e-06,
+      "loss": 0.6248,
+      "step": 11028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.206516710787295e-06,
+      "loss": 0.6809,
+      "step": 11029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204868035604035e-06,
+      "loss": 0.7104,
+      "step": 11030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.203219410837996e-06,
+      "loss": 0.6818,
+      "step": 11031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201570836535477e-06,
+      "loss": 0.7215,
+      "step": 11032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199922312742785e-06,
+      "loss": 0.7251,
+      "step": 11033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198273839506213e-06,
+      "loss": 0.7308,
+      "step": 11034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.196625416872066e-06,
+      "loss": 0.7003,
+      "step": 11035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194977044886636e-06,
+      "loss": 0.6052,
+      "step": 11036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.193328723596223e-06,
+      "loss": 0.6683,
+      "step": 11037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191680453047115e-06,
+      "loss": 0.6811,
+      "step": 11038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19003223328561e-06,
+      "loss": 0.679,
+      "step": 11039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188384064357994e-06,
+      "loss": 0.679,
+      "step": 11040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186735946310562e-06,
+      "loss": 0.7354,
+      "step": 11041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.185087879189598e-06,
+      "loss": 0.5891,
+      "step": 11042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183439863041387e-06,
+      "loss": 0.7245,
+      "step": 11043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181791897912226e-06,
+      "loss": 0.7202,
+      "step": 11044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180143983848388e-06,
+      "loss": 0.6553,
+      "step": 11045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178496120896156e-06,
+      "loss": 0.6085,
+      "step": 11046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176848309101815e-06,
+      "loss": 0.6783,
+      "step": 11047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175200548511641e-06,
+      "loss": 0.5929,
+      "step": 11048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173552839171915e-06,
+      "loss": 0.6852,
+      "step": 11049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.171905181128918e-06,
+      "loss": 0.783,
+      "step": 11050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.170257574428916e-06,
+      "loss": 0.7468,
+      "step": 11051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168610019118187e-06,
+      "loss": 0.6938,
+      "step": 11052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.166962515243004e-06,
+      "loss": 0.7461,
+      "step": 11053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165315062849639e-06,
+      "loss": 0.6791,
+      "step": 11054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16366766198436e-06,
+      "loss": 0.6711,
+      "step": 11055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.162020312693436e-06,
+      "loss": 0.7205,
+      "step": 11056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.160373015023136e-06,
+      "loss": 0.6992,
+      "step": 11057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158725769019719e-06,
+      "loss": 0.6804,
+      "step": 11058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157078574729451e-06,
+      "loss": 0.7381,
+      "step": 11059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155431432198597e-06,
+      "loss": 0.6816,
+      "step": 11060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153784341473416e-06,
+      "loss": 0.6151,
+      "step": 11061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152137302600168e-06,
+      "loss": 0.6801,
+      "step": 11062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150490315625114e-06,
+      "loss": 0.7048,
+      "step": 11063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148843380594504e-06,
+      "loss": 0.6985,
+      "step": 11064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147196497554595e-06,
+      "loss": 0.7165,
+      "step": 11065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145549666551643e-06,
+      "loss": 0.7154,
+      "step": 11066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143902887631898e-06,
+      "loss": 0.7357,
+      "step": 11067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14225616084161e-06,
+      "loss": 0.624,
+      "step": 11068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140609486227033e-06,
+      "loss": 0.6767,
+      "step": 11069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.138962863834414e-06,
+      "loss": 0.7848,
+      "step": 11070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.137316293709994e-06,
+      "loss": 0.7306,
+      "step": 11071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135669775900017e-06,
+      "loss": 0.6401,
+      "step": 11072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134023310450733e-06,
+      "loss": 0.7629,
+      "step": 11073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13237689740838e-06,
+      "loss": 0.705,
+      "step": 11074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130730536819197e-06,
+      "loss": 0.6918,
+      "step": 11075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129084228729434e-06,
+      "loss": 0.6236,
+      "step": 11076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127437973185309e-06,
+      "loss": 0.6775,
+      "step": 11077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125791770233071e-06,
+      "loss": 0.7435,
+      "step": 11078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.124145619918952e-06,
+      "loss": 0.6643,
+      "step": 11079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122499522289187e-06,
+      "loss": 0.7351,
+      "step": 11080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.120853477390002e-06,
+      "loss": 0.7207,
+      "step": 11081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119207485267633e-06,
+      "loss": 0.7053,
+      "step": 11082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.117561545968307e-06,
+      "loss": 0.7571,
+      "step": 11083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115915659538246e-06,
+      "loss": 0.663,
+      "step": 11084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.114269826023682e-06,
+      "loss": 0.6443,
+      "step": 11085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112624045470834e-06,
+      "loss": 0.6633,
+      "step": 11086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11097831792593e-06,
+      "loss": 0.7071,
+      "step": 11087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109332643435183e-06,
+      "loss": 0.722,
+      "step": 11088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107687022044828e-06,
+      "loss": 0.7232,
+      "step": 11089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.106041453801066e-06,
+      "loss": 0.6852,
+      "step": 11090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.104395938750123e-06,
+      "loss": 0.7035,
+      "step": 11091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102750476938211e-06,
+      "loss": 0.6796,
+      "step": 11092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10110506841154e-06,
+      "loss": 0.705,
+      "step": 11093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099459713216331e-06,
+      "loss": 0.6638,
+      "step": 11094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09781441139879e-06,
+      "loss": 0.6879,
+      "step": 11095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096169163005132e-06,
+      "loss": 0.6959,
+      "step": 11096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094523968081553e-06,
+      "loss": 0.737,
+      "step": 11097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092878826674265e-06,
+      "loss": 0.6841,
+      "step": 11098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091233738829476e-06,
+      "loss": 0.662,
+      "step": 11099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.089588704593383e-06,
+      "loss": 0.6815,
+      "step": 11100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.087943724012193e-06,
+      "loss": 0.7336,
+      "step": 11101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086298797132106e-06,
+      "loss": 0.6991,
+      "step": 11102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.084653923999315e-06,
+      "loss": 0.3493,
+      "step": 11103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083009104660021e-06,
+      "loss": 0.6864,
+      "step": 11104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08136433916042e-06,
+      "loss": 0.7733,
+      "step": 11105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079719627546706e-06,
+      "loss": 0.7057,
+      "step": 11106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078074969865068e-06,
+      "loss": 0.6653,
+      "step": 11107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.076430366161703e-06,
+      "loss": 0.6866,
+      "step": 11108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.074785816482798e-06,
+      "loss": 0.6644,
+      "step": 11109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07314132087454e-06,
+      "loss": 0.6344,
+      "step": 11110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071496879383115e-06,
+      "loss": 0.6833,
+      "step": 11111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.069852492054711e-06,
+      "loss": 0.7094,
+      "step": 11112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068208158935506e-06,
+      "loss": 0.7132,
+      "step": 11113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.066563880071688e-06,
+      "loss": 0.6741,
+      "step": 11114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06491965550944e-06,
+      "loss": 0.6633,
+      "step": 11115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06327548529493e-06,
+      "loss": 0.6768,
+      "step": 11116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06163136947434e-06,
+      "loss": 0.7248,
+      "step": 11117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05998730809385e-06,
+      "loss": 0.7394,
+      "step": 11118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058343301199632e-06,
+      "loss": 0.7169,
+      "step": 11119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056699348837857e-06,
+      "loss": 0.6504,
+      "step": 11120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055055451054698e-06,
+      "loss": 0.7311,
+      "step": 11121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053411607896327e-06,
+      "loss": 0.7042,
+      "step": 11122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051767819408907e-06,
+      "loss": 0.6763,
+      "step": 11123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050124085638607e-06,
+      "loss": 0.7033,
+      "step": 11124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048480406631591e-06,
+      "loss": 0.7026,
+      "step": 11125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046836782434024e-06,
+      "loss": 0.6238,
+      "step": 11126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04519321309207e-06,
+      "loss": 0.7258,
+      "step": 11127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043549698651888e-06,
+      "loss": 0.7313,
+      "step": 11128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041906239159636e-06,
+      "loss": 0.6409,
+      "step": 11129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04026283466147e-06,
+      "loss": 0.6521,
+      "step": 11130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.038619485203547e-06,
+      "loss": 0.6814,
+      "step": 11131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036976190832022e-06,
+      "loss": 0.6354,
+      "step": 11132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035332951593047e-06,
+      "loss": 0.6491,
+      "step": 11133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033689767532777e-06,
+      "loss": 0.6747,
+      "step": 11134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032046638697362e-06,
+      "loss": 0.7169,
+      "step": 11135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030403565132942e-06,
+      "loss": 0.6881,
+      "step": 11136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028760546885666e-06,
+      "loss": 0.6871,
+      "step": 11137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.027117584001685e-06,
+      "loss": 0.6854,
+      "step": 11138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.025474676527138e-06,
+      "loss": 0.6436,
+      "step": 11139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023831824508168e-06,
+      "loss": 0.6861,
+      "step": 11140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02218902799092e-06,
+      "loss": 0.6964,
+      "step": 11141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.02054628702152e-06,
+      "loss": 0.7067,
+      "step": 11142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018903601646117e-06,
+      "loss": 0.6777,
+      "step": 11143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017260971910843e-06,
+      "loss": 0.7222,
+      "step": 11144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01561839786183e-06,
+      "loss": 0.6898,
+      "step": 11145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013975879545216e-06,
+      "loss": 0.669,
+      "step": 11146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012333417007125e-06,
+      "loss": 0.6473,
+      "step": 11147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010691010293693e-06,
+      "loss": 0.6657,
+      "step": 11148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.009048659451044e-06,
+      "loss": 0.66,
+      "step": 11149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007406364525303e-06,
+      "loss": 0.6668,
+      "step": 11150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005764125562599e-06,
+      "loss": 0.7095,
+      "step": 11151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.004121942609051e-06,
+      "loss": 0.6042,
+      "step": 11152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002479815710781e-06,
+      "loss": 0.7147,
+      "step": 11153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000837744913919e-06,
+      "loss": 0.3436,
+      "step": 11154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999195730264569e-06,
+      "loss": 0.6914,
+      "step": 11155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997553771808853e-06,
+      "loss": 0.6569,
+      "step": 11156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995911869592888e-06,
+      "loss": 0.6542,
+      "step": 11157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994270023662784e-06,
+      "loss": 0.6737,
+      "step": 11158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992628234064658e-06,
+      "loss": 0.7079,
+      "step": 11159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990986500844619e-06,
+      "loss": 0.627,
+      "step": 11160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98934482404878e-06,
+      "loss": 0.6761,
+      "step": 11161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987703203723236e-06,
+      "loss": 0.618,
+      "step": 11162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986061639914103e-06,
+      "loss": 0.735,
+      "step": 11163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984420132667483e-06,
+      "loss": 0.3464,
+      "step": 11164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.982778682029478e-06,
+      "loss": 0.7593,
+      "step": 11165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98113728804619e-06,
+      "loss": 0.7338,
+      "step": 11166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979495950763717e-06,
+      "loss": 0.6432,
+      "step": 11167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.977854670228158e-06,
+      "loss": 0.6268,
+      "step": 11168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976213446485607e-06,
+      "loss": 0.6988,
+      "step": 11169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97457227958216e-06,
+      "loss": 0.7018,
+      "step": 11170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972931169563912e-06,
+      "loss": 0.6434,
+      "step": 11171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.971290116476951e-06,
+      "loss": 0.7575,
+      "step": 11172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96964912036737e-06,
+      "loss": 0.6829,
+      "step": 11173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968008181281257e-06,
+      "loss": 0.7103,
+      "step": 11174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966367299264696e-06,
+      "loss": 0.6016,
+      "step": 11175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964726474363771e-06,
+      "loss": 0.6544,
+      "step": 11176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963085706624572e-06,
+      "loss": 0.7691,
+      "step": 11177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961444996093172e-06,
+      "loss": 0.7142,
+      "step": 11178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959804342815657e-06,
+      "loss": 0.7142,
+      "step": 11179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958163746838111e-06,
+      "loss": 0.6815,
+      "step": 11180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9565232082066e-06,
+      "loss": 0.6188,
+      "step": 11181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9548827269672e-06,
+      "loss": 0.754,
+      "step": 11182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.953242303165992e-06,
+      "loss": 0.6567,
+      "step": 11183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951601936849044e-06,
+      "loss": 0.6701,
+      "step": 11184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.94996162806243e-06,
+      "loss": 0.6548,
+      "step": 11185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948321376852214e-06,
+      "loss": 0.6388,
+      "step": 11186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946681183264468e-06,
+      "loss": 0.6717,
+      "step": 11187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945041047345254e-06,
+      "loss": 0.7125,
+      "step": 11188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943400969140635e-06,
+      "loss": 0.6527,
+      "step": 11189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941760948696679e-06,
+      "loss": 0.6658,
+      "step": 11190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.940120986059441e-06,
+      "loss": 0.6976,
+      "step": 11191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938481081274985e-06,
+      "loss": 0.5927,
+      "step": 11192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.936841234389367e-06,
+      "loss": 0.7003,
+      "step": 11193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935201445448641e-06,
+      "loss": 0.654,
+      "step": 11194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.933561714498864e-06,
+      "loss": 0.7158,
+      "step": 11195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931922041586085e-06,
+      "loss": 0.6834,
+      "step": 11196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.930282426756358e-06,
+      "loss": 0.6887,
+      "step": 11197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928642870055732e-06,
+      "loss": 0.6963,
+      "step": 11198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927003371530254e-06,
+      "loss": 0.7513,
+      "step": 11199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.925363931225977e-06,
+      "loss": 0.7106,
+      "step": 11200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923724549188935e-06,
+      "loss": 0.6887,
+      "step": 11201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922085225465172e-06,
+      "loss": 0.6557,
+      "step": 11202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920445960100737e-06,
+      "loss": 0.5966,
+      "step": 11203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.918806753141663e-06,
+      "loss": 0.6453,
+      "step": 11204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917167604633992e-06,
+      "loss": 0.6559,
+      "step": 11205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915528514623763e-06,
+      "loss": 0.6548,
+      "step": 11206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913889483156999e-06,
+      "loss": 0.6696,
+      "step": 11207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912250510279743e-06,
+      "loss": 0.3079,
+      "step": 11208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.910611596038022e-06,
+      "loss": 0.6056,
+      "step": 11209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90897274047787e-06,
+      "loss": 0.6473,
+      "step": 11210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907333943645312e-06,
+      "loss": 0.6509,
+      "step": 11211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905695205586374e-06,
+      "loss": 0.6842,
+      "step": 11212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.904056526347084e-06,
+      "loss": 0.6907,
+      "step": 11213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902417905973463e-06,
+      "loss": 0.7133,
+      "step": 11214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.90077934451153e-06,
+      "loss": 0.6525,
+      "step": 11215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89914084200731e-06,
+      "loss": 0.6866,
+      "step": 11216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897502398506819e-06,
+      "loss": 0.6975,
+      "step": 11217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895864014056068e-06,
+      "loss": 0.6573,
+      "step": 11218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89422568870108e-06,
+      "loss": 0.6787,
+      "step": 11219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892587422487871e-06,
+      "loss": 0.6956,
+      "step": 11220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890949215462442e-06,
+      "loss": 0.7312,
+      "step": 11221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889311067670808e-06,
+      "loss": 0.6652,
+      "step": 11222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887672979158975e-06,
+      "loss": 0.7044,
+      "step": 11223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886034949972955e-06,
+      "loss": 0.6862,
+      "step": 11224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884396980158747e-06,
+      "loss": 0.6881,
+      "step": 11225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882759069762362e-06,
+      "loss": 0.663,
+      "step": 11226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881121218829789e-06,
+      "loss": 0.7134,
+      "step": 11227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879483427407038e-06,
+      "loss": 0.6861,
+      "step": 11228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.877845695540104e-06,
+      "loss": 0.7558,
+      "step": 11229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876208023274984e-06,
+      "loss": 0.6862,
+      "step": 11230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874570410657675e-06,
+      "loss": 0.6781,
+      "step": 11231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872932857734166e-06,
+      "loss": 0.651,
+      "step": 11232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871295364550453e-06,
+      "loss": 0.6711,
+      "step": 11233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869657931152522e-06,
+      "loss": 0.6702,
+      "step": 11234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868020557586363e-06,
+      "loss": 0.649,
+      "step": 11235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86638324389796e-06,
+      "loss": 0.5949,
+      "step": 11236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864745990133303e-06,
+      "loss": 0.7116,
+      "step": 11237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86310879633837e-06,
+      "loss": 0.6714,
+      "step": 11238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.861471662559148e-06,
+      "loss": 0.6285,
+      "step": 11239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859834588841608e-06,
+      "loss": 0.7015,
+      "step": 11240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858197575231737e-06,
+      "loss": 0.6267,
+      "step": 11241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856560621775507e-06,
+      "loss": 0.6924,
+      "step": 11242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85492372851889e-06,
+      "loss": 0.7189,
+      "step": 11243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853286895507865e-06,
+      "loss": 0.332,
+      "step": 11244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8516501227884e-06,
+      "loss": 0.6993,
+      "step": 11245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85001341040647e-06,
+      "loss": 0.7181,
+      "step": 11246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.848376758408034e-06,
+      "loss": 0.6517,
+      "step": 11247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846740166839062e-06,
+      "loss": 0.6875,
+      "step": 11248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84510363574552e-06,
+      "loss": 0.629,
+      "step": 11249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843467165173371e-06,
+      "loss": 0.673,
+      "step": 11250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841830755168574e-06,
+      "loss": 0.6797,
+      "step": 11251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84019440577709e-06,
+      "loss": 0.7311,
+      "step": 11252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.838558117044876e-06,
+      "loss": 0.682,
+      "step": 11253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836921889017889e-06,
+      "loss": 0.698,
+      "step": 11254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83528572174208e-06,
+      "loss": 0.642,
+      "step": 11255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.833649615263406e-06,
+      "loss": 0.6835,
+      "step": 11256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832013569627815e-06,
+      "loss": 0.6718,
+      "step": 11257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.830377584881259e-06,
+      "loss": 0.6927,
+      "step": 11258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828741661069685e-06,
+      "loss": 0.6885,
+      "step": 11259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.827105798239034e-06,
+      "loss": 0.7225,
+      "step": 11260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825469996435255e-06,
+      "loss": 0.7253,
+      "step": 11261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823834255704287e-06,
+      "loss": 0.6763,
+      "step": 11262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822198576092074e-06,
+      "loss": 0.7205,
+      "step": 11263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.820562957644553e-06,
+      "loss": 0.618,
+      "step": 11264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.818927400407666e-06,
+      "loss": 0.7309,
+      "step": 11265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81729190442734e-06,
+      "loss": 0.7301,
+      "step": 11266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815656469749512e-06,
+      "loss": 0.6599,
+      "step": 11267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814021096420114e-06,
+      "loss": 0.6561,
+      "step": 11268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812385784485079e-06,
+      "loss": 0.7394,
+      "step": 11269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810750533990334e-06,
+      "loss": 0.6496,
+      "step": 11270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809115344981807e-06,
+      "loss": 0.6748,
+      "step": 11271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807480217505423e-06,
+      "loss": 0.7187,
+      "step": 11272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8058451516071e-06,
+      "loss": 0.6888,
+      "step": 11273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804210147332766e-06,
+      "loss": 0.6952,
+      "step": 11274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80257520472834e-06,
+      "loss": 0.6768,
+      "step": 11275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800940323839738e-06,
+      "loss": 0.6538,
+      "step": 11276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.799305504712878e-06,
+      "loss": 0.6514,
+      "step": 11277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.797670747393674e-06,
+      "loss": 0.6621,
+      "step": 11278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.79603605192804e-06,
+      "loss": 0.6389,
+      "step": 11279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794401418361887e-06,
+      "loss": 0.6409,
+      "step": 11280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792766846741124e-06,
+      "loss": 0.7152,
+      "step": 11281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.791132337111657e-06,
+      "loss": 0.6931,
+      "step": 11282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789497889519396e-06,
+      "loss": 0.6677,
+      "step": 11283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787863504010239e-06,
+      "loss": 0.3287,
+      "step": 11284
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7862291806301e-06,
+      "loss": 0.6998,
+      "step": 11285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78459491942487e-06,
+      "loss": 0.6573,
+      "step": 11286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78296072044045e-06,
+      "loss": 0.7135,
+      "step": 11287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781326583722736e-06,
+      "loss": 0.6083,
+      "step": 11288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779692509317627e-06,
+      "loss": 0.6464,
+      "step": 11289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778058497271016e-06,
+      "loss": 0.6926,
+      "step": 11290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7764245476288e-06,
+      "loss": 0.7183,
+      "step": 11291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.696,
+      "step": 11292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.773156835741085e-06,
+      "loss": 0.6485,
+      "step": 11293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771523073587368e-06,
+      "loss": 0.7374,
+      "step": 11294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76988937402159e-06,
+      "loss": 0.6714,
+      "step": 11295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768255737089638e-06,
+      "loss": 0.681,
+      "step": 11296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766622162837391e-06,
+      "loss": 0.6676,
+      "step": 11297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764988651310731e-06,
+      "loss": 0.6337,
+      "step": 11298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763355202555532e-06,
+      "loss": 0.6727,
+      "step": 11299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761721816617673e-06,
+      "loss": 0.7072,
+      "step": 11300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760088493543026e-06,
+      "loss": 0.6777,
+      "step": 11301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.758455233377466e-06,
+      "loss": 0.7277,
+      "step": 11302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.756822036166865e-06,
+      "loss": 0.7203,
+      "step": 11303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.755188901957093e-06,
+      "loss": 0.6979,
+      "step": 11304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753555830794013e-06,
+      "loss": 0.6346,
+      "step": 11305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.751922822723491e-06,
+      "loss": 0.7283,
+      "step": 11306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750289877791394e-06,
+      "loss": 0.6735,
+      "step": 11307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74865699604358e-06,
+      "loss": 0.76,
+      "step": 11308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747024177525916e-06,
+      "loss": 0.3064,
+      "step": 11309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745391422284257e-06,
+      "loss": 0.673,
+      "step": 11310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74375873036446e-06,
+      "loss": 0.6546,
+      "step": 11311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742126101812377e-06,
+      "loss": 0.6111,
+      "step": 11312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740493536673866e-06,
+      "loss": 0.6751,
+      "step": 11313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738861034994775e-06,
+      "loss": 0.7145,
+      "step": 11314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737228596820954e-06,
+      "loss": 0.6982,
+      "step": 11315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.735596222198253e-06,
+      "loss": 0.6861,
+      "step": 11316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73396391117252e-06,
+      "loss": 0.5969,
+      "step": 11317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.732331663789592e-06,
+      "loss": 0.6772,
+      "step": 11318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730699480095316e-06,
+      "loss": 0.7634,
+      "step": 11319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729067360135534e-06,
+      "loss": 0.7206,
+      "step": 11320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.727435303956081e-06,
+      "loss": 0.6813,
+      "step": 11321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725803311602799e-06,
+      "loss": 0.7313,
+      "step": 11322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724171383121518e-06,
+      "loss": 0.6498,
+      "step": 11323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72253951855808e-06,
+      "loss": 0.691,
+      "step": 11324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.720907717958308e-06,
+      "loss": 0.6675,
+      "step": 11325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719275981368033e-06,
+      "loss": 0.6879,
+      "step": 11326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.717644308833087e-06,
+      "loss": 0.7088,
+      "step": 11327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716012700399294e-06,
+      "loss": 0.6227,
+      "step": 11328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71438115611248e-06,
+      "loss": 0.6624,
+      "step": 11329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.712749676018471e-06,
+      "loss": 0.5763,
+      "step": 11330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71111826016308e-06,
+      "loss": 0.6161,
+      "step": 11331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70948690859213e-06,
+      "loss": 0.6918,
+      "step": 11332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707855621351438e-06,
+      "loss": 0.6985,
+      "step": 11333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706224398486821e-06,
+      "loss": 0.7052,
+      "step": 11334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704593240044094e-06,
+      "loss": 0.7674,
+      "step": 11335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702962146069065e-06,
+      "loss": 0.6568,
+      "step": 11336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701331116607548e-06,
+      "loss": 0.7136,
+      "step": 11337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.699700151705348e-06,
+      "loss": 0.6811,
+      "step": 11338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698069251408271e-06,
+      "loss": 0.7823,
+      "step": 11339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.696438415762124e-06,
+      "loss": 0.6477,
+      "step": 11340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694807644812712e-06,
+      "loss": 0.6959,
+      "step": 11341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69317693860583e-06,
+      "loss": 0.7061,
+      "step": 11342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691546297187285e-06,
+      "loss": 0.6558,
+      "step": 11343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689915720602868e-06,
+      "loss": 0.6999,
+      "step": 11344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688285208898375e-06,
+      "loss": 0.7145,
+      "step": 11345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686654762119603e-06,
+      "loss": 0.6968,
+      "step": 11346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685024380312341e-06,
+      "loss": 0.6724,
+      "step": 11347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68339406352238e-06,
+      "loss": 0.716,
+      "step": 11348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68176381179551e-06,
+      "loss": 0.7045,
+      "step": 11349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68013362517752e-06,
+      "loss": 0.5829,
+      "step": 11350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678503503714188e-06,
+      "loss": 0.6805,
+      "step": 11351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6768734474513e-06,
+      "loss": 0.6158,
+      "step": 11352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675243456434632e-06,
+      "loss": 0.3345,
+      "step": 11353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673613530709973e-06,
+      "loss": 0.6559,
+      "step": 11354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.671983670323093e-06,
+      "loss": 0.6862,
+      "step": 11355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670353875319775e-06,
+      "loss": 0.6646,
+      "step": 11356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66872414574578e-06,
+      "loss": 0.7455,
+      "step": 11357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667094481646891e-06,
+      "loss": 0.683,
+      "step": 11358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.665464883068874e-06,
+      "loss": 0.6451,
+      "step": 11359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663835350057496e-06,
+      "loss": 0.6931,
+      "step": 11360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662205882658524e-06,
+      "loss": 0.3349,
+      "step": 11361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660576480917725e-06,
+      "loss": 0.5869,
+      "step": 11362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65894714488086e-06,
+      "loss": 0.7785,
+      "step": 11363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657317874593689e-06,
+      "loss": 0.7618,
+      "step": 11364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655688670101968e-06,
+      "loss": 0.6717,
+      "step": 11365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65405953145146e-06,
+      "loss": 0.6055,
+      "step": 11366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652430458687918e-06,
+      "loss": 0.7022,
+      "step": 11367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.650801451857092e-06,
+      "loss": 0.6213,
+      "step": 11368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649172511004741e-06,
+      "loss": 0.7187,
+      "step": 11369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647543636176606e-06,
+      "loss": 0.7376,
+      "step": 11370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645914827418442e-06,
+      "loss": 0.7077,
+      "step": 11371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64428608477599e-06,
+      "loss": 0.626,
+      "step": 11372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642657408294993e-06,
+      "loss": 0.7631,
+      "step": 11373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641028798021197e-06,
+      "loss": 0.7428,
+      "step": 11374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.639400254000344e-06,
+      "loss": 0.7007,
+      "step": 11375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637771776278173e-06,
+      "loss": 0.761,
+      "step": 11376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.636143364900412e-06,
+      "loss": 0.7093,
+      "step": 11377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634515019912803e-06,
+      "loss": 0.7504,
+      "step": 11378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632886741361077e-06,
+      "loss": 0.728,
+      "step": 11379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631258529290966e-06,
+      "loss": 0.7163,
+      "step": 11380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629630383748198e-06,
+      "loss": 0.3388,
+      "step": 11381
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628002304778505e-06,
+      "loss": 0.6373,
+      "step": 11382
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626374292427604e-06,
+      "loss": 0.7073,
+      "step": 11383
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.624746346741223e-06,
+      "loss": 0.6911,
+      "step": 11384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623118467765085e-06,
+      "loss": 0.6393,
+      "step": 11385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.621490655544908e-06,
+      "loss": 0.7139,
+      "step": 11386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6198629101264095e-06,
+      "loss": 0.7023,
+      "step": 11387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.618235231555308e-06,
+      "loss": 0.6785,
+      "step": 11388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.616607619877317e-06,
+      "loss": 0.6696,
+      "step": 11389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614980075138147e-06,
+      "loss": 0.6873,
+      "step": 11390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613352597383507e-06,
+      "loss": 0.8115,
+      "step": 11391
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611725186659111e-06,
+      "loss": 0.6927,
+      "step": 11392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61009784301066e-06,
+      "loss": 0.6196,
+      "step": 11393
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608470566483864e-06,
+      "loss": 0.6844,
+      "step": 11394
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606843357124426e-06,
+      "loss": 0.6852,
+      "step": 11395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60521621497804e-06,
+      "loss": 0.7225,
+      "step": 11396
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.603589140090412e-06,
+      "loss": 0.6318,
+      "step": 11397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601962132507233e-06,
+      "loss": 0.7155,
+      "step": 11398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6003351922742045e-06,
+      "loss": 0.6819,
+      "step": 11399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598708319437018e-06,
+      "loss": 0.6876,
+      "step": 11400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597081514041364e-06,
+      "loss": 0.6815,
+      "step": 11401
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5954547761329365e-06,
+      "loss": 0.663,
+      "step": 11402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593828105757417e-06,
+      "loss": 0.7019,
+      "step": 11403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5922015029604935e-06,
+      "loss": 0.6311,
+      "step": 11404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.590574967787852e-06,
+      "loss": 0.7184,
+      "step": 11405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588948500285172e-06,
+      "loss": 0.7097,
+      "step": 11406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5873221004981345e-06,
+      "loss": 0.6883,
+      "step": 11407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585695768472421e-06,
+      "loss": 0.6857,
+      "step": 11408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.584069504253703e-06,
+      "loss": 0.6307,
+      "step": 11409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582443307887657e-06,
+      "loss": 0.632,
+      "step": 11410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580817179419955e-06,
+      "loss": 0.7267,
+      "step": 11411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579191118896269e-06,
+      "loss": 0.6722,
+      "step": 11412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577565126362267e-06,
+      "loss": 0.7028,
+      "step": 11413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575939201863614e-06,
+      "loss": 0.724,
+      "step": 11414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5743133454459816e-06,
+      "loss": 0.7463,
+      "step": 11415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572687557155025e-06,
+      "loss": 0.7002,
+      "step": 11416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571061837036409e-06,
+      "loss": 0.6633,
+      "step": 11417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.569436185135789e-06,
+      "loss": 0.6758,
+      "step": 11418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567810601498828e-06,
+      "loss": 0.7119,
+      "step": 11419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5661850861711786e-06,
+      "loss": 0.6961,
+      "step": 11420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.564559639198499e-06,
+      "loss": 0.6794,
+      "step": 11421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56293426062643e-06,
+      "loss": 0.7029,
+      "step": 11422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56130895050063e-06,
+      "loss": 0.7082,
+      "step": 11423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559683708866743e-06,
+      "loss": 0.676,
+      "step": 11424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.558058535770416e-06,
+      "loss": 0.649,
+      "step": 11425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.556433431257293e-06,
+      "loss": 0.6317,
+      "step": 11426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.554808395373016e-06,
+      "loss": 0.7319,
+      "step": 11427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553183428163225e-06,
+      "loss": 0.6828,
+      "step": 11428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551558529673557e-06,
+      "loss": 0.7127,
+      "step": 11429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54993369994965e-06,
+      "loss": 0.5944,
+      "step": 11430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5483089390371345e-06,
+      "loss": 0.6404,
+      "step": 11431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546684246981646e-06,
+      "loss": 0.6785,
+      "step": 11432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545059623828815e-06,
+      "loss": 0.7422,
+      "step": 11433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54343506962427e-06,
+      "loss": 0.6523,
+      "step": 11434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541810584413635e-06,
+      "loss": 0.6801,
+      "step": 11435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.540186168242536e-06,
+      "loss": 0.6187,
+      "step": 11436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.538561821156594e-06,
+      "loss": 0.7238,
+      "step": 11437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5369375432014305e-06,
+      "loss": 0.7095,
+      "step": 11438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.535313334422666e-06,
+      "loss": 0.6766,
+      "step": 11439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533689194865917e-06,
+      "loss": 0.6265,
+      "step": 11440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532065124576802e-06,
+      "loss": 0.6455,
+      "step": 11441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5304411236009234e-06,
+      "loss": 0.6786,
+      "step": 11442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528817191983897e-06,
+      "loss": 0.6915,
+      "step": 11443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.527193329771334e-06,
+      "loss": 0.7015,
+      "step": 11444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525569537008841e-06,
+      "loss": 0.6288,
+      "step": 11445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523945813742022e-06,
+      "loss": 0.6953,
+      "step": 11446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.522322160016482e-06,
+      "loss": 0.731,
+      "step": 11447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.520698575877819e-06,
+      "loss": 0.6787,
+      "step": 11448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519075061371634e-06,
+      "loss": 0.6948,
+      "step": 11449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517451616543523e-06,
+      "loss": 0.6351,
+      "step": 11450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515828241439083e-06,
+      "loss": 0.6677,
+      "step": 11451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.514204936103906e-06,
+      "loss": 0.6699,
+      "step": 11452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.512581700583584e-06,
+      "loss": 0.6918,
+      "step": 11453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510958534923709e-06,
+      "loss": 0.626,
+      "step": 11454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.509335439169863e-06,
+      "loss": 0.6835,
+      "step": 11455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5077124133676345e-06,
+      "loss": 0.6453,
+      "step": 11456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506089457562608e-06,
+      "loss": 0.7227,
+      "step": 11457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504466571800363e-06,
+      "loss": 0.6127,
+      "step": 11458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.502843756126479e-06,
+      "loss": 0.3135,
+      "step": 11459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501221010586539e-06,
+      "loss": 0.6573,
+      "step": 11460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499598335226113e-06,
+      "loss": 0.7038,
+      "step": 11461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497975730090773e-06,
+      "loss": 0.6958,
+      "step": 11462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.496353195226094e-06,
+      "loss": 0.6173,
+      "step": 11463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4947307306776466e-06,
+      "loss": 0.6558,
+      "step": 11464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493108336490996e-06,
+      "loss": 0.6213,
+      "step": 11465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491486012711711e-06,
+      "loss": 0.3253,
+      "step": 11466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489863759385356e-06,
+      "loss": 0.7449,
+      "step": 11467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.488241576557488e-06,
+      "loss": 0.6767,
+      "step": 11468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48661946427367e-06,
+      "loss": 0.6619,
+      "step": 11469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484997422579458e-06,
+      "loss": 0.7414,
+      "step": 11470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.483375451520411e-06,
+      "loss": 0.6036,
+      "step": 11471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481753551142081e-06,
+      "loss": 0.7463,
+      "step": 11472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480131721490023e-06,
+      "loss": 0.6852,
+      "step": 11473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47850996260978e-06,
+      "loss": 0.6568,
+      "step": 11474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476888274546905e-06,
+      "loss": 0.7138,
+      "step": 11475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475266657346942e-06,
+      "loss": 0.6801,
+      "step": 11476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473645111055437e-06,
+      "loss": 0.6597,
+      "step": 11477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.472023635717931e-06,
+      "loss": 0.7421,
+      "step": 11478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470402231379961e-06,
+      "loss": 0.7401,
+      "step": 11479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468780898087075e-06,
+      "loss": 0.6461,
+      "step": 11480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4671596358847975e-06,
+      "loss": 0.637,
+      "step": 11481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465538444818667e-06,
+      "loss": 0.6743,
+      "step": 11482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4639173249342125e-06,
+      "loss": 0.636,
+      "step": 11483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462296276276969e-06,
+      "loss": 0.6962,
+      "step": 11484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460675298892462e-06,
+      "loss": 0.7164,
+      "step": 11485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459054392826224e-06,
+      "loss": 0.6057,
+      "step": 11486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.457433558123764e-06,
+      "loss": 0.5992,
+      "step": 11487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455812794830617e-06,
+      "loss": 0.6632,
+      "step": 11488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.454192102992297e-06,
+      "loss": 0.6743,
+      "step": 11489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.452571482654325e-06,
+      "loss": 0.7145,
+      "step": 11490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450950933862215e-06,
+      "loss": 0.6757,
+      "step": 11491
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449330456661483e-06,
+      "loss": 0.6655,
+      "step": 11492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447710051097642e-06,
+      "loss": 0.6995,
+      "step": 11493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446089717216196e-06,
+      "loss": 0.7202,
+      "step": 11494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.444469455062657e-06,
+      "loss": 0.6252,
+      "step": 11495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.442849264682532e-06,
+      "loss": 0.7005,
+      "step": 11496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.441229146121322e-06,
+      "loss": 0.7224,
+      "step": 11497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4396090994245295e-06,
+      "loss": 0.6924,
+      "step": 11498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.437989124637658e-06,
+      "loss": 0.6567,
+      "step": 11499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.436369221806201e-06,
+      "loss": 0.6646,
+      "step": 11500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434749390975655e-06,
+      "loss": 0.685,
+      "step": 11501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.433129632191513e-06,
+      "loss": 0.7092,
+      "step": 11502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431509945499269e-06,
+      "loss": 0.6911,
+      "step": 11503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429890330944412e-06,
+      "loss": 0.6835,
+      "step": 11504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428270788572432e-06,
+      "loss": 0.6947,
+      "step": 11505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426651318428815e-06,
+      "loss": 0.7203,
+      "step": 11506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.425031920559038e-06,
+      "loss": 0.7203,
+      "step": 11507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4234125950085855e-06,
+      "loss": 0.7039,
+      "step": 11508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.42179334182294e-06,
+      "loss": 0.7117,
+      "step": 11509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420174161047577e-06,
+      "loss": 0.6997,
+      "step": 11510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.418555052727974e-06,
+      "loss": 0.6448,
+      "step": 11511
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4169360169096036e-06,
+      "loss": 0.7294,
+      "step": 11512
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415317053637936e-06,
+      "loss": 0.7267,
+      "step": 11513
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.41369816295844e-06,
+      "loss": 0.6677,
+      "step": 11514
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.412079344916585e-06,
+      "loss": 0.6856,
+      "step": 11515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410460599557835e-06,
+      "loss": 0.6692,
+      "step": 11516
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408841926927654e-06,
+      "loss": 0.6861,
+      "step": 11517
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407223327071504e-06,
+      "loss": 0.6291,
+      "step": 11518
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.405604800034845e-06,
+      "loss": 0.69,
+      "step": 11519
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40398634586313e-06,
+      "loss": 0.7131,
+      "step": 11520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402367964601818e-06,
+      "loss": 0.6819,
+      "step": 11521
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40074965629636e-06,
+      "loss": 0.5919,
+      "step": 11522
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3991314209922085e-06,
+      "loss": 0.6966,
+      "step": 11523
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39751325873481e-06,
+      "loss": 0.7079,
+      "step": 11524
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.395895169569618e-06,
+      "loss": 0.6352,
+      "step": 11525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394277153542069e-06,
+      "loss": 0.7368,
+      "step": 11526
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39265921069761e-06,
+      "loss": 0.6341,
+      "step": 11527
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.391041341081679e-06,
+      "loss": 0.6406,
+      "step": 11528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3894235447397204e-06,
+      "loss": 0.707,
+      "step": 11529
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.387805821717166e-06,
+      "loss": 0.7318,
+      "step": 11530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386188172059452e-06,
+      "loss": 0.6968,
+      "step": 11531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.384570595812013e-06,
+      "loss": 0.6669,
+      "step": 11532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.382953093020276e-06,
+      "loss": 0.6022,
+      "step": 11533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.381335663729669e-06,
+      "loss": 0.6874,
+      "step": 11534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379718307985621e-06,
+      "loss": 0.7028,
+      "step": 11535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378101025833555e-06,
+      "loss": 0.7849,
+      "step": 11536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.376483817318893e-06,
+      "loss": 0.7132,
+      "step": 11537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3748666824870596e-06,
+      "loss": 0.6662,
+      "step": 11538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.373249621383466e-06,
+      "loss": 0.6504,
+      "step": 11539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.371632634053532e-06,
+      "loss": 0.7401,
+      "step": 11540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3700157205426695e-06,
+      "loss": 0.7036,
+      "step": 11541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36839888089629e-06,
+      "loss": 0.6779,
+      "step": 11542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366782115159807e-06,
+      "loss": 0.7186,
+      "step": 11543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.365165423378624e-06,
+      "loss": 0.7521,
+      "step": 11544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363548805598154e-06,
+      "loss": 0.6795,
+      "step": 11545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.361932261863791e-06,
+      "loss": 0.6627,
+      "step": 11546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360315792220942e-06,
+      "loss": 0.7031,
+      "step": 11547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358699396715001e-06,
+      "loss": 0.6908,
+      "step": 11548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357083075391373e-06,
+      "loss": 0.6191,
+      "step": 11549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35546682829545e-06,
+      "loss": 0.6446,
+      "step": 11550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.353850655472628e-06,
+      "loss": 0.6942,
+      "step": 11551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.352234556968289e-06,
+      "loss": 0.7139,
+      "step": 11552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35061853282783e-06,
+      "loss": 0.6901,
+      "step": 11553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.349002583096636e-06,
+      "loss": 0.7516,
+      "step": 11554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34738670782009e-06,
+      "loss": 0.6837,
+      "step": 11555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345770907043578e-06,
+      "loss": 0.3314,
+      "step": 11556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344155180812478e-06,
+      "loss": 0.6638,
+      "step": 11557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.342539529172172e-06,
+      "loss": 0.6863,
+      "step": 11558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340923952168031e-06,
+      "loss": 0.671,
+      "step": 11559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339308449845433e-06,
+      "loss": 0.6359,
+      "step": 11560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337693022249748e-06,
+      "loss": 0.6896,
+      "step": 11561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336077669426348e-06,
+      "loss": 0.7247,
+      "step": 11562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3344623914206004e-06,
+      "loss": 0.6988,
+      "step": 11563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332847188277874e-06,
+      "loss": 0.7706,
+      "step": 11564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.331232060043527e-06,
+      "loss": 0.6566,
+      "step": 11565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329617006762924e-06,
+      "loss": 0.6294,
+      "step": 11566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328002028481425e-06,
+      "loss": 0.7086,
+      "step": 11567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326387125244384e-06,
+      "loss": 0.7118,
+      "step": 11568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324772297097163e-06,
+      "loss": 0.7902,
+      "step": 11569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32315754408511e-06,
+      "loss": 0.6608,
+      "step": 11570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3215428662535834e-06,
+      "loss": 0.7069,
+      "step": 11571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3199282636479215e-06,
+      "loss": 0.7078,
+      "step": 11572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318313736313476e-06,
+      "loss": 0.6518,
+      "step": 11573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316699284295594e-06,
+      "loss": 0.6393,
+      "step": 11574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315084907639616e-06,
+      "loss": 0.6633,
+      "step": 11575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313470606390884e-06,
+      "loss": 0.6653,
+      "step": 11576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311856380594735e-06,
+      "loss": 0.6978,
+      "step": 11577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310242230296508e-06,
+      "loss": 0.6284,
+      "step": 11578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308628155541534e-06,
+      "loss": 0.6876,
+      "step": 11579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.307014156375145e-06,
+      "loss": 0.7122,
+      "step": 11580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3054002328426745e-06,
+      "loss": 0.6612,
+      "step": 11581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303786384989446e-06,
+      "loss": 0.7242,
+      "step": 11582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302172612860789e-06,
+      "loss": 0.6819,
+      "step": 11583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300558916502028e-06,
+      "loss": 0.6356,
+      "step": 11584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29894529595848e-06,
+      "loss": 0.703,
+      "step": 11585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297331751275467e-06,
+      "loss": 0.5991,
+      "step": 11586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295718282498306e-06,
+      "loss": 0.7408,
+      "step": 11587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29410488967231e-06,
+      "loss": 0.716,
+      "step": 11588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.292491572842792e-06,
+      "loss": 0.6618,
+      "step": 11589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2908783320550695e-06,
+      "loss": 0.7382,
+      "step": 11590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289265167354449e-06,
+      "loss": 0.6724,
+      "step": 11591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287652078786229e-06,
+      "loss": 0.7236,
+      "step": 11592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2860390663957205e-06,
+      "loss": 0.6211,
+      "step": 11593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284426130228225e-06,
+      "loss": 0.684,
+      "step": 11594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2828132703290435e-06,
+      "loss": 0.6344,
+      "step": 11595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2812004867434714e-06,
+      "loss": 0.7636,
+      "step": 11596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.27958777951681e-06,
+      "loss": 0.6794,
+      "step": 11597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.277975148694345e-06,
+      "loss": 0.6678,
+      "step": 11598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2763625943213736e-06,
+      "loss": 0.6685,
+      "step": 11599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274750116443185e-06,
+      "loss": 0.6822,
+      "step": 11600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.273137715105063e-06,
+      "loss": 0.6659,
+      "step": 11601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.271525390352297e-06,
+      "loss": 0.7275,
+      "step": 11602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269913142230166e-06,
+      "loss": 0.7362,
+      "step": 11603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2683009707839555e-06,
+      "loss": 0.6886,
+      "step": 11604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26668887605894e-06,
+      "loss": 0.7098,
+      "step": 11605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.265076858100397e-06,
+      "loss": 0.691,
+      "step": 11606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263464916953602e-06,
+      "loss": 0.6988,
+      "step": 11607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261853052663828e-06,
+      "loss": 0.7359,
+      "step": 11608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.260241265276339e-06,
+      "loss": 0.6272,
+      "step": 11609
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258629554836415e-06,
+      "loss": 0.6746,
+      "step": 11610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.25701792138931e-06,
+      "loss": 0.6715,
+      "step": 11611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2554063649802905e-06,
+      "loss": 0.6384,
+      "step": 11612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253794885654619e-06,
+      "loss": 0.6894,
+      "step": 11613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.252183483457558e-06,
+      "loss": 0.6556,
+      "step": 11614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.250572158434359e-06,
+      "loss": 0.6613,
+      "step": 11615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.248960910630281e-06,
+      "loss": 0.7184,
+      "step": 11616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.247349740090579e-06,
+      "loss": 0.6749,
+      "step": 11617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245738646860494e-06,
+      "loss": 0.3253,
+      "step": 11618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24412763098528e-06,
+      "loss": 0.7466,
+      "step": 11619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242516692510186e-06,
+      "loss": 0.6529,
+      "step": 11620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240905831480453e-06,
+      "loss": 0.7056,
+      "step": 11621
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.239295047941324e-06,
+      "loss": 0.647,
+      "step": 11622
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237684341938038e-06,
+      "loss": 0.6703,
+      "step": 11623
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236073713515832e-06,
+      "loss": 0.6888,
+      "step": 11624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2344631627199415e-06,
+      "loss": 0.6901,
+      "step": 11625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232852689595599e-06,
+      "loss": 0.7146,
+      "step": 11626
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2312422941880386e-06,
+      "loss": 0.6719,
+      "step": 11627
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229631976542486e-06,
+      "loss": 0.718,
+      "step": 11628
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2280217367041674e-06,
+      "loss": 0.6949,
+      "step": 11629
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226411574718314e-06,
+      "loss": 0.3271,
+      "step": 11630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.224801490630139e-06,
+      "loss": 0.7056,
+      "step": 11631
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223191484484866e-06,
+      "loss": 0.6776,
+      "step": 11632
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221581556327713e-06,
+      "loss": 0.6402,
+      "step": 11633
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219971706203894e-06,
+      "loss": 0.6634,
+      "step": 11634
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.218361934158627e-06,
+      "loss": 0.631,
+      "step": 11635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216752240237122e-06,
+      "loss": 0.7156,
+      "step": 11636
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.215142624484584e-06,
+      "loss": 0.5946,
+      "step": 11637
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213533086946221e-06,
+      "loss": 0.6279,
+      "step": 11638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211923627667241e-06,
+      "loss": 0.632,
+      "step": 11639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.210314246692845e-06,
+      "loss": 0.7172,
+      "step": 11640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208704944068232e-06,
+      "loss": 0.6711,
+      "step": 11641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207095719838602e-06,
+      "loss": 0.6713,
+      "step": 11642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.205486574049152e-06,
+      "loss": 0.694,
+      "step": 11643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203877506745071e-06,
+      "loss": 0.6793,
+      "step": 11644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.202268517971553e-06,
+      "loss": 0.7154,
+      "step": 11645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2006596077737875e-06,
+      "loss": 0.6982,
+      "step": 11646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199050776196963e-06,
+      "loss": 0.6405,
+      "step": 11647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.197442023286262e-06,
+      "loss": 0.6891,
+      "step": 11648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.19583334908687e-06,
+      "loss": 0.7178,
+      "step": 11649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1942247536439636e-06,
+      "loss": 0.7635,
+      "step": 11650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192616237002723e-06,
+      "loss": 0.7018,
+      "step": 11651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191007799208325e-06,
+      "loss": 0.6483,
+      "step": 11652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.189399440305942e-06,
+      "loss": 0.6787,
+      "step": 11653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1877911603407446e-06,
+      "loss": 0.7059,
+      "step": 11654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186182959357906e-06,
+      "loss": 0.6503,
+      "step": 11655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1845748374025955e-06,
+      "loss": 0.6694,
+      "step": 11656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182966794519969e-06,
+      "loss": 0.7017,
+      "step": 11657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181358830755192e-06,
+      "loss": 0.7284,
+      "step": 11658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17975094615343e-06,
+      "loss": 0.701,
+      "step": 11659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178143140759837e-06,
+      "loss": 0.7538,
+      "step": 11660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.176535414619571e-06,
+      "loss": 0.6483,
+      "step": 11661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1749277677777865e-06,
+      "loss": 0.6602,
+      "step": 11662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.173320200279632e-06,
+      "loss": 0.6995,
+      "step": 11663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17171271217026e-06,
+      "loss": 0.6835,
+      "step": 11664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.170105303494814e-06,
+      "loss": 0.6411,
+      "step": 11665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1684979742984424e-06,
+      "loss": 0.7005,
+      "step": 11666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1668907246262874e-06,
+      "loss": 0.6401,
+      "step": 11667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165283554523488e-06,
+      "loss": 0.6972,
+      "step": 11668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.163676464035186e-06,
+      "loss": 0.3194,
+      "step": 11669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162069453206511e-06,
+      "loss": 0.6675,
+      "step": 11670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1604625220826015e-06,
+      "loss": 0.6671,
+      "step": 11671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158855670708588e-06,
+      "loss": 0.6776,
+      "step": 11672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157248899129599e-06,
+      "loss": 0.7187,
+      "step": 11673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155642207390759e-06,
+      "loss": 0.6795,
+      "step": 11674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.154035595537203e-06,
+      "loss": 0.6902,
+      "step": 11675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1524290636140425e-06,
+      "loss": 0.6387,
+      "step": 11676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150822611666402e-06,
+      "loss": 0.6775,
+      "step": 11677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.149216239739396e-06,
+      "loss": 0.6595,
+      "step": 11678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147609947878146e-06,
+      "loss": 0.7191,
+      "step": 11679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146003736127762e-06,
+      "loss": 0.7427,
+      "step": 11680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.144397604533357e-06,
+      "loss": 0.7119,
+      "step": 11681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.142791553140045e-06,
+      "loss": 0.5869,
+      "step": 11682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1411855819929196e-06,
+      "loss": 0.6695,
+      "step": 11683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139579691137095e-06,
+      "loss": 0.7099,
+      "step": 11684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.137973880617673e-06,
+      "loss": 0.6955,
+      "step": 11685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.13636815047975e-06,
+      "loss": 0.6754,
+      "step": 11686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134762500768426e-06,
+      "loss": 0.8256,
+      "step": 11687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1331569315288e-06,
+      "loss": 0.7203,
+      "step": 11688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131551442805957e-06,
+      "loss": 0.7107,
+      "step": 11689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1299460346449944e-06,
+      "loss": 0.6654,
+      "step": 11690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128340707091e-06,
+      "loss": 0.697,
+      "step": 11691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.126735460189056e-06,
+      "loss": 0.6354,
+      "step": 11692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1251302939842526e-06,
+      "loss": 0.6154,
+      "step": 11693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123525208521667e-06,
+      "loss": 0.6942,
+      "step": 11694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.121920203846387e-06,
+      "loss": 0.6902,
+      "step": 11695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120315280003478e-06,
+      "loss": 0.687,
+      "step": 11696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118710437038023e-06,
+      "loss": 0.6032,
+      "step": 11697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117105674995092e-06,
+      "loss": 0.7632,
+      "step": 11698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.115500993919755e-06,
+      "loss": 0.6228,
+      "step": 11699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.113896393857083e-06,
+      "loss": 0.6592,
+      "step": 11700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112291874852146e-06,
+      "loss": 0.7207,
+      "step": 11701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1106874369499975e-06,
+      "loss": 0.6015,
+      "step": 11702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109083080195703e-06,
+      "loss": 0.6945,
+      "step": 11703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107478804634324e-06,
+      "loss": 0.6255,
+      "step": 11704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105874610310917e-06,
+      "loss": 0.6308,
+      "step": 11705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1042704972705355e-06,
+      "loss": 0.7049,
+      "step": 11706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102666465558232e-06,
+      "loss": 0.713,
+      "step": 11707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.101062515219059e-06,
+      "loss": 0.7329,
+      "step": 11708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0994586462980595e-06,
+      "loss": 0.6583,
+      "step": 11709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.097854858840283e-06,
+      "loss": 0.7312,
+      "step": 11710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0962511528907695e-06,
+      "loss": 0.692,
+      "step": 11711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0946475284945625e-06,
+      "loss": 0.6262,
+      "step": 11712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093043985696699e-06,
+      "loss": 0.2973,
+      "step": 11713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.091440524542219e-06,
+      "loss": 0.695,
+      "step": 11714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089837145076152e-06,
+      "loss": 0.7522,
+      "step": 11715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088233847343531e-06,
+      "loss": 0.7046,
+      "step": 11716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.086630631389386e-06,
+      "loss": 0.7361,
+      "step": 11717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085027497258744e-06,
+      "loss": 0.7172,
+      "step": 11718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083424444996627e-06,
+      "loss": 0.6596,
+      "step": 11719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.081821474648064e-06,
+      "loss": 0.3147,
+      "step": 11720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0802185862580756e-06,
+      "loss": 0.6564,
+      "step": 11721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0786157798716715e-06,
+      "loss": 0.6562,
+      "step": 11722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0770130555338686e-06,
+      "loss": 0.7231,
+      "step": 11723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.075410413289687e-06,
+      "loss": 0.6652,
+      "step": 11724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0738078531841335e-06,
+      "loss": 0.6974,
+      "step": 11725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072205375262218e-06,
+      "loss": 0.6931,
+      "step": 11726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070602979568948e-06,
+      "loss": 0.7491,
+      "step": 11727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0690006661493235e-06,
+      "loss": 0.3393,
+      "step": 11728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.067398435048349e-06,
+      "loss": 0.7332,
+      "step": 11729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065796286311024e-06,
+      "loss": 0.7365,
+      "step": 11730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064194219982345e-06,
+      "loss": 0.7385,
+      "step": 11731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.062592236107308e-06,
+      "loss": 0.6486,
+      "step": 11732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.060990334730906e-06,
+      "loss": 0.6942,
+      "step": 11733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059388515898129e-06,
+      "loss": 0.7057,
+      "step": 11734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.057786779653964e-06,
+      "loss": 0.6427,
+      "step": 11735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.056185126043393e-06,
+      "loss": 0.6709,
+      "step": 11736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.054583555111405e-06,
+      "loss": 0.6518,
+      "step": 11737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05298206690298e-06,
+      "loss": 0.6829,
+      "step": 11738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051380661463092e-06,
+      "loss": 0.3247,
+      "step": 11739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049779338836729e-06,
+      "loss": 0.7791,
+      "step": 11740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048178099068849e-06,
+      "loss": 0.7074,
+      "step": 11741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.046576942204435e-06,
+      "loss": 0.7245,
+      "step": 11742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.044975868288451e-06,
+      "loss": 0.7288,
+      "step": 11743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043374877365866e-06,
+      "loss": 0.7415,
+      "step": 11744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041773969481644e-06,
+      "loss": 0.6786,
+      "step": 11745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040173144680751e-06,
+      "loss": 0.7094,
+      "step": 11746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038572403008146e-06,
+      "loss": 0.6505,
+      "step": 11747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0369717445087785e-06,
+      "loss": 0.7037,
+      "step": 11748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035371169227612e-06,
+      "loss": 0.6667,
+      "step": 11749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033770677209597e-06,
+      "loss": 0.7097,
+      "step": 11750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.032170268499684e-06,
+      "loss": 0.6436,
+      "step": 11751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0305699431428225e-06,
+      "loss": 0.6903,
+      "step": 11752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028969701183961e-06,
+      "loss": 0.7094,
+      "step": 11753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027369542668037e-06,
+      "loss": 0.6488,
+      "step": 11754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0257694676399936e-06,
+      "loss": 0.7285,
+      "step": 11755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.024169476144772e-06,
+      "loss": 0.7654,
+      "step": 11756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022569568227307e-06,
+      "loss": 0.688,
+      "step": 11757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.020969743932535e-06,
+      "loss": 0.6971,
+      "step": 11758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019370003305383e-06,
+      "loss": 0.619,
+      "step": 11759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017770346390791e-06,
+      "loss": 0.7156,
+      "step": 11760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.016170773233675e-06,
+      "loss": 0.6826,
+      "step": 11761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014571283878963e-06,
+      "loss": 0.7008,
+      "step": 11762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012971878371579e-06,
+      "loss": 0.6524,
+      "step": 11763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011372556756441e-06,
+      "loss": 0.6996,
+      "step": 11764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00977331907847e-06,
+      "loss": 0.5864,
+      "step": 11765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.008174165382586e-06,
+      "loss": 0.6903,
+      "step": 11766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00657509571369e-06,
+      "loss": 0.7352,
+      "step": 11767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.004976110116697e-06,
+      "loss": 0.7836,
+      "step": 11768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003377208636518e-06,
+      "loss": 0.7441,
+      "step": 11769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001778391318059e-06,
+      "loss": 0.7039,
+      "step": 11770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000179658206223e-06,
+      "loss": 0.7431,
+      "step": 11771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.99858100934591e-06,
+      "loss": 0.623,
+      "step": 11772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996982444782021e-06,
+      "loss": 0.6758,
+      "step": 11773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9953839645594515e-06,
+      "loss": 0.7042,
+      "step": 11774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993785568723093e-06,
+      "loss": 0.6479,
+      "step": 11775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992187257317842e-06,
+      "loss": 0.651,
+      "step": 11776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990589030388585e-06,
+      "loss": 0.6602,
+      "step": 11777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988990887980209e-06,
+      "loss": 0.7073,
+      "step": 11778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.987392830137601e-06,
+      "loss": 0.7061,
+      "step": 11779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985794856905641e-06,
+      "loss": 0.7444,
+      "step": 11780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.984196968329209e-06,
+      "loss": 0.692,
+      "step": 11781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982599164453184e-06,
+      "loss": 0.713,
+      "step": 11782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.981001445322437e-06,
+      "loss": 0.6959,
+      "step": 11783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979403810981844e-06,
+      "loss": 0.6596,
+      "step": 11784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977806261476279e-06,
+      "loss": 0.711,
+      "step": 11785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976208796850608e-06,
+      "loss": 0.6602,
+      "step": 11786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974611417149692e-06,
+      "loss": 0.6726,
+      "step": 11787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9730141224183955e-06,
+      "loss": 0.6023,
+      "step": 11788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971416912701583e-06,
+      "loss": 0.6696,
+      "step": 11789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969819788044111e-06,
+      "loss": 0.6509,
+      "step": 11790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.968222748490836e-06,
+      "loss": 0.7341,
+      "step": 11791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.966625794086615e-06,
+      "loss": 0.6988,
+      "step": 11792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965028924876289e-06,
+      "loss": 0.7103,
+      "step": 11793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963432140904718e-06,
+      "loss": 0.651,
+      "step": 11794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961835442216743e-06,
+      "loss": 0.6692,
+      "step": 11795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960238828857207e-06,
+      "loss": 0.6938,
+      "step": 11796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.958642300870957e-06,
+      "loss": 0.7128,
+      "step": 11797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957045858302827e-06,
+      "loss": 0.7347,
+      "step": 11798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.95544950119766e-06,
+      "loss": 0.7101,
+      "step": 11799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953853229600284e-06,
+      "loss": 0.7464,
+      "step": 11800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952257043555534e-06,
+      "loss": 0.3524,
+      "step": 11801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950660943108239e-06,
+      "loss": 0.6408,
+      "step": 11802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9490649283032276e-06,
+      "loss": 0.6997,
+      "step": 11803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9474689991853205e-06,
+      "loss": 0.7586,
+      "step": 11804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.945873155799352e-06,
+      "loss": 0.6597,
+      "step": 11805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944277398190129e-06,
+      "loss": 0.6781,
+      "step": 11806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942681726402474e-06,
+      "loss": 0.6431,
+      "step": 11807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9410861404812016e-06,
+      "loss": 0.6429,
+      "step": 11808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939490640471125e-06,
+      "loss": 0.7034,
+      "step": 11809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.937895226417055e-06,
+      "loss": 0.6923,
+      "step": 11810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936299898363801e-06,
+      "loss": 0.7355,
+      "step": 11811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934704656356171e-06,
+      "loss": 0.661,
+      "step": 11812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9331095004389595e-06,
+      "loss": 0.7559,
+      "step": 11813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931514430656974e-06,
+      "loss": 0.6512,
+      "step": 11814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92991944705501e-06,
+      "loss": 0.7051,
+      "step": 11815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9283245496778655e-06,
+      "loss": 0.6812,
+      "step": 11816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926729738570333e-06,
+      "loss": 0.3295,
+      "step": 11817
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.925135013777207e-06,
+      "loss": 0.6729,
+      "step": 11818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92354037534327e-06,
+      "loss": 0.6785,
+      "step": 11819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9219458233133095e-06,
+      "loss": 0.7218,
+      "step": 11820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920351357732114e-06,
+      "loss": 0.726,
+      "step": 11821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.918756978644461e-06,
+      "loss": 0.7543,
+      "step": 11822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.917162686095129e-06,
+      "loss": 0.6472,
+      "step": 11823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9155684801288984e-06,
+      "loss": 0.6976,
+      "step": 11824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9139743607905405e-06,
+      "loss": 0.7844,
+      "step": 11825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9123803281248254e-06,
+      "loss": 0.7314,
+      "step": 11826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.910786382176527e-06,
+      "loss": 0.6925,
+      "step": 11827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9091925229904064e-06,
+      "loss": 0.6655,
+      "step": 11828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9075987506112305e-06,
+      "loss": 0.6519,
+      "step": 11829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906005065083763e-06,
+      "loss": 0.64,
+      "step": 11830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.904411466452767e-06,
+      "loss": 0.7339,
+      "step": 11831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90281795476299e-06,
+      "loss": 0.7359,
+      "step": 11832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90122453005919e-06,
+      "loss": 0.7231,
+      "step": 11833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899631192386121e-06,
+      "loss": 0.725,
+      "step": 11834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.898037941788534e-06,
+      "loss": 0.8203,
+      "step": 11835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896444778311173e-06,
+      "loss": 0.6763,
+      "step": 11836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894851701998784e-06,
+      "loss": 0.7026,
+      "step": 11837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8932587128961124e-06,
+      "loss": 0.7128,
+      "step": 11838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.891665811047894e-06,
+      "loss": 0.7029,
+      "step": 11839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890072996498867e-06,
+      "loss": 0.7115,
+      "step": 11840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888480269293766e-06,
+      "loss": 0.7598,
+      "step": 11841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886887629477326e-06,
+      "loss": 0.7115,
+      "step": 11842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.885295077094274e-06,
+      "loss": 0.6956,
+      "step": 11843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.883702612189344e-06,
+      "loss": 0.6592,
+      "step": 11844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882110234807253e-06,
+      "loss": 0.6475,
+      "step": 11845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.880517944992728e-06,
+      "loss": 0.6506,
+      "step": 11846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8789257427904875e-06,
+      "loss": 0.6876,
+      "step": 11847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877333628245253e-06,
+      "loss": 0.6491,
+      "step": 11848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.875741601401733e-06,
+      "loss": 0.7048,
+      "step": 11849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874149662304648e-06,
+      "loss": 0.7385,
+      "step": 11850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872557810998709e-06,
+      "loss": 0.7321,
+      "step": 11851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870966047528616e-06,
+      "loss": 0.6663,
+      "step": 11852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869374371939077e-06,
+      "loss": 0.625,
+      "step": 11853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8677827842747994e-06,
+      "loss": 0.6868,
+      "step": 11854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.866191284580479e-06,
+      "loss": 0.3271,
+      "step": 11855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.864599872900818e-06,
+      "loss": 0.6513,
+      "step": 11856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.863008549280513e-06,
+      "loss": 0.7274,
+      "step": 11857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8614173137642495e-06,
+      "loss": 0.7398,
+      "step": 11858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859826166396723e-06,
+      "loss": 0.6843,
+      "step": 11859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858235107222624e-06,
+      "loss": 0.6643,
+      "step": 11860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.856644136286634e-06,
+      "loss": 0.6957,
+      "step": 11861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855053253633439e-06,
+      "loss": 0.6134,
+      "step": 11862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853462459307718e-06,
+      "loss": 0.6983,
+      "step": 11863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851871753354154e-06,
+      "loss": 0.6881,
+      "step": 11864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850281135817415e-06,
+      "loss": 0.676,
+      "step": 11865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848690606742179e-06,
+      "loss": 0.7334,
+      "step": 11866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847100166173117e-06,
+      "loss": 0.7078,
+      "step": 11867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8455098141548945e-06,
+      "loss": 0.6287,
+      "step": 11868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8439195507321785e-06,
+      "loss": 0.6608,
+      "step": 11869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842329375949639e-06,
+      "loss": 0.6522,
+      "step": 11870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.840739289851927e-06,
+      "loss": 0.6316,
+      "step": 11871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839149292483707e-06,
+      "loss": 0.6423,
+      "step": 11872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837559383889631e-06,
+      "loss": 0.712,
+      "step": 11873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835969564114353e-06,
+      "loss": 0.6838,
+      "step": 11874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834379833202527e-06,
+      "loss": 0.637,
+      "step": 11875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832790191198799e-06,
+      "loss": 0.6836,
+      "step": 11876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8312006381478216e-06,
+      "loss": 0.6068,
+      "step": 11877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829611174094224e-06,
+      "loss": 0.68,
+      "step": 11878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828021799082658e-06,
+      "loss": 0.7096,
+      "step": 11879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82643251315776e-06,
+      "loss": 0.6725,
+      "step": 11880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824843316364165e-06,
+      "loss": 0.6864,
+      "step": 11881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823254208746507e-06,
+      "loss": 0.5997,
+      "step": 11882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.821665190349419e-06,
+      "loss": 0.7028,
+      "step": 11883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820076261217525e-06,
+      "loss": 0.6762,
+      "step": 11884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.818487421395452e-06,
+      "loss": 0.6907,
+      "step": 11885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.816898670927825e-06,
+      "loss": 0.6628,
+      "step": 11886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.815310009859264e-06,
+      "loss": 0.7285,
+      "step": 11887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813721438234389e-06,
+      "loss": 0.6877,
+      "step": 11888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812132956097814e-06,
+      "loss": 0.6725,
+      "step": 11889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.810544563494153e-06,
+      "loss": 0.6091,
+      "step": 11890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808956260468017e-06,
+      "loss": 0.6452,
+      "step": 11891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.807368047064014e-06,
+      "loss": 0.6896,
+      "step": 11892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805779923326748e-06,
+      "loss": 0.6626,
+      "step": 11893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.804191889300824e-06,
+      "loss": 0.673,
+      "step": 11894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.802603945030843e-06,
+      "loss": 0.6143,
+      "step": 11895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8010160905614095e-06,
+      "loss": 0.7169,
+      "step": 11896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799428325937107e-06,
+      "loss": 0.6909,
+      "step": 11897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797840651202533e-06,
+      "loss": 0.6085,
+      "step": 11898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.796253066402282e-06,
+      "loss": 0.632,
+      "step": 11899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.794665571580941e-06,
+      "loss": 0.6223,
+      "step": 11900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.793078166783092e-06,
+      "loss": 0.7199,
+      "step": 11901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791490852053322e-06,
+      "loss": 0.6692,
+      "step": 11902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7899036274362115e-06,
+      "loss": 0.6944,
+      "step": 11903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.788316492976337e-06,
+      "loss": 0.6933,
+      "step": 11904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786729448718272e-06,
+      "loss": 0.7475,
+      "step": 11905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785142494706594e-06,
+      "loss": 0.7026,
+      "step": 11906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78355563098587e-06,
+      "loss": 0.6925,
+      "step": 11907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781968857600668e-06,
+      "loss": 0.7288,
+      "step": 11908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780382174595558e-06,
+      "loss": 0.7213,
+      "step": 11909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.778795582015096e-06,
+      "loss": 0.6975,
+      "step": 11910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.777209079903846e-06,
+      "loss": 0.6093,
+      "step": 11911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7756226683063655e-06,
+      "loss": 0.6928,
+      "step": 11912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774036347267209e-06,
+      "loss": 0.6889,
+      "step": 11913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772450116830927e-06,
+      "loss": 0.653,
+      "step": 11914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770863977042075e-06,
+      "loss": 0.6799,
+      "step": 11915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.769277927945201e-06,
+      "loss": 0.6968,
+      "step": 11916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7676919695848445e-06,
+      "loss": 0.6374,
+      "step": 11917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766106102005547e-06,
+      "loss": 0.6569,
+      "step": 11918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7645203252518535e-06,
+      "loss": 0.694,
+      "step": 11919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.762934639368299e-06,
+      "loss": 0.6548,
+      "step": 11920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761349044399421e-06,
+      "loss": 0.7369,
+      "step": 11921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.759763540389752e-06,
+      "loss": 0.6237,
+      "step": 11922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758178127383813e-06,
+      "loss": 0.6098,
+      "step": 11923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756592805426141e-06,
+      "loss": 0.6835,
+      "step": 11924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755007574561255e-06,
+      "loss": 0.7168,
+      "step": 11925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753422434833681e-06,
+      "loss": 0.6559,
+      "step": 11926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751837386287936e-06,
+      "loss": 0.654,
+      "step": 11927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7502524289685375e-06,
+      "loss": 0.6438,
+      "step": 11928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748667562920002e-06,
+      "loss": 0.6556,
+      "step": 11929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.747082788186836e-06,
+      "loss": 0.6673,
+      "step": 11930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745498104813552e-06,
+      "loss": 0.6463,
+      "step": 11931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743913512844658e-06,
+      "loss": 0.6409,
+      "step": 11932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742329012324656e-06,
+      "loss": 0.6642,
+      "step": 11933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740744603298046e-06,
+      "loss": 0.669,
+      "step": 11934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739160285809336e-06,
+      "loss": 0.7105,
+      "step": 11935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73757605990301e-06,
+      "loss": 0.6491,
+      "step": 11936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7359919256235685e-06,
+      "loss": 0.698,
+      "step": 11937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734407883015502e-06,
+      "loss": 0.6964,
+      "step": 11938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732823932123296e-06,
+      "loss": 0.6729,
+      "step": 11939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731240072991441e-06,
+      "loss": 0.6668,
+      "step": 11940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729656305664418e-06,
+      "loss": 0.6606,
+      "step": 11941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728072630186715e-06,
+      "loss": 0.6913,
+      "step": 11942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726489046602795e-06,
+      "loss": 0.7137,
+      "step": 11943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.724905554957148e-06,
+      "loss": 0.6596,
+      "step": 11944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723322155294241e-06,
+      "loss": 0.6857,
+      "step": 11945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.721738847658546e-06,
+      "loss": 0.6862,
+      "step": 11946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720155632094529e-06,
+      "loss": 0.6895,
+      "step": 11947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718572508646659e-06,
+      "loss": 0.6437,
+      "step": 11948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716989477359399e-06,
+      "loss": 0.6317,
+      "step": 11949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715406538277205e-06,
+      "loss": 0.6672,
+      "step": 11950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713823691444536e-06,
+      "loss": 0.3319,
+      "step": 11951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.712240936905849e-06,
+      "loss": 0.665,
+      "step": 11952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7106582747055946e-06,
+      "loss": 0.7104,
+      "step": 11953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709075704888224e-06,
+      "loss": 0.7257,
+      "step": 11954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707493227498187e-06,
+      "loss": 0.7424,
+      "step": 11955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7059108425799226e-06,
+      "loss": 0.7138,
+      "step": 11956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704328550177877e-06,
+      "loss": 0.7136,
+      "step": 11957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.702746350336486e-06,
+      "loss": 0.6631,
+      "step": 11958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701164243100189e-06,
+      "loss": 0.7404,
+      "step": 11959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.69958222851342e-06,
+      "loss": 0.6488,
+      "step": 11960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6980003066206135e-06,
+      "loss": 0.7083,
+      "step": 11961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.696418477466199e-06,
+      "loss": 0.6822,
+      "step": 11962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694836741094593e-06,
+      "loss": 0.7599,
+      "step": 11963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693255097550231e-06,
+      "loss": 0.6038,
+      "step": 11964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691673546877527e-06,
+      "loss": 0.6783,
+      "step": 11965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690092089120903e-06,
+      "loss": 0.7267,
+      "step": 11966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.688510724324776e-06,
+      "loss": 0.6071,
+      "step": 11967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68692945253356e-06,
+      "loss": 0.3351,
+      "step": 11968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685348273791661e-06,
+      "loss": 0.6832,
+      "step": 11969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6837671881434885e-06,
+      "loss": 0.6417,
+      "step": 11970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682186195633451e-06,
+      "loss": 0.6628,
+      "step": 11971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680605296305951e-06,
+      "loss": 0.6708,
+      "step": 11972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679024490205387e-06,
+      "loss": 0.7039,
+      "step": 11973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.677443777376158e-06,
+      "loss": 0.6633,
+      "step": 11974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675863157862661e-06,
+      "loss": 0.7017,
+      "step": 11975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674282631709284e-06,
+      "loss": 0.6532,
+      "step": 11976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67270219896042e-06,
+      "loss": 0.6938,
+      "step": 11977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6711218596604555e-06,
+      "loss": 0.6881,
+      "step": 11978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669541613853774e-06,
+      "loss": 0.708,
+      "step": 11979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66796146158476e-06,
+      "loss": 0.7095,
+      "step": 11980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666381402897794e-06,
+      "loss": 0.6874,
+      "step": 11981
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664801437837249e-06,
+      "loss": 0.7089,
+      "step": 11982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663221566447501e-06,
+      "loss": 0.7335,
+      "step": 11983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.661641788772918e-06,
+      "loss": 0.7157,
+      "step": 11984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660062104857874e-06,
+      "loss": 0.6274,
+      "step": 11985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658482514746734e-06,
+      "loss": 0.738,
+      "step": 11986
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65690301848386e-06,
+      "loss": 0.3475,
+      "step": 11987
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.655323616113616e-06,
+      "loss": 0.6903,
+      "step": 11988
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.653744307680357e-06,
+      "loss": 0.687,
+      "step": 11989
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65216509322844e-06,
+      "loss": 0.6468,
+      "step": 11990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650585972802217e-06,
+      "loss": 0.702,
+      "step": 11991
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64900694644604e-06,
+      "loss": 0.6274,
+      "step": 11992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6474280142042555e-06,
+      "loss": 0.6863,
+      "step": 11993
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645849176121213e-06,
+      "loss": 0.6699,
+      "step": 11994
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644270432241249e-06,
+      "loss": 0.7074,
+      "step": 11995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.642691782608704e-06,
+      "loss": 0.7358,
+      "step": 11996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641113227267919e-06,
+      "loss": 0.6793,
+      "step": 11997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.639534766263225e-06,
+      "loss": 0.6391,
+      "step": 11998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6379563996389565e-06,
+      "loss": 0.6583,
+      "step": 11999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.636378127439438e-06,
+      "loss": 0.6604,
+      "step": 12000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634799949709007e-06,
+      "loss": 0.6132,
+      "step": 12001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.633221866491976e-06,
+      "loss": 0.6652,
+      "step": 12002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63164387783267e-06,
+      "loss": 0.6606,
+      "step": 12003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630065983775406e-06,
+      "loss": 0.6631,
+      "step": 12004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.628488184364504e-06,
+      "loss": 0.6433,
+      "step": 12005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.626910479644277e-06,
+      "loss": 0.7188,
+      "step": 12006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.625332869659036e-06,
+      "loss": 0.6872,
+      "step": 12007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623755354453082e-06,
+      "loss": 0.6927,
+      "step": 12008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622177934070727e-06,
+      "loss": 0.7573,
+      "step": 12009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620600608556272e-06,
+      "loss": 0.6343,
+      "step": 12010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619023377954018e-06,
+      "loss": 0.6194,
+      "step": 12011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617446242308261e-06,
+      "loss": 0.7374,
+      "step": 12012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615869201663296e-06,
+      "loss": 0.6819,
+      "step": 12013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614292256063417e-06,
+      "loss": 0.6971,
+      "step": 12014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.612715405552909e-06,
+      "loss": 0.6397,
+      "step": 12015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611138650176061e-06,
+      "loss": 0.6966,
+      "step": 12016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609561989977158e-06,
+      "loss": 0.659,
+      "step": 12017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60798542500048e-06,
+      "loss": 0.6828,
+      "step": 12018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6064089552903065e-06,
+      "loss": 0.6817,
+      "step": 12019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604832580890917e-06,
+      "loss": 0.7129,
+      "step": 12020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603256301846575e-06,
+      "loss": 0.7191,
+      "step": 12021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601680118201559e-06,
+      "loss": 0.7141,
+      "step": 12022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600104030000135e-06,
+      "loss": 0.617,
+      "step": 12023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598528037286566e-06,
+      "loss": 0.7618,
+      "step": 12024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596952140105119e-06,
+      "loss": 0.5928,
+      "step": 12025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.595376338500051e-06,
+      "loss": 0.6556,
+      "step": 12026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5938006325156246e-06,
+      "loss": 0.7722,
+      "step": 12027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5922250221960836e-06,
+      "loss": 0.6601,
+      "step": 12028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590649507585687e-06,
+      "loss": 0.6439,
+      "step": 12029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589074088728684e-06,
+      "loss": 0.6732,
+      "step": 12030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.587498765669319e-06,
+      "loss": 0.6451,
+      "step": 12031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585923538451837e-06,
+      "loss": 0.6984,
+      "step": 12032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584348407120481e-06,
+      "loss": 0.6654,
+      "step": 12033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582773371719484e-06,
+      "loss": 0.6844,
+      "step": 12034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581198432293087e-06,
+      "loss": 0.6899,
+      "step": 12035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.57962358888552e-06,
+      "loss": 0.6554,
+      "step": 12036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578048841541013e-06,
+      "loss": 0.6531,
+      "step": 12037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5764741903037945e-06,
+      "loss": 0.7089,
+      "step": 12038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.574899635218091e-06,
+      "loss": 0.6263,
+      "step": 12039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5733251763281246e-06,
+      "loss": 0.6481,
+      "step": 12040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.571750813678111e-06,
+      "loss": 0.7139,
+      "step": 12041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5701765473122705e-06,
+      "loss": 0.6496,
+      "step": 12042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568602377274816e-06,
+      "loss": 0.3561,
+      "step": 12043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.567028303609957e-06,
+      "loss": 0.5972,
+      "step": 12044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5654543263619065e-06,
+      "loss": 0.7001,
+      "step": 12045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.563880445574873e-06,
+      "loss": 0.6912,
+      "step": 12046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56230666129305e-06,
+      "loss": 0.6781,
+      "step": 12047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.560732973560643e-06,
+      "loss": 0.6954,
+      "step": 12048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559159382421848e-06,
+      "loss": 0.7288,
+      "step": 12049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557585887920864e-06,
+      "loss": 0.6904,
+      "step": 12050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.556012490101882e-06,
+      "loss": 0.7006,
+      "step": 12051
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55443918900909e-06,
+      "loss": 0.6675,
+      "step": 12052
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55286598468668e-06,
+      "loss": 0.6823,
+      "step": 12053
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5512928771788296e-06,
+      "loss": 0.6986,
+      "step": 12054
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549719866529723e-06,
+      "loss": 0.7603,
+      "step": 12055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5481469527835395e-06,
+      "loss": 0.7957,
+      "step": 12056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.546574135984455e-06,
+      "loss": 0.6519,
+      "step": 12057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545001416176642e-06,
+      "loss": 0.653,
+      "step": 12058
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543428793404274e-06,
+      "loss": 0.6857,
+      "step": 12059
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541856267711516e-06,
+      "loss": 0.7165,
+      "step": 12060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.540283839142534e-06,
+      "loss": 0.6418,
+      "step": 12061
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5387115077414885e-06,
+      "loss": 0.7078,
+      "step": 12062
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.537139273552544e-06,
+      "loss": 0.6678,
+      "step": 12063
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535567136619854e-06,
+      "loss": 0.6703,
+      "step": 12064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5339950969875706e-06,
+      "loss": 0.6627,
+      "step": 12065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532423154699855e-06,
+      "loss": 0.6523,
+      "step": 12066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.530851309800845e-06,
+      "loss": 0.6774,
+      "step": 12067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.529279562334692e-06,
+      "loss": 0.6944,
+      "step": 12068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527707912345535e-06,
+      "loss": 0.8023,
+      "step": 12069
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526136359877521e-06,
+      "loss": 0.6291,
+      "step": 12070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.524564904974786e-06,
+      "loss": 0.7786,
+      "step": 12071
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5229935476814664e-06,
+      "loss": 0.6201,
+      "step": 12072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521422288041688e-06,
+      "loss": 0.6045,
+      "step": 12073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519851126099586e-06,
+      "loss": 0.7299,
+      "step": 12074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518280061899285e-06,
+      "loss": 0.6718,
+      "step": 12075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.51670909548491e-06,
+      "loss": 0.692,
+      "step": 12076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515138226900583e-06,
+      "loss": 0.6554,
+      "step": 12077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5135674561904235e-06,
+      "loss": 0.6419,
+      "step": 12078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511996783398547e-06,
+      "loss": 0.6739,
+      "step": 12079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510426208569065e-06,
+      "loss": 0.6562,
+      "step": 12080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508855731746088e-06,
+      "loss": 0.6759,
+      "step": 12081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507285352973724e-06,
+      "loss": 0.662,
+      "step": 12082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5057150722960785e-06,
+      "loss": 0.307,
+      "step": 12083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504144889757253e-06,
+      "loss": 0.6938,
+      "step": 12084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502574805401351e-06,
+      "loss": 0.7544,
+      "step": 12085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501004819272462e-06,
+      "loss": 0.7063,
+      "step": 12086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499434931414683e-06,
+      "loss": 0.6175,
+      "step": 12087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497865141872107e-06,
+      "loss": 0.6562,
+      "step": 12088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.496295450688818e-06,
+      "loss": 0.6096,
+      "step": 12089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494725857908907e-06,
+      "loss": 0.6515,
+      "step": 12090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4931563635764535e-06,
+      "loss": 0.3291,
+      "step": 12091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.491586967735544e-06,
+      "loss": 0.6885,
+      "step": 12092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490017670430242e-06,
+      "loss": 0.7396,
+      "step": 12093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488448471704634e-06,
+      "loss": 0.6452,
+      "step": 12094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486879371602788e-06,
+      "loss": 0.7529,
+      "step": 12095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.485310370168771e-06,
+      "loss": 0.6769,
+      "step": 12096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483741467446654e-06,
+      "loss": 0.6006,
+      "step": 12097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482172663480497e-06,
+      "loss": 0.6774,
+      "step": 12098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48060395831436e-06,
+      "loss": 0.7519,
+      "step": 12099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4790353519923035e-06,
+      "loss": 0.6597,
+      "step": 12100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47746684455838e-06,
+      "loss": 0.7657,
+      "step": 12101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4758984360566444e-06,
+      "loss": 0.6603,
+      "step": 12102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474330126531145e-06,
+      "loss": 0.6593,
+      "step": 12103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472761916025927e-06,
+      "loss": 0.7396,
+      "step": 12104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4711938045850405e-06,
+      "loss": 0.3151,
+      "step": 12105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4696257922525186e-06,
+      "loss": 0.7089,
+      "step": 12106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4680578790724045e-06,
+      "loss": 0.6394,
+      "step": 12107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.466490065088733e-06,
+      "loss": 0.6923,
+      "step": 12108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464922350345534e-06,
+      "loss": 0.6948,
+      "step": 12109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463354734886842e-06,
+      "loss": 0.62,
+      "step": 12110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461787218756687e-06,
+      "loss": 0.7153,
+      "step": 12111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460219801999086e-06,
+      "loss": 0.6845,
+      "step": 12112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458652484658063e-06,
+      "loss": 0.6398,
+      "step": 12113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457085266777637e-06,
+      "loss": 0.7067,
+      "step": 12114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.455518148401827e-06,
+      "loss": 0.668,
+      "step": 12115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 0.68,
+      "step": 12116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452384210340099e-06,
+      "loss": 0.6489,
+      "step": 12117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4508173907422015e-06,
+      "loss": 0.7289,
+      "step": 12118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.449250670824951e-06,
+      "loss": 0.7073,
+      "step": 12119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447684050632355e-06,
+      "loss": 0.6732,
+      "step": 12120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4461175302084115e-06,
+      "loss": 0.6646,
+      "step": 12121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444551109597115e-06,
+      "loss": 0.6309,
+      "step": 12122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.442984788842463e-06,
+      "loss": 0.6721,
+      "step": 12123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441418567988444e-06,
+      "loss": 0.6062,
+      "step": 12124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.439852447079046e-06,
+      "loss": 0.7396,
+      "step": 12125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438286426158253e-06,
+      "loss": 0.6296,
+      "step": 12126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.436720505270051e-06,
+      "loss": 0.3198,
+      "step": 12127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.435154684458418e-06,
+      "loss": 0.6852,
+      "step": 12128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43358896376733e-06,
+      "loss": 0.7409,
+      "step": 12129
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432023343240761e-06,
+      "loss": 0.6974,
+      "step": 12130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4304578229226884e-06,
+      "loss": 0.6756,
+      "step": 12131
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428892402857072e-06,
+      "loss": 0.6757,
+      "step": 12132
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42732708308788e-06,
+      "loss": 0.6008,
+      "step": 12133
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4257618636590744e-06,
+      "loss": 0.6567,
+      "step": 12134
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.42419674461462e-06,
+      "loss": 0.6367,
+      "step": 12135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4226317259984705e-06,
+      "loss": 0.7286,
+      "step": 12136
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421066807854585e-06,
+      "loss": 0.7309,
+      "step": 12137
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419501990226904e-06,
+      "loss": 0.6527,
+      "step": 12138
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4179372731593845e-06,
+      "loss": 0.6624,
+      "step": 12139
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.416372656695973e-06,
+      "loss": 0.7247,
+      "step": 12140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414808140880609e-06,
+      "loss": 0.7165,
+      "step": 12141
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413243725757235e-06,
+      "loss": 0.6582,
+      "step": 12142
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4116794113697865e-06,
+      "loss": 0.7178,
+      "step": 12143
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4101151977622015e-06,
+      "loss": 0.7445,
+      "step": 12144
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408551084978407e-06,
+      "loss": 0.7087,
+      "step": 12145
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.406987073062336e-06,
+      "loss": 0.6616,
+      "step": 12146
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405423162057912e-06,
+      "loss": 0.6819,
+      "step": 12147
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.403859352009059e-06,
+      "loss": 0.6624,
+      "step": 12148
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402295642959698e-06,
+      "loss": 0.6876,
+      "step": 12149
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400732034953749e-06,
+      "loss": 0.6336,
+      "step": 12150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.399168528035121e-06,
+      "loss": 0.3295,
+      "step": 12151
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39760512224773e-06,
+      "loss": 0.5856,
+      "step": 12152
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396041817635482e-06,
+      "loss": 0.6539,
+      "step": 12153
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.394478614242285e-06,
+      "loss": 0.6402,
+      "step": 12154
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392915512112044e-06,
+      "loss": 0.5965,
+      "step": 12155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3913525112886575e-06,
+      "loss": 0.6542,
+      "step": 12156
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389789611816029e-06,
+      "loss": 0.674,
+      "step": 12157
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.388226813738043e-06,
+      "loss": 0.7424,
+      "step": 12158
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.386664117098596e-06,
+      "loss": 0.6808,
+      "step": 12159
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385101521941579e-06,
+      "loss": 0.6718,
+      "step": 12160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383539028310878e-06,
+      "loss": 0.7275,
+      "step": 12161
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381976636250374e-06,
+      "loss": 0.6949,
+      "step": 12162
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3804143458039535e-06,
+      "loss": 0.7875,
+      "step": 12163
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.378852157015486e-06,
+      "loss": 0.7018,
+      "step": 12164
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377290069928851e-06,
+      "loss": 0.7142,
+      "step": 12165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37572808458792e-06,
+      "loss": 0.3231,
+      "step": 12166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374166201036561e-06,
+      "loss": 0.6669,
+      "step": 12167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372604419318642e-06,
+      "loss": 0.6812,
+      "step": 12168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371042739478026e-06,
+      "loss": 0.6547,
+      "step": 12169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.369481161558576e-06,
+      "loss": 0.7452,
+      "step": 12170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367919685604144e-06,
+      "loss": 0.6647,
+      "step": 12171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366358311658589e-06,
+      "loss": 0.684,
+      "step": 12172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364797039765762e-06,
+      "loss": 0.6983,
+      "step": 12173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36323586996951e-06,
+      "loss": 0.7289,
+      "step": 12174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.361674802313682e-06,
+      "loss": 0.5985,
+      "step": 12175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360113836842125e-06,
+      "loss": 0.6519,
+      "step": 12176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358552973598673e-06,
+      "loss": 0.7039,
+      "step": 12177
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3569922126271665e-06,
+      "loss": 0.6612,
+      "step": 12178
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3554315539714364e-06,
+      "loss": 0.7107,
+      "step": 12179
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35387099767532e-06,
+      "loss": 0.6237,
+      "step": 12180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352310543782645e-06,
+      "loss": 0.6199,
+      "step": 12181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350750192337236e-06,
+      "loss": 0.69,
+      "step": 12182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34918994338292e-06,
+      "loss": 0.679,
+      "step": 12183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.347629796963513e-06,
+      "loss": 0.7831,
+      "step": 12184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3460697531228325e-06,
+      "loss": 0.7488,
+      "step": 12185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.344509811904696e-06,
+      "loss": 0.682,
+      "step": 12186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3429499733529135e-06,
+      "loss": 0.6861,
+      "step": 12187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3413902375112945e-06,
+      "loss": 0.6878,
+      "step": 12188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.339830604423647e-06,
+      "loss": 0.6761,
+      "step": 12189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338271074133771e-06,
+      "loss": 0.6566,
+      "step": 12190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336711646685467e-06,
+      "loss": 0.6931,
+      "step": 12191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335152322122533e-06,
+      "loss": 0.5988,
+      "step": 12192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333593100488764e-06,
+      "loss": 0.654,
+      "step": 12193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332033981827949e-06,
+      "loss": 0.6635,
+      "step": 12194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33047496618388e-06,
+      "loss": 0.652,
+      "step": 12195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.328916053600347e-06,
+      "loss": 0.6301,
+      "step": 12196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3273572441211216e-06,
+      "loss": 0.673,
+      "step": 12197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325798537789991e-06,
+      "loss": 0.6685,
+      "step": 12198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324239934650729e-06,
+      "loss": 0.6472,
+      "step": 12199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3226814347471125e-06,
+      "loss": 0.6823,
+      "step": 12200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3211230381229115e-06,
+      "loss": 0.7546,
+      "step": 12201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3195647448219e-06,
+      "loss": 0.6401,
+      "step": 12202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318006554887832e-06,
+      "loss": 0.6578,
+      "step": 12203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31644846836448e-06,
+      "loss": 0.6429,
+      "step": 12204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.314890485295598e-06,
+      "loss": 0.6898,
+      "step": 12205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313332605724946e-06,
+      "loss": 0.6594,
+      "step": 12206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311774829696276e-06,
+      "loss": 0.7296,
+      "step": 12207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3102171572533395e-06,
+      "loss": 0.6287,
+      "step": 12208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308659588439887e-06,
+      "loss": 0.675,
+      "step": 12209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30710212329966e-06,
+      "loss": 0.715,
+      "step": 12210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305544761876401e-06,
+      "loss": 0.7153,
+      "step": 12211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303987504213851e-06,
+      "loss": 0.6634,
+      "step": 12212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302430350355746e-06,
+      "loss": 0.7277,
+      "step": 12213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300873300345819e-06,
+      "loss": 0.7168,
+      "step": 12214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2993163542278025e-06,
+      "loss": 0.6409,
+      "step": 12215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.29775951204542e-06,
+      "loss": 0.6734,
+      "step": 12216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2962027738424e-06,
+      "loss": 0.6963,
+      "step": 12217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.294646139662462e-06,
+      "loss": 0.6927,
+      "step": 12218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293089609549325e-06,
+      "loss": 0.6901,
+      "step": 12219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291533183546708e-06,
+      "loss": 0.6589,
+      "step": 12220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.289976861698319e-06,
+      "loss": 0.6952,
+      "step": 12221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288420644047878e-06,
+      "loss": 0.6596,
+      "step": 12222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286864530639081e-06,
+      "loss": 0.6217,
+      "step": 12223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285308521515633e-06,
+      "loss": 0.6166,
+      "step": 12224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.283752616721242e-06,
+      "loss": 0.6819,
+      "step": 12225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2821968162996015e-06,
+      "loss": 0.6811,
+      "step": 12226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280641120294411e-06,
+      "loss": 0.6939,
+      "step": 12227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2790855287493605e-06,
+      "loss": 0.6865,
+      "step": 12228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277530041708139e-06,
+      "loss": 0.6791,
+      "step": 12229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275974659214433e-06,
+      "loss": 0.6832,
+      "step": 12230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274419381311928e-06,
+      "loss": 0.7211,
+      "step": 12231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2728642080443035e-06,
+      "loss": 0.7182,
+      "step": 12232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2713091394552385e-06,
+      "loss": 0.6798,
+      "step": 12233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2697541755884075e-06,
+      "loss": 0.689,
+      "step": 12234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268199316487482e-06,
+      "loss": 0.6987,
+      "step": 12235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266644562196131e-06,
+      "loss": 0.6292,
+      "step": 12236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265089912758021e-06,
+      "loss": 0.6497,
+      "step": 12237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263535368216816e-06,
+      "loss": 0.6558,
+      "step": 12238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261980928616172e-06,
+      "loss": 0.6772,
+      "step": 12239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26042659399975e-06,
+      "loss": 0.6462,
+      "step": 12240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2588723644112095e-06,
+      "loss": 0.6909,
+      "step": 12241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2573182398941915e-06,
+      "loss": 0.6664,
+      "step": 12242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255764220492349e-06,
+      "loss": 0.7071,
+      "step": 12243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254210306249325e-06,
+      "loss": 0.6938,
+      "step": 12244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252656497208767e-06,
+      "loss": 0.6623,
+      "step": 12245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.251102793414312e-06,
+      "loss": 0.6382,
+      "step": 12246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249549194909597e-06,
+      "loss": 0.7316,
+      "step": 12247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247995701738257e-06,
+      "loss": 0.6913,
+      "step": 12248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246442313943917e-06,
+      "loss": 0.6857,
+      "step": 12249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2448890315702106e-06,
+      "loss": 0.7074,
+      "step": 12250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243335854660759e-06,
+      "loss": 0.6644,
+      "step": 12251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.241782783259187e-06,
+      "loss": 0.7374,
+      "step": 12252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240229817409111e-06,
+      "loss": 0.6664,
+      "step": 12253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238676957154151e-06,
+      "loss": 0.6936,
+      "step": 12254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237124202537913e-06,
+      "loss": 0.6713,
+      "step": 12255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235571553604012e-06,
+      "loss": 0.7441,
+      "step": 12256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2340190103960515e-06,
+      "loss": 0.6405,
+      "step": 12257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232466572957638e-06,
+      "loss": 0.6876,
+      "step": 12258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230914241332374e-06,
+      "loss": 0.6747,
+      "step": 12259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.229362015563852e-06,
+      "loss": 0.7187,
+      "step": 12260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.227809895695678e-06,
+      "loss": 0.6779,
+      "step": 12261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226257881771431e-06,
+      "loss": 0.6567,
+      "step": 12262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224705973834708e-06,
+      "loss": 0.6296,
+      "step": 12263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223154171929089e-06,
+      "loss": 0.655,
+      "step": 12264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221602476098164e-06,
+      "loss": 0.604,
+      "step": 12265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220050886385512e-06,
+      "loss": 0.6506,
+      "step": 12266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21849940283471e-06,
+      "loss": 0.7344,
+      "step": 12267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.216948025489326e-06,
+      "loss": 0.7683,
+      "step": 12268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2153967543929374e-06,
+      "loss": 0.6979,
+      "step": 12269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213845589589111e-06,
+      "loss": 0.7445,
+      "step": 12270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212294531121414e-06,
+      "loss": 0.6991,
+      "step": 12271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.210743579033405e-06,
+      "loss": 0.6699,
+      "step": 12272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209192733368646e-06,
+      "loss": 0.746,
+      "step": 12273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.207641994170694e-06,
+      "loss": 0.6048,
+      "step": 12274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206091361483099e-06,
+      "loss": 0.6343,
+      "step": 12275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204540835349415e-06,
+      "loss": 0.7262,
+      "step": 12276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202990415813186e-06,
+      "loss": 0.7186,
+      "step": 12277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201440102917957e-06,
+      "loss": 0.6617,
+      "step": 12278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199889896707272e-06,
+      "loss": 0.6494,
+      "step": 12279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1983397972246686e-06,
+      "loss": 0.6991,
+      "step": 12280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1967898045136784e-06,
+      "loss": 0.5481,
+      "step": 12281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.195239918617837e-06,
+      "loss": 0.7338,
+      "step": 12282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193690139580674e-06,
+      "loss": 0.6565,
+      "step": 12283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192140467445712e-06,
+      "loss": 0.6375,
+      "step": 12284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.190590902256479e-06,
+      "loss": 0.6501,
+      "step": 12285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189041444056495e-06,
+      "loss": 0.6165,
+      "step": 12286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187492092889279e-06,
+      "loss": 0.691,
+      "step": 12287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.185942848798339e-06,
+      "loss": 0.6321,
+      "step": 12288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.184393711827187e-06,
+      "loss": 0.6837,
+      "step": 12289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1828446820193365e-06,
+      "loss": 0.6844,
+      "step": 12290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18129575941829e-06,
+      "loss": 0.7012,
+      "step": 12291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17974694406755e-06,
+      "loss": 0.6852,
+      "step": 12292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.178198236010619e-06,
+      "loss": 0.7546,
+      "step": 12293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.176649635290988e-06,
+      "loss": 0.7387,
+      "step": 12294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175101141952151e-06,
+      "loss": 0.6534,
+      "step": 12295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1735527560376e-06,
+      "loss": 0.7254,
+      "step": 12296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.172004477590823e-06,
+      "loss": 0.657,
+      "step": 12297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170456306655302e-06,
+      "loss": 0.3225,
+      "step": 12298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168908243274521e-06,
+      "loss": 0.6473,
+      "step": 12299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167360287491957e-06,
+      "loss": 0.6567,
+      "step": 12300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165812439351082e-06,
+      "loss": 0.6852,
+      "step": 12301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164264698895372e-06,
+      "loss": 0.7003,
+      "step": 12302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1627170661682954e-06,
+      "loss": 0.7086,
+      "step": 12303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161169541213317e-06,
+      "loss": 0.5728,
+      "step": 12304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159622124073898e-06,
+      "loss": 0.6711,
+      "step": 12305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1580748147935046e-06,
+      "loss": 0.6742,
+      "step": 12306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156527613415593e-06,
+      "loss": 0.2911,
+      "step": 12307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.15498051998361e-06,
+      "loss": 0.6366,
+      "step": 12308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153433534541009e-06,
+      "loss": 0.7393,
+      "step": 12309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151886657131243e-06,
+      "loss": 0.6677,
+      "step": 12310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150339887797752e-06,
+      "loss": 0.7215,
+      "step": 12311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14879322658398e-06,
+      "loss": 0.6548,
+      "step": 12312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1472466735333694e-06,
+      "loss": 0.6493,
+      "step": 12313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145700228689347e-06,
+      "loss": 0.688,
+      "step": 12314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144153892095352e-06,
+      "loss": 0.6963,
+      "step": 12315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142607663794811e-06,
+      "loss": 0.7316,
+      "step": 12316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141061543831153e-06,
+      "loss": 0.6914,
+      "step": 12317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1395155322478e-06,
+      "loss": 0.6458,
+      "step": 12318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.137969629088174e-06,
+      "loss": 0.7169,
+      "step": 12319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136423834395692e-06,
+      "loss": 0.6073,
+      "step": 12320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.134878148213768e-06,
+      "loss": 0.6448,
+      "step": 12321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133332570585813e-06,
+      "loss": 0.7597,
+      "step": 12322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131787101555235e-06,
+      "loss": 0.6419,
+      "step": 12323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1302417411654415e-06,
+      "loss": 0.7279,
+      "step": 12324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.128696489459831e-06,
+      "loss": 0.5975,
+      "step": 12325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127151346481811e-06,
+      "loss": 0.6798,
+      "step": 12326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125606312274768e-06,
+      "loss": 0.6716,
+      "step": 12327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.124061386882098e-06,
+      "loss": 0.6398,
+      "step": 12328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12251657034719e-06,
+      "loss": 0.6399,
+      "step": 12329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1209718627134365e-06,
+      "loss": 0.7214,
+      "step": 12330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119427264024216e-06,
+      "loss": 0.6838,
+      "step": 12331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117882774322912e-06,
+      "loss": 0.6539,
+      "step": 12332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.116338393652905e-06,
+      "loss": 0.68,
+      "step": 12333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114794122057561e-06,
+      "loss": 0.5985,
+      "step": 12334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113249959580259e-06,
+      "loss": 0.6976,
+      "step": 12335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111705906264364e-06,
+      "loss": 0.7197,
+      "step": 12336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.110161962153245e-06,
+      "loss": 0.6652,
+      "step": 12337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.108618127290263e-06,
+      "loss": 0.6609,
+      "step": 12338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.107074401718776e-06,
+      "loss": 0.693,
+      "step": 12339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105530785482141e-06,
+      "loss": 0.7508,
+      "step": 12340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.103987278623712e-06,
+      "loss": 0.7145,
+      "step": 12341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102443881186838e-06,
+      "loss": 0.6808,
+      "step": 12342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100900593214867e-06,
+      "loss": 0.6667,
+      "step": 12343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099357414751144e-06,
+      "loss": 0.6612,
+      "step": 12344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.097814345839005e-06,
+      "loss": 0.6045,
+      "step": 12345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096271386521799e-06,
+      "loss": 0.6573,
+      "step": 12346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094728536842849e-06,
+      "loss": 0.6953,
+      "step": 12347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0931857968454914e-06,
+      "loss": 0.6732,
+      "step": 12348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091643166573055e-06,
+      "loss": 0.7346,
+      "step": 12349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.090100646068864e-06,
+      "loss": 0.7144,
+      "step": 12350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0885582353762425e-06,
+      "loss": 0.6229,
+      "step": 12351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.087015934538515e-06,
+      "loss": 0.6566,
+      "step": 12352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085473743598988e-06,
+      "loss": 0.6515,
+      "step": 12353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083931662600977e-06,
+      "loss": 0.6642,
+      "step": 12354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0823896915877955e-06,
+      "loss": 0.6748,
+      "step": 12355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08084783060275e-06,
+      "loss": 0.6624,
+      "step": 12356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0793060796891436e-06,
+      "loss": 0.7174,
+      "step": 12357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077764438890277e-06,
+      "loss": 0.6525,
+      "step": 12358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07622290824945e-06,
+      "loss": 0.7258,
+      "step": 12359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.074681487809955e-06,
+      "loss": 0.7154,
+      "step": 12360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.073140177615083e-06,
+      "loss": 0.3307,
+      "step": 12361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.071598977708124e-06,
+      "loss": 0.5858,
+      "step": 12362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0700578881323636e-06,
+      "loss": 0.7332,
+      "step": 12363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0685169089310835e-06,
+      "loss": 0.664,
+      "step": 12364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066976040147565e-06,
+      "loss": 0.678,
+      "step": 12365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06543528182508e-06,
+      "loss": 0.7466,
+      "step": 12366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063894634006904e-06,
+      "loss": 0.6844,
+      "step": 12367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062354096736307e-06,
+      "loss": 0.6732,
+      "step": 12368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060813670056556e-06,
+      "loss": 0.7065,
+      "step": 12369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059273354010911e-06,
+      "loss": 0.6455,
+      "step": 12370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.057733148642637e-06,
+      "loss": 0.6932,
+      "step": 12371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056193053994994e-06,
+      "loss": 0.67,
+      "step": 12372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054653070111231e-06,
+      "loss": 0.7627,
+      "step": 12373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.053113197034597e-06,
+      "loss": 0.7212,
+      "step": 12374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.051573434808346e-06,
+      "loss": 0.6767,
+      "step": 12375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050033783475722e-06,
+      "loss": 0.6618,
+      "step": 12376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048494243079965e-06,
+      "loss": 0.3126,
+      "step": 12377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046954813664316e-06,
+      "loss": 0.6708,
+      "step": 12378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045415495272009e-06,
+      "loss": 0.7033,
+      "step": 12379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.043876287946276e-06,
+      "loss": 0.6404,
+      "step": 12380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042337191730348e-06,
+      "loss": 0.7153,
+      "step": 12381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040798206667451e-06,
+      "loss": 0.6195,
+      "step": 12382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039259332800806e-06,
+      "loss": 0.6383,
+      "step": 12383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037720570173637e-06,
+      "loss": 0.6572,
+      "step": 12384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03618191882916e-06,
+      "loss": 0.5909,
+      "step": 12385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034643378810586e-06,
+      "loss": 0.6498,
+      "step": 12386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033104950161127e-06,
+      "loss": 0.6474,
+      "step": 12387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031566632923992e-06,
+      "loss": 0.6137,
+      "step": 12388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0300284271423834e-06,
+      "loss": 0.6616,
+      "step": 12389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.028490332859502e-06,
+      "loss": 0.6687,
+      "step": 12390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0269523501185535e-06,
+      "loss": 0.6566,
+      "step": 12391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025414478962723e-06,
+      "loss": 0.6672,
+      "step": 12392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.023876719435206e-06,
+      "loss": 0.6709,
+      "step": 12393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022339071579191e-06,
+      "loss": 0.6787,
+      "step": 12394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.020801535437865e-06,
+      "loss": 0.6826,
+      "step": 12395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019264111054413e-06,
+      "loss": 0.6578,
+      "step": 12396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0177267984720074e-06,
+      "loss": 0.6148,
+      "step": 12397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016189597733835e-06,
+      "loss": 0.6658,
+      "step": 12398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014652508883054e-06,
+      "loss": 0.7597,
+      "step": 12399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.013115531962845e-06,
+      "loss": 0.6837,
+      "step": 12400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011578667016373e-06,
+      "loss": 0.6243,
+      "step": 12401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010041914086801e-06,
+      "loss": 0.6708,
+      "step": 12402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008505273217289e-06,
+      "loss": 0.6737,
+      "step": 12403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.006968744450996e-06,
+      "loss": 0.7017,
+      "step": 12404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.005432327831074e-06,
+      "loss": 0.6904,
+      "step": 12405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.003896023400675e-06,
+      "loss": 0.6299,
+      "step": 12406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.002359831202947e-06,
+      "loss": 0.742,
+      "step": 12407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000823751281035e-06,
+      "loss": 0.7726,
+      "step": 12408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.99928778367808e-06,
+      "loss": 0.7077,
+      "step": 12409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.99775192843722e-06,
+      "loss": 0.6733,
+      "step": 12410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996216185601596e-06,
+      "loss": 0.719,
+      "step": 12411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994680555214332e-06,
+      "loss": 0.6756,
+      "step": 12412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.99314503731856e-06,
+      "loss": 0.7017,
+      "step": 12413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991609631957405e-06,
+      "loss": 0.6503,
+      "step": 12414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9900743391739905e-06,
+      "loss": 0.6651,
+      "step": 12415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988539159011437e-06,
+      "loss": 0.6669,
+      "step": 12416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.987004091512865e-06,
+      "loss": 0.6597,
+      "step": 12417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985469136721378e-06,
+      "loss": 0.7242,
+      "step": 12418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983934294680089e-06,
+      "loss": 0.7424,
+      "step": 12419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9823995654321085e-06,
+      "loss": 0.6442,
+      "step": 12420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9808649490205384e-06,
+      "loss": 0.6821,
+      "step": 12421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9793304454884795e-06,
+      "loss": 0.6838,
+      "step": 12422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.977796054879028e-06,
+      "loss": 0.6663,
+      "step": 12423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976261777235282e-06,
+      "loss": 0.582,
+      "step": 12424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9747276126003265e-06,
+      "loss": 0.6434,
+      "step": 12425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.973193561017252e-06,
+      "loss": 0.6737,
+      "step": 12426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9716596225291425e-06,
+      "loss": 0.7177,
+      "step": 12427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9701257971790815e-06,
+      "loss": 0.6664,
+      "step": 12428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.968592085010146e-06,
+      "loss": 0.6876,
+      "step": 12429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9670584860654125e-06,
+      "loss": 0.7233,
+      "step": 12430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9655250003879495e-06,
+      "loss": 0.6596,
+      "step": 12431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963991628020829e-06,
+      "loss": 0.7708,
+      "step": 12432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9624583690071135e-06,
+      "loss": 0.6882,
+      "step": 12433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960925223389869e-06,
+      "loss": 0.6821,
+      "step": 12434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.959392191212149e-06,
+      "loss": 0.636,
+      "step": 12435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.957859272517017e-06,
+      "loss": 0.6982,
+      "step": 12436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956326467347525e-06,
+      "loss": 0.6714,
+      "step": 12437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.954793775746716e-06,
+      "loss": 0.6562,
+      "step": 12438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9532611977576405e-06,
+      "loss": 0.6258,
+      "step": 12439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951728733423341e-06,
+      "loss": 0.7052,
+      "step": 12440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950196382786859e-06,
+      "loss": 0.7337,
+      "step": 12441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948664145891232e-06,
+      "loss": 0.6842,
+      "step": 12442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.947132022779492e-06,
+      "loss": 0.7319,
+      "step": 12443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.945600013494669e-06,
+      "loss": 0.6801,
+      "step": 12444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944068118079791e-06,
+      "loss": 0.798,
+      "step": 12445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942536336577881e-06,
+      "loss": 0.6795,
+      "step": 12446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.941004669031962e-06,
+      "loss": 0.6703,
+      "step": 12447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9394731154850505e-06,
+      "loss": 0.6445,
+      "step": 12448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93794167598016e-06,
+      "loss": 0.6738,
+      "step": 12449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936410350560306e-06,
+      "loss": 0.7341,
+      "step": 12450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.93487913926849e-06,
+      "loss": 0.717,
+      "step": 12451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933348042147722e-06,
+      "loss": 0.614,
+      "step": 12452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931817059241002e-06,
+      "loss": 0.7403,
+      "step": 12453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930286190591328e-06,
+      "loss": 0.6719,
+      "step": 12454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.928755436241694e-06,
+      "loss": 0.7036,
+      "step": 12455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.927224796235099e-06,
+      "loss": 0.7126,
+      "step": 12456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.925694270614523e-06,
+      "loss": 0.6159,
+      "step": 12457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924163859422954e-06,
+      "loss": 0.6881,
+      "step": 12458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922633562703375e-06,
+      "loss": 0.7313,
+      "step": 12459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.921103380498767e-06,
+      "loss": 0.6059,
+      "step": 12460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919573312852105e-06,
+      "loss": 0.6746,
+      "step": 12461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.918043359806363e-06,
+      "loss": 0.717,
+      "step": 12462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.91651352140451e-06,
+      "loss": 0.7072,
+      "step": 12463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.914983797689507e-06,
+      "loss": 0.692,
+      "step": 12464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.913454188704324e-06,
+      "loss": 0.6149,
+      "step": 12465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.911924694491917e-06,
+      "loss": 0.6844,
+      "step": 12466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.910395315095246e-06,
+      "loss": 0.705,
+      "step": 12467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.908866050557259e-06,
+      "loss": 0.6324,
+      "step": 12468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9073369009209135e-06,
+      "loss": 0.6747,
+      "step": 12469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905807866229151e-06,
+      "loss": 0.6697,
+      "step": 12470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.904278946524915e-06,
+      "loss": 0.7516,
+      "step": 12471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90275014185115e-06,
+      "loss": 0.7099,
+      "step": 12472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901221452250789e-06,
+      "loss": 0.6936,
+      "step": 12473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89969287776677e-06,
+      "loss": 0.6591,
+      "step": 12474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898164418442018e-06,
+      "loss": 0.6899,
+      "step": 12475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896636074319471e-06,
+      "loss": 0.7717,
+      "step": 12476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.895107845442043e-06,
+      "loss": 0.7512,
+      "step": 12477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893579731852659e-06,
+      "loss": 0.7145,
+      "step": 12478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892051733594238e-06,
+      "loss": 0.6897,
+      "step": 12479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89052385070969e-06,
+      "loss": 0.662,
+      "step": 12480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888996083241931e-06,
+      "loss": 0.671,
+      "step": 12481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.887468431233874e-06,
+      "loss": 0.7075,
+      "step": 12482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885940894728412e-06,
+      "loss": 0.6594,
+      "step": 12483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884413473768452e-06,
+      "loss": 0.6242,
+      "step": 12484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.882886168396895e-06,
+      "loss": 0.6364,
+      "step": 12485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881358978656633e-06,
+      "loss": 0.6761,
+      "step": 12486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8798319045905595e-06,
+      "loss": 0.6128,
+      "step": 12487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878304946241562e-06,
+      "loss": 0.7345,
+      "step": 12488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87677810365253e-06,
+      "loss": 0.6246,
+      "step": 12489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87525137686634e-06,
+      "loss": 0.6904,
+      "step": 12490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873724765925872e-06,
+      "loss": 0.6781,
+      "step": 12491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872198270874004e-06,
+      "loss": 0.6722,
+      "step": 12492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.870671891753608e-06,
+      "loss": 0.6806,
+      "step": 12493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.869145628607551e-06,
+      "loss": 0.6505,
+      "step": 12494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.867619481478703e-06,
+      "loss": 0.6258,
+      "step": 12495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866093450409922e-06,
+      "loss": 0.6801,
+      "step": 12496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86456753544407e-06,
+      "loss": 0.7061,
+      "step": 12497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863041736624003e-06,
+      "loss": 0.7004,
+      "step": 12498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.861516053992571e-06,
+      "loss": 0.6332,
+      "step": 12499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.859990487592625e-06,
+      "loss": 0.7542,
+      "step": 12500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858465037467014e-06,
+      "loss": 0.6481,
+      "step": 12501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856939703658584e-06,
+      "loss": 0.6785,
+      "step": 12502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855414486210167e-06,
+      "loss": 0.6587,
+      "step": 12503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8538893851646e-06,
+      "loss": 0.7786,
+      "step": 12504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.85236440056472e-06,
+      "loss": 0.6741,
+      "step": 12505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850839532453356e-06,
+      "loss": 0.7391,
+      "step": 12506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849314780873335e-06,
+      "loss": 0.5638,
+      "step": 12507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847790145867485e-06,
+      "loss": 0.6381,
+      "step": 12508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8462656274786145e-06,
+      "loss": 0.6971,
+      "step": 12509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.84474122574955e-06,
+      "loss": 0.6815,
+      "step": 12510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843216940723101e-06,
+      "loss": 0.6871,
+      "step": 12511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.84169277244208e-06,
+      "loss": 0.6703,
+      "step": 12512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840168720949293e-06,
+      "loss": 0.6741,
+      "step": 12513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838644786287544e-06,
+      "loss": 0.6437,
+      "step": 12514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8371209684996345e-06,
+      "loss": 0.6575,
+      "step": 12515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.835597267628359e-06,
+      "loss": 0.7143,
+      "step": 12516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.834073683716515e-06,
+      "loss": 0.6736,
+      "step": 12517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832550216806889e-06,
+      "loss": 0.7239,
+      "step": 12518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831026866942272e-06,
+      "loss": 0.6678,
+      "step": 12519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.829503634165447e-06,
+      "loss": 0.6639,
+      "step": 12520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827980518519194e-06,
+      "loss": 0.6549,
+      "step": 12521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8264575200462904e-06,
+      "loss": 0.7091,
+      "step": 12522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.824934638789512e-06,
+      "loss": 0.699,
+      "step": 12523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823411874791628e-06,
+      "loss": 0.7341,
+      "step": 12524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.821889228095408e-06,
+      "loss": 0.634,
+      "step": 12525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820366698743614e-06,
+      "loss": 0.6959,
+      "step": 12526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818844286779008e-06,
+      "loss": 0.6747,
+      "step": 12527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8173219922443516e-06,
+      "loss": 0.7231,
+      "step": 12528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.815799815182393e-06,
+      "loss": 0.705,
+      "step": 12529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.814277755635884e-06,
+      "loss": 0.6427,
+      "step": 12530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.812755813647575e-06,
+      "loss": 0.6705,
+      "step": 12531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81123398926021e-06,
+      "loss": 0.6801,
+      "step": 12532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8097122825165265e-06,
+      "loss": 0.6469,
+      "step": 12533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.808190693459274e-06,
+      "loss": 0.6705,
+      "step": 12534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806669222131173e-06,
+      "loss": 0.7011,
+      "step": 12535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.805147868574964e-06,
+      "loss": 0.6706,
+      "step": 12536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80362663283337e-06,
+      "loss": 0.6125,
+      "step": 12537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802105514949119e-06,
+      "loss": 0.6906,
+      "step": 12538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80058451496493e-06,
+      "loss": 0.6638,
+      "step": 12539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799063632923523e-06,
+      "loss": 0.6753,
+      "step": 12540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797542868867611e-06,
+      "loss": 0.6815,
+      "step": 12541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796022222839906e-06,
+      "loss": 0.6394,
+      "step": 12542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794501694883117e-06,
+      "loss": 0.6942,
+      "step": 12543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7929812850399495e-06,
+      "loss": 0.6268,
+      "step": 12544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7914609933531016e-06,
+      "loss": 0.5604,
+      "step": 12545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789940819865275e-06,
+      "loss": 0.725,
+      "step": 12546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788420764619167e-06,
+      "loss": 0.674,
+      "step": 12547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786900827657462e-06,
+      "loss": 0.7038,
+      "step": 12548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78538100902285e-06,
+      "loss": 0.6795,
+      "step": 12549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783861308758019e-06,
+      "loss": 0.6857,
+      "step": 12550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.782341726905649e-06,
+      "loss": 0.6602,
+      "step": 12551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7808222635084186e-06,
+      "loss": 0.7374,
+      "step": 12552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.779302918608998e-06,
+      "loss": 0.6231,
+      "step": 12553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777783692250071e-06,
+      "loss": 0.6843,
+      "step": 12554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.776264584474294e-06,
+      "loss": 0.6935,
+      "step": 12555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.774745595324337e-06,
+      "loss": 0.7137,
+      "step": 12556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77322672484286e-06,
+      "loss": 0.7194,
+      "step": 12557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.771707973072522e-06,
+      "loss": 0.6797,
+      "step": 12558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770189340055978e-06,
+      "loss": 0.6921,
+      "step": 12559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768670825835885e-06,
+      "loss": 0.7269,
+      "step": 12560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767152430454875e-06,
+      "loss": 0.6639,
+      "step": 12561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765634153955609e-06,
+      "loss": 0.7349,
+      "step": 12562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764115996380724e-06,
+      "loss": 0.5939,
+      "step": 12563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762597957772856e-06,
+      "loss": 0.692,
+      "step": 12564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7610800381746425e-06,
+      "loss": 0.7274,
+      "step": 12565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7595622376287144e-06,
+      "loss": 0.6392,
+      "step": 12566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.758044556177703e-06,
+      "loss": 0.6912,
+      "step": 12567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756526993864225e-06,
+      "loss": 0.6501,
+      "step": 12568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755009550730908e-06,
+      "loss": 0.6625,
+      "step": 12569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753492226820367e-06,
+      "loss": 0.7504,
+      "step": 12570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751975022175221e-06,
+      "loss": 0.6602,
+      "step": 12571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750457936838078e-06,
+      "loss": 0.6684,
+      "step": 12572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748940970851548e-06,
+      "loss": 0.6594,
+      "step": 12573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.747424124258234e-06,
+      "loss": 0.6959,
+      "step": 12574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7459073971007384e-06,
+      "loss": 0.699,
+      "step": 12575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744390789421661e-06,
+      "loss": 0.6643,
+      "step": 12576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742874301263592e-06,
+      "loss": 0.674,
+      "step": 12577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741357932669128e-06,
+      "loss": 0.6644,
+      "step": 12578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739841683680855e-06,
+      "loss": 0.7536,
+      "step": 12579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.738325554341359e-06,
+      "loss": 0.715,
+      "step": 12580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736809544693218e-06,
+      "loss": 0.6145,
+      "step": 12581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.735293654779006e-06,
+      "loss": 0.6466,
+      "step": 12582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733777884641308e-06,
+      "loss": 0.6942,
+      "step": 12583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73226223432269e-06,
+      "loss": 0.6607,
+      "step": 12584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730746703865719e-06,
+      "loss": 0.6404,
+      "step": 12585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729231293312967e-06,
+      "loss": 0.6914,
+      "step": 12586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7277160027069825e-06,
+      "loss": 0.7178,
+      "step": 12587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726200832090331e-06,
+      "loss": 0.6218,
+      "step": 12588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724685781505563e-06,
+      "loss": 0.7868,
+      "step": 12589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7231708509952345e-06,
+      "loss": 0.6762,
+      "step": 12590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7216560406018885e-06,
+      "loss": 0.639,
+      "step": 12591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.720141350368072e-06,
+      "loss": 0.7269,
+      "step": 12592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7186267803363245e-06,
+      "loss": 0.7402,
+      "step": 12593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.717112330549185e-06,
+      "loss": 0.6677,
+      "step": 12594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715598001049185e-06,
+      "loss": 0.6539,
+      "step": 12595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714083791878857e-06,
+      "loss": 0.6548,
+      "step": 12596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.71256970308073e-06,
+      "loss": 0.6477,
+      "step": 12597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711055734697325e-06,
+      "loss": 0.651,
+      "step": 12598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709541886771167e-06,
+      "loss": 0.5611,
+      "step": 12599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708028159344767e-06,
+      "loss": 0.6817,
+      "step": 12600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.706514552460642e-06,
+      "loss": 0.6306,
+      "step": 12601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705001066161298e-06,
+      "loss": 0.6603,
+      "step": 12602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70348770048925e-06,
+      "loss": 0.6554,
+      "step": 12603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701974455486997e-06,
+      "loss": 0.7336,
+      "step": 12604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700461331197041e-06,
+      "loss": 0.7076,
+      "step": 12605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.698948327661882e-06,
+      "loss": 0.6823,
+      "step": 12606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697435444924006e-06,
+      "loss": 0.6693,
+      "step": 12607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6959226830259075e-06,
+      "loss": 0.6061,
+      "step": 12608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694410042010071e-06,
+      "loss": 0.5894,
+      "step": 12609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.692897521918982e-06,
+      "loss": 0.6854,
+      "step": 12610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691385122795119e-06,
+      "loss": 0.7203,
+      "step": 12611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689872844680959e-06,
+      "loss": 0.6594,
+      "step": 12612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.688360687618977e-06,
+      "loss": 0.699,
+      "step": 12613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.686848651651639e-06,
+      "loss": 0.6461,
+      "step": 12614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685336736821415e-06,
+      "loss": 0.6457,
+      "step": 12615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683824943170766e-06,
+      "loss": 0.7059,
+      "step": 12616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682313270742151e-06,
+      "loss": 0.678,
+      "step": 12617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680801719578026e-06,
+      "loss": 0.6169,
+      "step": 12618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679290289720851e-06,
+      "loss": 0.688,
+      "step": 12619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6777789812130635e-06,
+      "loss": 0.6652,
+      "step": 12620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.676267794097114e-06,
+      "loss": 0.676,
+      "step": 12621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.674756728415443e-06,
+      "loss": 0.65,
+      "step": 12622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673245784210497e-06,
+      "loss": 0.6898,
+      "step": 12623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671734961524707e-06,
+      "loss": 0.6421,
+      "step": 12624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6702242604005075e-06,
+      "loss": 0.678,
+      "step": 12625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668713680880322e-06,
+      "loss": 0.6654,
+      "step": 12626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667203223006579e-06,
+      "loss": 0.6917,
+      "step": 12627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665692886821701e-06,
+      "loss": 0.6579,
+      "step": 12628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664182672368105e-06,
+      "loss": 0.6338,
+      "step": 12629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.662672579688206e-06,
+      "loss": 0.678,
+      "step": 12630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.6846,
+      "step": 12631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.659652759819151e-06,
+      "loss": 0.6845,
+      "step": 12632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658143032714805e-06,
+      "loss": 0.6026,
+      "step": 12633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656633427553784e-06,
+      "loss": 0.6916,
+      "step": 12634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6551239443784856e-06,
+      "loss": 0.6144,
+      "step": 12635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653614583231304e-06,
+      "loss": 0.6317,
+      "step": 12636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.652105344154632e-06,
+      "loss": 0.6281,
+      "step": 12637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65059622719086e-06,
+      "loss": 0.6645,
+      "step": 12638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.649087232382364e-06,
+      "loss": 0.6373,
+      "step": 12639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647578359771531e-06,
+      "loss": 0.6114,
+      "step": 12640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646069609400735e-06,
+      "loss": 0.6869,
+      "step": 12641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.644560981312349e-06,
+      "loss": 0.6189,
+      "step": 12642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.643052475548752e-06,
+      "loss": 0.3141,
+      "step": 12643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6415440921523045e-06,
+      "loss": 0.6796,
+      "step": 12644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640035831165376e-06,
+      "loss": 0.6833,
+      "step": 12645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638527692630318e-06,
+      "loss": 0.5985,
+      "step": 12646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637019676589491e-06,
+      "loss": 0.5988,
+      "step": 12647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63551178308525e-06,
+      "loss": 0.6589,
+      "step": 12648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634004012159945e-06,
+      "loss": 0.6839,
+      "step": 12649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63249636385592e-06,
+      "loss": 0.7083,
+      "step": 12650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630988838215521e-06,
+      "loss": 0.6477,
+      "step": 12651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629481435281087e-06,
+      "loss": 0.6796,
+      "step": 12652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627974155094953e-06,
+      "loss": 0.694,
+      "step": 12653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626466997699451e-06,
+      "loss": 0.7571,
+      "step": 12654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624959963136914e-06,
+      "loss": 0.7729,
+      "step": 12655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623453051449664e-06,
+      "loss": 0.7044,
+      "step": 12656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621946262680026e-06,
+      "loss": 0.7043,
+      "step": 12657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620439596870321e-06,
+      "loss": 0.7153,
+      "step": 12658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.618933054062859e-06,
+      "loss": 0.7452,
+      "step": 12659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617426634299952e-06,
+      "loss": 0.6921,
+      "step": 12660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.615920337623912e-06,
+      "loss": 0.6874,
+      "step": 12661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614414164077044e-06,
+      "loss": 0.6777,
+      "step": 12662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6129081137016425e-06,
+      "loss": 0.6699,
+      "step": 12663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6114021865400185e-06,
+      "loss": 0.6918,
+      "step": 12664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609896382634463e-06,
+      "loss": 0.7034,
+      "step": 12665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6083907020272625e-06,
+      "loss": 0.6405,
+      "step": 12666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606885144760705e-06,
+      "loss": 0.7084,
+      "step": 12667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605379710877077e-06,
+      "loss": 0.6124,
+      "step": 12668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60387440041866e-06,
+      "loss": 0.7102,
+      "step": 12669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6023692134277296e-06,
+      "loss": 0.6693,
+      "step": 12670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.600864149946561e-06,
+      "loss": 0.6329,
+      "step": 12671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599359210017425e-06,
+      "loss": 0.6113,
+      "step": 12672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.597854393682587e-06,
+      "loss": 0.7252,
+      "step": 12673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596349700984314e-06,
+      "loss": 0.7024,
+      "step": 12674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594845131964861e-06,
+      "loss": 0.6374,
+      "step": 12675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593340686666489e-06,
+      "loss": 0.6687,
+      "step": 12676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5918363651314486e-06,
+      "loss": 0.7067,
+      "step": 12677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590332167401995e-06,
+      "loss": 0.6615,
+      "step": 12678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588828093520365e-06,
+      "loss": 0.696,
+      "step": 12679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.587324143528808e-06,
+      "loss": 0.7355,
+      "step": 12680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585820317469558e-06,
+      "loss": 0.6948,
+      "step": 12681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584316615384855e-06,
+      "loss": 0.6617,
+      "step": 12682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5828130373169274e-06,
+      "loss": 0.686,
+      "step": 12683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581309583308014e-06,
+      "loss": 0.7207,
+      "step": 12684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57980625340033e-06,
+      "loss": 0.6585,
+      "step": 12685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.578303047636099e-06,
+      "loss": 0.6836,
+      "step": 12686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5767999660575404e-06,
+      "loss": 0.6539,
+      "step": 12687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575297008706869e-06,
+      "loss": 0.6988,
+      "step": 12688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573794175626297e-06,
+      "loss": 0.7317,
+      "step": 12689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572291466858031e-06,
+      "loss": 0.6511,
+      "step": 12690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57078888244428e-06,
+      "loss": 0.6712,
+      "step": 12691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569286422427233e-06,
+      "loss": 0.6692,
+      "step": 12692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5677840868490975e-06,
+      "loss": 0.6276,
+      "step": 12693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566281875752066e-06,
+      "loss": 0.687,
+      "step": 12694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564779789178328e-06,
+      "loss": 0.6857,
+      "step": 12695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5632778271700705e-06,
+      "loss": 0.6948,
+      "step": 12696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56177598976948e-06,
+      "loss": 0.6304,
+      "step": 12697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560274277018729e-06,
+      "loss": 0.6949,
+      "step": 12698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.558772688959998e-06,
+      "loss": 0.7267,
+      "step": 12699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55727122563546e-06,
+      "loss": 0.7057,
+      "step": 12700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555769887087283e-06,
+      "loss": 0.6815,
+      "step": 12701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5542686733576365e-06,
+      "loss": 0.6494,
+      "step": 12702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552767584488674e-06,
+      "loss": 0.6524,
+      "step": 12703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55126662052257e-06,
+      "loss": 0.5729,
+      "step": 12704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5497657815014664e-06,
+      "loss": 0.6675,
+      "step": 12705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.54826506746752e-06,
+      "loss": 0.6702,
+      "step": 12706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546764478462876e-06,
+      "loss": 0.6324,
+      "step": 12707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545264014529685e-06,
+      "loss": 0.7194,
+      "step": 12708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543763675710082e-06,
+      "loss": 0.7191,
+      "step": 12709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.542263462046213e-06,
+      "loss": 0.6671,
+      "step": 12710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.540763373580203e-06,
+      "loss": 0.6068,
+      "step": 12711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539263410354183e-06,
+      "loss": 0.6777,
+      "step": 12712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5377635724102885e-06,
+      "loss": 0.6391,
+      "step": 12713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5362638597906385e-06,
+      "loss": 0.6661,
+      "step": 12714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.534764272537352e-06,
+      "loss": 0.6843,
+      "step": 12715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5332648106925495e-06,
+      "loss": 0.6719,
+      "step": 12716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.531765474298344e-06,
+      "loss": 0.5975,
+      "step": 12717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5302662633968395e-06,
+      "loss": 0.7161,
+      "step": 12718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528767178030146e-06,
+      "loss": 0.706,
+      "step": 12719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527268218240367e-06,
+      "loss": 0.7125,
+      "step": 12720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.525769384069598e-06,
+      "loss": 0.7617,
+      "step": 12721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524270675559938e-06,
+      "loss": 0.6788,
+      "step": 12722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522772092753478e-06,
+      "loss": 0.6813,
+      "step": 12723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521273635692306e-06,
+      "loss": 0.6324,
+      "step": 12724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519775304418507e-06,
+      "loss": 0.7286,
+      "step": 12725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518277098974163e-06,
+      "loss": 0.724,
+      "step": 12726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.516779019401353e-06,
+      "loss": 0.688,
+      "step": 12727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515281065742149e-06,
+      "loss": 0.6772,
+      "step": 12728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.513783238038621e-06,
+      "loss": 0.6096,
+      "step": 12729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512285536332844e-06,
+      "loss": 0.6757,
+      "step": 12730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510787960666871e-06,
+      "loss": 0.7065,
+      "step": 12731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5092905110827635e-06,
+      "loss": 0.6876,
+      "step": 12732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.507793187622585e-06,
+      "loss": 0.6378,
+      "step": 12733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5062959903283855e-06,
+      "loss": 0.5945,
+      "step": 12734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504798919242216e-06,
+      "loss": 0.6726,
+      "step": 12735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.503301974406122e-06,
+      "loss": 0.7354,
+      "step": 12736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501805155862142e-06,
+      "loss": 0.6884,
+      "step": 12737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500308463652317e-06,
+      "loss": 0.6702,
+      "step": 12738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498811897818685e-06,
+      "loss": 0.6899,
+      "step": 12739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497315458403275e-06,
+      "loss": 0.7083,
+      "step": 12740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4958191454481154e-06,
+      "loss": 0.7075,
+      "step": 12741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494322958995233e-06,
+      "loss": 0.715,
+      "step": 12742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492826899086648e-06,
+      "loss": 0.3178,
+      "step": 12743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491330965764377e-06,
+      "loss": 0.6897,
+      "step": 12744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489835159070435e-06,
+      "loss": 0.6328,
+      "step": 12745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488339479046832e-06,
+      "loss": 0.7077,
+      "step": 12746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486843925735576e-06,
+      "loss": 0.598,
+      "step": 12747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4853484991786686e-06,
+      "loss": 0.7359,
+      "step": 12748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.483853199418115e-06,
+      "loss": 0.7839,
+      "step": 12749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482358026495904e-06,
+      "loss": 0.6869,
+      "step": 12750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480862980454031e-06,
+      "loss": 0.688,
+      "step": 12751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.479368061334483e-06,
+      "loss": 0.6699,
+      "step": 12752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.47787326917925e-06,
+      "loss": 0.6401,
+      "step": 12753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.476378604030315e-06,
+      "loss": 0.6811,
+      "step": 12754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.474884065929651e-06,
+      "loss": 0.6434,
+      "step": 12755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473389654919241e-06,
+      "loss": 0.6111,
+      "step": 12756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4718953710410475e-06,
+      "loss": 0.7121,
+      "step": 12757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470401214337041e-06,
+      "loss": 0.6712,
+      "step": 12758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468907184849186e-06,
+      "loss": 0.5852,
+      "step": 12759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4674132826194446e-06,
+      "loss": 0.63,
+      "step": 12760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465919507689772e-06,
+      "loss": 0.7369,
+      "step": 12761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4644258601021225e-06,
+      "loss": 0.6813,
+      "step": 12762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.462932339898446e-06,
+      "loss": 0.6565,
+      "step": 12763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461438947120688e-06,
+      "loss": 0.7295,
+      "step": 12764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459945681810792e-06,
+      "loss": 0.6655,
+      "step": 12765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.458452544010697e-06,
+      "loss": 0.3167,
+      "step": 12766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456959533762338e-06,
+      "loss": 0.707,
+      "step": 12767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4554666511076485e-06,
+      "loss": 0.6004,
+      "step": 12768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45397389608856e-06,
+      "loss": 0.6906,
+      "step": 12769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452481268746988e-06,
+      "loss": 0.6582,
+      "step": 12770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4509887691248605e-06,
+      "loss": 0.6294,
+      "step": 12771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4494963972640935e-06,
+      "loss": 0.6119,
+      "step": 12772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.448004153206597e-06,
+      "loss": 0.6544,
+      "step": 12773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446512036994287e-06,
+      "loss": 0.72,
+      "step": 12774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445020048669075e-06,
+      "loss": 0.6314,
+      "step": 12775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443528188272854e-06,
+      "loss": 0.6837,
+      "step": 12776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442036455847528e-06,
+      "loss": 0.6492,
+      "step": 12777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440544851434992e-06,
+      "loss": 0.7145,
+      "step": 12778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439053375077141e-06,
+      "loss": 0.6384,
+      "step": 12779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437562026815862e-06,
+      "loss": 0.6884,
+      "step": 12780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43607080669304e-06,
+      "loss": 0.6566,
+      "step": 12781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434579714750559e-06,
+      "loss": 0.6474,
+      "step": 12782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433088751030293e-06,
+      "loss": 0.6758,
+      "step": 12783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431597915574121e-06,
+      "loss": 0.6721,
+      "step": 12784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430107208423909e-06,
+      "loss": 0.7554,
+      "step": 12785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42861662962153e-06,
+      "loss": 0.7068,
+      "step": 12786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427126179208844e-06,
+      "loss": 0.6777,
+      "step": 12787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.425635857227715e-06,
+      "loss": 0.6623,
+      "step": 12788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424145663719993e-06,
+      "loss": 0.6584,
+      "step": 12789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422655598727533e-06,
+      "loss": 0.6926,
+      "step": 12790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4211656622921874e-06,
+      "loss": 0.6769,
+      "step": 12791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419675854455799e-06,
+      "loss": 0.308,
+      "step": 12792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418186175260207e-06,
+      "loss": 0.5944,
+      "step": 12793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416696624747257e-06,
+      "loss": 0.6741,
+      "step": 12794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415207202958785e-06,
+      "loss": 0.6392,
+      "step": 12795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4137179099366135e-06,
+      "loss": 0.6329,
+      "step": 12796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412228745722574e-06,
+      "loss": 0.5793,
+      "step": 12797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410739710358491e-06,
+      "loss": 0.6566,
+      "step": 12798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.409250803886183e-06,
+      "loss": 0.6708,
+      "step": 12799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.407762026347469e-06,
+      "loss": 0.7334,
+      "step": 12800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406273377784162e-06,
+      "loss": 0.6513,
+      "step": 12801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.40478485823807e-06,
+      "loss": 0.7855,
+      "step": 12802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403296467751e-06,
+      "loss": 0.6802,
+      "step": 12803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4018082063647536e-06,
+      "loss": 0.7252,
+      "step": 12804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4003200741211295e-06,
+      "loss": 0.6989,
+      "step": 12805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.398832071061922e-06,
+      "loss": 0.6922,
+      "step": 12806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397344197228924e-06,
+      "loss": 0.7106,
+      "step": 12807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.395856452663927e-06,
+      "loss": 0.6952,
+      "step": 12808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.394368837408705e-06,
+      "loss": 0.6373,
+      "step": 12809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392881351505046e-06,
+      "loss": 0.6576,
+      "step": 12810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391393994994722e-06,
+      "loss": 0.7625,
+      "step": 12811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389906767919511e-06,
+      "loss": 0.6756,
+      "step": 12812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.388419670321177e-06,
+      "loss": 0.6124,
+      "step": 12813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386932702241496e-06,
+      "loss": 0.6663,
+      "step": 12814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38544586372222e-06,
+      "loss": 0.6995,
+      "step": 12815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383959154805112e-06,
+      "loss": 0.6583,
+      "step": 12816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3824725755319255e-06,
+      "loss": 0.6989,
+      "step": 12817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380986125944414e-06,
+      "loss": 0.6998,
+      "step": 12818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379499806084323e-06,
+      "loss": 0.6733,
+      "step": 12819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378013615993397e-06,
+      "loss": 0.6498,
+      "step": 12820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.376527555713382e-06,
+      "loss": 0.636,
+      "step": 12821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3750416252860006e-06,
+      "loss": 0.7067,
+      "step": 12822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373555824753e-06,
+      "loss": 0.6589,
+      "step": 12823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372070154156104e-06,
+      "loss": 0.6239,
+      "step": 12824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37058461353704e-06,
+      "loss": 0.6852,
+      "step": 12825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369099202937528e-06,
+      "loss": 0.6539,
+      "step": 12826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3676139223992926e-06,
+      "loss": 0.7313,
+      "step": 12827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36612877196404e-06,
+      "loss": 0.6642,
+      "step": 12828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.364643751673485e-06,
+      "loss": 0.6474,
+      "step": 12829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363158861569335e-06,
+      "loss": 0.6276,
+      "step": 12830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.361674101693297e-06,
+      "loss": 0.6988,
+      "step": 12831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360189472087067e-06,
+      "loss": 0.5681,
+      "step": 12832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358704972792339e-06,
+      "loss": 0.7587,
+      "step": 12833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357220603850819e-06,
+      "loss": 0.6542,
+      "step": 12834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.355736365304183e-06,
+      "loss": 0.6698,
+      "step": 12835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.354252257194121e-06,
+      "loss": 0.6944,
+      "step": 12836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352768279562315e-06,
+      "loss": 0.621,
+      "step": 12837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351284432450445e-06,
+      "loss": 0.6483,
+      "step": 12838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349800715900182e-06,
+      "loss": 0.7153,
+      "step": 12839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348317129953204e-06,
+      "loss": 0.6099,
+      "step": 12840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346833674651169e-06,
+      "loss": 0.6633,
+      "step": 12841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345350350035741e-06,
+      "loss": 0.65,
+      "step": 12842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343867156148588e-06,
+      "loss": 0.7595,
+      "step": 12843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342384093031361e-06,
+      "loss": 0.606,
+      "step": 12844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340901160725716e-06,
+      "loss": 0.6705,
+      "step": 12845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339418359273297e-06,
+      "loss": 0.7278,
+      "step": 12846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337935688715757e-06,
+      "loss": 0.7592,
+      "step": 12847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3364531490947295e-06,
+      "loss": 0.7148,
+      "step": 12848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334970740451854e-06,
+      "loss": 0.2935,
+      "step": 12849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333488462828766e-06,
+      "loss": 0.6557,
+      "step": 12850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.332006316267098e-06,
+      "loss": 0.6522,
+      "step": 12851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330524300808473e-06,
+      "loss": 0.6955,
+      "step": 12852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329042416494516e-06,
+      "loss": 0.7084,
+      "step": 12853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327560663366849e-06,
+      "loss": 0.702,
+      "step": 12854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326079041467083e-06,
+      "loss": 0.7038,
+      "step": 12855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324597550836834e-06,
+      "loss": 0.71,
+      "step": 12856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323116191517707e-06,
+      "loss": 0.6638,
+      "step": 12857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3216349635513125e-06,
+      "loss": 0.7033,
+      "step": 12858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320153866979245e-06,
+      "loss": 0.7335,
+      "step": 12859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31867290184311e-06,
+      "loss": 0.6606,
+      "step": 12860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3171920681844915e-06,
+      "loss": 0.7435,
+      "step": 12861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31571136604498e-06,
+      "loss": 0.6453,
+      "step": 12862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314230795466169e-06,
+      "loss": 0.6906,
+      "step": 12863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3127503564896385e-06,
+      "loss": 0.6456,
+      "step": 12864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311270049156967e-06,
+      "loss": 0.6262,
+      "step": 12865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309789873509732e-06,
+      "loss": 0.6932,
+      "step": 12866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3083098295895e-06,
+      "loss": 0.6552,
+      "step": 12867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.306829917437841e-06,
+      "loss": 0.6419,
+      "step": 12868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3053501370963186e-06,
+      "loss": 0.6702,
+      "step": 12869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303870488606494e-06,
+      "loss": 0.6869,
+      "step": 12870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302390972009924e-06,
+      "loss": 0.6771,
+      "step": 12871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300911587348162e-06,
+      "loss": 0.6381,
+      "step": 12872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.299432334662756e-06,
+      "loss": 0.6883,
+      "step": 12873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297953213995253e-06,
+      "loss": 0.699,
+      "step": 12874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296474225387194e-06,
+      "loss": 0.7036,
+      "step": 12875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294995368880117e-06,
+      "loss": 0.623,
+      "step": 12876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.293516644515558e-06,
+      "loss": 0.6844,
+      "step": 12877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292038052335045e-06,
+      "loss": 0.6357,
+      "step": 12878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290559592380111e-06,
+      "loss": 0.6155,
+      "step": 12879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28908126469227e-06,
+      "loss": 0.7186,
+      "step": 12880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.287603069313047e-06,
+      "loss": 0.6909,
+      "step": 12881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28612500628396e-06,
+      "loss": 0.7246,
+      "step": 12882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284647075646511e-06,
+      "loss": 0.6964,
+      "step": 12883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283169277442221e-06,
+      "loss": 0.6967,
+      "step": 12884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281691611712592e-06,
+      "loss": 0.6099,
+      "step": 12885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2802140784991255e-06,
+      "loss": 0.6746,
+      "step": 12886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278736677843311e-06,
+      "loss": 0.6529,
+      "step": 12887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277259409786648e-06,
+      "loss": 0.7219,
+      "step": 12888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275782274370626e-06,
+      "loss": 0.6263,
+      "step": 12889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2743052716367306e-06,
+      "loss": 0.6728,
+      "step": 12890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272828401626443e-06,
+      "loss": 0.6787,
+      "step": 12891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.271351664381244e-06,
+      "loss": 0.6853,
+      "step": 12892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2698750599426075e-06,
+      "loss": 0.6803,
+      "step": 12893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268398588352005e-06,
+      "loss": 0.6795,
+      "step": 12894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.266922249650903e-06,
+      "loss": 0.7045,
+      "step": 12895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265446043880767e-06,
+      "loss": 0.6683,
+      "step": 12896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263969971083056e-06,
+      "loss": 0.7313,
+      "step": 12897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2624940312992255e-06,
+      "loss": 0.6735,
+      "step": 12898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2610182245707335e-06,
+      "loss": 0.6494,
+      "step": 12899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2595425509390205e-06,
+      "loss": 0.6486,
+      "step": 12900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.258067010445535e-06,
+      "loss": 0.6681,
+      "step": 12901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256591603131719e-06,
+      "loss": 0.6894,
+      "step": 12902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255116329039005e-06,
+      "loss": 0.6069,
+      "step": 12903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.253641188208836e-06,
+      "loss": 0.5993,
+      "step": 12904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252166180682641e-06,
+      "loss": 0.658,
+      "step": 12905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.250691306501839e-06,
+      "loss": 0.5965,
+      "step": 12906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249216565707857e-06,
+      "loss": 0.6836,
+      "step": 12907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247741958342114e-06,
+      "loss": 0.7141,
+      "step": 12908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246267484446024e-06,
+      "loss": 0.6141,
+      "step": 12909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244793144060998e-06,
+      "loss": 0.6661,
+      "step": 12910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2433189372284455e-06,
+      "loss": 0.7013,
+      "step": 12911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.24184486398977e-06,
+      "loss": 0.6511,
+      "step": 12912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.240370924386371e-06,
+      "loss": 0.6401,
+      "step": 12913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.238897118459644e-06,
+      "loss": 0.7172,
+      "step": 12914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237423446250982e-06,
+      "loss": 0.715,
+      "step": 12915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235949907801776e-06,
+      "loss": 0.6593,
+      "step": 12916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234476503153409e-06,
+      "loss": 0.688,
+      "step": 12917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233003232347267e-06,
+      "loss": 0.5944,
+      "step": 12918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231530095424719e-06,
+      "loss": 0.7065,
+      "step": 12919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230057092427143e-06,
+      "loss": 0.7343,
+      "step": 12920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228584223395909e-06,
+      "loss": 0.6697,
+      "step": 12921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227111488372385e-06,
+      "loss": 0.6538,
+      "step": 12922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225638887397927e-06,
+      "loss": 0.6288,
+      "step": 12923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224166420513903e-06,
+      "loss": 0.6996,
+      "step": 12924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222694087761668e-06,
+      "loss": 0.6866,
+      "step": 12925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221221889182565e-06,
+      "loss": 0.5935,
+      "step": 12926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2197498248179455e-06,
+      "loss": 0.6814,
+      "step": 12927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218277894709154e-06,
+      "loss": 0.692,
+      "step": 12928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2168060988975285e-06,
+      "loss": 0.6713,
+      "step": 12929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2153344374244086e-06,
+      "loss": 0.5985,
+      "step": 12930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213862910331127e-06,
+      "loss": 0.6502,
+      "step": 12931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.212391517659001e-06,
+      "loss": 0.6636,
+      "step": 12932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210920259449369e-06,
+      "loss": 0.6741,
+      "step": 12933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209449135743547e-06,
+      "loss": 0.7048,
+      "step": 12934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.207978146582853e-06,
+      "loss": 0.6824,
+      "step": 12935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.206507292008601e-06,
+      "loss": 0.6465,
+      "step": 12936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.205036572062098e-06,
+      "loss": 0.6446,
+      "step": 12937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2035659867846554e-06,
+      "loss": 0.6795,
+      "step": 12938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2020955362175685e-06,
+      "loss": 0.6759,
+      "step": 12939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200625220402139e-06,
+      "loss": 0.703,
+      "step": 12940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1991550393796615e-06,
+      "loss": 0.6937,
+      "step": 12941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.197684993191426e-06,
+      "loss": 0.6675,
+      "step": 12942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196215081878717e-06,
+      "loss": 0.7007,
+      "step": 12943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194745305482828e-06,
+      "loss": 0.7033,
+      "step": 12944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.193275664045028e-06,
+      "loss": 0.7136,
+      "step": 12945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191806157606594e-06,
+      "loss": 0.7032,
+      "step": 12946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1903367862088024e-06,
+      "loss": 0.7202,
+      "step": 12947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188867549892918e-06,
+      "loss": 0.6797,
+      "step": 12948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.187398448700205e-06,
+      "loss": 0.7134,
+      "step": 12949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.185929482671926e-06,
+      "loss": 0.6444,
+      "step": 12950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1844606518493375e-06,
+      "loss": 0.7317,
+      "step": 12951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182991956273687e-06,
+      "loss": 0.6304,
+      "step": 12952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181523395986227e-06,
+      "loss": 0.7094,
+      "step": 12953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180054971028208e-06,
+      "loss": 0.7108,
+      "step": 12954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178586681440865e-06,
+      "loss": 0.6807,
+      "step": 12955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.177118527265438e-06,
+      "loss": 0.6579,
+      "step": 12956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175650508543164e-06,
+      "loss": 0.6311,
+      "step": 12957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.174182625315267e-06,
+      "loss": 0.7164,
+      "step": 12958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.172714877622975e-06,
+      "loss": 0.7149,
+      "step": 12959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171247265507511e-06,
+      "loss": 0.7226,
+      "step": 12960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169779789010095e-06,
+      "loss": 0.675,
+      "step": 12961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1683124481719405e-06,
+      "loss": 0.6969,
+      "step": 12962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16684524303426e-06,
+      "loss": 0.6687,
+      "step": 12963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165378173638258e-06,
+      "loss": 0.6988,
+      "step": 12964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16391124002514e-06,
+      "loss": 0.2986,
+      "step": 12965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162444442236105e-06,
+      "loss": 0.6864,
+      "step": 12966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16097778031235e-06,
+      "loss": 0.6453,
+      "step": 12967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.159511254295065e-06,
+      "loss": 0.6772,
+      "step": 12968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15804486422544e-06,
+      "loss": 0.677,
+      "step": 12969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1565786101446605e-06,
+      "loss": 0.6391,
+      "step": 12970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.155112492093902e-06,
+      "loss": 0.6143,
+      "step": 12971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.153646510114341e-06,
+      "loss": 0.7084,
+      "step": 12972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152180664247156e-06,
+      "loss": 0.6251,
+      "step": 12973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.150714954533514e-06,
+      "loss": 0.6672,
+      "step": 12974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.149249381014581e-06,
+      "loss": 0.6837,
+      "step": 12975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1477839437315146e-06,
+      "loss": 0.6287,
+      "step": 12976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146318642725481e-06,
+      "loss": 0.6654,
+      "step": 12977
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144853478037623e-06,
+      "loss": 0.6875,
+      "step": 12978
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.143388449709096e-06,
+      "loss": 0.6516,
+      "step": 12979
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.141923557781045e-06,
+      "loss": 0.6914,
+      "step": 12980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140458802294612e-06,
+      "loss": 0.6956,
+      "step": 12981
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138994183290937e-06,
+      "loss": 0.6829,
+      "step": 12982
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1375297008111544e-06,
+      "loss": 0.7019,
+      "step": 12983
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136065354896393e-06,
+      "loss": 0.688,
+      "step": 12984
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1346011455877834e-06,
+      "loss": 0.6718,
+      "step": 12985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133137072926444e-06,
+      "loss": 0.735,
+      "step": 12986
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131673136953497e-06,
+      "loss": 0.7094,
+      "step": 12987
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130209337710057e-06,
+      "loss": 0.6469,
+      "step": 12988
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.128745675237235e-06,
+      "loss": 0.6879,
+      "step": 12989
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1272821495761446e-06,
+      "loss": 0.7424,
+      "step": 12990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12581876076788e-06,
+      "loss": 0.6207,
+      "step": 12991
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124355508853543e-06,
+      "loss": 0.7189,
+      "step": 12992
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122892393874235e-06,
+      "loss": 0.7201,
+      "step": 12993
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121429415871047e-06,
+      "loss": 0.6765,
+      "step": 12994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119966574885065e-06,
+      "loss": 0.7166,
+      "step": 12995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1185038709573785e-06,
+      "loss": 0.6029,
+      "step": 12996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117041304129063e-06,
+      "loss": 0.6373,
+      "step": 12997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115578874441196e-06,
+      "loss": 0.703,
+      "step": 12998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114116581934854e-06,
+      "loss": 0.6779,
+      "step": 12999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112654426651102e-06,
+      "loss": 0.6938,
+      "step": 13000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111192408631008e-06,
+      "loss": 0.699,
+      "step": 13001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109730527915634e-06,
+      "loss": 0.5928,
+      "step": 13002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108268784546036e-06,
+      "loss": 0.6705,
+      "step": 13003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106807178563269e-06,
+      "loss": 0.7246,
+      "step": 13004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105345710008382e-06,
+      "loss": 0.7339,
+      "step": 13005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103884378922423e-06,
+      "loss": 0.66,
+      "step": 13006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1024231853464324e-06,
+      "loss": 0.6661,
+      "step": 13007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100962129321448e-06,
+      "loss": 0.6707,
+      "step": 13008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099501210888511e-06,
+      "loss": 0.7187,
+      "step": 13009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0980404300886425e-06,
+      "loss": 0.6728,
+      "step": 13010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096579786962872e-06,
+      "loss": 0.6615,
+      "step": 13011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0951192815522245e-06,
+      "loss": 0.7311,
+      "step": 13012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.093658913897715e-06,
+      "loss": 0.6522,
+      "step": 13013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0921986840403645e-06,
+      "loss": 0.6584,
+      "step": 13014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090738592021183e-06,
+      "loss": 0.643,
+      "step": 13015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089278637881181e-06,
+      "loss": 0.6697,
+      "step": 13016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0878188216613525e-06,
+      "loss": 0.7071,
+      "step": 13017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086359143402704e-06,
+      "loss": 0.6279,
+      "step": 13018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084899603146228e-06,
+      "loss": 0.6344,
+      "step": 13019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08344020093292e-06,
+      "loss": 0.6859,
+      "step": 13020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081980936803766e-06,
+      "loss": 0.6055,
+      "step": 13021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08052181079975e-06,
+      "loss": 0.7075,
+      "step": 13022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.079062822961854e-06,
+      "loss": 0.7166,
+      "step": 13023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077603973331054e-06,
+      "loss": 0.6688,
+      "step": 13024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.07614526194832e-06,
+      "loss": 0.7231,
+      "step": 13025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.074686688854622e-06,
+      "loss": 0.7039,
+      "step": 13026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0732282540909275e-06,
+      "loss": 0.6565,
+      "step": 13027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.071769957698194e-06,
+      "loss": 0.6267,
+      "step": 13028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.070311799717385e-06,
+      "loss": 0.7144,
+      "step": 13029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068853780189442e-06,
+      "loss": 0.6311,
+      "step": 13030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0673958991553226e-06,
+      "loss": 0.774,
+      "step": 13031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065938156655969e-06,
+      "loss": 0.7009,
+      "step": 13032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064480552732323e-06,
+      "loss": 0.6747,
+      "step": 13033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.063023087425324e-06,
+      "loss": 0.6022,
+      "step": 13034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061565760775906e-06,
+      "loss": 0.6235,
+      "step": 13035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060108572825e-06,
+      "loss": 0.6443,
+      "step": 13036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058651523613527e-06,
+      "loss": 0.7188,
+      "step": 13037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05719461318241e-06,
+      "loss": 0.6819,
+      "step": 13038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05573784157257e-06,
+      "loss": 0.6073,
+      "step": 13039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054281208824918e-06,
+      "loss": 0.6842,
+      "step": 13040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052824714980367e-06,
+      "loss": 0.6785,
+      "step": 13041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051368360079823e-06,
+      "loss": 0.3189,
+      "step": 13042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049912144164186e-06,
+      "loss": 0.7231,
+      "step": 13043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048456067274359e-06,
+      "loss": 0.6584,
+      "step": 13044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0470001294512315e-06,
+      "loss": 0.6895,
+      "step": 13045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045544330735698e-06,
+      "loss": 0.6352,
+      "step": 13046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044088671168643e-06,
+      "loss": 0.728,
+      "step": 13047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042633150790952e-06,
+      "loss": 0.6785,
+      "step": 13048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.041177769643507e-06,
+      "loss": 0.7014,
+      "step": 13049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039722527767175e-06,
+      "loss": 0.6618,
+      "step": 13050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038267425202829e-06,
+      "loss": 0.3344,
+      "step": 13051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.036812461991342e-06,
+      "loss": 0.7095,
+      "step": 13052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.035357638173568e-06,
+      "loss": 0.7289,
+      "step": 13053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033902953790375e-06,
+      "loss": 0.6906,
+      "step": 13054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032448408882622e-06,
+      "loss": 0.6998,
+      "step": 13055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03099400349115e-06,
+      "loss": 0.6786,
+      "step": 13056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02953973765681e-06,
+      "loss": 0.6616,
+      "step": 13057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028085611420448e-06,
+      "loss": 0.708,
+      "step": 13058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026631624822904e-06,
+      "loss": 0.7249,
+      "step": 13059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0251777779050114e-06,
+      "loss": 0.7098,
+      "step": 13060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0237240707076055e-06,
+      "loss": 0.6717,
+      "step": 13061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022270503271511e-06,
+      "loss": 0.6434,
+      "step": 13062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.020817075637554e-06,
+      "loss": 0.657,
+      "step": 13063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019363787846556e-06,
+      "loss": 0.6438,
+      "step": 13064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.017910639939332e-06,
+      "loss": 0.6306,
+      "step": 13065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016457631956693e-06,
+      "loss": 0.6658,
+      "step": 13066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.015004763939449e-06,
+      "loss": 0.3454,
+      "step": 13067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0135520359284094e-06,
+      "loss": 0.6733,
+      "step": 13068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012099447964366e-06,
+      "loss": 0.6496,
+      "step": 13069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010647000088118e-06,
+      "loss": 0.6713,
+      "step": 13070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009194692340461e-06,
+      "loss": 0.6716,
+      "step": 13071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007742524762182e-06,
+      "loss": 0.679,
+      "step": 13072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006290497394066e-06,
+      "loss": 0.6947,
+      "step": 13073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00483861027689e-06,
+      "loss": 0.6723,
+      "step": 13074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003386863451444e-06,
+      "loss": 0.6452,
+      "step": 13075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0019352569584876e-06,
+      "loss": 0.6579,
+      "step": 13076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000483790838795e-06,
+      "loss": 0.7199,
+      "step": 13077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99903246513313e-06,
+      "loss": 0.7473,
+      "step": 13078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997581279882256e-06,
+      "loss": 0.5676,
+      "step": 13079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996130235126928e-06,
+      "loss": 0.6427,
+      "step": 13080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.994679330907904e-06,
+      "loss": 0.6566,
+      "step": 13081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993228567265924e-06,
+      "loss": 0.6393,
+      "step": 13082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991777944241742e-06,
+      "loss": 0.6617,
+      "step": 13083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990327461876098e-06,
+      "loss": 0.7453,
+      "step": 13084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988877120209728e-06,
+      "loss": 0.6195,
+      "step": 13085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9874269192833645e-06,
+      "loss": 0.7223,
+      "step": 13086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98597685913774e-06,
+      "loss": 0.6551,
+      "step": 13087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.984526939813582e-06,
+      "loss": 0.5863,
+      "step": 13088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.983077161351606e-06,
+      "loss": 0.5934,
+      "step": 13089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981627523792533e-06,
+      "loss": 0.7133,
+      "step": 13090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980178027177075e-06,
+      "loss": 0.7173,
+      "step": 13091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978728671545945e-06,
+      "loss": 0.6816,
+      "step": 13092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977279456939847e-06,
+      "loss": 0.6418,
+      "step": 13093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975830383399485e-06,
+      "loss": 0.603,
+      "step": 13094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974381450965553e-06,
+      "loss": 0.6349,
+      "step": 13095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972932659678749e-06,
+      "loss": 0.6547,
+      "step": 13096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9714840095797615e-06,
+      "loss": 0.6848,
+      "step": 13097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970035500709276e-06,
+      "loss": 0.6728,
+      "step": 13098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.968587133107975e-06,
+      "loss": 0.7761,
+      "step": 13099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.967138906816538e-06,
+      "loss": 0.6722,
+      "step": 13100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965690821875641e-06,
+      "loss": 0.6582,
+      "step": 13101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964242878325945e-06,
+      "loss": 0.6938,
+      "step": 13102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962795076208124e-06,
+      "loss": 0.7482,
+      "step": 13103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961347415562841e-06,
+      "loss": 0.7328,
+      "step": 13104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959899896430751e-06,
+      "loss": 0.6875,
+      "step": 13105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.95845251885251e-06,
+      "loss": 0.6667,
+      "step": 13106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957005282868771e-06,
+      "loss": 0.662,
+      "step": 13107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955558188520174e-06,
+      "loss": 0.6311,
+      "step": 13108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954111235847365e-06,
+      "loss": 0.6594,
+      "step": 13109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952664424890982e-06,
+      "loss": 0.6515,
+      "step": 13110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9512177556916586e-06,
+      "loss": 0.6792,
+      "step": 13111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9497712282900265e-06,
+      "loss": 0.6669,
+      "step": 13112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948324842726713e-06,
+      "loss": 0.6861,
+      "step": 13113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946878599042337e-06,
+      "loss": 0.7057,
+      "step": 13114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945432497277522e-06,
+      "loss": 0.7049,
+      "step": 13115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9439865374728775e-06,
+      "loss": 0.6194,
+      "step": 13116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942540719669019e-06,
+      "loss": 0.5985,
+      "step": 13117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94109504390655e-06,
+      "loss": 0.6442,
+      "step": 13118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939649510226073e-06,
+      "loss": 0.6795,
+      "step": 13119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938204118668191e-06,
+      "loss": 0.6438,
+      "step": 13120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9367588692734905e-06,
+      "loss": 0.6415,
+      "step": 13121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935313762082567e-06,
+      "loss": 0.6393,
+      "step": 13122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933868797136002e-06,
+      "loss": 0.6597,
+      "step": 13123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.932423974474386e-06,
+      "loss": 0.6428,
+      "step": 13124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.930979294138296e-06,
+      "loss": 0.6745,
+      "step": 13125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.929534756168304e-06,
+      "loss": 0.7,
+      "step": 13126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928090360604985e-06,
+      "loss": 0.7129,
+      "step": 13127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926646107488897e-06,
+      "loss": 0.7184,
+      "step": 13128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925201996860609e-06,
+      "loss": 0.672,
+      "step": 13129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923758028760677e-06,
+      "loss": 0.713,
+      "step": 13130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922314203229658e-06,
+      "loss": 0.7088,
+      "step": 13131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.920870520308101e-06,
+      "loss": 0.6787,
+      "step": 13132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919426980036553e-06,
+      "loss": 0.7218,
+      "step": 13133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917983582455556e-06,
+      "loss": 0.6989,
+      "step": 13134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916540327605649e-06,
+      "loss": 0.3086,
+      "step": 13135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9150972155273664e-06,
+      "loss": 0.7027,
+      "step": 13136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9136542462612405e-06,
+      "loss": 0.6818,
+      "step": 13137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912211419847795e-06,
+      "loss": 0.6857,
+      "step": 13138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910768736327553e-06,
+      "loss": 0.7324,
+      "step": 13139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9093261957410395e-06,
+      "loss": 0.6431,
+      "step": 13140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.90788379812876e-06,
+      "loss": 0.6814,
+      "step": 13141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906441543531227e-06,
+      "loss": 0.6981,
+      "step": 13142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904999431988945e-06,
+      "loss": 0.7291,
+      "step": 13143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903557463542423e-06,
+      "loss": 0.6161,
+      "step": 13144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.902115638232157e-06,
+      "loss": 0.6202,
+      "step": 13145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900673956098644e-06,
+      "loss": 0.5833,
+      "step": 13146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.899232417182367e-06,
+      "loss": 0.6961,
+      "step": 13147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8977910215238165e-06,
+      "loss": 0.7034,
+      "step": 13148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8963497691634744e-06,
+      "loss": 0.6488,
+      "step": 13149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8949086601418196e-06,
+      "loss": 0.5613,
+      "step": 13150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8934676944993265e-06,
+      "loss": 0.6346,
+      "step": 13151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892026872276466e-06,
+      "loss": 0.6492,
+      "step": 13152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.890586193513701e-06,
+      "loss": 0.6553,
+      "step": 13153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889145658251497e-06,
+      "loss": 0.6654,
+      "step": 13154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887705266530313e-06,
+      "loss": 0.6906,
+      "step": 13155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886265018390599e-06,
+      "loss": 0.6242,
+      "step": 13156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8848249138728084e-06,
+      "loss": 0.6824,
+      "step": 13157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8833849530173875e-06,
+      "loss": 0.6906,
+      "step": 13158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8819451358647806e-06,
+      "loss": 0.6996,
+      "step": 13159
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880505462455418e-06,
+      "loss": 0.6592,
+      "step": 13160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879065932829738e-06,
+      "loss": 0.6945,
+      "step": 13161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877626547028171e-06,
+      "loss": 0.7252,
+      "step": 13162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876187305091139e-06,
+      "loss": 0.7106,
+      "step": 13163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874748207059071e-06,
+      "loss": 0.6619,
+      "step": 13164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8733092529723825e-06,
+      "loss": 0.673,
+      "step": 13165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871870442871488e-06,
+      "loss": 0.6841,
+      "step": 13166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870431776796792e-06,
+      "loss": 0.6362,
+      "step": 13167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868993254788704e-06,
+      "loss": 0.7258,
+      "step": 13168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8675548768876245e-06,
+      "loss": 0.7495,
+      "step": 13169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8661166431339525e-06,
+      "loss": 0.7285,
+      "step": 13170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.864678553568079e-06,
+      "loss": 0.6501,
+      "step": 13171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8632406082303966e-06,
+      "loss": 0.723,
+      "step": 13172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861802807161289e-06,
+      "loss": 0.6706,
+      "step": 13173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860365150401139e-06,
+      "loss": 0.7249,
+      "step": 13174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.858927637990322e-06,
+      "loss": 0.6584,
+      "step": 13175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857490269969211e-06,
+      "loss": 0.6716,
+      "step": 13176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8560530463781775e-06,
+      "loss": 0.6819,
+      "step": 13177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854615967257586e-06,
+      "loss": 0.7291,
+      "step": 13178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853179032647801e-06,
+      "loss": 0.6952,
+      "step": 13179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851742242589172e-06,
+      "loss": 0.6902,
+      "step": 13180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850305597122055e-06,
+      "loss": 0.6431,
+      "step": 13181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8488690962868e-06,
+      "loss": 0.6341,
+      "step": 13182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847432740123748e-06,
+      "loss": 0.6869,
+      "step": 13183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845996528673248e-06,
+      "loss": 0.3253,
+      "step": 13184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844560461975635e-06,
+      "loss": 0.6468,
+      "step": 13185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8431245400712345e-06,
+      "loss": 0.6666,
+      "step": 13186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841688763000381e-06,
+      "loss": 0.6622,
+      "step": 13187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8402531308033964e-06,
+      "loss": 0.6438,
+      "step": 13188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8388176435206024e-06,
+      "loss": 0.6315,
+      "step": 13189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837382301192316e-06,
+      "loss": 0.6579,
+      "step": 13190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835947103858848e-06,
+      "loss": 0.6977,
+      "step": 13191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834512051560508e-06,
+      "loss": 0.6685,
+      "step": 13192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8330771443376e-06,
+      "loss": 0.586,
+      "step": 13193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831642382230424e-06,
+      "loss": 0.593,
+      "step": 13194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830207765279275e-06,
+      "loss": 0.6877,
+      "step": 13195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828773293524447e-06,
+      "loss": 0.6236,
+      "step": 13196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.827338967006226e-06,
+      "loss": 0.7001,
+      "step": 13197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8259047857649e-06,
+      "loss": 0.6766,
+      "step": 13198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.824470749840742e-06,
+      "loss": 0.6395,
+      "step": 13199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.823036859274031e-06,
+      "loss": 0.7142,
+      "step": 13200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821603114105038e-06,
+      "loss": 0.6689,
+      "step": 13201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820169514374031e-06,
+      "loss": 0.6355,
+      "step": 13202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818736060121274e-06,
+      "loss": 0.7319,
+      "step": 13203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817302751387021e-06,
+      "loss": 0.6102,
+      "step": 13204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815869588211541e-06,
+      "loss": 0.6815,
+      "step": 13205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81443657063507e-06,
+      "loss": 0.6112,
+      "step": 13206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813003698697864e-06,
+      "loss": 0.6905,
+      "step": 13207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811570972440161e-06,
+      "loss": 0.5946,
+      "step": 13208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810138391902202e-06,
+      "loss": 0.6555,
+      "step": 13209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808705957124222e-06,
+      "loss": 0.6568,
+      "step": 13210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807273668146455e-06,
+      "loss": 0.2997,
+      "step": 13211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805841525009117e-06,
+      "loss": 0.6544,
+      "step": 13212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.80440952775244e-06,
+      "loss": 0.651,
+      "step": 13213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8029776764166415e-06,
+      "loss": 0.61,
+      "step": 13214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801545971041933e-06,
+      "loss": 0.6682,
+      "step": 13215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.800114411668527e-06,
+      "loss": 0.6713,
+      "step": 13216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798682998336628e-06,
+      "loss": 0.6362,
+      "step": 13217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7972517310864415e-06,
+      "loss": 0.671,
+      "step": 13218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.795820609958158e-06,
+      "loss": 0.641,
+      "step": 13219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794389634991977e-06,
+      "loss": 0.2874,
+      "step": 13220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792958806228086e-06,
+      "loss": 0.6674,
+      "step": 13221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.791528123706672e-06,
+      "loss": 0.694,
+      "step": 13222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790097587467915e-06,
+      "loss": 0.6362,
+      "step": 13223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788667197551995e-06,
+      "loss": 0.6895,
+      "step": 13224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7872369539990815e-06,
+      "loss": 0.6053,
+      "step": 13225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7858068568493465e-06,
+      "loss": 0.6976,
+      "step": 13226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784376906142953e-06,
+      "loss": 0.631,
+      "step": 13227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782947101920065e-06,
+      "loss": 0.6098,
+      "step": 13228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.781517444220836e-06,
+      "loss": 0.6552,
+      "step": 13229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78008793308542e-06,
+      "loss": 0.643,
+      "step": 13230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.778658568553969e-06,
+      "loss": 0.6827,
+      "step": 13231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777229350666621e-06,
+      "loss": 0.707,
+      "step": 13232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775800279463515e-06,
+      "loss": 0.6447,
+      "step": 13233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774371354984796e-06,
+      "loss": 0.6472,
+      "step": 13234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772942577270591e-06,
+      "loss": 0.6631,
+      "step": 13235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771513946361029e-06,
+      "loss": 0.6933,
+      "step": 13236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7700854622962365e-06,
+      "loss": 0.7058,
+      "step": 13237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768657125116327e-06,
+      "loss": 0.6323,
+      "step": 13238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.76722893486142e-06,
+      "loss": 0.6354,
+      "step": 13239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7658008915716245e-06,
+      "loss": 0.7239,
+      "step": 13240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764372995287051e-06,
+      "loss": 0.6517,
+      "step": 13241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762945246047801e-06,
+      "loss": 0.6518,
+      "step": 13242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.761517643893974e-06,
+      "loss": 0.638,
+      "step": 13243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760090188865665e-06,
+      "loss": 0.6652,
+      "step": 13244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758662881002965e-06,
+      "loss": 0.6573,
+      "step": 13245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757235720345959e-06,
+      "loss": 0.6493,
+      "step": 13246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.755808706934732e-06,
+      "loss": 0.6791,
+      "step": 13247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75438184080936e-06,
+      "loss": 0.6927,
+      "step": 13248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75295512200992e-06,
+      "loss": 0.6891,
+      "step": 13249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7515285505764844e-06,
+      "loss": 0.619,
+      "step": 13250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.750102126549111e-06,
+      "loss": 0.6501,
+      "step": 13251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748675849967867e-06,
+      "loss": 0.6487,
+      "step": 13252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747249720872806e-06,
+      "loss": 0.6862,
+      "step": 13253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.745823739303988e-06,
+      "loss": 0.6727,
+      "step": 13254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.744397905301459e-06,
+      "loss": 0.6953,
+      "step": 13255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742972218905266e-06,
+      "loss": 0.7481,
+      "step": 13256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741546680155451e-06,
+      "loss": 0.6744,
+      "step": 13257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7401212890920456e-06,
+      "loss": 0.6913,
+      "step": 13258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.738696045755086e-06,
+      "loss": 0.5848,
+      "step": 13259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7372709501846e-06,
+      "loss": 0.6207,
+      "step": 13260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735846002420614e-06,
+      "loss": 0.7262,
+      "step": 13261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7344212025031464e-06,
+      "loss": 0.6777,
+      "step": 13262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732996550472216e-06,
+      "loss": 0.6342,
+      "step": 13263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731572046367832e-06,
+      "loss": 0.7675,
+      "step": 13264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730147690230004e-06,
+      "loss": 0.6238,
+      "step": 13265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728723482098735e-06,
+      "loss": 0.6436,
+      "step": 13266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7272994220140254e-06,
+      "loss": 0.7264,
+      "step": 13267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.72587551001587e-06,
+      "loss": 0.6813,
+      "step": 13268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724451746144261e-06,
+      "loss": 0.7179,
+      "step": 13269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723028130439189e-06,
+      "loss": 0.7103,
+      "step": 13270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721604662940628e-06,
+      "loss": 0.6096,
+      "step": 13271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720181343688562e-06,
+      "loss": 0.7047,
+      "step": 13272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.718758172722963e-06,
+      "loss": 0.675,
+      "step": 13273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7173351500838075e-06,
+      "loss": 0.6985,
+      "step": 13274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7159122758110575e-06,
+      "loss": 0.6813,
+      "step": 13275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.714489549944679e-06,
+      "loss": 0.6589,
+      "step": 13276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713066972524625e-06,
+      "loss": 0.6887,
+      "step": 13277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.71164454359085e-06,
+      "loss": 0.683,
+      "step": 13278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710222263183306e-06,
+      "loss": 0.6841,
+      "step": 13279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7088001313419365e-06,
+      "loss": 0.6791,
+      "step": 13280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707378148106684e-06,
+      "loss": 0.6994,
+      "step": 13281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705956313517486e-06,
+      "loss": 0.6739,
+      "step": 13282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704534627614274e-06,
+      "loss": 0.6573,
+      "step": 13283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.703113090436978e-06,
+      "loss": 0.6382,
+      "step": 13284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701691702025523e-06,
+      "loss": 0.6635,
+      "step": 13285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700270462419828e-06,
+      "loss": 0.697,
+      "step": 13286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69884937165981e-06,
+      "loss": 0.635,
+      "step": 13287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6974284297853815e-06,
+      "loss": 0.6417,
+      "step": 13288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696007636836454e-06,
+      "loss": 0.6581,
+      "step": 13289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.694586992852922e-06,
+      "loss": 0.7062,
+      "step": 13290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693166497874691e-06,
+      "loss": 0.6686,
+      "step": 13291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691746151941656e-06,
+      "loss": 0.6646,
+      "step": 13292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690325955093704e-06,
+      "loss": 0.6729,
+      "step": 13293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.688905907370729e-06,
+      "loss": 0.6304,
+      "step": 13294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.687486008812611e-06,
+      "loss": 0.6623,
+      "step": 13295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686066259459231e-06,
+      "loss": 0.6465,
+      "step": 13296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684646659350457e-06,
+      "loss": 0.618,
+      "step": 13297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.683227208526162e-06,
+      "loss": 0.6997,
+      "step": 13298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.681807907026214e-06,
+      "loss": 0.686,
+      "step": 13299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680388754890473e-06,
+      "loss": 0.6575,
+      "step": 13300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678969752158797e-06,
+      "loss": 0.6124,
+      "step": 13301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677550898871039e-06,
+      "loss": 0.6351,
+      "step": 13302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6761321950670494e-06,
+      "loss": 0.6201,
+      "step": 13303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674713640786673e-06,
+      "loss": 0.6968,
+      "step": 13304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6732952360697505e-06,
+      "loss": 0.6751,
+      "step": 13305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671876980956117e-06,
+      "loss": 0.677,
+      "step": 13306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.670458875485607e-06,
+      "loss": 0.7288,
+      "step": 13307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669040919698048e-06,
+      "loss": 0.6784,
+      "step": 13308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.667623113633268e-06,
+      "loss": 0.7143,
+      "step": 13309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666205457331079e-06,
+      "loss": 0.7129,
+      "step": 13310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664787950831301e-06,
+      "loss": 0.6076,
+      "step": 13311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663370594173744e-06,
+      "loss": 0.7773,
+      "step": 13312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661953387398217e-06,
+      "loss": 0.6624,
+      "step": 13313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660536330544518e-06,
+      "loss": 0.6549,
+      "step": 13314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659119423652456e-06,
+      "loss": 0.625,
+      "step": 13315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.657702666761818e-06,
+      "loss": 0.6914,
+      "step": 13316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6562860599123935e-06,
+      "loss": 0.636,
+      "step": 13317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654869603143972e-06,
+      "loss": 0.6058,
+      "step": 13318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6534532964963355e-06,
+      "loss": 0.635,
+      "step": 13319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652037140009259e-06,
+      "loss": 0.598,
+      "step": 13320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6506211337225186e-06,
+      "loss": 0.6613,
+      "step": 13321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649205277675882e-06,
+      "loss": 0.5956,
+      "step": 13322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647789571909115e-06,
+      "loss": 0.7341,
+      "step": 13323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6463740164619776e-06,
+      "loss": 0.7082,
+      "step": 13324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644958611374229e-06,
+      "loss": 0.6177,
+      "step": 13325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.643543356685617e-06,
+      "loss": 0.6421,
+      "step": 13326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6421282524358946e-06,
+      "loss": 0.3116,
+      "step": 13327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6407132986648066e-06,
+      "loss": 0.6131,
+      "step": 13328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639298495412086e-06,
+      "loss": 0.6392,
+      "step": 13329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637883842717471e-06,
+      "loss": 0.7159,
+      "step": 13330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.636469340620695e-06,
+      "loss": 0.6485,
+      "step": 13331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6350549891614834e-06,
+      "loss": 0.6356,
+      "step": 13332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63364078837956e-06,
+      "loss": 0.6705,
+      "step": 13333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632226738314638e-06,
+      "loss": 0.7527,
+      "step": 13334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630812839006445e-06,
+      "loss": 0.6431,
+      "step": 13335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629399090494678e-06,
+      "loss": 0.743,
+      "step": 13336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627985492819049e-06,
+      "loss": 0.6348,
+      "step": 13337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626572046019258e-06,
+      "loss": 0.7239,
+      "step": 13338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625158750135002e-06,
+      "loss": 0.6402,
+      "step": 13339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623745605205975e-06,
+      "loss": 0.6264,
+      "step": 13340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622332611271869e-06,
+      "loss": 0.6593,
+      "step": 13341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.620919768372357e-06,
+      "loss": 0.6877,
+      "step": 13342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619507076547133e-06,
+      "loss": 0.6273,
+      "step": 13343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6180945358358674e-06,
+      "loss": 0.5873,
+      "step": 13344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.616682146278233e-06,
+      "loss": 0.6663,
+      "step": 13345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615269907913896e-06,
+      "loss": 0.706,
+      "step": 13346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613857820782523e-06,
+      "loss": 0.6876,
+      "step": 13347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612445884923773e-06,
+      "loss": 0.6914,
+      "step": 13348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611034100377297e-06,
+      "loss": 0.6914,
+      "step": 13349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6096224671827475e-06,
+      "loss": 0.6718,
+      "step": 13350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608210985379771e-06,
+      "loss": 0.7081,
+      "step": 13351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606799655008009e-06,
+      "loss": 0.6681,
+      "step": 13352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605388476107102e-06,
+      "loss": 0.6829,
+      "step": 13353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603977448716682e-06,
+      "loss": 0.6771,
+      "step": 13354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.602566572876379e-06,
+      "loss": 0.7297,
+      "step": 13355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601155848625818e-06,
+      "loss": 0.6678,
+      "step": 13356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599745276004619e-06,
+      "loss": 0.7015,
+      "step": 13357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5983348550523996e-06,
+      "loss": 0.6385,
+      "step": 13358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596924585808772e-06,
+      "loss": 0.6607,
+      "step": 13359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5955144683133444e-06,
+      "loss": 0.7078,
+      "step": 13360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.594104502605725e-06,
+      "loss": 0.6628,
+      "step": 13361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592694688725504e-06,
+      "loss": 0.6085,
+      "step": 13362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59128502671228e-06,
+      "loss": 0.7091,
+      "step": 13363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5898755166056485e-06,
+      "loss": 0.7232,
+      "step": 13364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588466158445194e-06,
+      "loss": 0.6214,
+      "step": 13365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5870569522704985e-06,
+      "loss": 0.6441,
+      "step": 13366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585647898121145e-06,
+      "loss": 0.6828,
+      "step": 13367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5842389960367e-06,
+      "loss": 0.6397,
+      "step": 13368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582830246056735e-06,
+      "loss": 0.6832,
+      "step": 13369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581421648220819e-06,
+      "loss": 0.6763,
+      "step": 13370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580013202568509e-06,
+      "loss": 0.666,
+      "step": 13371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5786049091393645e-06,
+      "loss": 0.6264,
+      "step": 13372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.577196767972938e-06,
+      "loss": 0.6849,
+      "step": 13373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575788779108777e-06,
+      "loss": 0.6992,
+      "step": 13374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5743809425864275e-06,
+      "loss": 0.6571,
+      "step": 13375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572973258445428e-06,
+      "loss": 0.6528,
+      "step": 13376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5715657267253134e-06,
+      "loss": 0.6949,
+      "step": 13377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.570158347465615e-06,
+      "loss": 0.695,
+      "step": 13378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568751120705861e-06,
+      "loss": 0.6312,
+      "step": 13379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567344046485574e-06,
+      "loss": 0.6486,
+      "step": 13380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565937124844275e-06,
+      "loss": 0.6349,
+      "step": 13381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564530355821471e-06,
+      "loss": 0.6173,
+      "step": 13382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563123739456673e-06,
+      "loss": 0.7036,
+      "step": 13383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.561717275789393e-06,
+      "loss": 0.7221,
+      "step": 13384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560310964859129e-06,
+      "loss": 0.6974,
+      "step": 13385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558904806705377e-06,
+      "loss": 0.6374,
+      "step": 13386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5574988013676345e-06,
+      "loss": 0.6752,
+      "step": 13387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556092948885382e-06,
+      "loss": 0.6979,
+      "step": 13388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554687249298108e-06,
+      "loss": 0.6507,
+      "step": 13389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.55328170264529e-06,
+      "loss": 0.657,
+      "step": 13390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5518763089664074e-06,
+      "loss": 0.5952,
+      "step": 13391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5504710683009275e-06,
+      "loss": 0.7285,
+      "step": 13392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549065980688319e-06,
+      "loss": 0.6385,
+      "step": 13393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547661046168046e-06,
+      "loss": 0.6917,
+      "step": 13394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546256264779565e-06,
+      "loss": 0.6245,
+      "step": 13395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54485163656233e-06,
+      "loss": 0.6527,
+      "step": 13396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54344716155579e-06,
+      "loss": 0.695,
+      "step": 13397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.542042839799392e-06,
+      "loss": 0.6417,
+      "step": 13398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540638671332577e-06,
+      "loss": 0.7271,
+      "step": 13399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5392346561947846e-06,
+      "loss": 0.6623,
+      "step": 13400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537830794425441e-06,
+      "loss": 0.6641,
+      "step": 13401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536427086063977e-06,
+      "loss": 0.5795,
+      "step": 13402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535023531149813e-06,
+      "loss": 0.6425,
+      "step": 13403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533620129722376e-06,
+      "loss": 0.7287,
+      "step": 13404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532216881821079e-06,
+      "loss": 0.6713,
+      "step": 13405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5308137874853304e-06,
+      "loss": 0.6783,
+      "step": 13406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529410846754543e-06,
+      "loss": 0.6891,
+      "step": 13407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52800805966811e-06,
+      "loss": 0.7126,
+      "step": 13408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.526605426265434e-06,
+      "loss": 0.6789,
+      "step": 13409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525202946585907e-06,
+      "loss": 0.6777,
+      "step": 13410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.523800620668921e-06,
+      "loss": 0.7004,
+      "step": 13411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52239844855386e-06,
+      "loss": 0.6072,
+      "step": 13412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520996430280105e-06,
+      "loss": 0.6911,
+      "step": 13413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.519594565887031e-06,
+      "loss": 0.6399,
+      "step": 13414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5181928554140105e-06,
+      "loss": 0.6147,
+      "step": 13415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516791298900414e-06,
+      "loss": 0.7184,
+      "step": 13416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5153898963856005e-06,
+      "loss": 0.6962,
+      "step": 13417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513988647908932e-06,
+      "loss": 0.65,
+      "step": 13418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512587553509764e-06,
+      "loss": 0.6802,
+      "step": 13419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511186613227448e-06,
+      "loss": 0.6686,
+      "step": 13420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509785827101325e-06,
+      "loss": 0.6839,
+      "step": 13421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5083851951707395e-06,
+      "loss": 0.6769,
+      "step": 13422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506984717475028e-06,
+      "loss": 0.6415,
+      "step": 13423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505584394053523e-06,
+      "loss": 0.6439,
+      "step": 13424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504184224945557e-06,
+      "loss": 0.6682,
+      "step": 13425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5027842101904575e-06,
+      "loss": 0.6724,
+      "step": 13426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501384349827536e-06,
+      "loss": 0.6817,
+      "step": 13427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4999846438961105e-06,
+      "loss": 0.6754,
+      "step": 13428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.498585092435495e-06,
+      "loss": 0.666,
+      "step": 13429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497185695484996e-06,
+      "loss": 0.6555,
+      "step": 13430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495786453083916e-06,
+      "loss": 0.5235,
+      "step": 13431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494387365271553e-06,
+      "loss": 0.7429,
+      "step": 13432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492988432087201e-06,
+      "loss": 0.6231,
+      "step": 13433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4915896535701505e-06,
+      "loss": 0.6701,
+      "step": 13434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.490191029759687e-06,
+      "loss": 0.6705,
+      "step": 13435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488792560695091e-06,
+      "loss": 0.3463,
+      "step": 13436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487394246415641e-06,
+      "loss": 0.3191,
+      "step": 13437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485996086960606e-06,
+      "loss": 0.6508,
+      "step": 13438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4845980823692605e-06,
+      "loss": 0.6683,
+      "step": 13439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483200232680859e-06,
+      "loss": 0.6952,
+      "step": 13440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481802537934666e-06,
+      "loss": 0.6579,
+      "step": 13441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.480404998169936e-06,
+      "loss": 0.6574,
+      "step": 13442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47900761342592e-06,
+      "loss": 0.672,
+      "step": 13443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47761038374186e-06,
+      "loss": 0.6494,
+      "step": 13444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.476213309157005e-06,
+      "loss": 0.6277,
+      "step": 13445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474816389710594e-06,
+      "loss": 0.6965,
+      "step": 13446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47341962544185e-06,
+      "loss": 0.7144,
+      "step": 13447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472023016390009e-06,
+      "loss": 0.5863,
+      "step": 13448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470626562594295e-06,
+      "loss": 0.6041,
+      "step": 13449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.469230264093924e-06,
+      "loss": 0.7458,
+      "step": 13450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.467834120928118e-06,
+      "loss": 0.7,
+      "step": 13451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466438133136083e-06,
+      "loss": 0.6929,
+      "step": 13452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46504230075703e-06,
+      "loss": 0.706,
+      "step": 13453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463646623830159e-06,
+      "loss": 0.6534,
+      "step": 13454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462251102394669e-06,
+      "loss": 0.6439,
+      "step": 13455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.460855736489754e-06,
+      "loss": 0.5876,
+      "step": 13456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459460526154605e-06,
+      "loss": 0.6345,
+      "step": 13457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458065471428406e-06,
+      "loss": 0.7455,
+      "step": 13458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456670572350341e-06,
+      "loss": 0.6917,
+      "step": 13459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45527582895958e-06,
+      "loss": 0.6424,
+      "step": 13460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453881241295298e-06,
+      "loss": 0.7079,
+      "step": 13461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452486809396663e-06,
+      "loss": 0.6169,
+      "step": 13462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45109253330284e-06,
+      "loss": 0.6454,
+      "step": 13463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449698413052981e-06,
+      "loss": 0.6558,
+      "step": 13464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4483044486862545e-06,
+      "loss": 0.6455,
+      "step": 13465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4469106402417985e-06,
+      "loss": 0.676,
+      "step": 13466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445516987758762e-06,
+      "loss": 0.6556,
+      "step": 13467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444123491276288e-06,
+      "loss": 0.7029,
+      "step": 13468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442730150833512e-06,
+      "loss": 0.7284,
+      "step": 13469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441336966469569e-06,
+      "loss": 0.7112,
+      "step": 13470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439943938223583e-06,
+      "loss": 0.6667,
+      "step": 13471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438551066134686e-06,
+      "loss": 0.6821,
+      "step": 13472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437158350241985e-06,
+      "loss": 0.6703,
+      "step": 13473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435765790584605e-06,
+      "loss": 0.6533,
+      "step": 13474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.434373387201654e-06,
+      "loss": 0.3251,
+      "step": 13475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43298114013224e-06,
+      "loss": 0.6792,
+      "step": 13476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4315890494154625e-06,
+      "loss": 0.6461,
+      "step": 13477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430197115090424e-06,
+      "loss": 0.6956,
+      "step": 13478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.42880533719621e-06,
+      "loss": 0.7128,
+      "step": 13479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427413715771913e-06,
+      "loss": 0.713,
+      "step": 13480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426022250856618e-06,
+      "loss": 0.5912,
+      "step": 13481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424630942489403e-06,
+      "loss": 0.6386,
+      "step": 13482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423239790709346e-06,
+      "loss": 0.5874,
+      "step": 13483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4218487955555145e-06,
+      "loss": 0.7082,
+      "step": 13484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420457957066985e-06,
+      "loss": 0.6402,
+      "step": 13485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419067275282809e-06,
+      "loss": 0.748,
+      "step": 13486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417676750242048e-06,
+      "loss": 0.6353,
+      "step": 13487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416286381983758e-06,
+      "loss": 0.6746,
+      "step": 13488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414896170546985e-06,
+      "loss": 0.7303,
+      "step": 13489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413506115970776e-06,
+      "loss": 0.6272,
+      "step": 13490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412116218294174e-06,
+      "loss": 0.6416,
+      "step": 13491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.410726477556209e-06,
+      "loss": 0.6287,
+      "step": 13492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4093368937959105e-06,
+      "loss": 0.7263,
+      "step": 13493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4079474670523146e-06,
+      "loss": 0.6362,
+      "step": 13494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40655819736444e-06,
+      "loss": 0.7055,
+      "step": 13495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405169084771305e-06,
+      "loss": 0.6342,
+      "step": 13496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403780129311923e-06,
+      "loss": 0.6626,
+      "step": 13497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402391331025309e-06,
+      "loss": 0.6703,
+      "step": 13498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401002689950458e-06,
+      "loss": 0.6751,
+      "step": 13499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399614206126377e-06,
+      "loss": 0.7071,
+      "step": 13500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398225879592061e-06,
+      "loss": 0.6553,
+      "step": 13501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.396837710386503e-06,
+      "loss": 0.661,
+      "step": 13502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395449698548688e-06,
+      "loss": 0.6696,
+      "step": 13503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394061844117602e-06,
+      "loss": 0.6646,
+      "step": 13504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392674147132221e-06,
+      "loss": 0.6867,
+      "step": 13505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3912866076315215e-06,
+      "loss": 0.6564,
+      "step": 13506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.389899225654471e-06,
+      "loss": 0.6275,
+      "step": 13507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3885120012400375e-06,
+      "loss": 0.7171,
+      "step": 13508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387124934427181e-06,
+      "loss": 0.7231,
+      "step": 13509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385738025254856e-06,
+      "loss": 0.6415,
+      "step": 13510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38435127376202e-06,
+      "loss": 0.6326,
+      "step": 13511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382964679987612e-06,
+      "loss": 0.6576,
+      "step": 13512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3815782439705774e-06,
+      "loss": 0.6594,
+      "step": 13513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.380191965749862e-06,
+      "loss": 0.6924,
+      "step": 13514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378805845364395e-06,
+      "loss": 0.6894,
+      "step": 13515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377419882853107e-06,
+      "loss": 0.6233,
+      "step": 13516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376034078254926e-06,
+      "loss": 0.5947,
+      "step": 13517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374648431608768e-06,
+      "loss": 0.6737,
+      "step": 13518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.373262942953552e-06,
+      "loss": 0.6091,
+      "step": 13519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371877612328191e-06,
+      "loss": 0.6974,
+      "step": 13520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3704924397715895e-06,
+      "loss": 0.738,
+      "step": 13521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.369107425322655e-06,
+      "loss": 0.6504,
+      "step": 13522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367722569020285e-06,
+      "loss": 0.675,
+      "step": 13523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366337870903373e-06,
+      "loss": 0.6403,
+      "step": 13524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36495333101081e-06,
+      "loss": 0.3141,
+      "step": 13525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363568949381481e-06,
+      "loss": 0.59,
+      "step": 13526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362184726054266e-06,
+      "loss": 0.6914,
+      "step": 13527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.360800661068043e-06,
+      "loss": 0.6461,
+      "step": 13528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359416754461685e-06,
+      "loss": 0.6523,
+      "step": 13529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358033006274063e-06,
+      "loss": 0.6843,
+      "step": 13530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356649416544031e-06,
+      "loss": 0.65,
+      "step": 13531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355265985310453e-06,
+      "loss": 0.6255,
+      "step": 13532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353882712612179e-06,
+      "loss": 0.6382,
+      "step": 13533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352499598488068e-06,
+      "loss": 0.6782,
+      "step": 13534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35111664297696e-06,
+      "loss": 0.6703,
+      "step": 13535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349733846117697e-06,
+      "loss": 0.6885,
+      "step": 13536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348351207949119e-06,
+      "loss": 0.7206,
+      "step": 13537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.34696872851005e-06,
+      "loss": 0.6546,
+      "step": 13538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345586407839324e-06,
+      "loss": 0.7124,
+      "step": 13539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344204245975761e-06,
+      "loss": 0.6364,
+      "step": 13540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.34282224295818e-06,
+      "loss": 0.6394,
+      "step": 13541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341440398825397e-06,
+      "loss": 0.7317,
+      "step": 13542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.340058713616222e-06,
+      "loss": 0.6324,
+      "step": 13543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338677187369458e-06,
+      "loss": 0.6292,
+      "step": 13544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337295820123908e-06,
+      "loss": 0.7298,
+      "step": 13545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335914611918368e-06,
+      "loss": 0.6305,
+      "step": 13546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.334533562791629e-06,
+      "loss": 0.7351,
+      "step": 13547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3331526727824794e-06,
+      "loss": 0.6758,
+      "step": 13548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331771941929702e-06,
+      "loss": 0.6195,
+      "step": 13549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.330391370272079e-06,
+      "loss": 0.6631,
+      "step": 13550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.329010957848377e-06,
+      "loss": 0.6497,
+      "step": 13551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32763070469737e-06,
+      "loss": 0.6085,
+      "step": 13552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326250610857823e-06,
+      "loss": 0.7324,
+      "step": 13553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324870676368491e-06,
+      "loss": 0.6589,
+      "step": 13554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32349090126814e-06,
+      "loss": 0.6386,
+      "step": 13555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322111285595521e-06,
+      "loss": 0.6307,
+      "step": 13556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320731829389373e-06,
+      "loss": 0.5708,
+      "step": 13557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 0.7008,
+      "step": 13558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317973395531471e-06,
+      "loss": 0.6291,
+      "step": 13559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.316594417957187e-06,
+      "loss": 0.6728,
+      "step": 13560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315215600004323e-06,
+      "loss": 0.6885,
+      "step": 13561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313836941711604e-06,
+      "loss": 0.6603,
+      "step": 13562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312458443117749e-06,
+      "loss": 0.7312,
+      "step": 13563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311080104261475e-06,
+      "loss": 0.6956,
+      "step": 13564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309701925181492e-06,
+      "loss": 0.7206,
+      "step": 13565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308323905916508e-06,
+      "loss": 0.6379,
+      "step": 13566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306946046505226e-06,
+      "loss": 0.6152,
+      "step": 13567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.305568346986342e-06,
+      "loss": 0.665,
+      "step": 13568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304190807398555e-06,
+      "loss": 0.6017,
+      "step": 13569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3028134277805445e-06,
+      "loss": 0.6734,
+      "step": 13570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301436208171001e-06,
+      "loss": 0.6918,
+      "step": 13571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300059148608602e-06,
+      "loss": 0.6897,
+      "step": 13572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298682249132025e-06,
+      "loss": 0.6944,
+      "step": 13573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297305509779935e-06,
+      "loss": 0.6534,
+      "step": 13574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295928930591009e-06,
+      "loss": 0.7263,
+      "step": 13575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.294552511603906e-06,
+      "loss": 0.6239,
+      "step": 13576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2931762528572775e-06,
+      "loss": 0.585,
+      "step": 13577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2918001543897805e-06,
+      "loss": 0.6672,
+      "step": 13578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290424216240062e-06,
+      "loss": 0.6204,
+      "step": 13579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.289048438446766e-06,
+      "loss": 0.6683,
+      "step": 13580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287672821048535e-06,
+      "loss": 0.6863,
+      "step": 13581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286297364084005e-06,
+      "loss": 0.6505,
+      "step": 13582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284922067591794e-06,
+      "loss": 0.6299,
+      "step": 13583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283546931610541e-06,
+      "loss": 0.7194,
+      "step": 13584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282171956178864e-06,
+      "loss": 0.6734,
+      "step": 13585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280797141335379e-06,
+      "loss": 0.6253,
+      "step": 13586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279422487118698e-06,
+      "loss": 0.6625,
+      "step": 13587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278047993567431e-06,
+      "loss": 0.6619,
+      "step": 13588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276673660720183e-06,
+      "loss": 0.3046,
+      "step": 13589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275299488615547e-06,
+      "loss": 0.689,
+      "step": 13590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.273925477292118e-06,
+      "loss": 0.6638,
+      "step": 13591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2725516267884895e-06,
+      "loss": 0.6343,
+      "step": 13592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271177937143245e-06,
+      "loss": 0.7524,
+      "step": 13593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269804408394962e-06,
+      "loss": 0.6526,
+      "step": 13594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268431040582227e-06,
+      "loss": 0.6849,
+      "step": 13595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2670578337436035e-06,
+      "loss": 0.6345,
+      "step": 13596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265684787917657e-06,
+      "loss": 0.6664,
+      "step": 13597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264311903142956e-06,
+      "loss": 0.6855,
+      "step": 13598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262939179458055e-06,
+      "loss": 0.6138,
+      "step": 13599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26156661690151e-06,
+      "loss": 0.6997,
+      "step": 13600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.260194215511869e-06,
+      "loss": 0.7036,
+      "step": 13601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258821975327679e-06,
+      "loss": 0.6754,
+      "step": 13602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2574498963874725e-06,
+      "loss": 0.701,
+      "step": 13603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.256077978729793e-06,
+      "loss": 0.6676,
+      "step": 13604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.254706222393168e-06,
+      "loss": 0.6106,
+      "step": 13605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253334627416124e-06,
+      "loss": 0.645,
+      "step": 13606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251963193837185e-06,
+      "loss": 0.5874,
+      "step": 13607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250591921694871e-06,
+      "loss": 0.7434,
+      "step": 13608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249220811027688e-06,
+      "loss": 0.6542,
+      "step": 13609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247849861874147e-06,
+      "loss": 0.6568,
+      "step": 13610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246479074272751e-06,
+      "loss": 0.354,
+      "step": 13611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245108448262002e-06,
+      "loss": 0.6958,
+      "step": 13612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2437379838803924e-06,
+      "loss": 0.6695,
+      "step": 13613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242367681166414e-06,
+      "loss": 0.3002,
+      "step": 13614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240997540158552e-06,
+      "loss": 0.6703,
+      "step": 13615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.239627560895287e-06,
+      "loss": 0.7247,
+      "step": 13616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238257743415095e-06,
+      "loss": 0.6211,
+      "step": 13617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23688808775645e-06,
+      "loss": 0.684,
+      "step": 13618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2355185939578184e-06,
+      "loss": 0.6374,
+      "step": 13619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234149262057663e-06,
+      "loss": 0.6605,
+      "step": 13620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2327800920944464e-06,
+      "loss": 0.6681,
+      "step": 13621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231411084106613e-06,
+      "loss": 0.6425,
+      "step": 13622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230042238132616e-06,
+      "loss": 0.7314,
+      "step": 13623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228673554210905e-06,
+      "loss": 0.6602,
+      "step": 13624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.227305032379916e-06,
+      "loss": 0.6821,
+      "step": 13625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225936672678086e-06,
+      "loss": 0.681,
+      "step": 13626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224568475143847e-06,
+      "loss": 0.6429,
+      "step": 13627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223200439815626e-06,
+      "loss": 0.6639,
+      "step": 13628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22183256673184e-06,
+      "loss": 0.7164,
+      "step": 13629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220464855930909e-06,
+      "loss": 0.7093,
+      "step": 13630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219097307451248e-06,
+      "loss": 0.7081,
+      "step": 13631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217729921331263e-06,
+      "loss": 0.6728,
+      "step": 13632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2163626976093586e-06,
+      "loss": 0.6554,
+      "step": 13633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214995636323933e-06,
+      "loss": 0.7215,
+      "step": 13634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2136287375133834e-06,
+      "loss": 0.3253,
+      "step": 13635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212262001216098e-06,
+      "loss": 0.6216,
+      "step": 13636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210895427470462e-06,
+      "loss": 0.6506,
+      "step": 13637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209529016314856e-06,
+      "loss": 0.6835,
+      "step": 13638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208162767787659e-06,
+      "loss": 0.6678,
+      "step": 13639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20679668192724e-06,
+      "loss": 0.6358,
+      "step": 13640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205430758771971e-06,
+      "loss": 0.7104,
+      "step": 13641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204064998360208e-06,
+      "loss": 0.6392,
+      "step": 13642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2026994007303075e-06,
+      "loss": 0.6838,
+      "step": 13643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201333965920632e-06,
+      "loss": 0.6809,
+      "step": 13644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199968693969526e-06,
+      "loss": 0.7587,
+      "step": 13645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198603584915334e-06,
+      "loss": 0.6993,
+      "step": 13646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197238638796399e-06,
+      "loss": 0.625,
+      "step": 13647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1958738556510495e-06,
+      "loss": 0.625,
+      "step": 13648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19450923551762e-06,
+      "loss": 0.6577,
+      "step": 13649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.193144778434436e-06,
+      "loss": 0.6895,
+      "step": 13650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191780484439818e-06,
+      "loss": 0.6639,
+      "step": 13651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190416353572085e-06,
+      "loss": 0.6822,
+      "step": 13652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189052385869549e-06,
+      "loss": 0.6735,
+      "step": 13653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187688581370516e-06,
+      "loss": 0.6816,
+      "step": 13654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186324940113291e-06,
+      "loss": 0.6345,
+      "step": 13655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1849614621361715e-06,
+      "loss": 0.7479,
+      "step": 13656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183598147477452e-06,
+      "loss": 0.6832,
+      "step": 13657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182234996175421e-06,
+      "loss": 0.6833,
+      "step": 13658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1808720082683665e-06,
+      "loss": 0.7119,
+      "step": 13659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179509183794568e-06,
+      "loss": 0.3052,
+      "step": 13660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178146522792296e-06,
+      "loss": 0.7152,
+      "step": 13661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176784025299826e-06,
+      "loss": 0.6711,
+      "step": 13662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1754216913554225e-06,
+      "loss": 0.689,
+      "step": 13663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174059520997344e-06,
+      "loss": 0.5802,
+      "step": 13664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172697514263857e-06,
+      "loss": 0.6857,
+      "step": 13665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17133567119321e-06,
+      "loss": 0.6327,
+      "step": 13666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169973991823652e-06,
+      "loss": 0.7192,
+      "step": 13667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.168612476193422e-06,
+      "loss": 0.6839,
+      "step": 13668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167251124340762e-06,
+      "loss": 0.6555,
+      "step": 13669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165889936303906e-06,
+      "loss": 0.6885,
+      "step": 13670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.164528912121084e-06,
+      "loss": 0.6757,
+      "step": 13671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1631680518305205e-06,
+      "loss": 0.6593,
+      "step": 13672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1618073554704355e-06,
+      "loss": 0.6923,
+      "step": 13673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160446823079046e-06,
+      "loss": 0.6156,
+      "step": 13674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159086454694563e-06,
+      "loss": 0.5969,
+      "step": 13675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157726250355193e-06,
+      "loss": 0.6462,
+      "step": 13676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156366210099137e-06,
+      "loss": 0.7228,
+      "step": 13677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155006333964595e-06,
+      "loss": 0.6237,
+      "step": 13678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153646621989759e-06,
+      "loss": 0.6374,
+      "step": 13679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152287074212818e-06,
+      "loss": 0.613,
+      "step": 13680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150927690671952e-06,
+      "loss": 0.6748,
+      "step": 13681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1495684714053405e-06,
+      "loss": 0.752,
+      "step": 13682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148209416451161e-06,
+      "loss": 0.688,
+      "step": 13683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1468505258475785e-06,
+      "loss": 0.6619,
+      "step": 13684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.145491799632765e-06,
+      "loss": 0.5837,
+      "step": 13685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.14413323784488e-06,
+      "loss": 0.6757,
+      "step": 13686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142774840522074e-06,
+      "loss": 0.6683,
+      "step": 13687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141416607702501e-06,
+      "loss": 0.6859,
+      "step": 13688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140058539424308e-06,
+      "loss": 0.7018,
+      "step": 13689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138700635725638e-06,
+      "loss": 0.6805,
+      "step": 13690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137342896644625e-06,
+      "loss": 0.6761,
+      "step": 13691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135985322219406e-06,
+      "loss": 0.6736,
+      "step": 13692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134627912488108e-06,
+      "loss": 0.6526,
+      "step": 13693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133270667488853e-06,
+      "loss": 0.7046,
+      "step": 13694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13191358725976e-06,
+      "loss": 0.6575,
+      "step": 13695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1305566718389446e-06,
+      "loss": 0.6118,
+      "step": 13696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129199921264518e-06,
+      "loss": 0.6793,
+      "step": 13697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12784333557458e-06,
+      "loss": 0.7132,
+      "step": 13698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126486914807241e-06,
+      "loss": 0.6903,
+      "step": 13699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1251306590005845e-06,
+      "loss": 0.6307,
+      "step": 13700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123774568192709e-06,
+      "loss": 0.718,
+      "step": 13701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122418642421698e-06,
+      "loss": 0.6852,
+      "step": 13702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.121062881725635e-06,
+      "loss": 0.688,
+      "step": 13703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119707286142594e-06,
+      "loss": 0.6877,
+      "step": 13704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118351855710654e-06,
+      "loss": 0.6819,
+      "step": 13705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116996590467882e-06,
+      "loss": 0.6734,
+      "step": 13706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115641490452336e-06,
+      "loss": 0.5516,
+      "step": 13707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114286555702079e-06,
+      "loss": 0.6834,
+      "step": 13708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.112931786255161e-06,
+      "loss": 0.6285,
+      "step": 13709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111577182149635e-06,
+      "loss": 0.6357,
+      "step": 13710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110222743423545e-06,
+      "loss": 0.6988,
+      "step": 13711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.108868470114935e-06,
+      "loss": 0.7026,
+      "step": 13712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107514362261828e-06,
+      "loss": 0.6993,
+      "step": 13713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106160419902267e-06,
+      "loss": 0.7359,
+      "step": 13714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104806643074276e-06,
+      "loss": 0.6489,
+      "step": 13715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103453031815874e-06,
+      "loss": 0.6714,
+      "step": 13716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.102099586165078e-06,
+      "loss": 0.6528,
+      "step": 13717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100746306159902e-06,
+      "loss": 0.6337,
+      "step": 13718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0993931918383556e-06,
+      "loss": 0.6675,
+      "step": 13719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0980402432384355e-06,
+      "loss": 0.6403,
+      "step": 13720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0966874603981435e-06,
+      "loss": 0.7247,
+      "step": 13721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095334843355472e-06,
+      "loss": 0.6582,
+      "step": 13722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093982392148411e-06,
+      "loss": 0.7103,
+      "step": 13723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.092630106814942e-06,
+      "loss": 0.6502,
+      "step": 13724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091277987393053e-06,
+      "loss": 0.6913,
+      "step": 13725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089926033920712e-06,
+      "loss": 0.6838,
+      "step": 13726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088574246435888e-06,
+      "loss": 0.6966,
+      "step": 13727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087222624976551e-06,
+      "loss": 0.6802,
+      "step": 13728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08587116958066e-06,
+      "loss": 0.683,
+      "step": 13729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084519880286171e-06,
+      "loss": 0.6373,
+      "step": 13730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083168757131036e-06,
+      "loss": 0.686,
+      "step": 13731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.081817800153207e-06,
+      "loss": 0.678,
+      "step": 13732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080467009390613e-06,
+      "loss": 0.6756,
+      "step": 13733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.079116384881205e-06,
+      "loss": 0.7015,
+      "step": 13734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077765926662911e-06,
+      "loss": 0.6263,
+      "step": 13735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.07641563477366e-06,
+      "loss": 0.6921,
+      "step": 13736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075065509251374e-06,
+      "loss": 0.6807,
+      "step": 13737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073715550133977e-06,
+      "loss": 0.6753,
+      "step": 13738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072365757459378e-06,
+      "loss": 0.6981,
+      "step": 13739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071016131265486e-06,
+      "loss": 0.6569,
+      "step": 13740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069666671590209e-06,
+      "loss": 0.3323,
+      "step": 13741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068317378471446e-06,
+      "loss": 0.6797,
+      "step": 13742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066968251947093e-06,
+      "loss": 0.6547,
+      "step": 13743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065619292055041e-06,
+      "loss": 0.687,
+      "step": 13744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064270498833176e-06,
+      "loss": 0.6775,
+      "step": 13745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062921872319381e-06,
+      "loss": 0.6201,
+      "step": 13746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.06157341255153e-06,
+      "loss": 0.71,
+      "step": 13747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.060225119567498e-06,
+      "loss": 0.6023,
+      "step": 13748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0588769934051495e-06,
+      "loss": 0.639,
+      "step": 13749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057529034102351e-06,
+      "loss": 0.6849,
+      "step": 13750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0561812416969585e-06,
+      "loss": 0.66,
+      "step": 13751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054833616226828e-06,
+      "loss": 0.6195,
+      "step": 13752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053486157729801e-06,
+      "loss": 0.6256,
+      "step": 13753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0521388662437285e-06,
+      "loss": 0.7072,
+      "step": 13754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050791741806449e-06,
+      "loss": 0.6485,
+      "step": 13755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0494447844557964e-06,
+      "loss": 0.6717,
+      "step": 13756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0480979942296e-06,
+      "loss": 0.6343,
+      "step": 13757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046751371165688e-06,
+      "loss": 0.6086,
+      "step": 13758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045404915301876e-06,
+      "loss": 0.6693,
+      "step": 13759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044058626675983e-06,
+      "loss": 0.6887,
+      "step": 13760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042712505325819e-06,
+      "loss": 0.6534,
+      "step": 13761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041366551289191e-06,
+      "loss": 0.6962,
+      "step": 13762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040020764603902e-06,
+      "loss": 0.6967,
+      "step": 13763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038675145307747e-06,
+      "loss": 0.6651,
+      "step": 13764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.037329693438521e-06,
+      "loss": 0.686,
+      "step": 13765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03598440903401e-06,
+      "loss": 0.7253,
+      "step": 13766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034639292131997e-06,
+      "loss": 0.5844,
+      "step": 13767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033294342770261e-06,
+      "loss": 0.6561,
+      "step": 13768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031949560986575e-06,
+      "loss": 0.6812,
+      "step": 13769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030604946818708e-06,
+      "loss": 0.6156,
+      "step": 13770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029260500304429e-06,
+      "loss": 0.6428,
+      "step": 13771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0279162214814886e-06,
+      "loss": 0.6224,
+      "step": 13772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026572110387645e-06,
+      "loss": 0.6779,
+      "step": 13773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025228167060646e-06,
+      "loss": 0.6249,
+      "step": 13774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023884391538244e-06,
+      "loss": 0.6244,
+      "step": 13775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022540783858177e-06,
+      "loss": 0.7327,
+      "step": 13776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.021197344058178e-06,
+      "loss": 0.6517,
+      "step": 13777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019854072175983e-06,
+      "loss": 0.658,
+      "step": 13778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018510968249313e-06,
+      "loss": 0.7128,
+      "step": 13779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017168032315891e-06,
+      "loss": 0.6502,
+      "step": 13780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015825264413436e-06,
+      "loss": 0.6441,
+      "step": 13781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014482664579659e-06,
+      "loss": 0.6984,
+      "step": 13782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013140232852268e-06,
+      "loss": 0.6819,
+      "step": 13783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011797969268966e-06,
+      "loss": 0.6451,
+      "step": 13784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010455873867452e-06,
+      "loss": 0.6398,
+      "step": 13785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009113946685418e-06,
+      "loss": 0.6855,
+      "step": 13786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0077721877605526e-06,
+      "loss": 0.5888,
+      "step": 13787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0064305971305416e-06,
+      "loss": 0.642,
+      "step": 13788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0050891748330636e-06,
+      "loss": 0.6914,
+      "step": 13789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003747920905791e-06,
+      "loss": 0.7005,
+      "step": 13790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.002406835386401e-06,
+      "loss": 0.7827,
+      "step": 13791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0010659183125475e-06,
+      "loss": 0.6943,
+      "step": 13792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999725169721898e-06,
+      "loss": 0.7148,
+      "step": 13793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998384589652103e-06,
+      "loss": 0.6789,
+      "step": 13794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99704417814082e-06,
+      "loss": 0.7238,
+      "step": 13795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995703935225692e-06,
+      "loss": 0.6514,
+      "step": 13796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994363860944365e-06,
+      "loss": 0.6495,
+      "step": 13797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9930239553344665e-06,
+      "loss": 0.6946,
+      "step": 13798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991684218433635e-06,
+      "loss": 0.6733,
+      "step": 13799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990344650279494e-06,
+      "loss": 0.6828,
+      "step": 13800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989005250909668e-06,
+      "loss": 0.6621,
+      "step": 13801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.987666020361775e-06,
+      "loss": 0.6215,
+      "step": 13802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986326958673428e-06,
+      "loss": 0.6517,
+      "step": 13803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984988065882235e-06,
+      "loss": 0.6248,
+      "step": 13804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983649342025798e-06,
+      "loss": 0.6983,
+      "step": 13805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982310787141719e-06,
+      "loss": 0.3256,
+      "step": 13806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9809724012675885e-06,
+      "loss": 0.6725,
+      "step": 13807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.979634184440998e-06,
+      "loss": 0.7281,
+      "step": 13808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978296136699531e-06,
+      "loss": 0.6968,
+      "step": 13809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9769582580807716e-06,
+      "loss": 0.6337,
+      "step": 13810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975620548622287e-06,
+      "loss": 0.7303,
+      "step": 13811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974283008361652e-06,
+      "loss": 0.7773,
+      "step": 13812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972945637336431e-06,
+      "loss": 0.6562,
+      "step": 13813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971608435584183e-06,
+      "loss": 0.6331,
+      "step": 13814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970271403142468e-06,
+      "loss": 0.6167,
+      "step": 13815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968934540048836e-06,
+      "loss": 0.704,
+      "step": 13816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967597846340838e-06,
+      "loss": 0.6443,
+      "step": 13817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9662613220560055e-06,
+      "loss": 0.6386,
+      "step": 13818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964924967231882e-06,
+      "loss": 0.6325,
+      "step": 13819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963588781905997e-06,
+      "loss": 0.6136,
+      "step": 13820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96225276611588e-06,
+      "loss": 0.7461,
+      "step": 13821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960916919899054e-06,
+      "loss": 0.7204,
+      "step": 13822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959581243293036e-06,
+      "loss": 0.6538,
+      "step": 13823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958245736335337e-06,
+      "loss": 0.666,
+      "step": 13824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95691039906347e-06,
+      "loss": 0.687,
+      "step": 13825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955575231514935e-06,
+      "loss": 0.6262,
+      "step": 13826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954240233727234e-06,
+      "loss": 0.6302,
+      "step": 13827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.952905405737858e-06,
+      "loss": 0.7721,
+      "step": 13828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951570747584298e-06,
+      "loss": 0.7074,
+      "step": 13829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950236259304041e-06,
+      "loss": 0.6896,
+      "step": 13830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9489019409345605e-06,
+      "loss": 0.6426,
+      "step": 13831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947567792513336e-06,
+      "loss": 0.6046,
+      "step": 13832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946233814077837e-06,
+      "loss": 0.6291,
+      "step": 13833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944900005665524e-06,
+      "loss": 0.6846,
+      "step": 13834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943566367313867e-06,
+      "loss": 0.6664,
+      "step": 13835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94223289906032e-06,
+      "loss": 0.6611,
+      "step": 13836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9408996009423285e-06,
+      "loss": 0.6948,
+      "step": 13837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939566472997343e-06,
+      "loss": 0.6544,
+      "step": 13838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9382335152628025e-06,
+      "loss": 0.6923,
+      "step": 13839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.936900727776146e-06,
+      "loss": 0.6837,
+      "step": 13840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935568110574805e-06,
+      "loss": 0.6772,
+      "step": 13841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934235663696207e-06,
+      "loss": 0.7262,
+      "step": 13842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932903387177773e-06,
+      "loss": 0.6734,
+      "step": 13843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931571281056923e-06,
+      "loss": 0.6471,
+      "step": 13844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930239345371067e-06,
+      "loss": 0.6439,
+      "step": 13845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928907580157617e-06,
+      "loss": 0.6801,
+      "step": 13846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927575985453973e-06,
+      "loss": 0.6444,
+      "step": 13847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.926244561297534e-06,
+      "loss": 0.6703,
+      "step": 13848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924913307725699e-06,
+      "loss": 0.7169,
+      "step": 13849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923582224775848e-06,
+      "loss": 0.6324,
+      "step": 13850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922251312485371e-06,
+      "loss": 0.6744,
+      "step": 13851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920920570891644e-06,
+      "loss": 0.5935,
+      "step": 13852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919590000032046e-06,
+      "loss": 0.6346,
+      "step": 13853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918259599943941e-06,
+      "loss": 0.6524,
+      "step": 13854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916929370664696e-06,
+      "loss": 0.6561,
+      "step": 13855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915599312231678e-06,
+      "loss": 0.7091,
+      "step": 13856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914269424682234e-06,
+      "loss": 0.6589,
+      "step": 13857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912939708053718e-06,
+      "loss": 0.5787,
+      "step": 13858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911610162383475e-06,
+      "loss": 0.6223,
+      "step": 13859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910280787708846e-06,
+      "loss": 0.6422,
+      "step": 13860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.908951584067166e-06,
+      "loss": 0.7192,
+      "step": 13861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907622551495773e-06,
+      "loss": 0.6597,
+      "step": 13862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9062936900319805e-06,
+      "loss": 0.6932,
+      "step": 13863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.904964999713123e-06,
+      "loss": 0.6893,
+      "step": 13864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9036364805765105e-06,
+      "loss": 0.7217,
+      "step": 13865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902308132659457e-06,
+      "loss": 0.6306,
+      "step": 13866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900979955999271e-06,
+      "loss": 0.6978,
+      "step": 13867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.899651950633254e-06,
+      "loss": 0.6533,
+      "step": 13868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898324116598708e-06,
+      "loss": 0.6892,
+      "step": 13869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896996453932918e-06,
+      "loss": 0.6681,
+      "step": 13870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895668962673175e-06,
+      "loss": 0.6103,
+      "step": 13871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894341642856763e-06,
+      "loss": 0.7451,
+      "step": 13872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89301449452096e-06,
+      "loss": 0.6675,
+      "step": 13873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891687517703042e-06,
+      "loss": 0.6871,
+      "step": 13874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890360712440276e-06,
+      "loss": 0.6608,
+      "step": 13875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889034078769925e-06,
+      "loss": 0.5933,
+      "step": 13876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887707616729249e-06,
+      "loss": 0.6777,
+      "step": 13877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886381326355503e-06,
+      "loss": 0.6613,
+      "step": 13878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885055207685937e-06,
+      "loss": 0.6434,
+      "step": 13879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883729260757793e-06,
+      "loss": 0.6207,
+      "step": 13880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882403485608316e-06,
+      "loss": 0.6186,
+      "step": 13881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881077882274739e-06,
+      "loss": 0.5982,
+      "step": 13882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8797524507942854e-06,
+      "loss": 0.6894,
+      "step": 13883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878427191204189e-06,
+      "loss": 0.623,
+      "step": 13884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877102103541669e-06,
+      "loss": 0.7237,
+      "step": 13885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8757771878439385e-06,
+      "loss": 0.6699,
+      "step": 13886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874452444148211e-06,
+      "loss": 0.6176,
+      "step": 13887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8731278724916955e-06,
+      "loss": 0.5925,
+      "step": 13888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871803472911586e-06,
+      "loss": 0.7094,
+      "step": 13889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8704792454450815e-06,
+      "loss": 0.676,
+      "step": 13890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8691551901293746e-06,
+      "loss": 0.6851,
+      "step": 13891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867831307001652e-06,
+      "loss": 0.6436,
+      "step": 13892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866507596099096e-06,
+      "loss": 0.6565,
+      "step": 13893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865184057458883e-06,
+      "loss": 0.7098,
+      "step": 13894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863860691118187e-06,
+      "loss": 0.6927,
+      "step": 13895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862537497114171e-06,
+      "loss": 0.6781,
+      "step": 13896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861214475484003e-06,
+      "loss": 0.6311,
+      "step": 13897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8598916262648376e-06,
+      "loss": 0.6691,
+      "step": 13898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858568949493829e-06,
+      "loss": 0.6894,
+      "step": 13899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857246445208125e-06,
+      "loss": 0.6606,
+      "step": 13900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855924113444871e-06,
+      "loss": 0.6566,
+      "step": 13901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854601954241201e-06,
+      "loss": 0.6891,
+      "step": 13902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853279967634249e-06,
+      "loss": 0.708,
+      "step": 13903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851958153661142e-06,
+      "loss": 0.634,
+      "step": 13904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85063651235901e-06,
+      "loss": 0.7524,
+      "step": 13905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849315043764971e-06,
+      "loss": 0.7046,
+      "step": 13906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847993747916135e-06,
+      "loss": 0.6299,
+      "step": 13907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846672624849619e-06,
+      "loss": 0.6506,
+      "step": 13908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845351674602517e-06,
+      "loss": 0.6979,
+      "step": 13909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844030897211933e-06,
+      "loss": 0.6642,
+      "step": 13910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842710292714963e-06,
+      "loss": 0.6554,
+      "step": 13911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841389861148696e-06,
+      "loss": 0.5979,
+      "step": 13912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840069602550217e-06,
+      "loss": 0.5851,
+      "step": 13913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.838749516956607e-06,
+      "loss": 0.6393,
+      "step": 13914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8374296044049395e-06,
+      "loss": 0.6874,
+      "step": 13915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836109864932286e-06,
+      "loss": 0.6675,
+      "step": 13916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8347902985757125e-06,
+      "loss": 0.6403,
+      "step": 13917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.833470905372278e-06,
+      "loss": 0.6223,
+      "step": 13918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8321516853590415e-06,
+      "loss": 0.7095,
+      "step": 13919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830832638573051e-06,
+      "loss": 0.6713,
+      "step": 13920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829513765051357e-06,
+      "loss": 0.706,
+      "step": 13921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828195064830994e-06,
+      "loss": 0.717,
+      "step": 13922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.826876537949003e-06,
+      "loss": 0.6287,
+      "step": 13923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.82555818444241e-06,
+      "loss": 0.7008,
+      "step": 13924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824240004348249e-06,
+      "loss": 0.7037,
+      "step": 13925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822921997703538e-06,
+      "loss": 0.6492,
+      "step": 13926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8216041645453e-06,
+      "loss": 0.7291,
+      "step": 13927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820286504910537e-06,
+      "loss": 0.633,
+      "step": 13928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8189690188362615e-06,
+      "loss": 0.718,
+      "step": 13929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8176517063594755e-06,
+      "loss": 0.6672,
+      "step": 13930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8163345675171745e-06,
+      "loss": 0.6157,
+      "step": 13931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8150176023463535e-06,
+      "loss": 0.6772,
+      "step": 13932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.813700810884e-06,
+      "loss": 0.6997,
+      "step": 13933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8123841931670945e-06,
+      "loss": 0.7021,
+      "step": 13934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8110677492326177e-06,
+      "loss": 0.6125,
+      "step": 13935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8097514791175417e-06,
+      "loss": 0.7062,
+      "step": 13936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8084353828588326e-06,
+      "loss": 0.6073,
+      "step": 13937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807119460493457e-06,
+      "loss": 0.6556,
+      "step": 13938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8058037120583703e-06,
+      "loss": 0.6869,
+      "step": 13939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8044881375905317e-06,
+      "loss": 0.6628,
+      "step": 13940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8031727371268813e-06,
+      "loss": 0.6836,
+      "step": 13941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8018575107043666e-06,
+      "loss": 0.6671,
+      "step": 13942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.800542458359927e-06,
+      "loss": 0.6874,
+      "step": 13943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7992275801304935e-06,
+      "loss": 0.7172,
+      "step": 13944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7979128760530003e-06,
+      "loss": 0.656,
+      "step": 13945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7965983461643696e-06,
+      "loss": 0.6925,
+      "step": 13946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7952839905015228e-06,
+      "loss": 0.7359,
+      "step": 13947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7939698091013687e-06,
+      "loss": 0.5834,
+      "step": 13948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7926558020008196e-06,
+      "loss": 0.684,
+      "step": 13949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79134196923678e-06,
+      "loss": 0.7548,
+      "step": 13950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79002831084615e-06,
+      "loss": 0.6917,
+      "step": 13951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7887148268658236e-06,
+      "loss": 0.6138,
+      "step": 13952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7874015173326907e-06,
+      "loss": 0.663,
+      "step": 13953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.786088382283638e-06,
+      "loss": 0.6072,
+      "step": 13954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.784775421755543e-06,
+      "loss": 0.6425,
+      "step": 13955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7834626357852833e-06,
+      "loss": 0.6774,
+      "step": 13956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7821500244097276e-06,
+      "loss": 0.6777,
+      "step": 13957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.780837587665742e-06,
+      "loss": 0.6393,
+      "step": 13958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7795253255901866e-06,
+      "loss": 0.6765,
+      "step": 13959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778213238219921e-06,
+      "loss": 0.7573,
+      "step": 13960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7769013255917874e-06,
+      "loss": 0.7203,
+      "step": 13961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7755895877426373e-06,
+      "loss": 0.7049,
+      "step": 13962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774278024709309e-06,
+      "loss": 0.6483,
+      "step": 13963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.772966636528641e-06,
+      "loss": 0.6545,
+      "step": 13964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7716554232374593e-06,
+      "loss": 0.6662,
+      "step": 13965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.770344384872601e-06,
+      "loss": 0.6199,
+      "step": 13966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7690335214708764e-06,
+      "loss": 0.5572,
+      "step": 13967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7677228330691053e-06,
+      "loss": 0.7366,
+      "step": 13968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7664123197041003e-06,
+      "loss": 0.7114,
+      "step": 13969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.6472,
+      "step": 13970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7637918182316047e-06,
+      "loss": 0.7065,
+      "step": 13971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7624818301977127e-06,
+      "loss": 0.7021,
+      "step": 13972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.761172017347783e-06,
+      "loss": 0.5919,
+      "step": 13973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7598623797186005e-06,
+      "loss": 0.6704,
+      "step": 13974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7585529173469494e-06,
+      "loss": 0.7397,
+      "step": 13975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7572436302696045e-06,
+      "loss": 0.663,
+      "step": 13976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.755934518523339e-06,
+      "loss": 0.6228,
+      "step": 13977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7546255821449183e-06,
+      "loss": 0.6644,
+      "step": 13978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75331682117111e-06,
+      "loss": 0.6679,
+      "step": 13979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7520082356386643e-06,
+      "loss": 0.6208,
+      "step": 13980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7506998255843363e-06,
+      "loss": 0.6762,
+      "step": 13981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749391591044873e-06,
+      "loss": 0.6975,
+      "step": 13982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7480835320570173e-06,
+      "loss": 0.6657,
+      "step": 13983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.746775648657507e-06,
+      "loss": 0.6353,
+      "step": 13984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745467940883071e-06,
+      "loss": 0.7465,
+      "step": 13985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7441604087704477e-06,
+      "loss": 0.6496,
+      "step": 13986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7428530523563478e-06,
+      "loss": 0.7044,
+      "step": 13987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7415458716774944e-06,
+      "loss": 0.6896,
+      "step": 13988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7402388667706e-06,
+      "loss": 0.6767,
+      "step": 13989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7389320376723713e-06,
+      "loss": 0.5878,
+      "step": 13990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737625384419513e-06,
+      "loss": 0.6059,
+      "step": 13991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736318907048726e-06,
+      "loss": 0.6081,
+      "step": 13992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.735012605596693e-06,
+      "loss": 0.6785,
+      "step": 13993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7337064801001122e-06,
+      "loss": 0.7018,
+      "step": 13994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7324005305956633e-06,
+      "loss": 0.5905,
+      "step": 13995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.731094757120025e-06,
+      "loss": 0.6892,
+      "step": 13996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.72978915970987e-06,
+      "loss": 0.6639,
+      "step": 13997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728483738401868e-06,
+      "loss": 0.7124,
+      "step": 13998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.727178493232685e-06,
+      "loss": 0.6832,
+      "step": 13999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7258734242389725e-06,
+      "loss": 0.6438,
+      "step": 14000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724568531457388e-06,
+      "loss": 0.6483,
+      "step": 14001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723263814924578e-06,
+      "loss": 0.644,
+      "step": 14002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7219592746771893e-06,
+      "loss": 0.6843,
+      "step": 14003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7206549107518586e-06,
+      "loss": 0.6483,
+      "step": 14004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719350723185221e-06,
+      "loss": 0.6378,
+      "step": 14005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.718046712013904e-06,
+      "loss": 0.697,
+      "step": 14006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7167428772745318e-06,
+      "loss": 0.5676,
+      "step": 14007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7154392190037237e-06,
+      "loss": 0.6912,
+      "step": 14008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7141357372380926e-06,
+      "loss": 0.6562,
+      "step": 14009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7128324320142494e-06,
+      "loss": 0.6533,
+      "step": 14010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.711529303368796e-06,
+      "loss": 0.7044,
+      "step": 14011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7102263513383363e-06,
+      "loss": 0.6622,
+      "step": 14012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7089235759594565e-06,
+      "loss": 0.6591,
+      "step": 14013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7076209772687467e-06,
+      "loss": 0.6244,
+      "step": 14014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.706318555302797e-06,
+      "loss": 0.7397,
+      "step": 14015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.705016310098184e-06,
+      "loss": 0.7194,
+      "step": 14016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.703714241691482e-06,
+      "loss": 0.6663,
+      "step": 14017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.702412350119263e-06,
+      "loss": 0.6355,
+      "step": 14018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7011106354180846e-06,
+      "loss": 0.6742,
+      "step": 14019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6998090976245106e-06,
+      "loss": 0.7142,
+      "step": 14020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.698507736775093e-06,
+      "loss": 0.6423,
+      "step": 14021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6972065529063837e-06,
+      "loss": 0.6405,
+      "step": 14022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6959055460549255e-06,
+      "loss": 0.7252,
+      "step": 14023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6946047162572586e-06,
+      "loss": 0.6688,
+      "step": 14024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6933040635499174e-06,
+      "loss": 0.6659,
+      "step": 14025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6920035879694317e-06,
+      "loss": 0.6273,
+      "step": 14026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6907032895523253e-06,
+      "loss": 0.6771,
+      "step": 14027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6894031683351183e-06,
+      "loss": 0.663,
+      "step": 14028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6881032243543246e-06,
+      "loss": 0.6823,
+      "step": 14029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6868034576464553e-06,
+      "loss": 0.6361,
+      "step": 14030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.685503868248017e-06,
+      "loss": 0.6022,
+      "step": 14031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6842044561955036e-06,
+      "loss": 0.7343,
+      "step": 14032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6829052215254125e-06,
+      "loss": 0.6544,
+      "step": 14033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6816061642742294e-06,
+      "loss": 0.6664,
+      "step": 14034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6803072844784473e-06,
+      "loss": 0.6536,
+      "step": 14035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6790085821745403e-06,
+      "loss": 0.6443,
+      "step": 14036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6777100573989855e-06,
+      "loss": 0.6792,
+      "step": 14037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6764117101882556e-06,
+      "loss": 0.6014,
+      "step": 14038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6751135405788074e-06,
+      "loss": 0.7101,
+      "step": 14039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.673815548607105e-06,
+      "loss": 0.6671,
+      "step": 14040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6725177343096017e-06,
+      "loss": 0.6367,
+      "step": 14041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6712200977227496e-06,
+      "loss": 0.6584,
+      "step": 14042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6699226388829933e-06,
+      "loss": 0.6811,
+      "step": 14043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6686253578267705e-06,
+      "loss": 0.578,
+      "step": 14044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6673282545905177e-06,
+      "loss": 0.739,
+      "step": 14045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666031329210663e-06,
+      "loss": 0.6947,
+      "step": 14046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6647345817236344e-06,
+      "loss": 0.6968,
+      "step": 14047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6634380121658484e-06,
+      "loss": 0.6894,
+      "step": 14048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6621416205737215e-06,
+      "loss": 0.6984,
+      "step": 14049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660845406983664e-06,
+      "loss": 0.6917,
+      "step": 14050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.659549371432083e-06,
+      "loss": 0.6243,
+      "step": 14051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6582535139553723e-06,
+      "loss": 0.5524,
+      "step": 14052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6569578345899293e-06,
+      "loss": 0.7258,
+      "step": 14053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.655662333372141e-06,
+      "loss": 0.6674,
+      "step": 14054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6543670103383986e-06,
+      "loss": 0.7325,
+      "step": 14055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6530718655250797e-06,
+      "loss": 0.7202,
+      "step": 14056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6517768989685608e-06,
+      "loss": 0.6785,
+      "step": 14057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650482110705206e-06,
+      "loss": 0.696,
+      "step": 14058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.649187500771384e-06,
+      "loss": 0.6605,
+      "step": 14059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.647893069203453e-06,
+      "loss": 0.6627,
+      "step": 14060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6465988160377695e-06,
+      "loss": 0.7092,
+      "step": 14061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645304741310682e-06,
+      "loss": 0.7102,
+      "step": 14062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6440108450585364e-06,
+      "loss": 0.6435,
+      "step": 14063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.642717127317671e-06,
+      "loss": 0.7185,
+      "step": 14064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.641423588124422e-06,
+      "loss": 0.6768,
+      "step": 14065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6401302275151173e-06,
+      "loss": 0.5833,
+      "step": 14066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6388370455260835e-06,
+      "loss": 0.6591,
+      "step": 14067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6375440421936404e-06,
+      "loss": 0.6329,
+      "step": 14068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6362512175541008e-06,
+      "loss": 0.6491,
+      "step": 14069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6349585716437794e-06,
+      "loss": 0.6364,
+      "step": 14070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.633666104498973e-06,
+      "loss": 0.6733,
+      "step": 14071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.632373816155985e-06,
+      "loss": 0.6576,
+      "step": 14072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6310817066511106e-06,
+      "loss": 0.7026,
+      "step": 14073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6297897760206347e-06,
+      "loss": 0.7093,
+      "step": 14074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6284980243008493e-06,
+      "loss": 0.7236,
+      "step": 14075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6272064515280314e-06,
+      "loss": 0.6735,
+      "step": 14076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625915057738457e-06,
+      "loss": 0.6753,
+      "step": 14077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.624623842968391e-06,
+      "loss": 0.7245,
+      "step": 14078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6233328072540995e-06,
+      "loss": 0.6111,
+      "step": 14079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6220419506318427e-06,
+      "loss": 0.658,
+      "step": 14080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.620751273137875e-06,
+      "loss": 0.6615,
+      "step": 14081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6194607748084456e-06,
+      "loss": 0.6935,
+      "step": 14082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618170455679798e-06,
+      "loss": 0.6113,
+      "step": 14083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6168803157881725e-06,
+      "loss": 0.6473,
+      "step": 14084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6155903551698034e-06,
+      "loss": 0.6589,
+      "step": 14085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6143005738609193e-06,
+      "loss": 0.6478,
+      "step": 14086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6130109718977457e-06,
+      "loss": 0.634,
+      "step": 14087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.611721549316499e-06,
+      "loss": 0.6495,
+      "step": 14088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6104323061533963e-06,
+      "loss": 0.6736,
+      "step": 14089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6091432424446494e-06,
+      "loss": 0.701,
+      "step": 14090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6078543582264534e-06,
+      "loss": 0.6271,
+      "step": 14091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6065656535350137e-06,
+      "loss": 0.6438,
+      "step": 14092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6052771284065212e-06,
+      "loss": 0.6865,
+      "step": 14093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.603988782877168e-06,
+      "loss": 0.6791,
+      "step": 14094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6027006169831325e-06,
+      "loss": 0.6275,
+      "step": 14095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6014126307606036e-06,
+      "loss": 0.7449,
+      "step": 14096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6001248242457453e-06,
+      "loss": 0.6525,
+      "step": 14097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5988371974747305e-06,
+      "loss": 0.5509,
+      "step": 14098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.597549750483722e-06,
+      "loss": 0.7088,
+      "step": 14099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5962624833088788e-06,
+      "loss": 0.6306,
+      "step": 14100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5949753959863543e-06,
+      "loss": 0.663,
+      "step": 14101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.593688488552297e-06,
+      "loss": 0.6643,
+      "step": 14102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5924017610428497e-06,
+      "loss": 0.6622,
+      "step": 14103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.591115213494153e-06,
+      "loss": 0.6119,
+      "step": 14104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.589828845942338e-06,
+      "loss": 0.681,
+      "step": 14105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588542658423534e-06,
+      "loss": 0.6917,
+      "step": 14106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5872566509738626e-06,
+      "loss": 0.6454,
+      "step": 14107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5859708236294456e-06,
+      "loss": 0.6547,
+      "step": 14108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5846851764263923e-06,
+      "loss": 0.6697,
+      "step": 14109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.583399709400817e-06,
+      "loss": 0.6037,
+      "step": 14110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5821144225888137e-06,
+      "loss": 0.6711,
+      "step": 14111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580829316026486e-06,
+      "loss": 0.6615,
+      "step": 14112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.579544389749925e-06,
+      "loss": 0.7036,
+      "step": 14113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578259643795219e-06,
+      "loss": 0.6366,
+      "step": 14114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.576975078198448e-06,
+      "loss": 0.6785,
+      "step": 14115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.575690692995699e-06,
+      "loss": 0.6793,
+      "step": 14116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5744064882230346e-06,
+      "loss": 0.6378,
+      "step": 14117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5731224639165265e-06,
+      "loss": 0.7105,
+      "step": 14118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.571838620112236e-06,
+      "loss": 0.7003,
+      "step": 14119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5705549568462217e-06,
+      "loss": 0.7264,
+      "step": 14120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.569271474154535e-06,
+      "loss": 0.5895,
+      "step": 14121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5679881720732246e-06,
+      "loss": 0.7201,
+      "step": 14122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5667050506383338e-06,
+      "loss": 0.732,
+      "step": 14123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.565422109885892e-06,
+      "loss": 0.7105,
+      "step": 14124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5641393498519396e-06,
+      "loss": 0.7093,
+      "step": 14125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5628567705725014e-06,
+      "loss": 0.6482,
+      "step": 14126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.561574372083598e-06,
+      "loss": 0.7229,
+      "step": 14127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5602921544212487e-06,
+      "loss": 0.6204,
+      "step": 14128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5590101176214654e-06,
+      "loss": 0.7617,
+      "step": 14129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.55772826172025e-06,
+      "loss": 0.6902,
+      "step": 14130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.556446586753608e-06,
+      "loss": 0.6766,
+      "step": 14131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5551650927575345e-06,
+      "loss": 0.6015,
+      "step": 14132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5538837797680214e-06,
+      "loss": 0.6784,
+      "step": 14133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.552602647821055e-06,
+      "loss": 0.6003,
+      "step": 14134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551321696952613e-06,
+      "loss": 0.6796,
+      "step": 14135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5500409271986824e-06,
+      "loss": 0.6739,
+      "step": 14136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.548760338595224e-06,
+      "loss": 0.6475,
+      "step": 14137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.547479931178206e-06,
+      "loss": 0.6058,
+      "step": 14138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546199704983592e-06,
+      "loss": 0.6158,
+      "step": 14139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5449196600473334e-06,
+      "loss": 0.7174,
+      "step": 14140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543639796405386e-06,
+      "loss": 0.6796,
+      "step": 14141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5423601140936947e-06,
+      "loss": 0.6636,
+      "step": 14142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5410806131481958e-06,
+      "loss": 0.6669,
+      "step": 14143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.539801293604824e-06,
+      "loss": 0.6668,
+      "step": 14144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5385221554995166e-06,
+      "loss": 0.6683,
+      "step": 14145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5372431988681944e-06,
+      "loss": 0.7117,
+      "step": 14146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5359644237467785e-06,
+      "loss": 0.6325,
+      "step": 14147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534685830171184e-06,
+      "loss": 0.601,
+      "step": 14148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.533407418177325e-06,
+      "loss": 0.6551,
+      "step": 14149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5321291878010976e-06,
+      "loss": 0.6782,
+      "step": 14150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5308511390784063e-06,
+      "loss": 0.6533,
+      "step": 14151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.529573272045146e-06,
+      "loss": 0.708,
+      "step": 14152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.528295586737205e-06,
+      "loss": 0.6927,
+      "step": 14153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.52701808319047e-06,
+      "loss": 0.7111,
+      "step": 14154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525740761440818e-06,
+      "loss": 0.6212,
+      "step": 14155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5244636215241236e-06,
+      "loss": 0.6307,
+      "step": 14156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.523186663476257e-06,
+      "loss": 0.6967,
+      "step": 14157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5219098873330827e-06,
+      "loss": 0.5922,
+      "step": 14158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5206332931304577e-06,
+      "loss": 0.6361,
+      "step": 14159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5193568809042365e-06,
+      "loss": 0.6901,
+      "step": 14160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.518080650690269e-06,
+      "loss": 0.6614,
+      "step": 14161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5168046025244005e-06,
+      "loss": 0.7082,
+      "step": 14162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515528736442464e-06,
+      "loss": 0.3022,
+      "step": 14163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.514253052480292e-06,
+      "loss": 0.6789,
+      "step": 14164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.512977550673721e-06,
+      "loss": 0.6488,
+      "step": 14165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511702231058568e-06,
+      "loss": 0.6674,
+      "step": 14166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510427093670653e-06,
+      "loss": 0.6078,
+      "step": 14167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.509152138545793e-06,
+      "loss": 0.6588,
+      "step": 14168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5078773657197873e-06,
+      "loss": 0.683,
+      "step": 14169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5066027752284427e-06,
+      "loss": 0.6458,
+      "step": 14170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5053283671075556e-06,
+      "loss": 0.679,
+      "step": 14171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.504054141392921e-06,
+      "loss": 0.6435,
+      "step": 14172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502780098120324e-06,
+      "loss": 0.696,
+      "step": 14173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.501506237325547e-06,
+      "loss": 0.283,
+      "step": 14174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5002325590443676e-06,
+      "loss": 0.6593,
+      "step": 14175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4989590633125583e-06,
+      "loss": 0.6773,
+      "step": 14176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4976857501658844e-06,
+      "loss": 0.6702,
+      "step": 14177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4964126196401094e-06,
+      "loss": 0.7365,
+      "step": 14178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4951396717709874e-06,
+      "loss": 0.6653,
+      "step": 14179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493866906594272e-06,
+      "loss": 0.6602,
+      "step": 14180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4925943241457117e-06,
+      "loss": 0.6754,
+      "step": 14181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.491321924461042e-06,
+      "loss": 0.7451,
+      "step": 14182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.490049707576001e-06,
+      "loss": 0.6976,
+      "step": 14183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488777673526317e-06,
+      "loss": 0.7174,
+      "step": 14184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487505822347722e-06,
+      "loss": 0.7349,
+      "step": 14185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4862341540759325e-06,
+      "loss": 0.6791,
+      "step": 14186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4849626687466665e-06,
+      "loss": 0.6508,
+      "step": 14187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483691366395635e-06,
+      "loss": 0.6942,
+      "step": 14188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4824202470585378e-06,
+      "loss": 0.6992,
+      "step": 14189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481149310771078e-06,
+      "loss": 0.6934,
+      "step": 14190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479878557568952e-06,
+      "loss": 0.6356,
+      "step": 14191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4786079874878464e-06,
+      "loss": 0.6568,
+      "step": 14192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4773376005634486e-06,
+      "loss": 0.6461,
+      "step": 14193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.476067396831436e-06,
+      "loss": 0.6359,
+      "step": 14194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4747973763274858e-06,
+      "loss": 0.5801,
+      "step": 14195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.473527539087264e-06,
+      "loss": 0.7013,
+      "step": 14196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472257885146436e-06,
+      "loss": 0.6386,
+      "step": 14197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4709884145406616e-06,
+      "loss": 0.6708,
+      "step": 14198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469719127305593e-06,
+      "loss": 0.7022,
+      "step": 14199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.468450023476879e-06,
+      "loss": 0.7,
+      "step": 14200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467181103090168e-06,
+      "loss": 0.6985,
+      "step": 14201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.46591236618109e-06,
+      "loss": 0.6437,
+      "step": 14202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.464643812785282e-06,
+      "loss": 0.6942,
+      "step": 14203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4633754429383726e-06,
+      "loss": 0.6168,
+      "step": 14204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4621072566759796e-06,
+      "loss": 0.6803,
+      "step": 14205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4608392540337287e-06,
+      "loss": 0.6078,
+      "step": 14206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.459571435047232e-06,
+      "loss": 0.633,
+      "step": 14207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4583037997520906e-06,
+      "loss": 0.6328,
+      "step": 14208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4570363481839097e-06,
+      "loss": 0.6483,
+      "step": 14209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4557690803782863e-06,
+      "loss": 0.6412,
+      "step": 14210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4545019963708126e-06,
+      "loss": 0.6116,
+      "step": 14211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.453235096197075e-06,
+      "loss": 0.6113,
+      "step": 14212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.451968379892655e-06,
+      "loss": 0.6327,
+      "step": 14213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4507018474931285e-06,
+      "loss": 0.7131,
+      "step": 14214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449435499034068e-06,
+      "loss": 0.754,
+      "step": 14215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448169334551039e-06,
+      "loss": 0.6351,
+      "step": 14216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4469033540796017e-06,
+      "loss": 0.6867,
+      "step": 14217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4456375576553124e-06,
+      "loss": 0.6857,
+      "step": 14218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4443719453137226e-06,
+      "loss": 0.6445,
+      "step": 14219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.44310651709038e-06,
+      "loss": 0.7303,
+      "step": 14220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4418412730208174e-06,
+      "loss": 0.7463,
+      "step": 14221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440576213140574e-06,
+      "loss": 0.7665,
+      "step": 14222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4393113374851807e-06,
+      "loss": 0.6834,
+      "step": 14223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4380466460901607e-06,
+      "loss": 0.6612,
+      "step": 14224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4367821389910303e-06,
+      "loss": 0.6302,
+      "step": 14225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4355178162233103e-06,
+      "loss": 0.6793,
+      "step": 14226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.434253677822512e-06,
+      "loss": 0.6685,
+      "step": 14227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4329897238241305e-06,
+      "loss": 0.6763,
+      "step": 14228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4317259542636683e-06,
+      "loss": 0.683,
+      "step": 14229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4304623691766193e-06,
+      "loss": 0.693,
+      "step": 14230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4291989685984718e-06,
+      "loss": 0.5897,
+      "step": 14231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4279357525647096e-06,
+      "loss": 0.7637,
+      "step": 14232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4266727211108107e-06,
+      "loss": 0.6845,
+      "step": 14233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4254098742722473e-06,
+      "loss": 0.6265,
+      "step": 14234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.424147212084488e-06,
+      "loss": 0.6036,
+      "step": 14235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422884734582994e-06,
+      "loss": 0.6726,
+      "step": 14236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.421622441803224e-06,
+      "loss": 0.6716,
+      "step": 14237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.42036033378063e-06,
+      "loss": 0.6444,
+      "step": 14238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4190984105506586e-06,
+      "loss": 0.6005,
+      "step": 14239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4178366721487544e-06,
+      "loss": 0.6159,
+      "step": 14240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416575118610348e-06,
+      "loss": 0.6434,
+      "step": 14241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415313749970874e-06,
+      "loss": 0.619,
+      "step": 14242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4140525662657597e-06,
+      "loss": 0.6121,
+      "step": 14243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4127915675304236e-06,
+      "loss": 0.7288,
+      "step": 14244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4115307538002796e-06,
+      "loss": 0.6194,
+      "step": 14245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4102701251107484e-06,
+      "loss": 0.6917,
+      "step": 14246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4090096814972263e-06,
+      "loss": 0.666,
+      "step": 14247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4077494229951146e-06,
+      "loss": 0.5934,
+      "step": 14248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.40648934963981e-06,
+      "loss": 0.6081,
+      "step": 14249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4052294614667012e-06,
+      "loss": 0.6106,
+      "step": 14250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4039697585111744e-06,
+      "loss": 0.6335,
+      "step": 14251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4027102408086076e-06,
+      "loss": 0.6599,
+      "step": 14252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.401450908394379e-06,
+      "loss": 0.6949,
+      "step": 14253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4001917613038482e-06,
+      "loss": 0.6363,
+      "step": 14254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3989327995723874e-06,
+      "loss": 0.7005,
+      "step": 14255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397674023235353e-06,
+      "loss": 0.7261,
+      "step": 14256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3964154323280985e-06,
+      "loss": 0.7157,
+      "step": 14257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3951570268859714e-06,
+      "loss": 0.7151,
+      "step": 14258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.393898806944318e-06,
+      "loss": 0.6312,
+      "step": 14259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3926407725384703e-06,
+      "loss": 0.6351,
+      "step": 14260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.391382923703763e-06,
+      "loss": 0.6373,
+      "step": 14261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.390125260475524e-06,
+      "loss": 0.6519,
+      "step": 14262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.388867782889076e-06,
+      "loss": 0.6631,
+      "step": 14263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3876104909797348e-06,
+      "loss": 0.6765,
+      "step": 14264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3863533847828103e-06,
+      "loss": 0.6425,
+      "step": 14265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.385096464333617e-06,
+      "loss": 0.6929,
+      "step": 14266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3838397296674473e-06,
+      "loss": 0.6906,
+      "step": 14267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.382583180819601e-06,
+      "loss": 0.6176,
+      "step": 14268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3813268178253687e-06,
+      "loss": 0.7651,
+      "step": 14269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3800706407200354e-06,
+      "loss": 0.6072,
+      "step": 14270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3788146495388818e-06,
+      "loss": 0.7013,
+      "step": 14271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.377558844317187e-06,
+      "loss": 0.6265,
+      "step": 14272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3763032250902127e-06,
+      "loss": 0.5951,
+      "step": 14273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3750477918932243e-06,
+      "loss": 0.6207,
+      "step": 14274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3737925447614892e-06,
+      "loss": 0.6859,
+      "step": 14275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.372537483730258e-06,
+      "loss": 0.7331,
+      "step": 14276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3712826088347783e-06,
+      "loss": 0.6752,
+      "step": 14277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3700279201102947e-06,
+      "loss": 0.6643,
+      "step": 14278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3687734175920505e-06,
+      "loss": 0.6558,
+      "step": 14279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3675191013152696e-06,
+      "loss": 0.6369,
+      "step": 14280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3662649713151864e-06,
+      "loss": 0.668,
+      "step": 14281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.365011027627021e-06,
+      "loss": 0.7008,
+      "step": 14282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.363757270285993e-06,
+      "loss": 0.7302,
+      "step": 14283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3625036993273142e-06,
+      "loss": 0.6603,
+      "step": 14284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3612503147861917e-06,
+      "loss": 0.6479,
+      "step": 14285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3599971166978287e-06,
+      "loss": 0.6072,
+      "step": 14286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.35874410509742e-06,
+      "loss": 0.6024,
+      "step": 14287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3574912800201574e-06,
+      "loss": 0.6624,
+      "step": 14288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3562386415012292e-06,
+      "loss": 0.6828,
+      "step": 14289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.354986189575814e-06,
+      "loss": 0.6874,
+      "step": 14290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.353733924279088e-06,
+      "loss": 0.7175,
+      "step": 14291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352481845646226e-06,
+      "loss": 0.6948,
+      "step": 14292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.351229953712386e-06,
+      "loss": 0.7131,
+      "step": 14293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3499782485127285e-06,
+      "loss": 0.6221,
+      "step": 14294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3487267300824155e-06,
+      "loss": 0.6395,
+      "step": 14295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347475398456591e-06,
+      "loss": 0.6349,
+      "step": 14296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3462242536704016e-06,
+      "loss": 0.6592,
+      "step": 14297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3449732957589876e-06,
+      "loss": 0.7735,
+      "step": 14298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3437225247574777e-06,
+      "loss": 0.6898,
+      "step": 14299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.342471940701004e-06,
+      "loss": 0.6987,
+      "step": 14300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3412215436246887e-06,
+      "loss": 0.6042,
+      "step": 14301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3399713335636508e-06,
+      "loss": 0.6699,
+      "step": 14302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3387213105530027e-06,
+      "loss": 0.7112,
+      "step": 14303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.337471474627851e-06,
+      "loss": 0.3016,
+      "step": 14304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3362218258232993e-06,
+      "loss": 0.6577,
+      "step": 14305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3349723641744447e-06,
+      "loss": 0.5958,
+      "step": 14306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3337230897163774e-06,
+      "loss": 0.6669,
+      "step": 14307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3324740024841863e-06,
+      "loss": 0.6099,
+      "step": 14308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3312251025129515e-06,
+      "loss": 0.6321,
+      "step": 14309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.329976389837748e-06,
+      "loss": 0.7051,
+      "step": 14310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3287278644936515e-06,
+      "loss": 0.6163,
+      "step": 14311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.32747952651572e-06,
+      "loss": 0.7134,
+      "step": 14312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3262313759390165e-06,
+      "loss": 0.6855,
+      "step": 14313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324983412798597e-06,
+      "loss": 0.7059,
+      "step": 14314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.323735637129508e-06,
+      "loss": 0.6623,
+      "step": 14315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3224880489668e-06,
+      "loss": 0.679,
+      "step": 14316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3212406483455083e-06,
+      "loss": 0.6552,
+      "step": 14317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3199934353006725e-06,
+      "loss": 0.6416,
+      "step": 14318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3187464098673115e-06,
+      "loss": 0.7001,
+      "step": 14319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.317499572080454e-06,
+      "loss": 0.6537,
+      "step": 14320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316252921975116e-06,
+      "loss": 0.6739,
+      "step": 14321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.315006459586313e-06,
+      "loss": 0.6074,
+      "step": 14322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3137601849490508e-06,
+      "loss": 0.6291,
+      "step": 14323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3125140980983326e-06,
+      "loss": 0.6895,
+      "step": 14324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3112681990691542e-06,
+      "loss": 0.6881,
+      "step": 14325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3100224878965094e-06,
+      "loss": 0.6639,
+      "step": 14326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3087769646153824e-06,
+      "loss": 0.6799,
+      "step": 14327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3075316292607552e-06,
+      "loss": 0.6801,
+      "step": 14328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3062864818676044e-06,
+      "loss": 0.6974,
+      "step": 14329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3050415224708997e-06,
+      "loss": 0.662,
+      "step": 14330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30379675110561e-06,
+      "loss": 0.7816,
+      "step": 14331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30255216780669e-06,
+      "loss": 0.6778,
+      "step": 14332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3013077726090958e-06,
+      "loss": 0.6385,
+      "step": 14333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3000635655477773e-06,
+      "loss": 0.6034,
+      "step": 14334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2988195466576754e-06,
+      "loss": 0.7127,
+      "step": 14335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2975757159737377e-06,
+      "loss": 0.646,
+      "step": 14336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2963320735308947e-06,
+      "loss": 0.6548,
+      "step": 14337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2950886193640695e-06,
+      "loss": 0.656,
+      "step": 14338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2938453535081895e-06,
+      "loss": 0.6247,
+      "step": 14339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.292602275998171e-06,
+      "loss": 0.6702,
+      "step": 14340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2913593868689277e-06,
+      "loss": 0.6426,
+      "step": 14341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2901166861553656e-06,
+      "loss": 0.6925,
+      "step": 14342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2888741738923867e-06,
+      "loss": 0.697,
+      "step": 14343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2876318501148884e-06,
+      "loss": 0.7053,
+      "step": 14344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2863897148577616e-06,
+      "loss": 0.6223,
+      "step": 14345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2851477681558928e-06,
+      "loss": 0.6168,
+      "step": 14346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.283906010044162e-06,
+      "loss": 0.6426,
+      "step": 14347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2826644405574448e-06,
+      "loss": 0.5767,
+      "step": 14348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.281423059730612e-06,
+      "loss": 0.6277,
+      "step": 14349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.28018186759853e-06,
+      "loss": 0.5875,
+      "step": 14350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278940864196054e-06,
+      "loss": 0.634,
+      "step": 14351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.27770004955804e-06,
+      "loss": 0.6502,
+      "step": 14352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276459423719337e-06,
+      "loss": 0.7317,
+      "step": 14353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2752189867147887e-06,
+      "loss": 0.6103,
+      "step": 14354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.273978738579232e-06,
+      "loss": 0.6804,
+      "step": 14355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2727386793475036e-06,
+      "loss": 0.6922,
+      "step": 14356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.271498809054433e-06,
+      "loss": 0.6652,
+      "step": 14357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.270259127734835e-06,
+      "loss": 0.687,
+      "step": 14358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2690196354235314e-06,
+      "loss": 0.6991,
+      "step": 14359
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2677803321553324e-06,
+      "loss": 0.6974,
+      "step": 14360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2665412179650457e-06,
+      "loss": 0.605,
+      "step": 14361
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2653022928874713e-06,
+      "loss": 0.6206,
+      "step": 14362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2640635569574095e-06,
+      "loss": 0.6567,
+      "step": 14363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2628250102096404e-06,
+      "loss": 0.6678,
+      "step": 14364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2615866526789584e-06,
+      "loss": 0.6626,
+      "step": 14365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2603484844001422e-06,
+      "loss": 0.6228,
+      "step": 14366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259110505407965e-06,
+      "loss": 0.6577,
+      "step": 14367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.257872715737196e-06,
+      "loss": 0.6581,
+      "step": 14368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2566351154225994e-06,
+      "loss": 0.6587,
+      "step": 14369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255397704498937e-06,
+      "loss": 0.602,
+      "step": 14370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2541604830009555e-06,
+      "loss": 0.6535,
+      "step": 14371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2529234509634068e-06,
+      "loss": 0.595,
+      "step": 14372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.251686608421033e-06,
+      "loss": 0.7261,
+      "step": 14373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2504499554085713e-06,
+      "loss": 0.6652,
+      "step": 14374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2492134919607523e-06,
+      "loss": 0.6639,
+      "step": 14375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247977218112309e-06,
+      "loss": 0.6279,
+      "step": 14376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246741133897956e-06,
+      "loss": 0.6735,
+      "step": 14377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2455052393524124e-06,
+      "loss": 0.6531,
+      "step": 14378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244269534510387e-06,
+      "loss": 0.6426,
+      "step": 14379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.243034019406587e-06,
+      "loss": 0.6582,
+      "step": 14380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2417986940757118e-06,
+      "loss": 0.6467,
+      "step": 14381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2405635585524566e-06,
+      "loss": 0.6931,
+      "step": 14382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.239328612871514e-06,
+      "loss": 0.6351,
+      "step": 14383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.238093857067558e-06,
+      "loss": 0.7689,
+      "step": 14384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.236859291175277e-06,
+      "loss": 0.6552,
+      "step": 14385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2356249152293416e-06,
+      "loss": 0.6557,
+      "step": 14386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2343907292644196e-06,
+      "loss": 0.6787,
+      "step": 14387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2331567333151736e-06,
+      "loss": 0.6699,
+      "step": 14388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2319229274162644e-06,
+      "loss": 0.6202,
+      "step": 14389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2306893116023384e-06,
+      "loss": 0.6437,
+      "step": 14390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.229455885908045e-06,
+      "loss": 0.6379,
+      "step": 14391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228222650368025e-06,
+      "loss": 0.6135,
+      "step": 14392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.226989605016915e-06,
+      "loss": 0.7042,
+      "step": 14393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225756749889346e-06,
+      "loss": 0.6974,
+      "step": 14394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.224524085019943e-06,
+      "loss": 0.6771,
+      "step": 14395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2232916104433264e-06,
+      "loss": 0.6116,
+      "step": 14396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2220593261941104e-06,
+      "loss": 0.6207,
+      "step": 14397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2208272323069044e-06,
+      "loss": 0.6304,
+      "step": 14398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.219595328816313e-06,
+      "loss": 0.6906,
+      "step": 14399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.218363615756933e-06,
+      "loss": 0.6271,
+      "step": 14400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217132093163361e-06,
+      "loss": 0.6943,
+      "step": 14401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2159007610701853e-06,
+      "loss": 0.7004,
+      "step": 14402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214669619511983e-06,
+      "loss": 0.6778,
+      "step": 14403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213438668523332e-06,
+      "loss": 0.6342,
+      "step": 14404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2122079081388103e-06,
+      "loss": 0.8127,
+      "step": 14405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.210977338392981e-06,
+      "loss": 0.6481,
+      "step": 14406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2097469593204043e-06,
+      "loss": 0.7041,
+      "step": 14407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2085167709556373e-06,
+      "loss": 0.6382,
+      "step": 14408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2072867733332346e-06,
+      "loss": 0.6484,
+      "step": 14409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2060569664877326e-06,
+      "loss": 0.6611,
+      "step": 14410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2048273504536765e-06,
+      "loss": 0.6849,
+      "step": 14411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.203597925265598e-06,
+      "loss": 0.6724,
+      "step": 14412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2023686909580287e-06,
+      "loss": 0.6868,
+      "step": 14413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.201139647565492e-06,
+      "loss": 0.697,
+      "step": 14414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199910795122505e-06,
+      "loss": 0.6814,
+      "step": 14415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.198682133663581e-06,
+      "loss": 0.6813,
+      "step": 14416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.197453663223228e-06,
+      "loss": 0.6067,
+      "step": 14417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.196225383835948e-06,
+      "loss": 0.6736,
+      "step": 14418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1949972955362383e-06,
+      "loss": 0.6647,
+      "step": 14419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1937693983585905e-06,
+      "loss": 0.6708,
+      "step": 14420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1925416923374897e-06,
+      "loss": 0.6583,
+      "step": 14421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1913141775074197e-06,
+      "loss": 0.6418,
+      "step": 14422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190086853902852e-06,
+      "loss": 0.6169,
+      "step": 14423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1888597215582527e-06,
+      "loss": 0.6252,
+      "step": 14424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1876327805080966e-06,
+      "loss": 0.7407,
+      "step": 14425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1864060307868383e-06,
+      "loss": 0.6791,
+      "step": 14426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185179472428931e-06,
+      "loss": 0.6568,
+      "step": 14427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1839531054688276e-06,
+      "loss": 0.6475,
+      "step": 14428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1827269299409645e-06,
+      "loss": 0.6643,
+      "step": 14429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1815009458797817e-06,
+      "loss": 0.622,
+      "step": 14430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.180275153319713e-06,
+      "loss": 0.6445,
+      "step": 14431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1790495522951848e-06,
+      "loss": 0.6721,
+      "step": 14432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.177824142840619e-06,
+      "loss": 0.6606,
+      "step": 14433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1765989249904307e-06,
+      "loss": 0.6335,
+      "step": 14434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1753738987790315e-06,
+      "loss": 0.7117,
+      "step": 14435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174149064240828e-06,
+      "loss": 0.6064,
+      "step": 14436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1729244214102194e-06,
+      "loss": 0.6888,
+      "step": 14437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171699970321601e-06,
+      "loss": 0.6275,
+      "step": 14438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.170475711009361e-06,
+      "loss": 0.6498,
+      "step": 14439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1692516435078847e-06,
+      "loss": 0.7076,
+      "step": 14440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1680277678515523e-06,
+      "loss": 0.6632,
+      "step": 14441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1668040840747315e-06,
+      "loss": 0.6974,
+      "step": 14442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.165580592211793e-06,
+      "loss": 0.6385,
+      "step": 14443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1643572922971e-06,
+      "loss": 0.684,
+      "step": 14444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1631341843650066e-06,
+      "loss": 0.6993,
+      "step": 14445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.161911268449869e-06,
+      "loss": 0.6385,
+      "step": 14446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.160688544586031e-06,
+      "loss": 0.6682,
+      "step": 14447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159466012807838e-06,
+      "loss": 0.7057,
+      "step": 14448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.158243673149617e-06,
+      "loss": 0.6357,
+      "step": 14449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1570215256457026e-06,
+      "loss": 0.6868,
+      "step": 14450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.155799570330419e-06,
+      "loss": 0.5606,
+      "step": 14451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154577807238085e-06,
+      "loss": 0.5658,
+      "step": 14452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1533562364030155e-06,
+      "loss": 0.6692,
+      "step": 14453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1521348578595178e-06,
+      "loss": 0.3136,
+      "step": 14454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1509136716418963e-06,
+      "loss": 0.6041,
+      "step": 14455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149692677784447e-06,
+      "loss": 0.6237,
+      "step": 14456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148471876321464e-06,
+      "loss": 0.7366,
+      "step": 14457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1472512672872334e-06,
+      "loss": 0.6072,
+      "step": 14458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1460308507160354e-06,
+      "loss": 0.7031,
+      "step": 14459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1448106266421485e-06,
+      "loss": 0.7106,
+      "step": 14460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1435905950998437e-06,
+      "loss": 0.5842,
+      "step": 14461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1423707561233818e-06,
+      "loss": 0.6759,
+      "step": 14462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1411511097470247e-06,
+      "loss": 0.6517,
+      "step": 14463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1399316560050287e-06,
+      "loss": 0.6671,
+      "step": 14464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.138712394931638e-06,
+      "loss": 0.6754,
+      "step": 14465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137493326561102e-06,
+      "loss": 0.6901,
+      "step": 14466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1362744509276567e-06,
+      "loss": 0.6627,
+      "step": 14467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135055768065538e-06,
+      "loss": 0.631,
+      "step": 14468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1338372780089666e-06,
+      "loss": 0.7484,
+      "step": 14469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132618980792168e-06,
+      "loss": 0.6765,
+      "step": 14470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1314008764493586e-06,
+      "loss": 0.6204,
+      "step": 14471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.13018296501475e-06,
+      "loss": 0.6766,
+      "step": 14472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.128965246522546e-06,
+      "loss": 0.6752,
+      "step": 14473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1277477210069486e-06,
+      "loss": 0.6192,
+      "step": 14474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1265303885021514e-06,
+      "loss": 0.6493,
+      "step": 14475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1253132490423454e-06,
+      "loss": 0.6672,
+      "step": 14476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1240963026617142e-06,
+      "loss": 0.6215,
+      "step": 14477
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1228795493944353e-06,
+      "loss": 0.6764,
+      "step": 14478
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.121662989274683e-06,
+      "loss": 0.6733,
+      "step": 14479
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1204466223366238e-06,
+      "loss": 0.6103,
+      "step": 14480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119230448614424e-06,
+      "loss": 0.6349,
+      "step": 14481
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1180144681422355e-06,
+      "loss": 0.6673,
+      "step": 14482
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1167986809542095e-06,
+      "loss": 0.6389,
+      "step": 14483
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.115583087084495e-06,
+      "loss": 0.7285,
+      "step": 14484
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.114367686567228e-06,
+      "loss": 0.6873,
+      "step": 14485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.113152479436551e-06,
+      "loss": 0.6867,
+      "step": 14486
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.111937465726592e-06,
+      "loss": 0.6447,
+      "step": 14487
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1107226454714713e-06,
+      "loss": 0.6393,
+      "step": 14488
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1095080187053084e-06,
+      "loss": 0.6495,
+      "step": 14489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1082935854622186e-06,
+      "loss": 0.6238,
+      "step": 14490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1070793457763106e-06,
+      "loss": 0.6362,
+      "step": 14491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105865299681684e-06,
+      "loss": 0.6886,
+      "step": 14492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.104651447212438e-06,
+      "loss": 0.6066,
+      "step": 14493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1034377884026647e-06,
+      "loss": 0.6083,
+      "step": 14494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.102224323286449e-06,
+      "loss": 0.6826,
+      "step": 14495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.101011051897872e-06,
+      "loss": 0.6134,
+      "step": 14496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0997979742710095e-06,
+      "loss": 0.6227,
+      "step": 14497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0985850904399305e-06,
+      "loss": 0.7075,
+      "step": 14498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0973724004387017e-06,
+      "loss": 0.6509,
+      "step": 14499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0961599043013823e-06,
+      "loss": 0.6718,
+      "step": 14500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.094947602062023e-06,
+      "loss": 0.6844,
+      "step": 14501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0937354937546717e-06,
+      "loss": 0.6871,
+      "step": 14502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0925235794133725e-06,
+      "loss": 0.6876,
+      "step": 14503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0913118590721624e-06,
+      "loss": 0.6378,
+      "step": 14504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0901003327650747e-06,
+      "loss": 0.7025,
+      "step": 14505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.088889000526131e-06,
+      "loss": 0.6278,
+      "step": 14506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087677862389361e-06,
+      "loss": 0.6195,
+      "step": 14507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086466918388773e-06,
+      "loss": 0.6701,
+      "step": 14508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0852561685583783e-06,
+      "loss": 0.6152,
+      "step": 14509
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.084045612932183e-06,
+      "loss": 0.2843,
+      "step": 14510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.082835251544184e-06,
+      "loss": 0.5915,
+      "step": 14511
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081625084428377e-06,
+      "loss": 0.6699,
+      "step": 14512
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080415111618752e-06,
+      "loss": 0.6782,
+      "step": 14513
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0792053331492822e-06,
+      "loss": 0.69,
+      "step": 14514
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0779957490539557e-06,
+      "loss": 0.6785,
+      "step": 14515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0767863593667404e-06,
+      "loss": 0.6357,
+      "step": 14516
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0755771641216013e-06,
+      "loss": 0.2728,
+      "step": 14517
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0743681633525025e-06,
+      "loss": 0.6828,
+      "step": 14518
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.073159357093396e-06,
+      "loss": 0.7018,
+      "step": 14519
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.071950745378237e-06,
+      "loss": 0.347,
+      "step": 14520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.070742328240963e-06,
+      "loss": 0.6089,
+      "step": 14521
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0695341057155158e-06,
+      "loss": 0.69,
+      "step": 14522
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0683260778358305e-06,
+      "loss": 0.6707,
+      "step": 14523
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.067118244635833e-06,
+      "loss": 0.6419,
+      "step": 14524
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.065910606149447e-06,
+      "loss": 0.6877,
+      "step": 14525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0647031624105906e-06,
+      "loss": 0.6416,
+      "step": 14526
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063495913453174e-06,
+      "loss": 0.6722,
+      "step": 14527
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0622888593111044e-06,
+      "loss": 0.6232,
+      "step": 14528
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0610820000182817e-06,
+      "loss": 0.3118,
+      "step": 14529
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.059875335608601e-06,
+      "loss": 0.6565,
+      "step": 14530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0586688661159537e-06,
+      "loss": 0.7169,
+      "step": 14531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.057462591574223e-06,
+      "loss": 0.667,
+      "step": 14532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0562565120172906e-06,
+      "loss": 0.6794,
+      "step": 14533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0550506274790216e-06,
+      "loss": 0.6837,
+      "step": 14534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.053844937993292e-06,
+      "loss": 0.6307,
+      "step": 14535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.052639443593962e-06,
+      "loss": 0.6465,
+      "step": 14536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0514341443148878e-06,
+      "loss": 0.6782,
+      "step": 14537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0502290401899217e-06,
+      "loss": 0.6578,
+      "step": 14538
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.049024131252912e-06,
+      "loss": 0.6752,
+      "step": 14539
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.047819417537694e-06,
+      "loss": 0.725,
+      "step": 14540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.046614899078104e-06,
+      "loss": 0.7021,
+      "step": 14541
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0454105759079745e-06,
+      "loss": 0.5927,
+      "step": 14542
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0442064480611267e-06,
+      "loss": 0.5977,
+      "step": 14543
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0430025155713795e-06,
+      "loss": 0.6163,
+      "step": 14544
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0417987784725477e-06,
+      "loss": 0.6106,
+      "step": 14545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.040595236798438e-06,
+      "loss": 0.6725,
+      "step": 14546
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039391890582851e-06,
+      "loss": 0.6127,
+      "step": 14547
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0381887398595867e-06,
+      "loss": 0.7011,
+      "step": 14548
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0369857846624327e-06,
+      "loss": 0.6132,
+      "step": 14549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0357830250251763e-06,
+      "loss": 0.6929,
+      "step": 14550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.034580460981599e-06,
+      "loss": 0.6929,
+      "step": 14551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0333780925654756e-06,
+      "loss": 0.7041,
+      "step": 14552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0321759198105707e-06,
+      "loss": 0.7057,
+      "step": 14553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0309739427506503e-06,
+      "loss": 0.6264,
+      "step": 14554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0297721614194697e-06,
+      "loss": 0.3089,
+      "step": 14555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.028570575850789e-06,
+      "loss": 0.6945,
+      "step": 14556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0273691860783517e-06,
+      "loss": 0.7268,
+      "step": 14557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.026167992135898e-06,
+      "loss": 0.6346,
+      "step": 14558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024966994057168e-06,
+      "loss": 0.7189,
+      "step": 14559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0237661918758876e-06,
+      "loss": 0.6463,
+      "step": 14560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.022565585625783e-06,
+      "loss": 0.6434,
+      "step": 14561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0213651753405737e-06,
+      "loss": 0.6303,
+      "step": 14562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.020164961053976e-06,
+      "loss": 0.3097,
+      "step": 14563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.018964942799697e-06,
+      "loss": 0.5934,
+      "step": 14564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017765120611441e-06,
+      "loss": 0.6082,
+      "step": 14565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.016565494522904e-06,
+      "loss": 0.661,
+      "step": 14566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0153660645677794e-06,
+      "loss": 0.6887,
+      "step": 14567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014166830779753e-06,
+      "loss": 0.3233,
+      "step": 14568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.012967793192506e-06,
+      "loss": 0.6545,
+      "step": 14569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0117689518397153e-06,
+      "loss": 0.762,
+      "step": 14570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0105703067550495e-06,
+      "loss": 0.6707,
+      "step": 14571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0093718579721764e-06,
+      "loss": 0.6917,
+      "step": 14572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.008173605524749e-06,
+      "loss": 0.6557,
+      "step": 14573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.006975549446425e-06,
+      "loss": 0.7022,
+      "step": 14574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0057776897708478e-06,
+      "loss": 0.7047,
+      "step": 14575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.004580026531667e-06,
+      "loss": 0.7268,
+      "step": 14576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0033825597625167e-06,
+      "loss": 0.6652,
+      "step": 14577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0021852894970306e-06,
+      "loss": 0.7291,
+      "step": 14578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0009882157688296e-06,
+      "loss": 0.6803,
+      "step": 14579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9997913386115364e-06,
+      "loss": 0.6415,
+      "step": 14580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9985946580587653e-06,
+      "loss": 0.6519,
+      "step": 14581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997398174144127e-06,
+      "loss": 0.7312,
+      "step": 14582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9962018869012245e-06,
+      "loss": 0.6392,
+      "step": 14583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995005796363657e-06,
+      "loss": 0.676,
+      "step": 14584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993809902565017e-06,
+      "loss": 0.6576,
+      "step": 14585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.992614205538892e-06,
+      "loss": 0.6965,
+      "step": 14586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9914187053188624e-06,
+      "loss": 0.6544,
+      "step": 14587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9902234019385056e-06,
+      "loss": 0.7059,
+      "step": 14588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989028295431393e-06,
+      "loss": 0.6855,
+      "step": 14589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9878333858310883e-06,
+      "loss": 0.6586,
+      "step": 14590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.986638673171156e-06,
+      "loss": 0.6271,
+      "step": 14591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.985444157485142e-06,
+      "loss": 0.7303,
+      "step": 14592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.984249838806599e-06,
+      "loss": 0.5919,
+      "step": 14593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.98305571716907e-06,
+      "loss": 0.6978,
+      "step": 14594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.98186179260609e-06,
+      "loss": 0.6742,
+      "step": 14595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9806680651511964e-06,
+      "loss": 0.6027,
+      "step": 14596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9794745348379127e-06,
+      "loss": 0.7252,
+      "step": 14597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9782812016997642e-06,
+      "loss": 0.5899,
+      "step": 14598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977088065770257e-06,
+      "loss": 0.6943,
+      "step": 14599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975895127082907e-06,
+      "loss": 0.6935,
+      "step": 14600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9747023856712187e-06,
+      "loss": 0.6879,
+      "step": 14601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.973509841568688e-06,
+      "loss": 0.7237,
+      "step": 14602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9723174948088106e-06,
+      "loss": 0.6292,
+      "step": 14603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.971125345425073e-06,
+      "loss": 0.6787,
+      "step": 14604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.969933393450958e-06,
+      "loss": 0.6214,
+      "step": 14605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.968741638919942e-06,
+      "loss": 0.7057,
+      "step": 14606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9675500818654956e-06,
+      "loss": 0.6498,
+      "step": 14607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9663587223210844e-06,
+      "loss": 0.6752,
+      "step": 14608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.965167560320169e-06,
+      "loss": 0.6942,
+      "step": 14609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9639765958962043e-06,
+      "loss": 0.6538,
+      "step": 14610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9627858290826407e-06,
+      "loss": 0.6281,
+      "step": 14611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.961595259912915e-06,
+      "loss": 0.643,
+      "step": 14612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9604048884204706e-06,
+      "loss": 0.6365,
+      "step": 14613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.959214714638737e-06,
+      "loss": 0.6656,
+      "step": 14614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9580247386011398e-06,
+      "loss": 0.6112,
+      "step": 14615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9568349603411052e-06,
+      "loss": 0.705,
+      "step": 14616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.955645379892048e-06,
+      "loss": 0.6669,
+      "step": 14617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9544559972873743e-06,
+      "loss": 0.726,
+      "step": 14618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9532668125604902e-06,
+      "loss": 0.6796,
+      "step": 14619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9520778257447935e-06,
+      "loss": 0.6924,
+      "step": 14620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9508890368736798e-06,
+      "loss": 0.6233,
+      "step": 14621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9497004459805356e-06,
+      "loss": 0.6558,
+      "step": 14622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9485120530987422e-06,
+      "loss": 0.707,
+      "step": 14623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.947323858261678e-06,
+      "loss": 0.6823,
+      "step": 14624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9461358615027136e-06,
+      "loss": 0.6725,
+      "step": 14625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.944948062855214e-06,
+      "loss": 0.6783,
+      "step": 14626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9437604623525394e-06,
+      "loss": 0.6408,
+      "step": 14627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9425730600280445e-06,
+      "loss": 0.6484,
+      "step": 14628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9413858559150776e-06,
+      "loss": 0.6818,
+      "step": 14629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9401988500469846e-06,
+      "loss": 0.6723,
+      "step": 14630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939012042457099e-06,
+      "loss": 0.6273,
+      "step": 14631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9378254331787535e-06,
+      "loss": 0.6712,
+      "step": 14632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9366390222452756e-06,
+      "loss": 0.6548,
+      "step": 14633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.935452809689987e-06,
+      "loss": 0.7267,
+      "step": 14634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934266795546203e-06,
+      "loss": 0.6665,
+      "step": 14635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.933080979847229e-06,
+      "loss": 0.6451,
+      "step": 14636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.93189536262638e-06,
+      "loss": 0.6399,
+      "step": 14637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9307099439169453e-06,
+      "loss": 0.6974,
+      "step": 14638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9295247237522195e-06,
+      "loss": 0.6596,
+      "step": 14639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9283397021654914e-06,
+      "loss": 0.708,
+      "step": 14640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9271548791900437e-06,
+      "loss": 0.6349,
+      "step": 14641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9259702548591505e-06,
+      "loss": 0.687,
+      "step": 14642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9247858292060883e-06,
+      "loss": 0.6163,
+      "step": 14643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.923601602264111e-06,
+      "loss": 0.652,
+      "step": 14644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9224175740664893e-06,
+      "loss": 0.5721,
+      "step": 14645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.921233744646472e-06,
+      "loss": 0.6643,
+      "step": 14646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9200501140373096e-06,
+      "loss": 0.6945,
+      "step": 14647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9188666822722433e-06,
+      "loss": 0.7032,
+      "step": 14648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.917683449384512e-06,
+      "loss": 0.6664,
+      "step": 14649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9165004154073494e-06,
+      "loss": 0.7049,
+      "step": 14650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9153175803739766e-06,
+      "loss": 0.6192,
+      "step": 14651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.914134944317615e-06,
+      "loss": 0.6475,
+      "step": 14652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9129525072714813e-06,
+      "loss": 0.2961,
+      "step": 14653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9117702692687854e-06,
+      "loss": 0.7244,
+      "step": 14654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.910588230342729e-06,
+      "loss": 0.6424,
+      "step": 14655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9094063905265124e-06,
+      "loss": 0.6334,
+      "step": 14656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9082247498533267e-06,
+      "loss": 0.6417,
+      "step": 14657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9070433083563597e-06,
+      "loss": 0.6686,
+      "step": 14658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.905862066068793e-06,
+      "loss": 0.6094,
+      "step": 14659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.904681023023801e-06,
+      "loss": 0.6367,
+      "step": 14660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903500179254556e-06,
+      "loss": 0.6776,
+      "step": 14661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.90231953479422e-06,
+      "loss": 0.6252,
+      "step": 14662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9011390896759584e-06,
+      "loss": 0.7023,
+      "step": 14663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.899958843932915e-06,
+      "loss": 0.6547,
+      "step": 14664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8987787975982405e-06,
+      "loss": 0.6199,
+      "step": 14665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.897598950705082e-06,
+      "loss": 0.679,
+      "step": 14666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8964193032865717e-06,
+      "loss": 0.6722,
+      "step": 14667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.895239855375843e-06,
+      "loss": 0.6513,
+      "step": 14668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894060607006024e-06,
+      "loss": 0.6278,
+      "step": 14669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8928815582102265e-06,
+      "loss": 0.6774,
+      "step": 14670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89170270902157e-06,
+      "loss": 0.7549,
+      "step": 14671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890524059473161e-06,
+      "loss": 0.5954,
+      "step": 14672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.889345609598104e-06,
+      "loss": 0.6795,
+      "step": 14673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.888167359429497e-06,
+      "loss": 0.6651,
+      "step": 14674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8869893090004296e-06,
+      "loss": 0.7197,
+      "step": 14675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.88581145834399e-06,
+      "loss": 0.7146,
+      "step": 14676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.884633807493258e-06,
+      "loss": 0.663,
+      "step": 14677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.883456356481308e-06,
+      "loss": 0.6849,
+      "step": 14678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8822791053412093e-06,
+      "loss": 0.6175,
+      "step": 14679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8811020541060264e-06,
+      "loss": 0.6937,
+      "step": 14680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8799252028088177e-06,
+      "loss": 0.6321,
+      "step": 14681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8787485514826374e-06,
+      "loss": 0.6447,
+      "step": 14682
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8775721001605274e-06,
+      "loss": 0.6553,
+      "step": 14683
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8763958488755304e-06,
+      "loss": 0.7253,
+      "step": 14684
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.875219797660681e-06,
+      "loss": 0.6236,
+      "step": 14685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8740439465490145e-06,
+      "loss": 0.6306,
+      "step": 14686
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.872868295573552e-06,
+      "loss": 0.7033,
+      "step": 14687
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.871692844767313e-06,
+      "loss": 0.6496,
+      "step": 14688
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.870517594163312e-06,
+      "loss": 0.6713,
+      "step": 14689
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8693425437945534e-06,
+      "loss": 0.7033,
+      "step": 14690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8681676936940397e-06,
+      "loss": 0.6377,
+      "step": 14691
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8669930438947667e-06,
+      "loss": 0.7086,
+      "step": 14692
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8658185944297266e-06,
+      "loss": 0.7266,
+      "step": 14693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864644345331905e-06,
+      "loss": 0.67,
+      "step": 14694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.86347029663428e-06,
+      "loss": 0.665,
+      "step": 14695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8622964483698255e-06,
+      "loss": 0.6761,
+      "step": 14696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8611228005715098e-06,
+      "loss": 0.6566,
+      "step": 14697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8599493532722954e-06,
+      "loss": 0.7175,
+      "step": 14698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.858776106505139e-06,
+      "loss": 0.7111,
+      "step": 14699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8576030603029926e-06,
+      "loss": 0.6744,
+      "step": 14700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.856430214698801e-06,
+      "loss": 0.7139,
+      "step": 14701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8552575697255067e-06,
+      "loss": 0.6747,
+      "step": 14702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8540851254160406e-06,
+      "loss": 0.6006,
+      "step": 14703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8529128818033314e-06,
+      "loss": 0.7006,
+      "step": 14704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8517408389203003e-06,
+      "loss": 0.6241,
+      "step": 14705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850568996799872e-06,
+      "loss": 0.666,
+      "step": 14706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8493973554749534e-06,
+      "loss": 0.6437,
+      "step": 14707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8482259149784542e-06,
+      "loss": 0.6968,
+      "step": 14708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8470546753432692e-06,
+      "loss": 0.6891,
+      "step": 14709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845883636602297e-06,
+      "loss": 0.6536,
+      "step": 14710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8447127987884247e-06,
+      "loss": 0.6685,
+      "step": 14711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.843542161934538e-06,
+      "loss": 0.6796,
+      "step": 14712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.842371726073514e-06,
+      "loss": 0.6934,
+      "step": 14713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841201491238226e-06,
+      "loss": 0.343,
+      "step": 14714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8400314574615385e-06,
+      "loss": 0.6408,
+      "step": 14715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8388616247763135e-06,
+      "loss": 0.6422,
+      "step": 14716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8376919932154077e-06,
+      "loss": 0.6641,
+      "step": 14717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8365225628116687e-06,
+      "loss": 0.7692,
+      "step": 14718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.835353333597942e-06,
+      "loss": 0.6178,
+      "step": 14719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.834184305607066e-06,
+      "loss": 0.6811,
+      "step": 14720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8330154788718767e-06,
+      "loss": 0.5952,
+      "step": 14721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.831846853425193e-06,
+      "loss": 0.7065,
+      "step": 14722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8306784292998423e-06,
+      "loss": 0.6331,
+      "step": 14723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8295102065286385e-06,
+      "loss": 0.7166,
+      "step": 14724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828342185144389e-06,
+      "loss": 0.7163,
+      "step": 14725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8271743651799046e-06,
+      "loss": 0.6977,
+      "step": 14726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826006746667982e-06,
+      "loss": 0.6533,
+      "step": 14727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8248393296414156e-06,
+      "loss": 0.6277,
+      "step": 14728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.823672114132988e-06,
+      "loss": 0.6575,
+      "step": 14729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8225051001754844e-06,
+      "loss": 0.6691,
+      "step": 14730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8213382878016795e-06,
+      "loss": 0.6274,
+      "step": 14731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8201716770443453e-06,
+      "loss": 0.6719,
+      "step": 14732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.819005267936247e-06,
+      "loss": 0.6325,
+      "step": 14733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8178390605101414e-06,
+      "loss": 0.313,
+      "step": 14734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816673054798784e-06,
+      "loss": 0.643,
+      "step": 14735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8155072508349236e-06,
+      "loss": 0.6546,
+      "step": 14736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8143416486513e-06,
+      "loss": 0.6945,
+      "step": 14737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.813176248280651e-06,
+      "loss": 0.7247,
+      "step": 14738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8120110497557072e-06,
+      "loss": 0.6672,
+      "step": 14739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.810846053109195e-06,
+      "loss": 0.6975,
+      "step": 14740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8096812583738354e-06,
+      "loss": 0.6241,
+      "step": 14741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.808516665582336e-06,
+      "loss": 0.6888,
+      "step": 14742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8073522747674096e-06,
+      "loss": 0.6898,
+      "step": 14743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8061880859617574e-06,
+      "loss": 0.6872,
+      "step": 14744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.805024099198077e-06,
+      "loss": 0.6073,
+      "step": 14745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.803860314509057e-06,
+      "loss": 0.6727,
+      "step": 14746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.80269673192739e-06,
+      "loss": 0.658,
+      "step": 14747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.801533351485749e-06,
+      "loss": 0.6686,
+      "step": 14748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.80037017321681e-06,
+      "loss": 0.6528,
+      "step": 14749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.799207197153241e-06,
+      "loss": 0.6548,
+      "step": 14750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.798044423327706e-06,
+      "loss": 0.6351,
+      "step": 14751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7968818517728614e-06,
+      "loss": 0.6768,
+      "step": 14752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7957194825213597e-06,
+      "loss": 0.6349,
+      "step": 14753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7945573156058447e-06,
+      "loss": 0.6111,
+      "step": 14754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7933953510589573e-06,
+      "loss": 0.6962,
+      "step": 14755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7922335889133336e-06,
+      "loss": 0.6931,
+      "step": 14756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7910720292015993e-06,
+      "loss": 0.6344,
+      "step": 14757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7899106719563794e-06,
+      "loss": 0.7211,
+      "step": 14758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7887495172102908e-06,
+      "loss": 0.6049,
+      "step": 14759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7875885649959488e-06,
+      "loss": 0.6982,
+      "step": 14760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786427815345951e-06,
+      "loss": 0.6452,
+      "step": 14761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7852672682929037e-06,
+      "loss": 0.6667,
+      "step": 14762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7841069238693997e-06,
+      "loss": 0.676,
+      "step": 14763
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782946782108028e-06,
+      "loss": 0.6039,
+      "step": 14764
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.781786843041372e-06,
+      "loss": 0.6228,
+      "step": 14765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7806271067020065e-06,
+      "loss": 0.6564,
+      "step": 14766
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7794675731225117e-06,
+      "loss": 0.6887,
+      "step": 14767
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7783082423354456e-06,
+      "loss": 0.6613,
+      "step": 14768
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777149114373371e-06,
+      "loss": 0.6165,
+      "step": 14769
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7759901892688424e-06,
+      "loss": 0.6207,
+      "step": 14770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7748314670544087e-06,
+      "loss": 0.6543,
+      "step": 14771
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7736729477626147e-06,
+      "loss": 0.6598,
+      "step": 14772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7725146314259987e-06,
+      "loss": 0.6673,
+      "step": 14773
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7713565180770854e-06,
+      "loss": 0.639,
+      "step": 14774
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.770198607748409e-06,
+      "loss": 0.6482,
+      "step": 14775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7690409004724883e-06,
+      "loss": 0.627,
+      "step": 14776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7678833962818364e-06,
+      "loss": 0.6744,
+      "step": 14777
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7667260952089635e-06,
+      "loss": 0.7012,
+      "step": 14778
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.765568997286374e-06,
+      "loss": 0.7057,
+      "step": 14779
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7644121025465664e-06,
+      "loss": 0.6979,
+      "step": 14780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.763255411022028e-06,
+      "loss": 0.686,
+      "step": 14781
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.762098922745248e-06,
+      "loss": 0.2978,
+      "step": 14782
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760942637748706e-06,
+      "loss": 0.6199,
+      "step": 14783
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7597865560648774e-06,
+      "loss": 0.6617,
+      "step": 14784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.758630677726233e-06,
+      "loss": 0.6705,
+      "step": 14785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.757475002765233e-06,
+      "loss": 0.7114,
+      "step": 14786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7563195312143387e-06,
+      "loss": 0.6491,
+      "step": 14787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7551642631059994e-06,
+      "loss": 0.6792,
+      "step": 14788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7540091984726625e-06,
+      "loss": 0.694,
+      "step": 14789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7528543373467687e-06,
+      "loss": 0.5927,
+      "step": 14790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.751699679760753e-06,
+      "loss": 0.7224,
+      "step": 14791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7505452257470443e-06,
+      "loss": 0.7147,
+      "step": 14792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.749390975338069e-06,
+      "loss": 0.6528,
+      "step": 14793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.748236928566238e-06,
+      "loss": 0.6381,
+      "step": 14794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.747083085463965e-06,
+      "loss": 0.7069,
+      "step": 14795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.745929446063661e-06,
+      "loss": 0.6103,
+      "step": 14796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.744776010397724e-06,
+      "loss": 0.6338,
+      "step": 14797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7436227784985493e-06,
+      "loss": 0.7085,
+      "step": 14798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7424697503985275e-06,
+      "loss": 0.654,
+      "step": 14799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7413169261300375e-06,
+      "loss": 0.6403,
+      "step": 14800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.74016430572546e-06,
+      "loss": 0.6233,
+      "step": 14801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.739011889217166e-06,
+      "loss": 0.6372,
+      "step": 14802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.737859676637523e-06,
+      "loss": 0.6995,
+      "step": 14803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73670766801889e-06,
+      "loss": 0.6605,
+      "step": 14804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735555863393622e-06,
+      "loss": 0.7399,
+      "step": 14805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7344042627940693e-06,
+      "loss": 0.6779,
+      "step": 14806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.733252866252574e-06,
+      "loss": 0.679,
+      "step": 14807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7321016738014737e-06,
+      "loss": 0.6574,
+      "step": 14808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7309506854731017e-06,
+      "loss": 0.6092,
+      "step": 14809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7297999012997824e-06,
+      "loss": 0.6049,
+      "step": 14810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7286493213138364e-06,
+      "loss": 0.6976,
+      "step": 14811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.727498945547582e-06,
+      "loss": 0.6043,
+      "step": 14812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7263487740333216e-06,
+      "loss": 0.6423,
+      "step": 14813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.725198806803362e-06,
+      "loss": 0.6317,
+      "step": 14814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.724049043889997e-06,
+      "loss": 0.616,
+      "step": 14815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722899485325524e-06,
+      "loss": 0.659,
+      "step": 14816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.721750131142228e-06,
+      "loss": 0.63,
+      "step": 14817
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7206009813723865e-06,
+      "loss": 0.6664,
+      "step": 14818
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719452036048279e-06,
+      "loss": 0.617,
+      "step": 14819
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.718303295202167e-06,
+      "loss": 0.7228,
+      "step": 14820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.717154758866316e-06,
+      "loss": 0.669,
+      "step": 14821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7160064270729846e-06,
+      "loss": 0.6704,
+      "step": 14822
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7148582998544247e-06,
+      "loss": 0.712,
+      "step": 14823
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7137103772428806e-06,
+      "loss": 0.6194,
+      "step": 14824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7125626592705934e-06,
+      "loss": 0.5755,
+      "step": 14825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7114151459697957e-06,
+      "loss": 0.7139,
+      "step": 14826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7102678373727176e-06,
+      "loss": 0.6533,
+      "step": 14827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7091207335115822e-06,
+      "loss": 0.5981,
+      "step": 14828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7079738344186044e-06,
+      "loss": 0.6188,
+      "step": 14829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.706827140125998e-06,
+      "loss": 0.6857,
+      "step": 14830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705680650665967e-06,
+      "loss": 0.6768,
+      "step": 14831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7045343660707146e-06,
+      "loss": 0.6798,
+      "step": 14832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.703388286372428e-06,
+      "loss": 0.7301,
+      "step": 14833
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7022424116032996e-06,
+      "loss": 0.6606,
+      "step": 14834
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701096741795508e-06,
+      "loss": 0.6818,
+      "step": 14835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6999512769812363e-06,
+      "loss": 0.6119,
+      "step": 14836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6988060171926534e-06,
+      "loss": 0.6503,
+      "step": 14837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.697660962461923e-06,
+      "loss": 0.6713,
+      "step": 14838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.696516112821208e-06,
+      "loss": 0.6705,
+      "step": 14839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6953714683026566e-06,
+      "loss": 0.6671,
+      "step": 14840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69422702893842e-06,
+      "loss": 0.6655,
+      "step": 14841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6930827947606387e-06,
+      "loss": 0.7211,
+      "step": 14842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6919387658014506e-06,
+      "loss": 0.6378,
+      "step": 14843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.690794942092987e-06,
+      "loss": 0.6934,
+      "step": 14844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6896513236673705e-06,
+      "loss": 0.6794,
+      "step": 14845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6885079105567224e-06,
+      "loss": 0.6433,
+      "step": 14846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.687364702793155e-06,
+      "loss": 0.6703,
+      "step": 14847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6862217004087753e-06,
+      "loss": 0.7123,
+      "step": 14848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.685078903435686e-06,
+      "loss": 0.6689,
+      "step": 14849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6839363119059834e-06,
+      "loss": 0.6094,
+      "step": 14850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6827939258517575e-06,
+      "loss": 0.6515,
+      "step": 14851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6816517453050948e-06,
+      "loss": 0.5336,
+      "step": 14852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6805097702980697e-06,
+      "loss": 0.6214,
+      "step": 14853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6793680008627578e-06,
+      "loss": 0.6768,
+      "step": 14854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678226437031226e-06,
+      "loss": 0.6379,
+      "step": 14855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677085078835532e-06,
+      "loss": 0.6202,
+      "step": 14856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.675943926307738e-06,
+      "loss": 0.6014,
+      "step": 14857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6748029794798934e-06,
+      "loss": 0.6733,
+      "step": 14858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6736622383840383e-06,
+      "loss": 0.7236,
+      "step": 14859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.672521703052211e-06,
+      "loss": 0.6229,
+      "step": 14860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.671381373516446e-06,
+      "loss": 0.6588,
+      "step": 14861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6702412498087702e-06,
+      "loss": 0.6123,
+      "step": 14862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6691013319612035e-06,
+      "loss": 0.724,
+      "step": 14863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.667961620005762e-06,
+      "loss": 0.6664,
+      "step": 14864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6668221139744543e-06,
+      "loss": 0.7847,
+      "step": 14865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.665682813899285e-06,
+      "loss": 0.664,
+      "step": 14866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6645437198122503e-06,
+      "loss": 0.6517,
+      "step": 14867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663404831745344e-06,
+      "loss": 0.6579,
+      "step": 14868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6622661497305513e-06,
+      "loss": 0.5919,
+      "step": 14869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6611276737998538e-06,
+      "loss": 0.6574,
+      "step": 14870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.659989403985227e-06,
+      "loss": 0.6169,
+      "step": 14871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6588513403186366e-06,
+      "loss": 0.6646,
+      "step": 14872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.657713482832046e-06,
+      "loss": 0.603,
+      "step": 14873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6565758315574152e-06,
+      "loss": 0.6602,
+      "step": 14874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.655438386526694e-06,
+      "loss": 0.6938,
+      "step": 14875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654301147771825e-06,
+      "loss": 0.6346,
+      "step": 14876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6531641153247556e-06,
+      "loss": 0.6416,
+      "step": 14877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.652027289217418e-06,
+      "loss": 0.7324,
+      "step": 14878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.650890669481736e-06,
+      "loss": 0.6028,
+      "step": 14879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.649754256149635e-06,
+      "loss": 0.7068,
+      "step": 14880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6486180492530313e-06,
+      "loss": 0.644,
+      "step": 14881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6474820488238363e-06,
+      "loss": 0.7114,
+      "step": 14882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6463462548939546e-06,
+      "loss": 0.6642,
+      "step": 14883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.645210667495286e-06,
+      "loss": 0.6192,
+      "step": 14884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.644075286659724e-06,
+      "loss": 0.6366,
+      "step": 14885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6429401124191568e-06,
+      "loss": 0.6707,
+      "step": 14886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.641805144805466e-06,
+      "loss": 0.6506,
+      "step": 14887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6406703838505268e-06,
+      "loss": 0.6953,
+      "step": 14888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6395358295862105e-06,
+      "loss": 0.6817,
+      "step": 14889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6384014820443817e-06,
+      "loss": 0.7437,
+      "step": 14890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6372673412569027e-06,
+      "loss": 0.6273,
+      "step": 14891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6361334072556187e-06,
+      "loss": 0.64,
+      "step": 14892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6349996800723822e-06,
+      "loss": 0.668,
+      "step": 14893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6338661597390325e-06,
+      "loss": 0.6658,
+      "step": 14894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6327328462874056e-06,
+      "loss": 0.6084,
+      "step": 14895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.631599739749329e-06,
+      "loss": 0.6732,
+      "step": 14896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6304668401566334e-06,
+      "loss": 0.636,
+      "step": 14897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6293341475411307e-06,
+      "loss": 0.6271,
+      "step": 14898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6282016619346353e-06,
+      "loss": 0.5977,
+      "step": 14899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.627069383368953e-06,
+      "loss": 0.6753,
+      "step": 14900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6259373118758846e-06,
+      "loss": 0.6073,
+      "step": 14901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6248054474872254e-06,
+      "loss": 0.6345,
+      "step": 14902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.623673790234763e-06,
+      "loss": 0.6797,
+      "step": 14903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622542340150286e-06,
+      "loss": 0.6316,
+      "step": 14904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6214110972655604e-06,
+      "loss": 0.646,
+      "step": 14905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620280061612368e-06,
+      "loss": 0.6302,
+      "step": 14906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.619149233222471e-06,
+      "loss": 0.6958,
+      "step": 14907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.618018612127631e-06,
+      "loss": 0.6922,
+      "step": 14908
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6168881983595994e-06,
+      "loss": 0.6744,
+      "step": 14909
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6157579919501285e-06,
+      "loss": 0.6144,
+      "step": 14910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.614627992930956e-06,
+      "loss": 0.6462,
+      "step": 14911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6134982013338197e-06,
+      "loss": 0.66,
+      "step": 14912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6123686171904517e-06,
+      "loss": 0.7022,
+      "step": 14913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.611239240532576e-06,
+      "loss": 0.6529,
+      "step": 14914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6101100713919124e-06,
+      "loss": 0.6271,
+      "step": 14915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6089811098001716e-06,
+      "loss": 0.695,
+      "step": 14916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607852355789069e-06,
+      "loss": 0.715,
+      "step": 14917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.606723809390297e-06,
+      "loss": 0.6464,
+      "step": 14918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.605595470635556e-06,
+      "loss": 0.6239,
+      "step": 14919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6044673395565345e-06,
+      "loss": 0.6845,
+      "step": 14920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.603339416184918e-06,
+      "loss": 0.5877,
+      "step": 14921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602211700552384e-06,
+      "loss": 0.6867,
+      "step": 14922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6010841926906074e-06,
+      "loss": 0.6677,
+      "step": 14923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.59995689263125e-06,
+      "loss": 0.6888,
+      "step": 14924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598829800405972e-06,
+      "loss": 0.5927,
+      "step": 14925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597702916046434e-06,
+      "loss": 0.6386,
+      "step": 14926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.596576239584283e-06,
+      "loss": 0.6539,
+      "step": 14927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5954497710511616e-06,
+      "loss": 0.7418,
+      "step": 14928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594323510478709e-06,
+      "loss": 0.3026,
+      "step": 14929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5931974578985575e-06,
+      "loss": 0.6532,
+      "step": 14930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5920716133423275e-06,
+      "loss": 0.6414,
+      "step": 14931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5909459768416424e-06,
+      "loss": 0.6345,
+      "step": 14932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5898205484281156e-06,
+      "loss": 0.6405,
+      "step": 14933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.588695328133357e-06,
+      "loss": 0.6918,
+      "step": 14934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5875703159889665e-06,
+      "loss": 0.697,
+      "step": 14935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.586445512026543e-06,
+      "loss": 0.613,
+      "step": 14936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585320916277676e-06,
+      "loss": 0.3059,
+      "step": 14937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58419652877395e-06,
+      "loss": 0.6344,
+      "step": 14938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5830723495469455e-06,
+      "loss": 0.639,
+      "step": 14939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.581948378628235e-06,
+      "loss": 0.6758,
+      "step": 14940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5808246160493844e-06,
+      "loss": 0.6754,
+      "step": 14941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579701061841957e-06,
+      "loss": 0.7012,
+      "step": 14942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5785777160375102e-06,
+      "loss": 0.6276,
+      "step": 14943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5774545786675887e-06,
+      "loss": 0.6425,
+      "step": 14944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576331649763736e-06,
+      "loss": 0.6699,
+      "step": 14945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5752089293574966e-06,
+      "loss": 0.6597,
+      "step": 14946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.574086417480398e-06,
+      "loss": 0.633,
+      "step": 14947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5729641141639693e-06,
+      "loss": 0.6232,
+      "step": 14948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.571842019439731e-06,
+      "loss": 0.6243,
+      "step": 14949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5707201333391942e-06,
+      "loss": 0.702,
+      "step": 14950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569598455893869e-06,
+      "loss": 0.6575,
+      "step": 14951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.568476987135259e-06,
+      "loss": 0.689,
+      "step": 14952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.567355727094861e-06,
+      "loss": 0.6852,
+      "step": 14953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.566234675804167e-06,
+      "loss": 0.683,
+      "step": 14954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5651138332946602e-06,
+      "loss": 0.619,
+      "step": 14955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563993199597823e-06,
+      "loss": 0.6365,
+      "step": 14956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5628727747451266e-06,
+      "loss": 0.6782,
+      "step": 14957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.56175255876804e-06,
+      "loss": 0.6382,
+      "step": 14958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5606325516980247e-06,
+      "loss": 0.6307,
+      "step": 14959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5595127535665365e-06,
+      "loss": 0.632,
+      "step": 14960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5583931644050254e-06,
+      "loss": 0.6387,
+      "step": 14961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.557273784244938e-06,
+      "loss": 0.738,
+      "step": 14962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5561546131177094e-06,
+      "loss": 0.6217,
+      "step": 14963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5550356510547725e-06,
+      "loss": 0.7072,
+      "step": 14964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5539168980875518e-06,
+      "loss": 0.6584,
+      "step": 14965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.552798354247473e-06,
+      "loss": 0.608,
+      "step": 14966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55168001956595e-06,
+      "loss": 0.6943,
+      "step": 14967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5505618940743905e-06,
+      "loss": 0.6755,
+      "step": 14968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5494439778042e-06,
+      "loss": 0.651,
+      "step": 14969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5483262707867717e-06,
+      "loss": 0.6639,
+      "step": 14970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547208773053499e-06,
+      "loss": 0.568,
+      "step": 14971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5460914846357678e-06,
+      "loss": 0.64,
+      "step": 14972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.544974405564956e-06,
+      "loss": 0.6849,
+      "step": 14973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5438575358724394e-06,
+      "loss": 0.6175,
+      "step": 14974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5427408755895853e-06,
+      "loss": 0.6199,
+      "step": 14975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541624424747755e-06,
+      "loss": 0.6662,
+      "step": 14976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.540508183378305e-06,
+      "loss": 0.6279,
+      "step": 14977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5393921515125864e-06,
+      "loss": 0.7087,
+      "step": 14978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.538276329181942e-06,
+      "loss": 0.6777,
+      "step": 14979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5371607164177116e-06,
+      "loss": 0.7078,
+      "step": 14980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5360453132512265e-06,
+      "loss": 0.6749,
+      "step": 14981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5349301197138177e-06,
+      "loss": 0.7006,
+      "step": 14982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5338151358367992e-06,
+      "loss": 0.6938,
+      "step": 14983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532700361651489e-06,
+      "loss": 0.6527,
+      "step": 14984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5315857971891955e-06,
+      "loss": 0.599,
+      "step": 14985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5304714424812203e-06,
+      "loss": 0.6732,
+      "step": 14986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5293572975588654e-06,
+      "loss": 0.6832,
+      "step": 14987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5282433624534233e-06,
+      "loss": 0.697,
+      "step": 14988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5271296371961717e-06,
+      "loss": 0.6881,
+      "step": 14989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5260161218183956e-06,
+      "loss": 0.7412,
+      "step": 14990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524902816351367e-06,
+      "loss": 0.6817,
+      "step": 14991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.523789720826353e-06,
+      "loss": 0.6585,
+      "step": 14992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.522676835274617e-06,
+      "loss": 0.7161,
+      "step": 14993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.521564159727413e-06,
+      "loss": 0.6236,
+      "step": 14994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.520451694215994e-06,
+      "loss": 0.6373,
+      "step": 14995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.519339438771603e-06,
+      "loss": 0.6748,
+      "step": 14996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5182273934254765e-06,
+      "loss": 0.655,
+      "step": 14997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.517115558208849e-06,
+      "loss": 0.6104,
+      "step": 14998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516003933152946e-06,
+      "loss": 0.583,
+      "step": 14999
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 0.644,
+      "step": 15000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5137813136481935e-06,
+      "loss": 0.726,
+      "step": 15001
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512670319261765e-06,
+      "loss": 0.6498,
+      "step": 15002
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511559535160908e-06,
+      "loss": 0.7012,
+      "step": 15003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.510448961376819e-06,
+      "loss": 0.6575,
+      "step": 15004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5093385979406902e-06,
+      "loss": 0.6197,
+      "step": 15005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5082284448837034e-06,
+      "loss": 0.6502,
+      "step": 15006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5071185022370427e-06,
+      "loss": 0.7263,
+      "step": 15007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5060087700318824e-06,
+      "loss": 0.6806,
+      "step": 15008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.504899248299385e-06,
+      "loss": 0.6658,
+      "step": 15009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503789937070712e-06,
+      "loss": 0.6659,
+      "step": 15010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5026808363770217e-06,
+      "loss": 0.6132,
+      "step": 15011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5015719462494635e-06,
+      "loss": 0.7128,
+      "step": 15012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5004632667191797e-06,
+      "loss": 0.6892,
+      "step": 15013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.499354797817312e-06,
+      "loss": 0.6238,
+      "step": 15014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4982465395749833e-06,
+      "loss": 0.6453,
+      "step": 15015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4971384920233286e-06,
+      "loss": 0.6785,
+      "step": 15016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4960306551934655e-06,
+      "loss": 0.6953,
+      "step": 15017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4949230291165073e-06,
+      "loss": 0.6556,
+      "step": 15018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4938156138235625e-06,
+      "loss": 0.6257,
+      "step": 15019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492708409345733e-06,
+      "loss": 0.6652,
+      "step": 15020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.49160141571412e-06,
+      "loss": 0.6708,
+      "step": 15021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4904946329598058e-06,
+      "loss": 0.6445,
+      "step": 15022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.48938806111388e-06,
+      "loss": 0.2908,
+      "step": 15023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.488281700207419e-06,
+      "loss": 0.653,
+      "step": 15024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.487175550271498e-06,
+      "loss": 0.5869,
+      "step": 15025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4860696113371786e-06,
+      "loss": 0.684,
+      "step": 15026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4849638834355315e-06,
+      "loss": 0.6238,
+      "step": 15027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4838583665976034e-06,
+      "loss": 0.66,
+      "step": 15028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.482753060854446e-06,
+      "loss": 0.6528,
+      "step": 15029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481647966237101e-06,
+      "loss": 0.6265,
+      "step": 15030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4805430827766077e-06,
+      "loss": 0.6313,
+      "step": 15031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.479438410503995e-06,
+      "loss": 0.5943,
+      "step": 15032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47833394945029e-06,
+      "loss": 0.6427,
+      "step": 15033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.477229699646515e-06,
+      "loss": 0.5925,
+      "step": 15034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4761256611236737e-06,
+      "loss": 0.6447,
+      "step": 15035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.475021833912783e-06,
+      "loss": 0.6849,
+      "step": 15036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4739182180448406e-06,
+      "loss": 0.7058,
+      "step": 15037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4728148135508434e-06,
+      "loss": 0.6241,
+      "step": 15038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4717116204617797e-06,
+      "loss": 0.6333,
+      "step": 15039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.470608638808638e-06,
+      "loss": 0.3171,
+      "step": 15040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4695058686223896e-06,
+      "loss": 0.6023,
+      "step": 15041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4684033099340075e-06,
+      "loss": 0.2842,
+      "step": 15042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4673009627744604e-06,
+      "loss": 0.6309,
+      "step": 15043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.466198827174707e-06,
+      "loss": 0.7,
+      "step": 15044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4650969031657015e-06,
+      "loss": 0.6251,
+      "step": 15045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.463995190778391e-06,
+      "loss": 0.6537,
+      "step": 15046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4628936900437183e-06,
+      "loss": 0.6765,
+      "step": 15047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461792400992621e-06,
+      "loss": 0.67,
+      "step": 15048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4606913236560283e-06,
+      "loss": 0.6618,
+      "step": 15049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4595904580648633e-06,
+      "loss": 0.6391,
+      "step": 15050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4584898042500472e-06,
+      "loss": 0.6643,
+      "step": 15051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4573893622424896e-06,
+      "loss": 0.6231,
+      "step": 15052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4562891320731007e-06,
+      "loss": 0.65,
+      "step": 15053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.455189113772777e-06,
+      "loss": 0.7256,
+      "step": 15054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4540893073724114e-06,
+      "loss": 0.6934,
+      "step": 15055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4529897129028977e-06,
+      "loss": 0.679,
+      "step": 15056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451890330395118e-06,
+      "loss": 0.6812,
+      "step": 15057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4507911598799462e-06,
+      "loss": 0.6245,
+      "step": 15058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.449692201388256e-06,
+      "loss": 0.6591,
+      "step": 15059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4485934549509126e-06,
+      "loss": 0.6098,
+      "step": 15060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4474949205987706e-06,
+      "loss": 0.6174,
+      "step": 15061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4463965983626857e-06,
+      "loss": 0.6635,
+      "step": 15062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.445298488273504e-06,
+      "loss": 0.6568,
+      "step": 15063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444200590362068e-06,
+      "loss": 0.674,
+      "step": 15064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4431029046592105e-06,
+      "loss": 0.626,
+      "step": 15065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.442005431195763e-06,
+      "loss": 0.6901,
+      "step": 15066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440908170002546e-06,
+      "loss": 0.6553,
+      "step": 15067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4398111211103793e-06,
+      "loss": 0.7289,
+      "step": 15068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.438714284550073e-06,
+      "loss": 0.6926,
+      "step": 15069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4376176603524316e-06,
+      "loss": 0.6867,
+      "step": 15070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.436521248548256e-06,
+      "loss": 0.6227,
+      "step": 15071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4354250491683373e-06,
+      "loss": 0.6371,
+      "step": 15072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434329062243467e-06,
+      "loss": 0.6818,
+      "step": 15073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.433233287804421e-06,
+      "loss": 0.6037,
+      "step": 15074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4321377258819747e-06,
+      "loss": 0.7056,
+      "step": 15075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4310423765069023e-06,
+      "loss": 0.6507,
+      "step": 15076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4299472397099654e-06,
+      "loss": 0.6361,
+      "step": 15077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42885231552192e-06,
+      "loss": 0.6934,
+      "step": 15078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.427757603973523e-06,
+      "loss": 0.6491,
+      "step": 15079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4266631050955126e-06,
+      "loss": 0.719,
+      "step": 15080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4255688189186312e-06,
+      "loss": 0.6716,
+      "step": 15081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4244747454736117e-06,
+      "loss": 0.7155,
+      "step": 15082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4233808847911835e-06,
+      "loss": 0.6511,
+      "step": 15083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4222872369020676e-06,
+      "loss": 0.5844,
+      "step": 15084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4211938018369783e-06,
+      "loss": 0.5821,
+      "step": 15085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4201005796266274e-06,
+      "loss": 0.7332,
+      "step": 15086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.419007570301718e-06,
+      "loss": 0.6834,
+      "step": 15087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4179147738929475e-06,
+      "loss": 0.6362,
+      "step": 15088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4168221904310073e-06,
+      "loss": 0.6365,
+      "step": 15089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.415729819946583e-06,
+      "loss": 0.6411,
+      "step": 15090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.414637662470356e-06,
+      "loss": 0.6066,
+      "step": 15091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413545718033001e-06,
+      "loss": 0.5765,
+      "step": 15092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412453986665182e-06,
+      "loss": 0.7095,
+      "step": 15093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411362468397561e-06,
+      "loss": 0.6156,
+      "step": 15094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4102711632607978e-06,
+      "loss": 0.6724,
+      "step": 15095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409180071285536e-06,
+      "loss": 0.6625,
+      "step": 15096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4080891925024265e-06,
+      "loss": 0.7175,
+      "step": 15097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4069985269421047e-06,
+      "loss": 0.6685,
+      "step": 15098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4059080746352035e-06,
+      "loss": 0.7232,
+      "step": 15099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404817835612346e-06,
+      "loss": 0.5847,
+      "step": 15100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.403727809904154e-06,
+      "loss": 0.6076,
+      "step": 15101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4026379975412394e-06,
+      "loss": 0.6635,
+      "step": 15102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401548398554213e-06,
+      "loss": 0.6937,
+      "step": 15103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400459012973675e-06,
+      "loss": 0.6637,
+      "step": 15104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.399369840830221e-06,
+      "loss": 0.6787,
+      "step": 15105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3982808821544436e-06,
+      "loss": 0.6952,
+      "step": 15106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3971921369769237e-06,
+      "loss": 0.6592,
+      "step": 15107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3961036053282414e-06,
+      "loss": 0.6821,
+      "step": 15108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3950152872389675e-06,
+      "loss": 0.628,
+      "step": 15109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3939271827396683e-06,
+      "loss": 0.6453,
+      "step": 15110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3928392918609032e-06,
+      "loss": 0.6725,
+      "step": 15111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.391751614633231e-06,
+      "loss": 0.6836,
+      "step": 15112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390664151087192e-06,
+      "loss": 0.5783,
+      "step": 15113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389576901253331e-06,
+      "loss": 0.6785,
+      "step": 15114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3884898651621857e-06,
+      "loss": 0.7151,
+      "step": 15115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387403042844282e-06,
+      "loss": 0.6194,
+      "step": 15116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3863164343301493e-06,
+      "loss": 0.6919,
+      "step": 15117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.385230039650306e-06,
+      "loss": 0.6592,
+      "step": 15118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.384143858835258e-06,
+      "loss": 0.6227,
+      "step": 15119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3830578919155155e-06,
+      "loss": 0.6863,
+      "step": 15120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3819721389215767e-06,
+      "loss": 0.6237,
+      "step": 15121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.380886599883937e-06,
+      "loss": 0.6615,
+      "step": 15122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3798012748330834e-06,
+      "loss": 0.6101,
+      "step": 15123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378716163799497e-06,
+      "loss": 0.6652,
+      "step": 15124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377631266813656e-06,
+      "loss": 0.6755,
+      "step": 15125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3765465839060285e-06,
+      "loss": 0.6327,
+      "step": 15126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375462115107079e-06,
+      "loss": 0.6351,
+      "step": 15127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3743778604472654e-06,
+      "loss": 0.6009,
+      "step": 15128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.37329381995704e-06,
+      "loss": 0.5772,
+      "step": 15129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3722099936668475e-06,
+      "loss": 0.6534,
+      "step": 15130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3711263816071307e-06,
+      "loss": 0.6952,
+      "step": 15131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3700429838083192e-06,
+      "loss": 0.6996,
+      "step": 15132
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.368959800300842e-06,
+      "loss": 0.7096,
+      "step": 15133
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3678768311151225e-06,
+      "loss": 0.6669,
+      "step": 15134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3667940762815746e-06,
+      "loss": 0.6837,
+      "step": 15135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.365711535830606e-06,
+      "loss": 0.6462,
+      "step": 15136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364629209792626e-06,
+      "loss": 0.6465,
+      "step": 15137
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3635470981980324e-06,
+      "loss": 0.7137,
+      "step": 15138
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.362465201077212e-06,
+      "loss": 0.6613,
+      "step": 15139
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3613835184605527e-06,
+      "loss": 0.6696,
+      "step": 15140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3603020503784337e-06,
+      "loss": 0.6919,
+      "step": 15141
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.359220796861228e-06,
+      "loss": 0.6159,
+      "step": 15142
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.358139757939305e-06,
+      "loss": 0.6868,
+      "step": 15143
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.357058933643027e-06,
+      "loss": 0.6468,
+      "step": 15144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3559783240027432e-06,
+      "loss": 0.6579,
+      "step": 15145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3548979290488104e-06,
+      "loss": 0.6483,
+      "step": 15146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.35381774881157e-06,
+      "loss": 0.6137,
+      "step": 15147
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352737783321357e-06,
+      "loss": 0.6913,
+      "step": 15148
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351658032608506e-06,
+      "loss": 0.6499,
+      "step": 15149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3505784967033397e-06,
+      "loss": 0.6459,
+      "step": 15150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.349499175636183e-06,
+      "loss": 0.6597,
+      "step": 15151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3484200694373406e-06,
+      "loss": 0.6616,
+      "step": 15152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3473411781371245e-06,
+      "loss": 0.6882,
+      "step": 15153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3462625017658356e-06,
+      "loss": 0.6926,
+      "step": 15154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3451840403537694e-06,
+      "loss": 0.6747,
+      "step": 15155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3441057939312105e-06,
+      "loss": 0.6369,
+      "step": 15156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3430277625284524e-06,
+      "loss": 0.6437,
+      "step": 15157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3419499461757632e-06,
+      "loss": 0.597,
+      "step": 15158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3408723449034165e-06,
+      "loss": 0.6652,
+      "step": 15159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.339794958741678e-06,
+      "loss": 0.6802,
+      "step": 15160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3387177877208055e-06,
+      "loss": 0.6466,
+      "step": 15161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.337640831871052e-06,
+      "loss": 0.5928,
+      "step": 15162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336564091222666e-06,
+      "loss": 0.7225,
+      "step": 15163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3354875658058896e-06,
+      "loss": 0.6277,
+      "step": 15164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.334411255650949e-06,
+      "loss": 0.6252,
+      "step": 15165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333335160788082e-06,
+      "loss": 0.7136,
+      "step": 15166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3322592812475074e-06,
+      "loss": 0.6664,
+      "step": 15167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3311836170594448e-06,
+      "loss": 0.5907,
+      "step": 15168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3301081682541018e-06,
+      "loss": 0.6154,
+      "step": 15169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.329032934861687e-06,
+      "loss": 0.6733,
+      "step": 15170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3279579169123923e-06,
+      "loss": 0.6173,
+      "step": 15171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3268831144364147e-06,
+      "loss": 0.6925,
+      "step": 15172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3258085274639385e-06,
+      "loss": 0.6737,
+      "step": 15173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3247341560251456e-06,
+      "loss": 0.6887,
+      "step": 15174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3236600001502095e-06,
+      "loss": 0.6513,
+      "step": 15175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3225860598692997e-06,
+      "loss": 0.6675,
+      "step": 15176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321512335212577e-06,
+      "loss": 0.6803,
+      "step": 15177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3204388262101974e-06,
+      "loss": 0.6682,
+      "step": 15178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3193655328923116e-06,
+      "loss": 0.6254,
+      "step": 15179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.318292455289063e-06,
+      "loss": 0.6392,
+      "step": 15180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3172195934305907e-06,
+      "loss": 0.6017,
+      "step": 15181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3161469473470267e-06,
+      "loss": 0.7222,
+      "step": 15182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315074517068494e-06,
+      "loss": 0.6677,
+      "step": 15183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.314002302625119e-06,
+      "loss": 0.7074,
+      "step": 15184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312930304047004e-06,
+      "loss": 0.6617,
+      "step": 15185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3118585213642674e-06,
+      "loss": 0.6698,
+      "step": 15186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.310786954607006e-06,
+      "loss": 0.6602,
+      "step": 15187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3097156038053158e-06,
+      "loss": 0.6328,
+      "step": 15188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3086444689892872e-06,
+      "loss": 0.617,
+      "step": 15189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3075735501890062e-06,
+      "loss": 0.6797,
+      "step": 15190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.306502847434543e-06,
+      "loss": 0.6448,
+      "step": 15191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3054323607559736e-06,
+      "loss": 0.6715,
+      "step": 15192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3043620901833617e-06,
+      "loss": 0.6809,
+      "step": 15193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.303292035746767e-06,
+      "loss": 0.6232,
+      "step": 15194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3022221974762427e-06,
+      "loss": 0.6674,
+      "step": 15195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.301152575401836e-06,
+      "loss": 0.7,
+      "step": 15196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3000831695535873e-06,
+      "loss": 0.6325,
+      "step": 15197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.299013979961532e-06,
+      "loss": 0.5859,
+      "step": 15198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2979450066556976e-06,
+      "loss": 0.7188,
+      "step": 15199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296876249666107e-06,
+      "loss": 0.6003,
+      "step": 15200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.295807709022778e-06,
+      "loss": 0.6414,
+      "step": 15201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.294739384755721e-06,
+      "loss": 0.635,
+      "step": 15202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.293671276894942e-06,
+      "loss": 0.6625,
+      "step": 15203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2926033854704355e-06,
+      "loss": 0.6695,
+      "step": 15204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2915357105121938e-06,
+      "loss": 0.6422,
+      "step": 15205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290468252050204e-06,
+      "loss": 0.7012,
+      "step": 15206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.289401010114449e-06,
+      "loss": 0.6469,
+      "step": 15207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.288333984734901e-06,
+      "loss": 0.6987,
+      "step": 15208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2872671759415277e-06,
+      "loss": 0.6407,
+      "step": 15209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2862005837642943e-06,
+      "loss": 0.687,
+      "step": 15210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285134208233151e-06,
+      "loss": 0.6904,
+      "step": 15211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2840680493780498e-06,
+      "loss": 0.6042,
+      "step": 15212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2830021072289345e-06,
+      "loss": 0.3276,
+      "step": 15213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2819363818157426e-06,
+      "loss": 0.7741,
+      "step": 15214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280870873168406e-06,
+      "loss": 0.6634,
+      "step": 15215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2798055813168496e-06,
+      "loss": 0.6342,
+      "step": 15216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2787405062909927e-06,
+      "loss": 0.7759,
+      "step": 15217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2776756481207483e-06,
+      "loss": 0.6583,
+      "step": 15218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2766110068360247e-06,
+      "loss": 0.6478,
+      "step": 15219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.275546582466721e-06,
+      "loss": 0.663,
+      "step": 15220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2744823750427338e-06,
+      "loss": 0.6458,
+      "step": 15221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2734183845939504e-06,
+      "loss": 0.6176,
+      "step": 15222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.272354611150258e-06,
+      "loss": 0.6693,
+      "step": 15223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2712910547415266e-06,
+      "loss": 0.6314,
+      "step": 15224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2702277153976304e-06,
+      "loss": 0.6096,
+      "step": 15225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.26916459314843e-06,
+      "loss": 0.6501,
+      "step": 15226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268101688023788e-06,
+      "loss": 0.6172,
+      "step": 15227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267039000053558e-06,
+      "loss": 0.7194,
+      "step": 15228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.265976529267585e-06,
+      "loss": 0.7114,
+      "step": 15229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2649142756957053e-06,
+      "loss": 0.649,
+      "step": 15230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2638522393677564e-06,
+      "loss": 0.6497,
+      "step": 15231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262790420313564e-06,
+      "loss": 0.6292,
+      "step": 15232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2617288185629504e-06,
+      "loss": 0.6874,
+      "step": 15233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2606674341457333e-06,
+      "loss": 0.667,
+      "step": 15234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2596062670917196e-06,
+      "loss": 0.6066,
+      "step": 15235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2585453174307137e-06,
+      "loss": 0.7118,
+      "step": 15236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2574845851925144e-06,
+      "loss": 0.6653,
+      "step": 15237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2564240704069095e-06,
+      "loss": 0.5981,
+      "step": 15238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.255363773103688e-06,
+      "loss": 0.6309,
+      "step": 15239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2543036933126262e-06,
+      "loss": 0.6899,
+      "step": 15240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2532438310634984e-06,
+      "loss": 0.6967,
+      "step": 15241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2521841863860728e-06,
+      "loss": 0.6132,
+      "step": 15242
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251124759310106e-06,
+      "loss": 0.6492,
+      "step": 15243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2500655498653544e-06,
+      "loss": 0.6777,
+      "step": 15244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2490065580815666e-06,
+      "loss": 0.6892,
+      "step": 15245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2479477839884835e-06,
+      "loss": 0.6483,
+      "step": 15246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246889227615844e-06,
+      "loss": 0.6226,
+      "step": 15247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2458308889933778e-06,
+      "loss": 0.6,
+      "step": 15248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2447727681508112e-06,
+      "loss": 0.6913,
+      "step": 15249
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2437148651178564e-06,
+      "loss": 0.7048,
+      "step": 15250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.242657179924228e-06,
+      "loss": 0.6658,
+      "step": 15251
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2415997125996312e-06,
+      "loss": 0.3551,
+      "step": 15252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.240542463173766e-06,
+      "loss": 0.6444,
+      "step": 15253
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239485431676325e-06,
+      "loss": 0.6099,
+      "step": 15254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2384286181369964e-06,
+      "loss": 0.69,
+      "step": 15255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.237372022585461e-06,
+      "loss": 0.7072,
+      "step": 15256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2363156450513946e-06,
+      "loss": 0.6224,
+      "step": 15257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2352594855644648e-06,
+      "loss": 0.6073,
+      "step": 15258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234203544154335e-06,
+      "loss": 0.6337,
+      "step": 15259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233147820850663e-06,
+      "loss": 0.6914,
+      "step": 15260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2320923156830964e-06,
+      "loss": 0.6119,
+      "step": 15261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2310370286812855e-06,
+      "loss": 0.614,
+      "step": 15262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229981959874862e-06,
+      "loss": 0.6314,
+      "step": 15263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2289271092934596e-06,
+      "loss": 0.6235,
+      "step": 15264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227872476966706e-06,
+      "loss": 0.6357,
+      "step": 15265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2268180629242164e-06,
+      "loss": 0.6454,
+      "step": 15266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2257638671956128e-06,
+      "loss": 0.621,
+      "step": 15267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224709889810501e-06,
+      "loss": 0.6035,
+      "step": 15268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.223656130798477e-06,
+      "loss": 0.6543,
+      "step": 15269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.222602590189139e-06,
+      "loss": 0.6381,
+      "step": 15270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2215492680120777e-06,
+      "loss": 0.6871,
+      "step": 15271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2204961642968747e-06,
+      "loss": 0.7231,
+      "step": 15272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.219443279073107e-06,
+      "loss": 0.6521,
+      "step": 15273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2183906123703457e-06,
+      "loss": 0.6439,
+      "step": 15274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2173381642181558e-06,
+      "loss": 0.7339,
+      "step": 15275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.216285934646095e-06,
+      "loss": 0.7146,
+      "step": 15276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2152339236837173e-06,
+      "loss": 0.5869,
+      "step": 15277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214182131360567e-06,
+      "loss": 0.6906,
+      "step": 15278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2131305577061856e-06,
+      "loss": 0.6687,
+      "step": 15279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212079202750107e-06,
+      "loss": 0.6229,
+      "step": 15280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.211028066521862e-06,
+      "loss": 0.5612,
+      "step": 15281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209977149050966e-06,
+      "loss": 0.6877,
+      "step": 15282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2089264503669374e-06,
+      "loss": 0.6266,
+      "step": 15283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2078759704992856e-06,
+      "loss": 0.6198,
+      "step": 15284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2068257094775137e-06,
+      "loss": 0.6646,
+      "step": 15285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2057756673311205e-06,
+      "loss": 0.7314,
+      "step": 15286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2047258440895924e-06,
+      "loss": 0.622,
+      "step": 15287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.203676239782424e-06,
+      "loss": 0.6741,
+      "step": 15288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2026268544390838e-06,
+      "loss": 0.6418,
+      "step": 15289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2015776880890483e-06,
+      "loss": 0.6917,
+      "step": 15290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2005287407617837e-06,
+      "loss": 0.6164,
+      "step": 15291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.199480012486751e-06,
+      "loss": 0.761,
+      "step": 15292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198431503293402e-06,
+      "loss": 0.6905,
+      "step": 15293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1973832132111906e-06,
+      "loss": 0.5879,
+      "step": 15294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1963351422695477e-06,
+      "loss": 0.704,
+      "step": 15295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1952872904979184e-06,
+      "loss": 0.6902,
+      "step": 15296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1942396579257298e-06,
+      "loss": 0.694,
+      "step": 15297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193192244582404e-06,
+      "loss": 0.7181,
+      "step": 15298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1921450504973586e-06,
+      "loss": 0.3055,
+      "step": 15299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1910980757000045e-06,
+      "loss": 0.6259,
+      "step": 15300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.190051320219749e-06,
+      "loss": 0.6259,
+      "step": 15301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1890047840859862e-06,
+      "loss": 0.6769,
+      "step": 15302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1879584673281116e-06,
+      "loss": 0.6364,
+      "step": 15303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18691236997551e-06,
+      "loss": 0.7118,
+      "step": 15304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.185866492057563e-06,
+      "loss": 0.7036,
+      "step": 15305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1848208336036434e-06,
+      "loss": 0.7291,
+      "step": 15306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1837753946431204e-06,
+      "loss": 0.608,
+      "step": 15307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1827301752053553e-06,
+      "loss": 0.6438,
+      "step": 15308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.6311,
+      "step": 15309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.180640395015513e-06,
+      "loss": 0.6786,
+      "step": 15310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.179595834322129e-06,
+      "loss": 0.6738,
+      "step": 15311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178551493268888e-06,
+      "loss": 0.6021,
+      "step": 15312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1775073718851203e-06,
+      "loss": 0.6379,
+      "step": 15313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176463470200153e-06,
+      "loss": 0.6674,
+      "step": 15314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.175419788243298e-06,
+      "loss": 0.6453,
+      "step": 15315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1743763260438757e-06,
+      "loss": 0.6465,
+      "step": 15316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173333083631187e-06,
+      "loss": 0.6697,
+      "step": 15317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1722900610345354e-06,
+      "loss": 0.6586,
+      "step": 15318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1712472582832124e-06,
+      "loss": 0.6518,
+      "step": 15319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1702046754065086e-06,
+      "loss": 0.6253,
+      "step": 15320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1691623124337003e-06,
+      "loss": 0.6851,
+      "step": 15321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1681201693940667e-06,
+      "loss": 0.6514,
+      "step": 15322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.167078246316875e-06,
+      "loss": 0.7063,
+      "step": 15323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16603654323139e-06,
+      "loss": 0.6627,
+      "step": 15324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1649950601668667e-06,
+      "loss": 0.6588,
+      "step": 15325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163953797152557e-06,
+      "loss": 0.6283,
+      "step": 15326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1629127542177052e-06,
+      "loss": 0.6854,
+      "step": 15327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1618719313915494e-06,
+      "loss": 0.5776,
+      "step": 15328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16083132870332e-06,
+      "loss": 0.6168,
+      "step": 15329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159790946182245e-06,
+      "loss": 0.6143,
+      "step": 15330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.158750783857545e-06,
+      "loss": 0.6821,
+      "step": 15331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1577108417584303e-06,
+      "loss": 0.592,
+      "step": 15332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1566711199141133e-06,
+      "loss": 0.6993,
+      "step": 15333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1556316183537886e-06,
+      "loss": 0.6578,
+      "step": 15334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1545923371066534e-06,
+      "loss": 0.6147,
+      "step": 15335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153553276201896e-06,
+      "loss": 0.6707,
+      "step": 15336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152514435668702e-06,
+      "loss": 0.6854,
+      "step": 15337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.151475815536247e-06,
+      "loss": 0.6035,
+      "step": 15338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1504374158336995e-06,
+      "loss": 0.6528,
+      "step": 15339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1493992365902273e-06,
+      "loss": 0.6138,
+      "step": 15340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1483612778349814e-06,
+      "loss": 0.6791,
+      "step": 15341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.147323539597118e-06,
+      "loss": 0.6166,
+      "step": 15342
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1462860219057824e-06,
+      "loss": 0.6254,
+      "step": 15343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1452487247901123e-06,
+      "loss": 0.66,
+      "step": 15344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.144211648279241e-06,
+      "loss": 0.709,
+      "step": 15345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.143174792402297e-06,
+      "loss": 0.6359,
+      "step": 15346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1421381571883992e-06,
+      "loss": 0.669,
+      "step": 15347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1411017426666627e-06,
+      "loss": 0.6473,
+      "step": 15348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140065548866196e-06,
+      "loss": 0.6258,
+      "step": 15349
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1390295758160993e-06,
+      "loss": 0.6904,
+      "step": 15350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.137993823545472e-06,
+      "loss": 0.616,
+      "step": 15351
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136958292083401e-06,
+      "loss": 0.6469,
+      "step": 15352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1359229814589733e-06,
+      "loss": 0.7434,
+      "step": 15353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1348878917012593e-06,
+      "loss": 0.6605,
+      "step": 15354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133853022839336e-06,
+      "loss": 0.6192,
+      "step": 15355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1328183749022626e-06,
+      "loss": 0.6768,
+      "step": 15356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131783947919104e-06,
+      "loss": 0.6142,
+      "step": 15357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1307497419189106e-06,
+      "loss": 0.3259,
+      "step": 15358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1297157569307305e-06,
+      "loss": 0.6151,
+      "step": 15359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.128681992983598e-06,
+      "loss": 0.651,
+      "step": 15360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1276484501065508e-06,
+      "loss": 0.6954,
+      "step": 15361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.126615128328616e-06,
+      "loss": 0.649,
+      "step": 15362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125582027678814e-06,
+      "loss": 0.606,
+      "step": 15363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1245491481861615e-06,
+      "loss": 0.6184,
+      "step": 15364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1235164898796667e-06,
+      "loss": 0.6015,
+      "step": 15365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122484052788333e-06,
+      "loss": 0.6478,
+      "step": 15366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1214518369411554e-06,
+      "loss": 0.6168,
+      "step": 15367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.120419842367125e-06,
+      "loss": 0.6591,
+      "step": 15368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119388069095226e-06,
+      "loss": 0.6972,
+      "step": 15369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1183565171544374e-06,
+      "loss": 0.5761,
+      "step": 15370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.11732518657373e-06,
+      "loss": 0.6612,
+      "step": 15371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1162940773820707e-06,
+      "loss": 0.6633,
+      "step": 15372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1152631896084154e-06,
+      "loss": 0.6476,
+      "step": 15373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114232523281717e-06,
+      "loss": 0.701,
+      "step": 15374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.113202078430926e-06,
+      "loss": 0.6597,
+      "step": 15375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112171855084978e-06,
+      "loss": 0.6451,
+      "step": 15376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1111418532728133e-06,
+      "loss": 0.6798,
+      "step": 15377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.110112073023357e-06,
+      "loss": 0.6488,
+      "step": 15378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1090825143655347e-06,
+      "loss": 0.6774,
+      "step": 15379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1080531773282554e-06,
+      "loss": 0.6403,
+      "step": 15380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1070240619404313e-06,
+      "loss": 0.6837,
+      "step": 15381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1059951682309678e-06,
+      "loss": 0.66,
+      "step": 15382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.104966496228759e-06,
+      "loss": 0.6951,
+      "step": 15383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1039380459626967e-06,
+      "loss": 0.6894,
+      "step": 15384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102909817461667e-06,
+      "loss": 0.6983,
+      "step": 15385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.101881810754546e-06,
+      "loss": 0.5795,
+      "step": 15386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.100854025870207e-06,
+      "loss": 0.6664,
+      "step": 15387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0998264628375164e-06,
+      "loss": 0.7174,
+      "step": 15388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098799121685332e-06,
+      "loss": 0.6542,
+      "step": 15389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0977720024425086e-06,
+      "loss": 0.6343,
+      "step": 15390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.096745105137894e-06,
+      "loss": 0.5982,
+      "step": 15391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0957184298003296e-06,
+      "loss": 0.6222,
+      "step": 15392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.094691976458647e-06,
+      "loss": 0.6965,
+      "step": 15393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0936657451416765e-06,
+      "loss": 0.7214,
+      "step": 15394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09263973587824e-06,
+      "loss": 0.6967,
+      "step": 15395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0916139486971545e-06,
+      "loss": 0.681,
+      "step": 15396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0905883836272256e-06,
+      "loss": 0.6971,
+      "step": 15397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0895630406972657e-06,
+      "loss": 0.6821,
+      "step": 15398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0885379199360646e-06,
+      "loss": 0.6395,
+      "step": 15399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087513021372415e-06,
+      "loss": 0.6954,
+      "step": 15400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0864883450351013e-06,
+      "loss": 0.6042,
+      "step": 15401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0854638909529047e-06,
+      "loss": 0.6852,
+      "step": 15402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0844396591545944e-06,
+      "loss": 0.6539,
+      "step": 15403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0834156496689385e-06,
+      "loss": 0.714,
+      "step": 15404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0823918625246963e-06,
+      "loss": 0.5953,
+      "step": 15405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0813682977506213e-06,
+      "loss": 0.6238,
+      "step": 15406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.080344955375461e-06,
+      "loss": 0.6712,
+      "step": 15407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0793218354279556e-06,
+      "loss": 0.6682,
+      "step": 15408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.07829893793684e-06,
+      "loss": 0.2889,
+      "step": 15409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0772762629308452e-06,
+      "loss": 0.3159,
+      "step": 15410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0762538104386943e-06,
+      "loss": 0.6661,
+      "step": 15411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.075231580489098e-06,
+      "loss": 0.6283,
+      "step": 15412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0742095731107693e-06,
+      "loss": 0.6861,
+      "step": 15413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0731877883324114e-06,
+      "loss": 0.7145,
+      "step": 15414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0721662261827214e-06,
+      "loss": 0.6344,
+      "step": 15415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0711448866903915e-06,
+      "loss": 0.6751,
+      "step": 15416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070123769884104e-06,
+      "loss": 0.5973,
+      "step": 15417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0691028757925435e-06,
+      "loss": 0.6015,
+      "step": 15418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.068082204444377e-06,
+      "loss": 0.5731,
+      "step": 15419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0670617558682714e-06,
+      "loss": 0.6592,
+      "step": 15420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.066041530092887e-06,
+      "loss": 0.6449,
+      "step": 15421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0650215271468776e-06,
+      "loss": 0.6933,
+      "step": 15422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.06400174705889e-06,
+      "loss": 0.3222,
+      "step": 15423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0629821898575684e-06,
+      "loss": 0.621,
+      "step": 15424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0619628555715397e-06,
+      "loss": 0.6284,
+      "step": 15425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.06094374422944e-06,
+      "loss": 0.2832,
+      "step": 15426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0599248558598884e-06,
+      "loss": 0.7009,
+      "step": 15427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0589061904915033e-06,
+      "loss": 0.6855,
+      "step": 15428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0578877481528914e-06,
+      "loss": 0.6445,
+      "step": 15429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.056869528872658e-06,
+      "loss": 0.6666,
+      "step": 15430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0558515326794026e-06,
+      "loss": 0.6649,
+      "step": 15431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.054833759601711e-06,
+      "loss": 0.7303,
+      "step": 15432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05381620966817e-06,
+      "loss": 0.6648,
+      "step": 15433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0527988829073587e-06,
+      "loss": 0.6532,
+      "step": 15434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0517817793478477e-06,
+      "loss": 0.6037,
+      "step": 15435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.050764899018206e-06,
+      "loss": 0.5796,
+      "step": 15436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0497482419469905e-06,
+      "loss": 0.6456,
+      "step": 15437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.048731808162755e-06,
+      "loss": 0.6058,
+      "step": 15438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0477155976940466e-06,
+      "loss": 0.595,
+      "step": 15439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046699610569407e-06,
+      "loss": 0.5935,
+      "step": 15440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04568384681737e-06,
+      "loss": 0.6066,
+      "step": 15441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0446683064664654e-06,
+      "loss": 0.7046,
+      "step": 15442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043652989545212e-06,
+      "loss": 0.6397,
+      "step": 15443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0426378960821304e-06,
+      "loss": 0.6387,
+      "step": 15444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0416230261057245e-06,
+      "loss": 0.6343,
+      "step": 15445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0406083796444976e-06,
+      "loss": 0.6326,
+      "step": 15446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.039593956726951e-06,
+      "loss": 0.6849,
+      "step": 15447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0385797573815722e-06,
+      "loss": 0.6351,
+      "step": 15448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037565781636849e-06,
+      "loss": 0.653,
+      "step": 15449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036552029521257e-06,
+      "loss": 0.598,
+      "step": 15450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0355385010632666e-06,
+      "loss": 0.6398,
+      "step": 15451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0345251962913447e-06,
+      "loss": 0.6263,
+      "step": 15452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0335121152339498e-06,
+      "loss": 0.6907,
+      "step": 15453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0324992579195358e-06,
+      "loss": 0.6461,
+      "step": 15454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0314866243765486e-06,
+      "loss": 0.615,
+      "step": 15455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0304742146334287e-06,
+      "loss": 0.6476,
+      "step": 15456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0294620287186106e-06,
+      "loss": 0.693,
+      "step": 15457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0284500666605223e-06,
+      "loss": 0.6621,
+      "step": 15458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027438328487583e-06,
+      "loss": 0.6675,
+      "step": 15459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0264268142282117e-06,
+      "loss": 0.6454,
+      "step": 15460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0254155239108143e-06,
+      "loss": 0.612,
+      "step": 15461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0244044575637945e-06,
+      "loss": 0.6732,
+      "step": 15462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023393615215551e-06,
+      "loss": 0.3307,
+      "step": 15463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0223829968944685e-06,
+      "loss": 0.6445,
+      "step": 15464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0213726026289336e-06,
+      "loss": 0.6626,
+      "step": 15465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0203624324473213e-06,
+      "loss": 0.684,
+      "step": 15466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.019352486378008e-06,
+      "loss": 0.6693,
+      "step": 15467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.018342764449356e-06,
+      "loss": 0.7114,
+      "step": 15468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0173332666897227e-06,
+      "loss": 0.6543,
+      "step": 15469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0163239931274647e-06,
+      "loss": 0.689,
+      "step": 15470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0153149437909214e-06,
+      "loss": 0.619,
+      "step": 15471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0143061187084366e-06,
+      "loss": 0.6779,
+      "step": 15472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0132975179083412e-06,
+      "loss": 0.6147,
+      "step": 15473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012289141418965e-06,
+      "loss": 0.6918,
+      "step": 15474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.011280989268628e-06,
+      "loss": 0.661,
+      "step": 15475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0102730614856434e-06,
+      "loss": 0.6145,
+      "step": 15476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.00926535809832e-06,
+      "loss": 0.6798,
+      "step": 15477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0082578791349604e-06,
+      "loss": 0.6411,
+      "step": 15478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.00725062462386e-06,
+      "loss": 0.6844,
+      "step": 15479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006243594593308e-06,
+      "loss": 0.3059,
+      "step": 15480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0052367890715864e-06,
+      "loss": 0.6332,
+      "step": 15481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0042302080869726e-06,
+      "loss": 0.6482,
+      "step": 15482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0032238516677393e-06,
+      "loss": 0.6357,
+      "step": 15483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0022177198421465e-06,
+      "loss": 0.6516,
+      "step": 15484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.001211812638453e-06,
+      "loss": 0.6225,
+      "step": 15485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0002061300849086e-06,
+      "loss": 0.6197,
+      "step": 15486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9992006722097636e-06,
+      "loss": 0.6124,
+      "step": 15487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.998195439041253e-06,
+      "loss": 0.6475,
+      "step": 15488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9971904306076127e-06,
+      "loss": 0.5623,
+      "step": 15489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9961856469370656e-06,
+      "loss": 0.6891,
+      "step": 15490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9951810880578303e-06,
+      "loss": 0.2998,
+      "step": 15491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994176753998124e-06,
+      "loss": 0.6659,
+      "step": 15492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.993172644786151e-06,
+      "loss": 0.6769,
+      "step": 15493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9921687604501137e-06,
+      "loss": 0.6768,
+      "step": 15494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9911651010182074e-06,
+      "loss": 0.6483,
+      "step": 15495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9901616665186197e-06,
+      "loss": 0.6142,
+      "step": 15496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9891584569795318e-06,
+      "loss": 0.6503,
+      "step": 15497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9881554724291206e-06,
+      "loss": 0.657,
+      "step": 15498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9871527128955546e-06,
+      "loss": 0.6076,
+      "step": 15499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986150178406998e-06,
+      "loss": 0.6397,
+      "step": 15500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.985147868991606e-06,
+      "loss": 0.6743,
+      "step": 15501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9841457846775313e-06,
+      "loss": 0.6598,
+      "step": 15502
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9831439254929153e-06,
+      "loss": 0.6009,
+      "step": 15503
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982142291465896e-06,
+      "loss": 0.6146,
+      "step": 15504
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981140882624605e-06,
+      "loss": 0.6942,
+      "step": 15505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9801396989971654e-06,
+      "loss": 0.6272,
+      "step": 15506
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9791387406117024e-06,
+      "loss": 0.6406,
+      "step": 15507
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978138007496323e-06,
+      "loss": 0.722,
+      "step": 15508
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9771374996791394e-06,
+      "loss": 0.6698,
+      "step": 15509
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.976137217188242e-06,
+      "loss": 0.6113,
+      "step": 15510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.975137160051731e-06,
+      "loss": 0.3066,
+      "step": 15511
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9741373282976908e-06,
+      "loss": 0.6563,
+      "step": 15512
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9731377219542027e-06,
+      "loss": 0.7372,
+      "step": 15513
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9721383410493424e-06,
+      "loss": 0.656,
+      "step": 15514
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.971139185611176e-06,
+      "loss": 0.296,
+      "step": 15515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970140255667766e-06,
+      "loss": 0.602,
+      "step": 15516
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.96914155124717e-06,
+      "loss": 0.677,
+      "step": 15517
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.968143072377435e-06,
+      "loss": 0.59,
+      "step": 15518
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.967144819086604e-06,
+      "loss": 0.6756,
+      "step": 15519
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966146791402713e-06,
+      "loss": 0.6615,
+      "step": 15520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9651489893537936e-06,
+      "loss": 0.6797,
+      "step": 15521
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9641514129678716e-06,
+      "loss": 0.6205,
+      "step": 15522
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963154062272958e-06,
+      "loss": 0.6918,
+      "step": 15523
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9621569372970693e-06,
+      "loss": 0.5612,
+      "step": 15524
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9611600380682074e-06,
+      "loss": 0.6377,
+      "step": 15525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9601633646143724e-06,
+      "loss": 0.7548,
+      "step": 15526
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.959166916963553e-06,
+      "loss": 0.6824,
+      "step": 15527
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.958170695143743e-06,
+      "loss": 0.7492,
+      "step": 15528
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9571746991829145e-06,
+      "loss": 0.653,
+      "step": 15529
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956178929109043e-06,
+      "loss": 0.5718,
+      "step": 15530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9551833849500946e-06,
+      "loss": 0.6792,
+      "step": 15531
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9541880667340294e-06,
+      "loss": 0.726,
+      "step": 15532
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9531929744888023e-06,
+      "loss": 0.685,
+      "step": 15533
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.952198108242361e-06,
+      "loss": 0.6898,
+      "step": 15534
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9512034680226466e-06,
+      "loss": 0.6862,
+      "step": 15535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9502090538575947e-06,
+      "loss": 0.6634,
+      "step": 15536
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.949214865775133e-06,
+      "loss": 0.5832,
+      "step": 15537
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.948220903803184e-06,
+      "loss": 0.5958,
+      "step": 15538
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.947227167969663e-06,
+      "loss": 0.6555,
+      "step": 15539
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.946233658302481e-06,
+      "loss": 0.605,
+      "step": 15540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945240374829541e-06,
+      "loss": 0.715,
+      "step": 15541
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944247317578741e-06,
+      "loss": 0.6655,
+      "step": 15542
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943254486577969e-06,
+      "loss": 0.6201,
+      "step": 15543
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9422618818551087e-06,
+      "loss": 0.633,
+      "step": 15544
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94126950343804e-06,
+      "loss": 0.6012,
+      "step": 15545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940277351354635e-06,
+      "loss": 0.7172,
+      "step": 15546
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9392854256327533e-06,
+      "loss": 0.6266,
+      "step": 15547
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938293726300263e-06,
+      "loss": 0.6079,
+      "step": 15548
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9373022533850104e-06,
+      "loss": 0.6623,
+      "step": 15549
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9363110069148416e-06,
+      "loss": 0.6634,
+      "step": 15550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9353199869175978e-06,
+      "loss": 0.6546,
+      "step": 15551
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9343291934211107e-06,
+      "loss": 0.6443,
+      "step": 15552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9333386264532095e-06,
+      "loss": 0.6172,
+      "step": 15553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.932348286041713e-06,
+      "loss": 0.6194,
+      "step": 15554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.931358172214438e-06,
+      "loss": 0.6911,
+      "step": 15555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9303682849991867e-06,
+      "loss": 0.5998,
+      "step": 15556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9293786244237656e-06,
+      "loss": 0.6341,
+      "step": 15557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9283891905159693e-06,
+      "loss": 0.6755,
+      "step": 15558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927399983303585e-06,
+      "loss": 0.6139,
+      "step": 15559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9264110028143965e-06,
+      "loss": 0.6234,
+      "step": 15560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.925422249076182e-06,
+      "loss": 0.6556,
+      "step": 15561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9244337221167066e-06,
+      "loss": 0.5772,
+      "step": 15562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9234454219637346e-06,
+      "loss": 0.312,
+      "step": 15563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9224573486450225e-06,
+      "loss": 0.612,
+      "step": 15564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9214695021883232e-06,
+      "loss": 0.6464,
+      "step": 15565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.92048188262138e-06,
+      "loss": 0.6555,
+      "step": 15566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919494489971928e-06,
+      "loss": 0.7352,
+      "step": 15567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9185073242677067e-06,
+      "loss": 0.631,
+      "step": 15568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9175203855364323e-06,
+      "loss": 0.6409,
+      "step": 15569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916533673805826e-06,
+      "loss": 0.6397,
+      "step": 15570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9155471891036024e-06,
+      "loss": 0.6165,
+      "step": 15571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9145609314574652e-06,
+      "loss": 0.6326,
+      "step": 15572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913574900895114e-06,
+      "loss": 0.6915,
+      "step": 15573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9125890974442475e-06,
+      "loss": 0.6831,
+      "step": 15574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9116035211325433e-06,
+      "loss": 0.6421,
+      "step": 15575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9106181719876835e-06,
+      "loss": 0.6308,
+      "step": 15576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9096330500373484e-06,
+      "loss": 0.6111,
+      "step": 15577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908648155309202e-06,
+      "loss": 0.6634,
+      "step": 15578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.907663487830905e-06,
+      "loss": 0.6765,
+      "step": 15579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9066790476301133e-06,
+      "loss": 0.7075,
+      "step": 15580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9056948347344783e-06,
+      "loss": 0.6482,
+      "step": 15581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.904710849171636e-06,
+      "loss": 0.6443,
+      "step": 15582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9037270909692252e-06,
+      "loss": 0.6855,
+      "step": 15583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.902743560154875e-06,
+      "loss": 0.6405,
+      "step": 15584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9017602567562077e-06,
+      "loss": 0.6461,
+      "step": 15585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9007771808008424e-06,
+      "loss": 0.7037,
+      "step": 15586
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.899794332316387e-06,
+      "loss": 0.6246,
+      "step": 15587
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8988117113304449e-06,
+      "loss": 0.658,
+      "step": 15588
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8978293178706165e-06,
+      "loss": 0.6154,
+      "step": 15589
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8968471519644893e-06,
+      "loss": 0.6658,
+      "step": 15590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8958652136396505e-06,
+      "loss": 0.6909,
+      "step": 15591
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.894883502923678e-06,
+      "loss": 0.6616,
+      "step": 15592
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8939020198441416e-06,
+      "loss": 0.7486,
+      "step": 15593
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8929207644286119e-06,
+      "loss": 0.712,
+      "step": 15594
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.891939736704641e-06,
+      "loss": 0.6592,
+      "step": 15595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8909589366997827e-06,
+      "loss": 0.6474,
+      "step": 15596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8899783644415881e-06,
+      "loss": 0.6434,
+      "step": 15597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8889980199575942e-06,
+      "loss": 0.7005,
+      "step": 15598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8880179032753355e-06,
+      "loss": 0.6363,
+      "step": 15599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8870380144223399e-06,
+      "loss": 0.6572,
+      "step": 15600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8860583534261245e-06,
+      "loss": 0.6519,
+      "step": 15601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8850789203142051e-06,
+      "loss": 0.6925,
+      "step": 15602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8840997151140905e-06,
+      "loss": 0.6009,
+      "step": 15603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8831207378532823e-06,
+      "loss": 0.718,
+      "step": 15604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.882141988559274e-06,
+      "loss": 0.632,
+      "step": 15605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.881163467259557e-06,
+      "loss": 0.6346,
+      "step": 15606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.88018517398161e-06,
+      "loss": 0.6854,
+      "step": 15607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8792071087529118e-06,
+      "loss": 0.6443,
+      "step": 15608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.878229271600931e-06,
+      "loss": 0.6323,
+      "step": 15609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8772516625531311e-06,
+      "loss": 0.7005,
+      "step": 15610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8762742816369672e-06,
+      "loss": 0.5533,
+      "step": 15611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8752971288798915e-06,
+      "loss": 0.6341,
+      "step": 15612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874320204309349e-06,
+      "loss": 0.6395,
+      "step": 15613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8733435079527728e-06,
+      "loss": 0.6412,
+      "step": 15614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.872367039837596e-06,
+      "loss": 0.7126,
+      "step": 15615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8713907999912407e-06,
+      "loss": 0.7236,
+      "step": 15616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8704147884411305e-06,
+      "loss": 0.7237,
+      "step": 15617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8694390052146737e-06,
+      "loss": 0.6071,
+      "step": 15618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8684634503392784e-06,
+      "loss": 0.6719,
+      "step": 15619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8674881238423426e-06,
+      "loss": 0.7473,
+      "step": 15620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8665130257512566e-06,
+      "loss": 0.6148,
+      "step": 15621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.865538156093407e-06,
+      "loss": 0.6977,
+      "step": 15622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.864563514896175e-06,
+      "loss": 0.3117,
+      "step": 15623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8635891021869324e-06,
+      "loss": 0.6865,
+      "step": 15624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8626149179930474e-06,
+      "loss": 0.6625,
+      "step": 15625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8616409623418807e-06,
+      "loss": 0.6716,
+      "step": 15626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8606672352607856e-06,
+      "loss": 0.612,
+      "step": 15627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8596937367771094e-06,
+      "loss": 0.6957,
+      "step": 15628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8587204669181935e-06,
+      "loss": 0.6778,
+      "step": 15629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8577474257113736e-06,
+      "loss": 0.681,
+      "step": 15630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.856774613183977e-06,
+      "loss": 0.6892,
+      "step": 15631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.855802029363326e-06,
+      "loss": 0.6517,
+      "step": 15632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8548296742767401e-06,
+      "loss": 0.6339,
+      "step": 15633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8538575479515208e-06,
+      "loss": 0.6239,
+      "step": 15634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8528856504149738e-06,
+      "loss": 0.711,
+      "step": 15635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8519139816943976e-06,
+      "loss": 0.7084,
+      "step": 15636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8509425418170768e-06,
+      "loss": 0.6808,
+      "step": 15637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8499713308103018e-06,
+      "loss": 0.6545,
+      "step": 15638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8490003487013486e-06,
+      "loss": 0.718,
+      "step": 15639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.848029595517482e-06,
+      "loss": 0.6623,
+      "step": 15640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8470590712859703e-06,
+      "loss": 0.632,
+      "step": 15641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8460887760340706e-06,
+      "loss": 0.639,
+      "step": 15642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8451187097890333e-06,
+      "loss": 0.7122,
+      "step": 15643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8441488725781043e-06,
+      "loss": 0.7073,
+      "step": 15644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8431792644285207e-06,
+      "loss": 0.6797,
+      "step": 15645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8422098853675162e-06,
+      "loss": 0.6058,
+      "step": 15646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8412407354223149e-06,
+      "loss": 0.7498,
+      "step": 15647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.840271814620137e-06,
+      "loss": 0.6546,
+      "step": 15648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8393031229881931e-06,
+      "loss": 0.652,
+      "step": 15649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8383346605536923e-06,
+      "loss": 0.6673,
+      "step": 15650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837366427343832e-06,
+      "loss": 0.6347,
+      "step": 15651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8363984233858102e-06,
+      "loss": 0.6266,
+      "step": 15652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8354306487068063e-06,
+      "loss": 0.6259,
+      "step": 15653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8344631033340044e-06,
+      "loss": 0.6441,
+      "step": 15654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8334957872945792e-06,
+      "loss": 0.6491,
+      "step": 15655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8325287006156978e-06,
+      "loss": 0.66,
+      "step": 15656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8315618433245186e-06,
+      "loss": 0.6545,
+      "step": 15657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.830595215448201e-06,
+      "loss": 0.6522,
+      "step": 15658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8296288170138944e-06,
+      "loss": 0.6141,
+      "step": 15659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8286626480487336e-06,
+      "loss": 0.645,
+      "step": 15660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827696708579859e-06,
+      "loss": 0.7339,
+      "step": 15661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8267309986343972e-06,
+      "loss": 0.6562,
+      "step": 15662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.825765518239473e-06,
+      "loss": 0.3176,
+      "step": 15663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8248002674221999e-06,
+      "loss": 0.6186,
+      "step": 15664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8238352462096897e-06,
+      "loss": 0.6559,
+      "step": 15665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8228704546290443e-06,
+      "loss": 0.6873,
+      "step": 15666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8219058927073607e-06,
+      "loss": 0.626,
+      "step": 15667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8209415604717285e-06,
+      "loss": 0.6278,
+      "step": 15668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8199774579492324e-06,
+      "loss": 0.6182,
+      "step": 15669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8190135851669488e-06,
+      "loss": 0.6617,
+      "step": 15670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8180499421519504e-06,
+      "loss": 0.6122,
+      "step": 15671
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8170865289313021e-06,
+      "loss": 0.6664,
+      "step": 15672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8161233455320571e-06,
+      "loss": 0.6646,
+      "step": 15673
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.815160391981271e-06,
+      "loss": 0.6695,
+      "step": 15674
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.814197668305987e-06,
+      "loss": 0.7121,
+      "step": 15675
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8132351745332445e-06,
+      "loss": 0.6473,
+      "step": 15676
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8122729106900726e-06,
+      "loss": 0.605,
+      "step": 15677
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.811310876803506e-06,
+      "loss": 0.7034,
+      "step": 15678
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8103490729005546e-06,
+      "loss": 0.6409,
+      "step": 15679
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809387499008235e-06,
+      "loss": 0.6436,
+      "step": 15680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8084261551535531e-06,
+      "loss": 0.654,
+      "step": 15681
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8074650413635086e-06,
+      "loss": 0.6144,
+      "step": 15682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8065041576650944e-06,
+      "loss": 0.6366,
+      "step": 15683
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8055435040852987e-06,
+      "loss": 0.6157,
+      "step": 15684
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8045830806511033e-06,
+      "loss": 0.6484,
+      "step": 15685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8036228873894745e-06,
+      "loss": 0.7901,
+      "step": 15686
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8026629243273885e-06,
+      "loss": 0.7343,
+      "step": 15687
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8017031914918037e-06,
+      "loss": 0.595,
+      "step": 15688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.800743688909673e-06,
+      "loss": 0.6823,
+      "step": 15689
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799784416607947e-06,
+      "loss": 0.6283,
+      "step": 15690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7988253746135686e-06,
+      "loss": 0.6896,
+      "step": 15691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7978665629534675e-06,
+      "loss": 0.6565,
+      "step": 15692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7969079816545754e-06,
+      "loss": 0.6776,
+      "step": 15693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.795949630743815e-06,
+      "loss": 0.589,
+      "step": 15694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.794991510248102e-06,
+      "loss": 0.6302,
+      "step": 15695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7940336201943454e-06,
+      "loss": 0.6605,
+      "step": 15696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.793075960609445e-06,
+      "loss": 0.6825,
+      "step": 15697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.792118531520306e-06,
+      "loss": 0.6238,
+      "step": 15698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7911613329538102e-06,
+      "loss": 0.655,
+      "step": 15699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7902043649368428e-06,
+      "loss": 0.5948,
+      "step": 15700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.789247627496281e-06,
+      "loss": 0.71,
+      "step": 15701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.788291120658997e-06,
+      "loss": 0.6384,
+      "step": 15702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7873348444518523e-06,
+      "loss": 0.6335,
+      "step": 15703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7863787989017078e-06,
+      "loss": 0.6509,
+      "step": 15704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7854229840354109e-06,
+      "loss": 0.6538,
+      "step": 15705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7844673998798045e-06,
+      "loss": 0.6159,
+      "step": 15706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7835120464617317e-06,
+      "loss": 0.6243,
+      "step": 15707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782556923808022e-06,
+      "loss": 0.5872,
+      "step": 15708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7816020319455019e-06,
+      "loss": 0.6494,
+      "step": 15709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7806473709009875e-06,
+      "loss": 0.6723,
+      "step": 15710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7796929407012953e-06,
+      "loss": 0.6219,
+      "step": 15711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7787387413732248e-06,
+      "loss": 0.6224,
+      "step": 15712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7777847729435792e-06,
+      "loss": 0.6585,
+      "step": 15713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7768310354391505e-06,
+      "loss": 0.6716,
+      "step": 15714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7758775288867237e-06,
+      "loss": 0.6701,
+      "step": 15715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7749242533130794e-06,
+      "loss": 0.6196,
+      "step": 15716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7739712087449911e-06,
+      "loss": 0.6393,
+      "step": 15717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7730183952092262e-06,
+      "loss": 0.6263,
+      "step": 15718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7720658127325442e-06,
+      "loss": 0.7231,
+      "step": 15719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.771113461341698e-06,
+      "loss": 0.3216,
+      "step": 15720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7701613410634367e-06,
+      "loss": 0.634,
+      "step": 15721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7692094519244996e-06,
+      "loss": 0.7129,
+      "step": 15722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.768257793951621e-06,
+      "loss": 0.3473,
+      "step": 15723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767306367171533e-06,
+      "loss": 0.6842,
+      "step": 15724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7663551716109495e-06,
+      "loss": 0.6165,
+      "step": 15725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.765404207296587e-06,
+      "loss": 0.6377,
+      "step": 15726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.764453474255159e-06,
+      "loss": 0.649,
+      "step": 15727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7635029725133635e-06,
+      "loss": 0.6188,
+      "step": 15728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.762552702097896e-06,
+      "loss": 0.6179,
+      "step": 15729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7616026630354498e-06,
+      "loss": 0.7114,
+      "step": 15730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7606528553526991e-06,
+      "loss": 0.6974,
+      "step": 15731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7597032790763257e-06,
+      "loss": 0.6617,
+      "step": 15732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7587539342329963e-06,
+      "loss": 0.6883,
+      "step": 15733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7578048208493748e-06,
+      "loss": 0.665,
+      "step": 15734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7568559389521166e-06,
+      "loss": 0.6446,
+      "step": 15735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7559072885678729e-06,
+      "loss": 0.683,
+      "step": 15736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7549588697232866e-06,
+      "loss": 0.6132,
+      "step": 15737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7540106824449943e-06,
+      "loss": 0.6705,
+      "step": 15738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7530627267596267e-06,
+      "loss": 0.5659,
+      "step": 15739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7521150026938071e-06,
+      "loss": 0.6213,
+      "step": 15740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7511675102741532e-06,
+      "loss": 0.6951,
+      "step": 15741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7502202495272757e-06,
+      "loss": 0.6721,
+      "step": 15742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7492732204797825e-06,
+      "loss": 0.6387,
+      "step": 15743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7483264231582642e-06,
+      "loss": 0.7085,
+      "step": 15744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7473798575893152e-06,
+      "loss": 0.7236,
+      "step": 15745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7464335237995223e-06,
+      "loss": 0.6397,
+      "step": 15746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7454874218154584e-06,
+      "loss": 0.6338,
+      "step": 15747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7445415516637033e-06,
+      "loss": 0.6589,
+      "step": 15748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7435959133708169e-06,
+      "loss": 0.644,
+      "step": 15749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7426505069633614e-06,
+      "loss": 0.6896,
+      "step": 15750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7417053324678856e-06,
+      "loss": 0.6447,
+      "step": 15751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.740760389910935e-06,
+      "loss": 0.6371,
+      "step": 15752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7398156793190513e-06,
+      "loss": 0.6701,
+      "step": 15753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7388712007187658e-06,
+      "loss": 0.6873,
+      "step": 15754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737926954136605e-06,
+      "loss": 0.6443,
+      "step": 15755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736982939599089e-06,
+      "loss": 0.687,
+      "step": 15756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7360391571327295e-06,
+      "loss": 0.6473,
+      "step": 15757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7350956067640356e-06,
+      "loss": 0.687,
+      "step": 15758
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.734152288519505e-06,
+      "loss": 0.7237,
+      "step": 15759
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.733209202425632e-06,
+      "loss": 0.6269,
+      "step": 15760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7322663485089042e-06,
+      "loss": 0.6297,
+      "step": 15761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7313237267958016e-06,
+      "loss": 0.6644,
+      "step": 15762
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.730381337312802e-06,
+      "loss": 0.6663,
+      "step": 15763
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7294391800863664e-06,
+      "loss": 0.6898,
+      "step": 15764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.728497255142958e-06,
+      "loss": 0.6395,
+      "step": 15765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7275555625090324e-06,
+      "loss": 0.6633,
+      "step": 15766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.726614102211035e-06,
+      "loss": 0.7503,
+      "step": 15767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7256728742754114e-06,
+      "loss": 0.5984,
+      "step": 15768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7247318787285972e-06,
+      "loss": 0.6394,
+      "step": 15769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7237911155970155e-06,
+      "loss": 0.6866,
+      "step": 15770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.722850584907091e-06,
+      "loss": 0.6343,
+      "step": 15771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7219102866852378e-06,
+      "loss": 0.6681,
+      "step": 15772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.720970220957866e-06,
+      "loss": 0.714,
+      "step": 15773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.720030387751377e-06,
+      "loss": 0.6119,
+      "step": 15774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7190907870921668e-06,
+      "loss": 0.5746,
+      "step": 15775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.718151419006625e-06,
+      "loss": 0.687,
+      "step": 15776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7172122835211337e-06,
+      "loss": 0.6342,
+      "step": 15777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.716273380662068e-06,
+      "loss": 0.6159,
+      "step": 15778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7153347104557993e-06,
+      "loss": 0.6767,
+      "step": 15779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7143962729286912e-06,
+      "loss": 0.6546,
+      "step": 15780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7134580681070977e-06,
+      "loss": 0.6549,
+      "step": 15781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7125200960173726e-06,
+      "loss": 0.6951,
+      "step": 15782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7115823566858546e-06,
+      "loss": 0.6809,
+      "step": 15783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7106448501388827e-06,
+      "loss": 0.7568,
+      "step": 15784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7097075764027871e-06,
+      "loss": 0.6694,
+      "step": 15785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7087705355038919e-06,
+      "loss": 0.6734,
+      "step": 15786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707833727468512e-06,
+      "loss": 0.623,
+      "step": 15787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068971523229639e-06,
+      "loss": 0.6302,
+      "step": 15788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70596081009355e-06,
+      "loss": 0.6547,
+      "step": 15789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050247008065634e-06,
+      "loss": 0.5868,
+      "step": 15790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704088824488299e-06,
+      "loss": 0.6803,
+      "step": 15791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7031531811650403e-06,
+      "loss": 0.6346,
+      "step": 15792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702217770863065e-06,
+      "loss": 0.6478,
+      "step": 15793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7012825936086452e-06,
+      "loss": 0.7045,
+      "step": 15794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7003476494280492e-06,
+      "loss": 0.6483,
+      "step": 15795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6994129383475267e-06,
+      "loss": 0.6456,
+      "step": 15796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6984784603933368e-06,
+      "loss": 0.7118,
+      "step": 15797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6975442155917233e-06,
+      "loss": 0.7156,
+      "step": 15798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6966102039689246e-06,
+      "loss": 0.66,
+      "step": 15799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6956764255511727e-06,
+      "loss": 0.621,
+      "step": 15800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6947428803646936e-06,
+      "loss": 0.6335,
+      "step": 15801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6938095684357093e-06,
+      "loss": 0.6647,
+      "step": 15802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.692876489790426e-06,
+      "loss": 0.6555,
+      "step": 15803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.691943644455052e-06,
+      "loss": 0.6477,
+      "step": 15804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6910110324557893e-06,
+      "loss": 0.6688,
+      "step": 15805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6900786538188285e-06,
+      "loss": 0.69,
+      "step": 15806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.689146508570354e-06,
+      "loss": 0.698,
+      "step": 15807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6882145967365525e-06,
+      "loss": 0.6751,
+      "step": 15808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6872829183435913e-06,
+      "loss": 0.6674,
+      "step": 15809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.686351473417639e-06,
+      "loss": 0.5766,
+      "step": 15810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6854202619848549e-06,
+      "loss": 0.6329,
+      "step": 15811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6844892840713932e-06,
+      "loss": 0.6426,
+      "step": 15812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6835585397034015e-06,
+      "loss": 0.6633,
+      "step": 15813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6826280289070186e-06,
+      "loss": 0.6077,
+      "step": 15814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.681697751708382e-06,
+      "loss": 0.6376,
+      "step": 15815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6807677081336116e-06,
+      "loss": 0.5889,
+      "step": 15816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.679837898208836e-06,
+      "loss": 0.6114,
+      "step": 15817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6789083219601654e-06,
+      "loss": 0.6755,
+      "step": 15818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.677978979413708e-06,
+      "loss": 0.3173,
+      "step": 15819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6770498705955663e-06,
+      "loss": 0.6417,
+      "step": 15820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6761209955318359e-06,
+      "loss": 0.6451,
+      "step": 15821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6751923542485994e-06,
+      "loss": 0.5961,
+      "step": 15822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6742639467719423e-06,
+      "loss": 0.6956,
+      "step": 15823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6733357731279375e-06,
+      "loss": 0.5745,
+      "step": 15824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6724078333426551e-06,
+      "loss": 0.5974,
+      "step": 15825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6714801274421555e-06,
+      "loss": 0.3013,
+      "step": 15826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6705526554524941e-06,
+      "loss": 0.641,
+      "step": 15827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6696254173997196e-06,
+      "loss": 0.6352,
+      "step": 15828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6686984133098737e-06,
+      "loss": 0.6414,
+      "step": 15829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.667771643208992e-06,
+      "loss": 0.6315,
+      "step": 15830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6668451071231028e-06,
+      "loss": 0.6974,
+      "step": 15831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6659188050782294e-06,
+      "loss": 0.5973,
+      "step": 15832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6649927371003871e-06,
+      "loss": 0.6446,
+      "step": 15833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6640669032155866e-06,
+      "loss": 0.6587,
+      "step": 15834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6631413034498268e-06,
+      "loss": 0.5795,
+      "step": 15835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6622159378291036e-06,
+      "loss": 0.6212,
+      "step": 15836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6612908063794098e-06,
+      "loss": 0.6214,
+      "step": 15837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6603659091267266e-06,
+      "loss": 0.5896,
+      "step": 15838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.659441246097031e-06,
+      "loss": 0.6648,
+      "step": 15839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6585168173162914e-06,
+      "loss": 0.6787,
+      "step": 15840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6575926228104743e-06,
+      "loss": 0.6394,
+      "step": 15841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6566686626055307e-06,
+      "loss": 0.6039,
+      "step": 15842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6557449367274125e-06,
+      "loss": 0.7065,
+      "step": 15843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.654821445202064e-06,
+      "loss": 0.6112,
+      "step": 15844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6538981880554217e-06,
+      "loss": 0.7371,
+      "step": 15845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6529751653134163e-06,
+      "loss": 0.7261,
+      "step": 15846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6520523770019704e-06,
+      "loss": 0.7348,
+      "step": 15847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6511298231470007e-06,
+      "loss": 0.7238,
+      "step": 15848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.650207503774418e-06,
+      "loss": 0.6758,
+      "step": 15849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.649285418910127e-06,
+      "loss": 0.5639,
+      "step": 15850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6483635685800248e-06,
+      "loss": 0.6089,
+      "step": 15851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6474419528100005e-06,
+      "loss": 0.6226,
+      "step": 15852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6465205716259391e-06,
+      "loss": 0.6118,
+      "step": 15853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.645599425053721e-06,
+      "loss": 0.6051,
+      "step": 15854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6446785131192123e-06,
+      "loss": 0.6112,
+      "step": 15855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6437578358482765e-06,
+      "loss": 0.635,
+      "step": 15856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6428373932667762e-06,
+      "loss": 0.6724,
+      "step": 15857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.64191718540056e-06,
+      "loss": 0.6493,
+      "step": 15858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6409972122754736e-06,
+      "loss": 0.6372,
+      "step": 15859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6400774739173565e-06,
+      "loss": 0.6037,
+      "step": 15860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6391579703520344e-06,
+      "loss": 0.6476,
+      "step": 15861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6382387016053358e-06,
+      "loss": 0.6705,
+      "step": 15862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6373196677030779e-06,
+      "loss": 0.6256,
+      "step": 15863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636400868671073e-06,
+      "loss": 0.6746,
+      "step": 15864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6354823045351254e-06,
+      "loss": 0.6705,
+      "step": 15865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.634563975321034e-06,
+      "loss": 0.6245,
+      "step": 15866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6336458810545909e-06,
+      "loss": 0.6506,
+      "step": 15867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6327280217615793e-06,
+      "loss": 0.691,
+      "step": 15868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6318103974677802e-06,
+      "loss": 0.6408,
+      "step": 15869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6308930081989648e-06,
+      "loss": 0.6003,
+      "step": 15870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6299758539808974e-06,
+      "loss": 0.5995,
+      "step": 15871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6290589348393371e-06,
+      "loss": 0.6275,
+      "step": 15872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6281422508000399e-06,
+      "loss": 0.6495,
+      "step": 15873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6272258018887455e-06,
+      "loss": 0.6202,
+      "step": 15874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.626309588131194e-06,
+      "loss": 0.6233,
+      "step": 15875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6253936095531199e-06,
+      "loss": 0.6245,
+      "step": 15876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6244778661802463e-06,
+      "loss": 0.6997,
+      "step": 15877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6235623580382953e-06,
+      "loss": 0.6352,
+      "step": 15878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6226470851529785e-06,
+      "loss": 0.6259,
+      "step": 15879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6217320475500043e-06,
+      "loss": 0.6793,
+      "step": 15880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6208172452550664e-06,
+      "loss": 0.6891,
+      "step": 15881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6199026782938608e-06,
+      "loss": 0.6733,
+      "step": 15882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.618988346692073e-06,
+      "loss": 0.651,
+      "step": 15883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6180742504753822e-06,
+      "loss": 0.7169,
+      "step": 15884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6171603896694622e-06,
+      "loss": 0.6091,
+      "step": 15885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6162467642999779e-06,
+      "loss": 0.6168,
+      "step": 15886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6153333743925902e-06,
+      "loss": 0.6697,
+      "step": 15887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.614420219972952e-06,
+      "loss": 0.6223,
+      "step": 15888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6135073010667091e-06,
+      "loss": 0.6537,
+      "step": 15889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6125946176995021e-06,
+      "loss": 0.634,
+      "step": 15890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6116821698969632e-06,
+      "loss": 0.6975,
+      "step": 15891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.610769957684719e-06,
+      "loss": 0.6852,
+      "step": 15892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.609857981088394e-06,
+      "loss": 0.6368,
+      "step": 15893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6089462401335932e-06,
+      "loss": 0.6245,
+      "step": 15894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6080347348459292e-06,
+      "loss": 0.6261,
+      "step": 15895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6071234652509993e-06,
+      "loss": 0.6442,
+      "step": 15896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6062124313743965e-06,
+      "loss": 0.6592,
+      "step": 15897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6053016332417114e-06,
+      "loss": 0.607,
+      "step": 15898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6043910708785248e-06,
+      "loss": 0.6815,
+      "step": 15899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.603480744310406e-06,
+      "loss": 0.6201,
+      "step": 15900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6025706535629227e-06,
+      "loss": 0.6415,
+      "step": 15901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6016607986616373e-06,
+      "loss": 0.6742,
+      "step": 15902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6007511796321018e-06,
+      "loss": 0.6105,
+      "step": 15903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5998417964998637e-06,
+      "loss": 0.6985,
+      "step": 15904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5989326492904643e-06,
+      "loss": 0.6646,
+      "step": 15905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5980237380294373e-06,
+      "loss": 0.6904,
+      "step": 15906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5971150627423093e-06,
+      "loss": 0.6078,
+      "step": 15907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5962066234546014e-06,
+      "loss": 0.6756,
+      "step": 15908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5952984201918275e-06,
+      "loss": 0.6553,
+      "step": 15909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5943904529794952e-06,
+      "loss": 0.6465,
+      "step": 15910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5934827218431059e-06,
+      "loss": 0.671,
+      "step": 15911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5925752268081517e-06,
+      "loss": 0.6854,
+      "step": 15912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5916679679001234e-06,
+      "loss": 0.6839,
+      "step": 15913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5907609451444972e-06,
+      "loss": 0.6805,
+      "step": 15914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5898541585667503e-06,
+      "loss": 0.6863,
+      "step": 15915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5889476081923494e-06,
+      "loss": 0.6721,
+      "step": 15916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5880412940467538e-06,
+      "loss": 0.6542,
+      "step": 15917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5871352161554221e-06,
+      "loss": 0.6483,
+      "step": 15918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5862293745438017e-06,
+      "loss": 0.6579,
+      "step": 15919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5853237692373291e-06,
+      "loss": 0.6524,
+      "step": 15920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5844184002614417e-06,
+      "loss": 0.6598,
+      "step": 15921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5835132676415666e-06,
+      "loss": 0.6558,
+      "step": 15922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5826083714031249e-06,
+      "loss": 0.6674,
+      "step": 15923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5817037115715307e-06,
+      "loss": 0.6315,
+      "step": 15924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5807992881721924e-06,
+      "loss": 0.6202,
+      "step": 15925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579895101230512e-06,
+      "loss": 0.6882,
+      "step": 15926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5789911507718824e-06,
+      "loss": 0.6077,
+      "step": 15927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5780874368216937e-06,
+      "loss": 0.6071,
+      "step": 15928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.577183959405325e-06,
+      "loss": 0.7398,
+      "step": 15929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.576280718548151e-06,
+      "loss": 0.651,
+      "step": 15930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5753777142755422e-06,
+      "loss": 0.6373,
+      "step": 15931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5744749466128596e-06,
+      "loss": 0.6291,
+      "step": 15932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5735724155854538e-06,
+      "loss": 0.6848,
+      "step": 15933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5726701212186757e-06,
+      "loss": 0.6436,
+      "step": 15934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5717680635378674e-06,
+      "loss": 0.6269,
+      "step": 15935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5708662425683619e-06,
+      "loss": 0.2758,
+      "step": 15936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.569964658335489e-06,
+      "loss": 0.6829,
+      "step": 15937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.569063310864567e-06,
+      "loss": 0.7577,
+      "step": 15938
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.568162200180917e-06,
+      "loss": 0.6907,
+      "step": 15939
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5672613263098424e-06,
+      "loss": 0.621,
+      "step": 15940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5663606892766447e-06,
+      "loss": 0.5756,
+      "step": 15941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5654602891066196e-06,
+      "loss": 0.6293,
+      "step": 15942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5645601258250564e-06,
+      "loss": 0.6757,
+      "step": 15943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5636601994572342e-06,
+      "loss": 0.6133,
+      "step": 15944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5627605100284326e-06,
+      "loss": 0.6608,
+      "step": 15945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5618610575639115e-06,
+      "loss": 0.5997,
+      "step": 15946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5609618420889406e-06,
+      "loss": 0.6656,
+      "step": 15947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5600628636287718e-06,
+      "loss": 0.7039,
+      "step": 15948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.559164122208653e-06,
+      "loss": 0.6975,
+      "step": 15949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5582656178538258e-06,
+      "loss": 0.6316,
+      "step": 15950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5573673505895259e-06,
+      "loss": 0.707,
+      "step": 15951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5564693204409841e-06,
+      "loss": 0.6032,
+      "step": 15952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5555715274334172e-06,
+      "loss": 0.6129,
+      "step": 15953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5546739715920411e-06,
+      "loss": 0.6191,
+      "step": 15954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.553776652942066e-06,
+      "loss": 0.6071,
+      "step": 15955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5528795715086932e-06,
+      "loss": 0.6964,
+      "step": 15956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5519827273171173e-06,
+      "loss": 0.6104,
+      "step": 15957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5510861203925253e-06,
+      "loss": 0.6179,
+      "step": 15958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5501897507601016e-06,
+      "loss": 0.612,
+      "step": 15959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5492936184450191e-06,
+      "loss": 0.662,
+      "step": 15960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5483977234724478e-06,
+      "loss": 0.6859,
+      "step": 15961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5475020658675476e-06,
+      "loss": 0.3053,
+      "step": 15962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5466066456554752e-06,
+      "loss": 0.6569,
+      "step": 15963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5457114628613778e-06,
+      "loss": 0.6984,
+      "step": 15964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.544816517510399e-06,
+      "loss": 0.7007,
+      "step": 15965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5439218096276687e-06,
+      "loss": 0.6677,
+      "step": 15966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.543027339238321e-06,
+      "loss": 0.636,
+      "step": 15967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.542133106367475e-06,
+      "loss": 0.6377,
+      "step": 15968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5412391110402458e-06,
+      "loss": 0.6749,
+      "step": 15969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5403453532817413e-06,
+      "loss": 0.671,
+      "step": 15970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539451833117067e-06,
+      "loss": 0.6071,
+      "step": 15971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5385585505713129e-06,
+      "loss": 0.6726,
+      "step": 15972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5376655056695688e-06,
+      "loss": 0.665,
+      "step": 15973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5367726984369158e-06,
+      "loss": 0.6588,
+      "step": 15974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5358801288984304e-06,
+      "loss": 0.5921,
+      "step": 15975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5349877970791804e-06,
+      "loss": 0.6625,
+      "step": 15976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.534095703004227e-06,
+      "loss": 0.7084,
+      "step": 15977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5332038466986253e-06,
+      "loss": 0.731,
+      "step": 15978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5323122281874236e-06,
+      "loss": 0.6694,
+      "step": 15979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5314208474956627e-06,
+      "loss": 0.6802,
+      "step": 15980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5305297046483791e-06,
+      "loss": 0.626,
+      "step": 15981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5296387996705998e-06,
+      "loss": 0.6629,
+      "step": 15982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5287481325873466e-06,
+      "loss": 0.6447,
+      "step": 15983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5278577034236374e-06,
+      "loss": 0.6468,
+      "step": 15984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5269675122044746e-06,
+      "loss": 0.6862,
+      "step": 15985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5260775589548626e-06,
+      "loss": 0.5753,
+      "step": 15986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5251878436997935e-06,
+      "loss": 0.6591,
+      "step": 15987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5242983664642596e-06,
+      "loss": 0.585,
+      "step": 15988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.523409127273242e-06,
+      "loss": 0.6578,
+      "step": 15989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5225201261517142e-06,
+      "loss": 0.6414,
+      "step": 15990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.521631363124646e-06,
+      "loss": 0.6634,
+      "step": 15991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5207428382169942e-06,
+      "loss": 0.6729,
+      "step": 15992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5198545514537177e-06,
+      "loss": 0.6791,
+      "step": 15993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5189665028597622e-06,
+      "loss": 0.6191,
+      "step": 15994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5180786924600699e-06,
+      "loss": 0.6778,
+      "step": 15995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5171911202795753e-06,
+      "loss": 0.6378,
+      "step": 15996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5163037863432073e-06,
+      "loss": 0.6898,
+      "step": 15997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.515416690675885e-06,
+      "loss": 0.597,
+      "step": 15998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.514529833302525e-06,
+      "loss": 0.6902,
+      "step": 15999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513643214248035e-06,
+      "loss": 0.7043,
+      "step": 16000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5127568335373145e-06,
+      "loss": 0.6547,
+      "step": 16001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5118706911952597e-06,
+      "loss": 0.645,
+      "step": 16002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5109847872467576e-06,
+      "loss": 0.6565,
+      "step": 16003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5100991217166916e-06,
+      "loss": 0.6231,
+      "step": 16004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5092136946299308e-06,
+      "loss": 0.709,
+      "step": 16005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5083285060113474e-06,
+      "loss": 0.6984,
+      "step": 16006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5074435558857981e-06,
+      "loss": 0.6424,
+      "step": 16007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.506558844278141e-06,
+      "loss": 0.6797,
+      "step": 16008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5056743712132237e-06,
+      "loss": 0.6485,
+      "step": 16009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.504790136715889e-06,
+      "loss": 0.6891,
+      "step": 16010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5039061408109656e-06,
+      "loss": 0.6328,
+      "step": 16011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5030223835232826e-06,
+      "loss": 0.6711,
+      "step": 16012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5021388648776624e-06,
+      "loss": 0.6763,
+      "step": 16013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5012555848989185e-06,
+      "loss": 0.7046,
+      "step": 16014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5003725436118588e-06,
+      "loss": 0.6852,
+      "step": 16015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4994897410412822e-06,
+      "loss": 0.623,
+      "step": 16016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4986071772119848e-06,
+      "loss": 0.6547,
+      "step": 16017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.497724852148753e-06,
+      "loss": 0.3297,
+      "step": 16018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4968427658763661e-06,
+      "loss": 0.6909,
+      "step": 16019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4959609184196e-06,
+      "loss": 0.6397,
+      "step": 16020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4950793098032213e-06,
+      "loss": 0.7217,
+      "step": 16021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4941979400519902e-06,
+      "loss": 0.6703,
+      "step": 16022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4933168091906615e-06,
+      "loss": 0.6891,
+      "step": 16023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4924359172439794e-06,
+      "loss": 0.6513,
+      "step": 16024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491555264236686e-06,
+      "loss": 0.6211,
+      "step": 16025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4906748501935153e-06,
+      "loss": 0.6403,
+      "step": 16026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4897946751391901e-06,
+      "loss": 0.6406,
+      "step": 16027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4889147390984382e-06,
+      "loss": 0.6863,
+      "step": 16028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4880350420959678e-06,
+      "loss": 0.7069,
+      "step": 16029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4871555841564889e-06,
+      "loss": 0.6606,
+      "step": 16030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4862763653046985e-06,
+      "loss": 0.6097,
+      "step": 16031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4853973855652892e-06,
+      "loss": 0.6681,
+      "step": 16032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4845186449629511e-06,
+      "loss": 0.5661,
+      "step": 16033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4836401435223613e-06,
+      "loss": 0.6295,
+      "step": 16034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4827618812681943e-06,
+      "loss": 0.6429,
+      "step": 16035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.481883858225116e-06,
+      "loss": 0.6075,
+      "step": 16036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4810060744177868e-06,
+      "loss": 0.6386,
+      "step": 16037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.480128529870859e-06,
+      "loss": 0.681,
+      "step": 16038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4792512246089797e-06,
+      "loss": 0.6193,
+      "step": 16039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4783741586567868e-06,
+      "loss": 0.3148,
+      "step": 16040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4774973320389151e-06,
+      "loss": 0.2932,
+      "step": 16041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4766207447799897e-06,
+      "loss": 0.6571,
+      "step": 16042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4757443969046337e-06,
+      "loss": 0.6263,
+      "step": 16043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4748682884374531e-06,
+      "loss": 0.6184,
+      "step": 16044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4739924194030576e-06,
+      "loss": 0.6805,
+      "step": 16045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4731167898260457e-06,
+      "loss": 0.6638,
+      "step": 16046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.472241399731007e-06,
+      "loss": 0.6273,
+      "step": 16047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.471366249142534e-06,
+      "loss": 0.7095,
+      "step": 16048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4704913380852037e-06,
+      "loss": 0.6807,
+      "step": 16049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4696166665835853e-06,
+      "loss": 0.709,
+      "step": 16050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.468742234662246e-06,
+      "loss": 0.6302,
+      "step": 16051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4678680423457447e-06,
+      "loss": 0.6351,
+      "step": 16052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4669940896586343e-06,
+      "loss": 0.6246,
+      "step": 16053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4661203766254583e-06,
+      "loss": 0.6748,
+      "step": 16054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4652469032707572e-06,
+      "loss": 0.6528,
+      "step": 16055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4643736696190625e-06,
+      "loss": 0.6614,
+      "step": 16056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.463500675694899e-06,
+      "loss": 0.6492,
+      "step": 16057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4626279215227857e-06,
+      "loss": 0.6537,
+      "step": 16058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4617554071272344e-06,
+      "loss": 0.6476,
+      "step": 16059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.46088313253275e-06,
+      "loss": 0.6181,
+      "step": 16060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4600110977638292e-06,
+      "loss": 0.691,
+      "step": 16061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4591393028449685e-06,
+      "loss": 0.6044,
+      "step": 16062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4582677478006468e-06,
+      "loss": 0.6747,
+      "step": 16063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4573964326553447e-06,
+      "loss": 0.6609,
+      "step": 16064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4565253574335336e-06,
+      "loss": 0.6245,
+      "step": 16065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4556545221596774e-06,
+      "loss": 0.6116,
+      "step": 16066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4547839268582353e-06,
+      "loss": 0.6666,
+      "step": 16067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4539135715536546e-06,
+      "loss": 0.6458,
+      "step": 16068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4530434562703877e-06,
+      "loss": 0.6199,
+      "step": 16069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4521735810328651e-06,
+      "loss": 0.657,
+      "step": 16070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4513039458655199e-06,
+      "loss": 0.6073,
+      "step": 16071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4504345507927753e-06,
+      "loss": 0.6317,
+      "step": 16072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.449565395839051e-06,
+      "loss": 0.6037,
+      "step": 16073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4486964810287551e-06,
+      "loss": 0.6338,
+      "step": 16074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4478278063862961e-06,
+      "loss": 0.6068,
+      "step": 16075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.446959371936062e-06,
+      "loss": 0.6461,
+      "step": 16076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4460911777024521e-06,
+      "loss": 0.6619,
+      "step": 16077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4452232237098473e-06,
+      "loss": 0.5834,
+      "step": 16078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4443555099826245e-06,
+      "loss": 0.6059,
+      "step": 16079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4434880365451532e-06,
+      "loss": 0.7221,
+      "step": 16080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.442620803421797e-06,
+      "loss": 0.6262,
+      "step": 16081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4417538106369155e-06,
+      "loss": 0.7793,
+      "step": 16082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4408870582148538e-06,
+      "loss": 0.7162,
+      "step": 16083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4400205461799576e-06,
+      "loss": 0.6473,
+      "step": 16084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4391542745565634e-06,
+      "loss": 0.6931,
+      "step": 16085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4382882433689994e-06,
+      "loss": 0.6566,
+      "step": 16086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4374224526415903e-06,
+      "loss": 0.6483,
+      "step": 16087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4365569023986525e-06,
+      "loss": 0.6013,
+      "step": 16088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4356915926644942e-06,
+      "loss": 0.6412,
+      "step": 16089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4348265234634185e-06,
+      "loss": 0.6893,
+      "step": 16090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4339616948197211e-06,
+      "loss": 0.6631,
+      "step": 16091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4330971067576915e-06,
+      "loss": 0.6168,
+      "step": 16092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4322327593016117e-06,
+      "loss": 0.6384,
+      "step": 16093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4313686524757587e-06,
+      "loss": 0.6338,
+      "step": 16094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.430504786304402e-06,
+      "loss": 0.6895,
+      "step": 16095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4296411608118e-06,
+      "loss": 0.6349,
+      "step": 16096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4287777760222078e-06,
+      "loss": 0.692,
+      "step": 16097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.427914631959879e-06,
+      "loss": 0.7584,
+      "step": 16098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4270517286490526e-06,
+      "loss": 0.666,
+      "step": 16099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4261890661139631e-06,
+      "loss": 0.678,
+      "step": 16100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4253266443788427e-06,
+      "loss": 0.7028,
+      "step": 16101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4244644634679084e-06,
+      "loss": 0.6998,
+      "step": 16102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.423602523405375e-06,
+      "loss": 0.7394,
+      "step": 16103
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4227408242154539e-06,
+      "loss": 0.6756,
+      "step": 16104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.421879365922343e-06,
+      "loss": 0.6728,
+      "step": 16105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4210181485502394e-06,
+      "loss": 0.592,
+      "step": 16106
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4201571721233287e-06,
+      "loss": 0.625,
+      "step": 16107
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4192964366657936e-06,
+      "loss": 0.6039,
+      "step": 16108
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.418435942201808e-06,
+      "loss": 0.6362,
+      "step": 16109
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.417575688755538e-06,
+      "loss": 0.6963,
+      "step": 16110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4167156763511458e-06,
+      "loss": 0.6586,
+      "step": 16111
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4158559050127852e-06,
+      "loss": 0.7141,
+      "step": 16112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4149963747646033e-06,
+      "loss": 0.6543,
+      "step": 16113
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4141370856307413e-06,
+      "loss": 0.6606,
+      "step": 16114
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4132780376353295e-06,
+      "loss": 0.6476,
+      "step": 16115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4124192308024965e-06,
+      "loss": 0.7152,
+      "step": 16116
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4115606651563607e-06,
+      "loss": 0.6647,
+      "step": 16117
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4107023407210397e-06,
+      "loss": 0.697,
+      "step": 16118
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4098442575206372e-06,
+      "loss": 0.6613,
+      "step": 16119
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4089864155792542e-06,
+      "loss": 0.6612,
+      "step": 16120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4081288149209838e-06,
+      "loss": 0.6469,
+      "step": 16121
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4072714555699096e-06,
+      "loss": 0.3145,
+      "step": 16122
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4064143375501115e-06,
+      "loss": 0.6232,
+      "step": 16123
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4055574608856638e-06,
+      "loss": 0.6393,
+      "step": 16124
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4047008256006323e-06,
+      "loss": 0.6547,
+      "step": 16125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4038444317190735e-06,
+      "loss": 0.6954,
+      "step": 16126
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4029882792650429e-06,
+      "loss": 0.6755,
+      "step": 16127
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4021323682625842e-06,
+      "loss": 0.6769,
+      "step": 16128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4012766987357362e-06,
+      "loss": 0.6845,
+      "step": 16129
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4004212707085319e-06,
+      "loss": 0.6687,
+      "step": 16130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3995660842049952e-06,
+      "loss": 0.6944,
+      "step": 16131
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.398711139249145e-06,
+      "loss": 0.7365,
+      "step": 16132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3978564358649926e-06,
+      "loss": 0.654,
+      "step": 16133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.397001974076546e-06,
+      "loss": 0.6748,
+      "step": 16134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3961477539077972e-06,
+      "loss": 0.6802,
+      "step": 16135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3952937753827411e-06,
+      "loss": 0.7329,
+      "step": 16136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3944400385253588e-06,
+      "loss": 0.6357,
+      "step": 16137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.393586543359633e-06,
+      "loss": 0.6611,
+      "step": 16138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3927332899095325e-06,
+      "loss": 0.7292,
+      "step": 16139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3918802781990227e-06,
+      "loss": 0.7054,
+      "step": 16140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3910275082520575e-06,
+      "loss": 0.633,
+      "step": 16141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3901749800925878e-06,
+      "loss": 0.6808,
+      "step": 16142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3893226937445602e-06,
+      "loss": 0.6115,
+      "step": 16143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3884706492319088e-06,
+      "loss": 0.6652,
+      "step": 16144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3876188465785667e-06,
+      "loss": 0.6675,
+      "step": 16145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3867672858084536e-06,
+      "loss": 0.6645,
+      "step": 16146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3859159669454891e-06,
+      "loss": 0.6733,
+      "step": 16147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3850648900135811e-06,
+      "loss": 0.6796,
+      "step": 16148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3842140550366334e-06,
+      "loss": 0.6725,
+      "step": 16149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3833634620385417e-06,
+      "loss": 0.71,
+      "step": 16150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3825131110431954e-06,
+      "loss": 0.7027,
+      "step": 16151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3816630020744782e-06,
+      "loss": 0.7481,
+      "step": 16152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.380813135156266e-06,
+      "loss": 0.7035,
+      "step": 16153
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3799635103124255e-06,
+      "loss": 0.5892,
+      "step": 16154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3791141275668196e-06,
+      "loss": 0.6663,
+      "step": 16155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.378264986943303e-06,
+      "loss": 0.6273,
+      "step": 16156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.377416088465724e-06,
+      "loss": 0.6679,
+      "step": 16157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3765674321579281e-06,
+      "loss": 0.7124,
+      "step": 16158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3757190180437473e-06,
+      "loss": 0.6996,
+      "step": 16159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3748708461470128e-06,
+      "loss": 0.6886,
+      "step": 16160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3740229164915409e-06,
+      "loss": 0.6378,
+      "step": 16161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3731752291011492e-06,
+      "loss": 0.6893,
+      "step": 16162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.372327783999644e-06,
+      "loss": 0.6787,
+      "step": 16163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3714805812108279e-06,
+      "loss": 0.6427,
+      "step": 16164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3706336207584948e-06,
+      "loss": 0.717,
+      "step": 16165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3697869026664302e-06,
+      "loss": 0.613,
+      "step": 16166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3689404269584172e-06,
+      "loss": 0.6538,
+      "step": 16167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.368094193658227e-06,
+      "loss": 0.6338,
+      "step": 16168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3672482027896295e-06,
+      "loss": 0.6484,
+      "step": 16169
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366402454376381e-06,
+      "loss": 0.6432,
+      "step": 16170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3655569484422382e-06,
+      "loss": 0.6623,
+      "step": 16171
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3647116850109453e-06,
+      "loss": 0.6372,
+      "step": 16172
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3638666641062458e-06,
+      "loss": 0.6419,
+      "step": 16173
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3630218857518673e-06,
+      "loss": 0.5878,
+      "step": 16174
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3621773499715385e-06,
+      "loss": 0.694,
+      "step": 16175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3613330567889782e-06,
+      "loss": 0.7186,
+      "step": 16176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3604890062278985e-06,
+      "loss": 0.6539,
+      "step": 16177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359645198312003e-06,
+      "loss": 0.7397,
+      "step": 16178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3588016330649978e-06,
+      "loss": 0.683,
+      "step": 16179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3579583105105677e-06,
+      "loss": 0.6226,
+      "step": 16180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3571152306724e-06,
+      "loss": 0.6836,
+      "step": 16181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.356272393574174e-06,
+      "loss": 0.661,
+      "step": 16182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3554297992395593e-06,
+      "loss": 0.6314,
+      "step": 16183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3545874476922227e-06,
+      "loss": 0.5956,
+      "step": 16184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3537453389558197e-06,
+      "loss": 0.6395,
+      "step": 16185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3529034730540036e-06,
+      "loss": 0.6603,
+      "step": 16186
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3520618500104177e-06,
+      "loss": 0.6691,
+      "step": 16187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3512204698486996e-06,
+      "loss": 0.6852,
+      "step": 16188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3503793325924807e-06,
+      "loss": 0.696,
+      "step": 16189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.349538438265383e-06,
+      "loss": 0.6424,
+      "step": 16190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3486977868910246e-06,
+      "loss": 0.6652,
+      "step": 16191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.347857378493017e-06,
+      "loss": 0.6845,
+      "step": 16192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3470172130949599e-06,
+      "loss": 0.6558,
+      "step": 16193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3461772907204518e-06,
+      "loss": 0.6534,
+      "step": 16194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3453376113930828e-06,
+      "loss": 0.674,
+      "step": 16195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.344498175136435e-06,
+      "loss": 0.6091,
+      "step": 16196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3436589819740842e-06,
+      "loss": 0.68,
+      "step": 16197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3428200319295981e-06,
+      "loss": 0.6504,
+      "step": 16198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3419813250265457e-06,
+      "loss": 0.6889,
+      "step": 16199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3411428612884758e-06,
+      "loss": 0.6325,
+      "step": 16200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3403046407389386e-06,
+      "loss": 0.7426,
+      "step": 16201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3394666634014775e-06,
+      "loss": 0.6047,
+      "step": 16202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3386289292996258e-06,
+      "loss": 0.6473,
+      "step": 16203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3377914384569124e-06,
+      "loss": 0.6687,
+      "step": 16204
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3369541908968608e-06,
+      "loss": 0.7229,
+      "step": 16205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.336117186642979e-06,
+      "loss": 0.696,
+      "step": 16206
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3352804257187813e-06,
+      "loss": 0.6985,
+      "step": 16207
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3344439081477667e-06,
+      "loss": 0.5977,
+      "step": 16208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3336076339534288e-06,
+      "loss": 0.6511,
+      "step": 16209
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3327716031592553e-06,
+      "loss": 0.6216,
+      "step": 16210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3319358157887252e-06,
+      "loss": 0.6982,
+      "step": 16211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3311002718653164e-06,
+      "loss": 0.7058,
+      "step": 16212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3302649714124893e-06,
+      "loss": 0.6495,
+      "step": 16213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.329429914453707e-06,
+      "loss": 0.6628,
+      "step": 16214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3285951010124222e-06,
+      "loss": 0.6208,
+      "step": 16215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3277605311120801e-06,
+      "loss": 0.631,
+      "step": 16216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3269262047761222e-06,
+      "loss": 0.6443,
+      "step": 16217
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3260921220279787e-06,
+      "loss": 0.3175,
+      "step": 16218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3252582828910775e-06,
+      "loss": 0.6367,
+      "step": 16219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3244246873888356e-06,
+      "loss": 0.6525,
+      "step": 16220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.323591335544665e-06,
+      "loss": 0.6402,
+      "step": 16221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3227582273819727e-06,
+      "loss": 0.6403,
+      "step": 16222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3219253629241545e-06,
+      "loss": 0.6732,
+      "step": 16223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.321092742194604e-06,
+      "loss": 0.5695,
+      "step": 16224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3202603652167068e-06,
+      "loss": 0.6296,
+      "step": 16225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3194282320138362e-06,
+      "loss": 0.6511,
+      "step": 16226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.318596342609363e-06,
+      "loss": 0.6262,
+      "step": 16227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3177646970266566e-06,
+      "loss": 0.5878,
+      "step": 16228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3169332952890712e-06,
+      "loss": 0.6665,
+      "step": 16229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3161021374199568e-06,
+      "loss": 0.6754,
+      "step": 16230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.31527122344266e-06,
+      "loss": 0.607,
+      "step": 16231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3144405533805138e-06,
+      "loss": 0.6486,
+      "step": 16232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.313610127256848e-06,
+      "loss": 0.6387,
+      "step": 16233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3127799450949862e-06,
+      "loss": 0.5968,
+      "step": 16234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3119500069182466e-06,
+      "loss": 0.692,
+      "step": 16235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.311120312749936e-06,
+      "loss": 0.6862,
+      "step": 16236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.310290862613357e-06,
+      "loss": 0.7643,
+      "step": 16237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3094616565318064e-06,
+      "loss": 0.6675,
+      "step": 16238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3086326945285721e-06,
+      "loss": 0.6721,
+      "step": 16239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3078039766269357e-06,
+      "loss": 0.6876,
+      "step": 16240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306975502850173e-06,
+      "loss": 0.6395,
+      "step": 16241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3061472732215518e-06,
+      "loss": 0.6439,
+      "step": 16242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3053192877643329e-06,
+      "loss": 0.7083,
+      "step": 16243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3044915465017738e-06,
+      "loss": 0.6584,
+      "step": 16244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3036640494571162e-06,
+      "loss": 0.7054,
+      "step": 16245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3028367966536048e-06,
+      "loss": 0.6197,
+      "step": 16246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30200978811447e-06,
+      "loss": 0.6217,
+      "step": 16247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301183023862943e-06,
+      "loss": 0.6364,
+      "step": 16248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.300356503922242e-06,
+      "loss": 0.6133,
+      "step": 16249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299530228315581e-06,
+      "loss": 0.6421,
+      "step": 16250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2987041970661686e-06,
+      "loss": 0.5973,
+      "step": 16251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2978784101971975e-06,
+      "loss": 0.6683,
+      "step": 16252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2970528677318662e-06,
+      "loss": 0.6702,
+      "step": 16253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.296227569693358e-06,
+      "loss": 0.6164,
+      "step": 16254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2954025161048522e-06,
+      "loss": 0.6906,
+      "step": 16255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2945777069895216e-06,
+      "loss": 0.6416,
+      "step": 16256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2937531423705308e-06,
+      "loss": 0.6141,
+      "step": 16257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.292928822271038e-06,
+      "loss": 0.6673,
+      "step": 16258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2921047467141957e-06,
+      "loss": 0.6347,
+      "step": 16259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2912809157231465e-06,
+      "loss": 0.6708,
+      "step": 16260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2904573293210298e-06,
+      "loss": 0.6125,
+      "step": 16261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.289633987530977e-06,
+      "loss": 0.6386,
+      "step": 16262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.28881089037611e-06,
+      "loss": 0.5659,
+      "step": 16263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2879880378795496e-06,
+      "loss": 0.7151,
+      "step": 16264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.287165430064401e-06,
+      "loss": 0.6485,
+      "step": 16265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2863430669537701e-06,
+      "loss": 0.6801,
+      "step": 16266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.285520948570752e-06,
+      "loss": 0.6272,
+      "step": 16267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2846990749384392e-06,
+      "loss": 0.6971,
+      "step": 16268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2838774460799119e-06,
+      "loss": 0.6254,
+      "step": 16269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2830560620182486e-06,
+      "loss": 0.632,
+      "step": 16270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2822349227765185e-06,
+      "loss": 0.6755,
+      "step": 16271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2814140283777788e-06,
+      "loss": 0.6288,
+      "step": 16272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2805933788450886e-06,
+      "loss": 0.6398,
+      "step": 16273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2797729742014952e-06,
+      "loss": 0.6165,
+      "step": 16274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2789528144700413e-06,
+      "loss": 0.6476,
+      "step": 16275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2781328996737596e-06,
+      "loss": 0.6361,
+      "step": 16276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2773132298356794e-06,
+      "loss": 0.648,
+      "step": 16277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.27649380497882e-06,
+      "loss": 0.6317,
+      "step": 16278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2756746251261976e-06,
+      "loss": 0.6273,
+      "step": 16279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.274855690300818e-06,
+      "loss": 0.5917,
+      "step": 16280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2740370005256809e-06,
+      "loss": 0.6749,
+      "step": 16281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2732185558237808e-06,
+      "loss": 0.6201,
+      "step": 16282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.272400356218103e-06,
+      "loss": 0.5937,
+      "step": 16283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2715824017316302e-06,
+      "loss": 0.6901,
+      "step": 16284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2707646923873306e-06,
+      "loss": 0.6208,
+      "step": 16285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2699472282081716e-06,
+      "loss": 0.6262,
+      "step": 16286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2691300092171134e-06,
+      "loss": 0.7332,
+      "step": 16287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2683130354371033e-06,
+      "loss": 0.6267,
+      "step": 16288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2674963068910927e-06,
+      "loss": 0.6243,
+      "step": 16289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2666798236020207e-06,
+      "loss": 0.6757,
+      "step": 16290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2658635855928114e-06,
+      "loss": 0.6074,
+      "step": 16291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2650475928863948e-06,
+      "loss": 0.7089,
+      "step": 16292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2642318455056858e-06,
+      "loss": 0.6431,
+      "step": 16293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2634163434735968e-06,
+      "loss": 0.6181,
+      "step": 16294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.26260108681303e-06,
+      "loss": 0.6711,
+      "step": 16295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2617860755468846e-06,
+      "loss": 0.6281,
+      "step": 16296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.260971309698048e-06,
+      "loss": 0.5143,
+      "step": 16297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2601567892894063e-06,
+      "loss": 0.6862,
+      "step": 16298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.259342514343832e-06,
+      "loss": 0.6643,
+      "step": 16299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2585284848841972e-06,
+      "loss": 0.602,
+      "step": 16300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2577147009333645e-06,
+      "loss": 0.5953,
+      "step": 16301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.256901162514188e-06,
+      "loss": 0.6615,
+      "step": 16302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2560878696495183e-06,
+      "loss": 0.6427,
+      "step": 16303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2552748223621947e-06,
+      "loss": 0.6155,
+      "step": 16304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2544620206750523e-06,
+      "loss": 0.6371,
+      "step": 16305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.253649464610921e-06,
+      "loss": 0.6811,
+      "step": 16306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.25283715419262e-06,
+      "loss": 0.6044,
+      "step": 16307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.252025089442962e-06,
+      "loss": 0.7265,
+      "step": 16308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.251213270384759e-06,
+      "loss": 0.6643,
+      "step": 16309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2504016970408105e-06,
+      "loss": 0.6919,
+      "step": 16310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2495903694339061e-06,
+      "loss": 0.6746,
+      "step": 16311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2487792875868354e-06,
+      "loss": 0.6646,
+      "step": 16312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2479684515223766e-06,
+      "loss": 0.616,
+      "step": 16313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2471578612633028e-06,
+      "loss": 0.623,
+      "step": 16314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2463475168323802e-06,
+      "loss": 0.6864,
+      "step": 16315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2455374182523684e-06,
+      "loss": 0.6413,
+      "step": 16316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2447275655460178e-06,
+      "loss": 0.6751,
+      "step": 16317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.243917958736075e-06,
+      "loss": 0.6929,
+      "step": 16318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2431085978452783e-06,
+      "loss": 0.5959,
+      "step": 16319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2422994828963576e-06,
+      "loss": 0.6427,
+      "step": 16320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2414906139120385e-06,
+      "loss": 0.6395,
+      "step": 16321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.240681990915038e-06,
+      "loss": 0.6272,
+      "step": 16322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2398736139280687e-06,
+      "loss": 0.6451,
+      "step": 16323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2390654829738302e-06,
+      "loss": 0.6519,
+      "step": 16324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2382575980750221e-06,
+      "loss": 0.6668,
+      "step": 16325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2374499592543331e-06,
+      "loss": 0.6668,
+      "step": 16326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2366425665344473e-06,
+      "loss": 0.6754,
+      "step": 16327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2358354199380373e-06,
+      "loss": 0.6479,
+      "step": 16328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2350285194877798e-06,
+      "loss": 0.6853,
+      "step": 16329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2342218652063309e-06,
+      "loss": 0.6924,
+      "step": 16330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2334154571163459e-06,
+      "loss": 0.6311,
+      "step": 16331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2326092952404756e-06,
+      "loss": 0.626,
+      "step": 16332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2318033796013607e-06,
+      "loss": 0.5712,
+      "step": 16333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2309977102216352e-06,
+      "loss": 0.6546,
+      "step": 16334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2301922871239268e-06,
+      "loss": 0.7424,
+      "step": 16335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2293871103308596e-06,
+      "loss": 0.61,
+      "step": 16336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2285821798650388e-06,
+      "loss": 0.6924,
+      "step": 16337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2277774957490807e-06,
+      "loss": 0.652,
+      "step": 16338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2269730580055806e-06,
+      "loss": 0.6773,
+      "step": 16339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2261688666571315e-06,
+      "loss": 0.6085,
+      "step": 16340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.225364921726322e-06,
+      "loss": 0.6432,
+      "step": 16341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2245612232357307e-06,
+      "loss": 0.6119,
+      "step": 16342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2237577712079274e-06,
+      "loss": 0.6598,
+      "step": 16343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2229545656654784e-06,
+      "loss": 0.6862,
+      "step": 16344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2221516066309435e-06,
+      "loss": 0.6765,
+      "step": 16345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2213488941268726e-06,
+      "loss": 0.7015,
+      "step": 16346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2205464281758117e-06,
+      "loss": 0.6481,
+      "step": 16347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2197442088002964e-06,
+      "loss": 0.6703,
+      "step": 16348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.218942236022862e-06,
+      "loss": 0.5547,
+      "step": 16349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2181405098660282e-06,
+      "loss": 0.6543,
+      "step": 16350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2173390303523124e-06,
+      "loss": 0.6965,
+      "step": 16351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.216537797504226e-06,
+      "loss": 0.6146,
+      "step": 16352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2157368113442703e-06,
+      "loss": 0.7134,
+      "step": 16353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2149360718949431e-06,
+      "loss": 0.5216,
+      "step": 16354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2141355791787347e-06,
+      "loss": 0.6555,
+      "step": 16355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213335333218123e-06,
+      "loss": 0.6182,
+      "step": 16356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.212535334035584e-06,
+      "loss": 0.6782,
+      "step": 16357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2117355816535903e-06,
+      "loss": 0.6288,
+      "step": 16358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.210936076094601e-06,
+      "loss": 0.6627,
+      "step": 16359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.21013681738107e-06,
+      "loss": 0.6605,
+      "step": 16360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2093378055354466e-06,
+      "loss": 0.6255,
+      "step": 16361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.208539040580171e-06,
+      "loss": 0.7217,
+      "step": 16362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2077405225376748e-06,
+      "loss": 0.6849,
+      "step": 16363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2069422514303875e-06,
+      "loss": 0.6519,
+      "step": 16364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2061442272807266e-06,
+      "loss": 0.6591,
+      "step": 16365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2053464501111056e-06,
+      "loss": 0.635,
+      "step": 16366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.204548919943932e-06,
+      "loss": 0.6897,
+      "step": 16367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2037516368016044e-06,
+      "loss": 0.6517,
+      "step": 16368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.202954600706514e-06,
+      "loss": 0.6826,
+      "step": 16369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2021578116810461e-06,
+      "loss": 0.6811,
+      "step": 16370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2013612697475797e-06,
+      "loss": 0.7513,
+      "step": 16371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2005649749284853e-06,
+      "loss": 0.6592,
+      "step": 16372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1997689272461276e-06,
+      "loss": 0.555,
+      "step": 16373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1989731267228643e-06,
+      "loss": 0.6304,
+      "step": 16374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1981775733810474e-06,
+      "loss": 0.6772,
+      "step": 16375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1973822672430168e-06,
+      "loss": 0.6275,
+      "step": 16376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1965872083311093e-06,
+      "loss": 0.6604,
+      "step": 16377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1957923966676576e-06,
+      "loss": 0.6636,
+      "step": 16378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1949978322749833e-06,
+      "loss": 0.6286,
+      "step": 16379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1942035151754018e-06,
+      "loss": 0.7072,
+      "step": 16380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1934094453912237e-06,
+      "loss": 0.6797,
+      "step": 16381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1926156229447472e-06,
+      "loss": 0.6357,
+      "step": 16382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1918220478582686e-06,
+      "loss": 0.6763,
+      "step": 16383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.191028720154077e-06,
+      "loss": 0.7263,
+      "step": 16384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1902356398544534e-06,
+      "loss": 0.6284,
+      "step": 16385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.18944280698167e-06,
+      "loss": 0.7335,
+      "step": 16386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1886502215579964e-06,
+      "loss": 0.6257,
+      "step": 16387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.187857883605691e-06,
+      "loss": 0.6419,
+      "step": 16388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1870657931470076e-06,
+      "loss": 0.6029,
+      "step": 16389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.186273950204193e-06,
+      "loss": 0.6902,
+      "step": 16390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1854823547994855e-06,
+      "loss": 0.6665,
+      "step": 16391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1846910069551186e-06,
+      "loss": 0.5683,
+      "step": 16392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1838999066933165e-06,
+      "loss": 0.6179,
+      "step": 16393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183109054036301e-06,
+      "loss": 0.6398,
+      "step": 16394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1823184490062778e-06,
+      "loss": 0.6945,
+      "step": 16395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1815280916254557e-06,
+      "loss": 0.6173,
+      "step": 16396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.180737981916028e-06,
+      "loss": 0.6487,
+      "step": 16397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1799481199001916e-06,
+      "loss": 0.6422,
+      "step": 16398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1791585056001265e-06,
+      "loss": 0.6277,
+      "step": 16399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.178369139038009e-06,
+      "loss": 0.6571,
+      "step": 16400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1775800202360133e-06,
+      "loss": 0.6384,
+      "step": 16401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1767911492162965e-06,
+      "loss": 0.6165,
+      "step": 16402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1760025260010165e-06,
+      "loss": 0.6225,
+      "step": 16403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1752141506123238e-06,
+      "loss": 0.6372,
+      "step": 16404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.174426023072357e-06,
+      "loss": 0.6735,
+      "step": 16405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.173638143403254e-06,
+      "loss": 0.6078,
+      "step": 16406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1728505116271417e-06,
+      "loss": 0.6831,
+      "step": 16407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.172063127766142e-06,
+      "loss": 0.6936,
+      "step": 16408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1712759918423677e-06,
+      "loss": 0.714,
+      "step": 16409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1704891038779264e-06,
+      "loss": 0.6841,
+      "step": 16410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1697024638949184e-06,
+      "loss": 0.6289,
+      "step": 16411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.168916071915437e-06,
+      "loss": 0.6644,
+      "step": 16412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1681299279615689e-06,
+      "loss": 0.6743,
+      "step": 16413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1673440320553941e-06,
+      "loss": 0.6518,
+      "step": 16414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.166558384218982e-06,
+      "loss": 0.5794,
+      "step": 16415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1657729844743992e-06,
+      "loss": 0.6332,
+      "step": 16416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1649878328437037e-06,
+      "loss": 0.612,
+      "step": 16417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1642029293489466e-06,
+      "loss": 0.6375,
+      "step": 16418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1634182740121746e-06,
+      "loss": 0.7323,
+      "step": 16419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1626338668554272e-06,
+      "loss": 0.5818,
+      "step": 16420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1618497079007285e-06,
+      "loss": 0.6603,
+      "step": 16421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1610657971701044e-06,
+      "loss": 0.628,
+      "step": 16422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1602821346855731e-06,
+      "loss": 0.6914,
+      "step": 16423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1594987204691433e-06,
+      "loss": 0.6786,
+      "step": 16424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1587155545428163e-06,
+      "loss": 0.6871,
+      "step": 16425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1579326369285903e-06,
+      "loss": 0.6541,
+      "step": 16426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1571499676484522e-06,
+      "loss": 0.672,
+      "step": 16427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1563675467243829e-06,
+      "loss": 0.6201,
+      "step": 16428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1555853741783596e-06,
+      "loss": 0.656,
+      "step": 16429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154803450032348e-06,
+      "loss": 0.6331,
+      "step": 16430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1540217743083093e-06,
+      "loss": 0.6078,
+      "step": 16431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1532403470281983e-06,
+      "loss": 0.7264,
+      "step": 16432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1524591682139619e-06,
+      "loss": 0.6492,
+      "step": 16433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.151678237887538e-06,
+      "loss": 0.7039,
+      "step": 16434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.150897556070859e-06,
+      "loss": 0.5965,
+      "step": 16435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.150117122785853e-06,
+      "loss": 0.7002,
+      "step": 16436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1493369380544383e-06,
+      "loss": 0.7326,
+      "step": 16437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1485570018985248e-06,
+      "loss": 0.656,
+      "step": 16438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1477773143400205e-06,
+      "loss": 0.6154,
+      "step": 16439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1469978754008238e-06,
+      "loss": 0.6918,
+      "step": 16440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1462186851028222e-06,
+      "loss": 0.6562,
+      "step": 16441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 0.5955,
+      "step": 16442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.144661050517939e-06,
+      "loss": 0.6306,
+      "step": 16443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.143882606274803e-06,
+      "loss": 0.6149,
+      "step": 16444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1431044107603584e-06,
+      "loss": 0.6517,
+      "step": 16445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1423264639964627e-06,
+      "loss": 0.6106,
+      "step": 16446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1415487660049573e-06,
+      "loss": 0.5873,
+      "step": 16447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1407713168076939e-06,
+      "loss": 0.6718,
+      "step": 16448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1399941164265016e-06,
+      "loss": 0.6076,
+      "step": 16449
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1392171648832117e-06,
+      "loss": 0.6506,
+      "step": 16450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1384404621996425e-06,
+      "loss": 0.6715,
+      "step": 16451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1376640083976098e-06,
+      "loss": 0.6786,
+      "step": 16452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1368878034989227e-06,
+      "loss": 0.6503,
+      "step": 16453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1361118475253774e-06,
+      "loss": 0.6347,
+      "step": 16454
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1353361404987683e-06,
+      "loss": 0.631,
+      "step": 16455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1345606824408817e-06,
+      "loss": 0.6194,
+      "step": 16456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1337854733734977e-06,
+      "loss": 0.6125,
+      "step": 16457
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1330105133183843e-06,
+      "loss": 0.6297,
+      "step": 16458
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1322358022973145e-06,
+      "loss": 0.6626,
+      "step": 16459
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1314613403320408e-06,
+      "loss": 0.6505,
+      "step": 16460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1306871274443143e-06,
+      "loss": 0.6393,
+      "step": 16461
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1299131636558813e-06,
+      "loss": 0.6448,
+      "step": 16462
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1291394489884778e-06,
+      "loss": 0.646,
+      "step": 16463
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.128365983463835e-06,
+      "loss": 0.6253,
+      "step": 16464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1275927671036746e-06,
+      "loss": 0.682,
+      "step": 16465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.126819799929716e-06,
+      "loss": 0.6653,
+      "step": 16466
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1260470819636614e-06,
+      "loss": 0.664,
+      "step": 16467
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1252746132272197e-06,
+      "loss": 0.625,
+      "step": 16468
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.124502393742084e-06,
+      "loss": 0.6074,
+      "step": 16469
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1237304235299418e-06,
+      "loss": 0.655,
+      "step": 16470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1229587026124755e-06,
+      "loss": 0.6948,
+      "step": 16471
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1221872310113612e-06,
+      "loss": 0.6878,
+      "step": 16472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1214160087482607e-06,
+      "loss": 0.6277,
+      "step": 16473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1206450358448373e-06,
+      "loss": 0.6008,
+      "step": 16474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.119874312322744e-06,
+      "loss": 0.682,
+      "step": 16475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1191038382036267e-06,
+      "loss": 0.5892,
+      "step": 16476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1183336135091261e-06,
+      "loss": 0.6155,
+      "step": 16477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1175636382608712e-06,
+      "loss": 0.7031,
+      "step": 16478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1167939124804894e-06,
+      "loss": 0.6139,
+      "step": 16479
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.116024436189599e-06,
+      "loss": 0.6921,
+      "step": 16480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1152552094098102e-06,
+      "loss": 0.5985,
+      "step": 16481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.114486232162728e-06,
+      "loss": 0.7145,
+      "step": 16482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.113717504469949e-06,
+      "loss": 0.5793,
+      "step": 16483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.112949026353063e-06,
+      "loss": 0.6259,
+      "step": 16484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1121807978336552e-06,
+      "loss": 0.6436,
+      "step": 16485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1114128189332984e-06,
+      "loss": 0.6543,
+      "step": 16486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.110645089673561e-06,
+      "loss": 0.6629,
+      "step": 16487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10987761007601e-06,
+      "loss": 0.5942,
+      "step": 16488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1091103801621971e-06,
+      "loss": 0.7183,
+      "step": 16489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1083433999536719e-06,
+      "loss": 0.6552,
+      "step": 16490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1075766694719737e-06,
+      "loss": 0.7064,
+      "step": 16491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1068101887386395e-06,
+      "loss": 0.6586,
+      "step": 16492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1060439577751925e-06,
+      "loss": 0.5921,
+      "step": 16493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1052779766031552e-06,
+      "loss": 0.2832,
+      "step": 16494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1045122452440383e-06,
+      "loss": 0.6519,
+      "step": 16495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.103746763719351e-06,
+      "loss": 0.7011,
+      "step": 16496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1029815320505888e-06,
+      "loss": 0.6292,
+      "step": 16497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1022165502592464e-06,
+      "loss": 0.6311,
+      "step": 16498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1014518183668077e-06,
+      "loss": 0.6393,
+      "step": 16499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10068733639475e-06,
+      "loss": 0.6677,
+      "step": 16500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0999231043645454e-06,
+      "loss": 0.675,
+      "step": 16501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0991591222976573e-06,
+      "loss": 0.6877,
+      "step": 16502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0983953902155419e-06,
+      "loss": 0.6979,
+      "step": 16503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.097631908139649e-06,
+      "loss": 0.5774,
+      "step": 16504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0968686760914248e-06,
+      "loss": 0.635,
+      "step": 16505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0961056940922987e-06,
+      "loss": 0.7223,
+      "step": 16506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0953429621637024e-06,
+      "loss": 0.6351,
+      "step": 16507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.09458048032706e-06,
+      "loss": 0.7039,
+      "step": 16508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.093818248603783e-06,
+      "loss": 0.6596,
+      "step": 16509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0930562670152811e-06,
+      "loss": 0.5391,
+      "step": 16510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0922945355829574e-06,
+      "loss": 0.6467,
+      "step": 16511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0915330543281987e-06,
+      "loss": 0.5906,
+      "step": 16512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0907718232723962e-06,
+      "loss": 0.6985,
+      "step": 16513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0900108424369283e-06,
+      "loss": 0.6715,
+      "step": 16514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0892501118431676e-06,
+      "loss": 0.5918,
+      "step": 16515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0884896315124793e-06,
+      "loss": 0.6215,
+      "step": 16516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0877294014662233e-06,
+      "loss": 0.7051,
+      "step": 16517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0869694217257498e-06,
+      "loss": 0.639,
+      "step": 16518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0862096923124032e-06,
+      "loss": 0.6674,
+      "step": 16519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.085450213247522e-06,
+      "loss": 0.6723,
+      "step": 16520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0846909845524356e-06,
+      "loss": 0.6814,
+      "step": 16521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0839320062484671e-06,
+      "loss": 0.6606,
+      "step": 16522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0831732783569337e-06,
+      "loss": 0.6821,
+      "step": 16523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0824148008991463e-06,
+      "loss": 0.5896,
+      "step": 16524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0816565738964023e-06,
+      "loss": 0.6811,
+      "step": 16525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0808985973700003e-06,
+      "loss": 0.5914,
+      "step": 16526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0801408713412287e-06,
+      "loss": 0.7119,
+      "step": 16527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.079383395831365e-06,
+      "loss": 0.6078,
+      "step": 16528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0786261708616875e-06,
+      "loss": 0.7078,
+      "step": 16529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0778691964534626e-06,
+      "loss": 0.6157,
+      "step": 16530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0771124726279514e-06,
+      "loss": 0.6203,
+      "step": 16531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.076355999406402e-06,
+      "loss": 0.6814,
+      "step": 16532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0755997768100646e-06,
+      "loss": 0.6558,
+      "step": 16533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0748438048601774e-06,
+      "loss": 0.6736,
+      "step": 16534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0740880835779698e-06,
+      "loss": 0.6571,
+      "step": 16535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.07333261298467e-06,
+      "loss": 0.6381,
+      "step": 16536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0725773931014948e-06,
+      "loss": 0.6514,
+      "step": 16537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0718224239496533e-06,
+      "loss": 0.6577,
+      "step": 16538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0710677055503515e-06,
+      "loss": 0.6858,
+      "step": 16539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0703132379247838e-06,
+      "loss": 0.6507,
+      "step": 16540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.069559021094142e-06,
+      "loss": 0.6527,
+      "step": 16541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.068805055079607e-06,
+      "loss": 0.6058,
+      "step": 16542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0680513399023562e-06,
+      "loss": 0.6487,
+      "step": 16543
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0672978755835572e-06,
+      "loss": 0.7239,
+      "step": 16544
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.066544662144371e-06,
+      "loss": 0.625,
+      "step": 16545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0657916996059503e-06,
+      "loss": 0.6683,
+      "step": 16546
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.065038987989445e-06,
+      "loss": 0.6157,
+      "step": 16547
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0642865273159931e-06,
+      "loss": 0.5796,
+      "step": 16548
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0635343176067313e-06,
+      "loss": 0.6354,
+      "step": 16549
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0627823588827858e-06,
+      "loss": 0.6421,
+      "step": 16550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0620306511652723e-06,
+      "loss": 0.6397,
+      "step": 16551
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0612791944753032e-06,
+      "loss": 0.6547,
+      "step": 16552
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0605279888339858e-06,
+      "loss": 0.6733,
+      "step": 16553
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0597770342624169e-06,
+      "loss": 0.7221,
+      "step": 16554
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0590263307816883e-06,
+      "loss": 0.6034,
+      "step": 16555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0582758784128822e-06,
+      "loss": 0.6939,
+      "step": 16556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.057525677177077e-06,
+      "loss": 0.5948,
+      "step": 16557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0567757270953428e-06,
+      "loss": 0.6432,
+      "step": 16558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0560260281887402e-06,
+      "loss": 0.6169,
+      "step": 16559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0552765804783272e-06,
+      "loss": 0.6373,
+      "step": 16560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0545273839851522e-06,
+      "loss": 0.7245,
+      "step": 16561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0537784387302552e-06,
+      "loss": 0.6479,
+      "step": 16562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0530297447346728e-06,
+      "loss": 0.6202,
+      "step": 16563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.05228130201943e-06,
+      "loss": 0.6845,
+      "step": 16564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0515331106055492e-06,
+      "loss": 0.6718,
+      "step": 16565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0507851705140416e-06,
+      "loss": 0.653,
+      "step": 16566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0500374817659164e-06,
+      "loss": 0.5709,
+      "step": 16567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0492900443821685e-06,
+      "loss": 0.6158,
+      "step": 16568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0485428583837943e-06,
+      "loss": 0.6555,
+      "step": 16569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.04779592379178e-06,
+      "loss": 0.6279,
+      "step": 16570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0470492406270994e-06,
+      "loss": 0.618,
+      "step": 16571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0463028089107252e-06,
+      "loss": 0.6966,
+      "step": 16572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0455566286636209e-06,
+      "loss": 0.6736,
+      "step": 16573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0448106999067453e-06,
+      "loss": 0.6866,
+      "step": 16574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0440650226610449e-06,
+      "loss": 0.6861,
+      "step": 16575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0433195969474675e-06,
+      "loss": 0.596,
+      "step": 16576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0425744227869406e-06,
+      "loss": 0.6869,
+      "step": 16577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0418295002004008e-06,
+      "loss": 0.6595,
+      "step": 16578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.041084829208766e-06,
+      "loss": 0.7308,
+      "step": 16579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0403404098329506e-06,
+      "loss": 0.6419,
+      "step": 16580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0395962420938632e-06,
+      "loss": 0.6631,
+      "step": 16581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.038852326012404e-06,
+      "loss": 0.6256,
+      "step": 16582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0381086616094671e-06,
+      "loss": 0.6857,
+      "step": 16583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.037365248905936e-06,
+      "loss": 0.7071,
+      "step": 16584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0366220879226918e-06,
+      "loss": 0.6312,
+      "step": 16585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0358791786806055e-06,
+      "loss": 0.6941,
+      "step": 16586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0351365212005427e-06,
+      "loss": 0.6887,
+      "step": 16587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0343941155033598e-06,
+      "loss": 0.6264,
+      "step": 16588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0336519616099127e-06,
+      "loss": 0.6558,
+      "step": 16589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.032910059541039e-06,
+      "loss": 0.6808,
+      "step": 16590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0321684093175788e-06,
+      "loss": 0.6592,
+      "step": 16591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0314270109603608e-06,
+      "loss": 0.6397,
+      "step": 16592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0306858644902086e-06,
+      "loss": 0.6347,
+      "step": 16593
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0299449699279352e-06,
+      "loss": 0.5763,
+      "step": 16594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.029204327294352e-06,
+      "loss": 0.2793,
+      "step": 16595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.02846393661026e-06,
+      "loss": 0.6077,
+      "step": 16596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0277237978964494e-06,
+      "loss": 0.6754,
+      "step": 16597
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0269839111737113e-06,
+      "loss": 0.7314,
+      "step": 16598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0262442764628255e-06,
+      "loss": 0.6643,
+      "step": 16599
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0255048937845647e-06,
+      "loss": 0.6941,
+      "step": 16600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0247657631596942e-06,
+      "loss": 0.7128,
+      "step": 16601
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0240268846089751e-06,
+      "loss": 0.6431,
+      "step": 16602
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0232882581531556e-06,
+      "loss": 0.6336,
+      "step": 16603
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0225498838129822e-06,
+      "loss": 0.6003,
+      "step": 16604
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0218117616091928e-06,
+      "loss": 0.6066,
+      "step": 16605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0210738915625174e-06,
+      "loss": 0.5988,
+      "step": 16606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0203362736936795e-06,
+      "loss": 0.6838,
+      "step": 16607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.019598908023396e-06,
+      "loss": 0.6768,
+      "step": 16608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0188617945723755e-06,
+      "loss": 0.6156,
+      "step": 16609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0181249333613207e-06,
+      "loss": 0.7074,
+      "step": 16610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0173883244109272e-06,
+      "loss": 0.6342,
+      "step": 16611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0166519677418817e-06,
+      "loss": 0.5872,
+      "step": 16612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0159158633748666e-06,
+      "loss": 0.6893,
+      "step": 16613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0151800113305542e-06,
+      "loss": 0.6472,
+      "step": 16614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0144444116296149e-06,
+      "loss": 0.7059,
+      "step": 16615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0137090642927039e-06,
+      "loss": 0.5957,
+      "step": 16616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0129739693404738e-06,
+      "loss": 0.6353,
+      "step": 16617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0122391267935738e-06,
+      "loss": 0.6264,
+      "step": 16618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0115045366726406e-06,
+      "loss": 0.647,
+      "step": 16619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0107701989983044e-06,
+      "loss": 0.7083,
+      "step": 16620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0100361137911918e-06,
+      "loss": 0.5858,
+      "step": 16621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.00930228107192e-06,
+      "loss": 0.655,
+      "step": 16622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0085687008610955e-06,
+      "loss": 0.6362,
+      "step": 16623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0078353731793245e-06,
+      "loss": 0.6433,
+      "step": 16624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0071022980472013e-06,
+      "loss": 0.626,
+      "step": 16625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0063694754853149e-06,
+      "loss": 0.5657,
+      "step": 16626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0056369055142478e-06,
+      "loss": 0.6927,
+      "step": 16627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0049045881545749e-06,
+      "loss": 0.678,
+      "step": 16628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.004172523426863e-06,
+      "loss": 0.6458,
+      "step": 16629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.003440711351672e-06,
+      "loss": 0.6695,
+      "step": 16630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0027091519495557e-06,
+      "loss": 0.6627,
+      "step": 16631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.00197784524106e-06,
+      "loss": 0.6084,
+      "step": 16632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0012467912467249e-06,
+      "loss": 0.7283,
+      "step": 16633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0005159899870831e-06,
+      "loss": 0.6745,
+      "step": 16634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.99785441482659e-07,
+      "loss": 0.6635,
+      "step": 16635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.990551457539687e-07,
+      "loss": 0.752,
+      "step": 16636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.983251028215235e-07,
+      "loss": 0.7009,
+      "step": 16637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.975953127058257e-07,
+      "loss": 0.6515,
+      "step": 16638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.96865775427377e-07,
+      "loss": 0.6317,
+      "step": 16639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.961364910066617e-07,
+      "loss": 0.6336,
+      "step": 16640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.95407459464165e-07,
+      "loss": 0.6531,
+      "step": 16641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.946786808203634e-07,
+      "loss": 0.6357,
+      "step": 16642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.93950155095721e-07,
+      "loss": 0.6891,
+      "step": 16643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.932218823106998e-07,
+      "loss": 0.6415,
+      "step": 16644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.92493862485755e-07,
+      "loss": 0.6815,
+      "step": 16645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.917660956413332e-07,
+      "loss": 0.7183,
+      "step": 16646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.910385817978741e-07,
+      "loss": 0.6618,
+      "step": 16647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.6746,
+      "step": 16648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.89584313195565e-07,
+      "loss": 0.6357,
+      "step": 16649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.888575584775605e-07,
+      "loss": 0.6429,
+      "step": 16650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.881310568422065e-07,
+      "loss": 0.6629,
+      "step": 16651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.874048083099064e-07,
+      "loss": 0.6486,
+      "step": 16652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.866788129010574e-07,
+      "loss": 0.7324,
+      "step": 16653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85953070636051e-07,
+      "loss": 0.6052,
+      "step": 16654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.852275815352697e-07,
+      "loss": 0.6395,
+      "step": 16655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.845023456190873e-07,
+      "loss": 0.6949,
+      "step": 16656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.837773629078728e-07,
+      "loss": 0.6835,
+      "step": 16657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.830526334219882e-07,
+      "loss": 0.6315,
+      "step": 16658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.823281571817888e-07,
+      "loss": 0.6245,
+      "step": 16659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.81603934207621e-07,
+      "loss": 0.6694,
+      "step": 16660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.808799645198285e-07,
+      "loss": 0.6219,
+      "step": 16661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.801562481387383e-07,
+      "loss": 0.6771,
+      "step": 16662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.794327850846797e-07,
+      "loss": 0.666,
+      "step": 16663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.787095753779708e-07,
+      "loss": 0.6385,
+      "step": 16664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.779866190389242e-07,
+      "loss": 0.6024,
+      "step": 16665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.772639160878438e-07,
+      "loss": 0.6577,
+      "step": 16666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.765414665450267e-07,
+      "loss": 0.6355,
+      "step": 16667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.758192704307633e-07,
+      "loss": 0.6591,
+      "step": 16668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.750973277653387e-07,
+      "loss": 0.6646,
+      "step": 16669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.743756385690273e-07,
+      "loss": 0.6612,
+      "step": 16670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.736542028620976e-07,
+      "loss": 0.605,
+      "step": 16671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.729330206648135e-07,
+      "loss": 0.6354,
+      "step": 16672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.722120919974276e-07,
+      "loss": 0.6555,
+      "step": 16673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.714914168801915e-07,
+      "loss": 0.6437,
+      "step": 16674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.707709953333399e-07,
+      "loss": 0.2807,
+      "step": 16675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.700508273771093e-07,
+      "loss": 0.6654,
+      "step": 16676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.693309130317262e-07,
+      "loss": 0.6485,
+      "step": 16677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.686112523174085e-07,
+      "loss": 0.7136,
+      "step": 16678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.678918452543695e-07,
+      "loss": 0.6389,
+      "step": 16679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.671726918628144e-07,
+      "loss": 0.6646,
+      "step": 16680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.664537921629425e-07,
+      "loss": 0.7178,
+      "step": 16681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.657351461749408e-07,
+      "loss": 0.6804,
+      "step": 16682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.650167539189937e-07,
+      "loss": 0.6834,
+      "step": 16683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.642986154152788e-07,
+      "loss": 0.6253,
+      "step": 16684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.63580730683965e-07,
+      "loss": 0.6445,
+      "step": 16685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.628630997452137e-07,
+      "loss": 0.6408,
+      "step": 16686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.62145722619182e-07,
+      "loss": 0.6841,
+      "step": 16687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.614285993260153e-07,
+      "loss": 0.6479,
+      "step": 16688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.60711729885856e-07,
+      "loss": 0.6556,
+      "step": 16689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59995114318837e-07,
+      "loss": 0.6854,
+      "step": 16690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.592787526450853e-07,
+      "loss": 0.6685,
+      "step": 16691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.585626448847207e-07,
+      "loss": 0.6721,
+      "step": 16692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.578467910578538e-07,
+      "loss": 0.6972,
+      "step": 16693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.571311911845938e-07,
+      "loss": 0.6301,
+      "step": 16694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.564158452850326e-07,
+      "loss": 0.6183,
+      "step": 16695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55700753379265e-07,
+      "loss": 0.6801,
+      "step": 16696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.54985915487374e-07,
+      "loss": 0.6898,
+      "step": 16697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.542713316294338e-07,
+      "loss": 0.6654,
+      "step": 16698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.53557001825518e-07,
+      "loss": 0.6415,
+      "step": 16699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.528429260956895e-07,
+      "loss": 0.637,
+      "step": 16700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.52129104459999e-07,
+      "loss": 0.6678,
+      "step": 16701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.514155369384958e-07,
+      "loss": 0.694,
+      "step": 16702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.507022235512231e-07,
+      "loss": 0.6407,
+      "step": 16703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.499891643182124e-07,
+      "loss": 0.668,
+      "step": 16704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.492763592594912e-07,
+      "loss": 0.6779,
+      "step": 16705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.485638083950787e-07,
+      "loss": 0.6237,
+      "step": 16706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.478515117449883e-07,
+      "loss": 0.674,
+      "step": 16707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471394693292235e-07,
+      "loss": 0.6447,
+      "step": 16708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.464276811677842e-07,
+      "loss": 0.6578,
+      "step": 16709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.457161472806597e-07,
+      "loss": 0.5989,
+      "step": 16710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.450048676878354e-07,
+      "loss": 0.7053,
+      "step": 16711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.442938424092862e-07,
+      "loss": 0.7122,
+      "step": 16712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.435830714649851e-07,
+      "loss": 0.6755,
+      "step": 16713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.428725548748907e-07,
+      "loss": 0.6535,
+      "step": 16714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.421622926589579e-07,
+      "loss": 0.6906,
+      "step": 16715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.414522848371377e-07,
+      "loss": 0.5943,
+      "step": 16716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.407425314293694e-07,
+      "loss": 0.6506,
+      "step": 16717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.400330324555873e-07,
+      "loss": 0.7195,
+      "step": 16718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.393237879357153e-07,
+      "loss": 0.6745,
+      "step": 16719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.386147978896787e-07,
+      "loss": 0.6389,
+      "step": 16720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.37906062337386e-07,
+      "loss": 0.636,
+      "step": 16721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.371975812987422e-07,
+      "loss": 0.6565,
+      "step": 16722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.364893547936471e-07,
+      "loss": 0.6632,
+      "step": 16723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.357813828419893e-07,
+      "loss": 0.6698,
+      "step": 16724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.35073665463655e-07,
+      "loss": 0.6205,
+      "step": 16725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.343662026785217e-07,
+      "loss": 0.7205,
+      "step": 16726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.336589945064523e-07,
+      "loss": 0.7366,
+      "step": 16727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.329520409673165e-07,
+      "loss": 0.6296,
+      "step": 16728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.322453420809663e-07,
+      "loss": 0.6957,
+      "step": 16729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.315388978672502e-07,
+      "loss": 0.5955,
+      "step": 16730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.308327083460078e-07,
+      "loss": 0.6293,
+      "step": 16731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.301267735370756e-07,
+      "loss": 0.6664,
+      "step": 16732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.294210934602798e-07,
+      "loss": 0.5625,
+      "step": 16733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.28715668135437e-07,
+      "loss": 0.6608,
+      "step": 16734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.2801049758236e-07,
+      "loss": 0.6843,
+      "step": 16735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.273055818208554e-07,
+      "loss": 0.6738,
+      "step": 16736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.266009208707205e-07,
+      "loss": 0.7053,
+      "step": 16737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.258965147517462e-07,
+      "loss": 0.7007,
+      "step": 16738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.251923634837156e-07,
+      "loss": 0.6387,
+      "step": 16739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.244884670864062e-07,
+      "loss": 0.6683,
+      "step": 16740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.237848255795867e-07,
+      "loss": 0.6868,
+      "step": 16741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.230814389830189e-07,
+      "loss": 0.708,
+      "step": 16742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.223783073164583e-07,
+      "loss": 0.6451,
+      "step": 16743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.216754305996522e-07,
+      "loss": 0.7124,
+      "step": 16744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.209728088523417e-07,
+      "loss": 0.6709,
+      "step": 16745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.202704420942621e-07,
+      "loss": 0.6441,
+      "step": 16746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.195683303451341e-07,
+      "loss": 0.7054,
+      "step": 16747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.188664736246832e-07,
+      "loss": 0.6295,
+      "step": 16748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.181648719526181e-07,
+      "loss": 0.6763,
+      "step": 16749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.174635253486441e-07,
+      "loss": 0.6243,
+      "step": 16750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.167624338324599e-07,
+      "loss": 0.6685,
+      "step": 16751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.160615974237563e-07,
+      "loss": 0.5992,
+      "step": 16752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.153610161422144e-07,
+      "loss": 0.675,
+      "step": 16753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.146606900075117e-07,
+      "loss": 0.7234,
+      "step": 16754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.139606190393169e-07,
+      "loss": 0.6222,
+      "step": 16755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.132608032572921e-07,
+      "loss": 0.6122,
+      "step": 16756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.125612426810914e-07,
+      "loss": 0.6005,
+      "step": 16757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.118619373303639e-07,
+      "loss": 0.6404,
+      "step": 16758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.111628872247491e-07,
+      "loss": 0.6959,
+      "step": 16759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.104640923838792e-07,
+      "loss": 0.7453,
+      "step": 16760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.097655528273819e-07,
+      "loss": 0.5876,
+      "step": 16761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.090672685748747e-07,
+      "loss": 0.6474,
+      "step": 16762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.083692396459698e-07,
+      "loss": 0.657,
+      "step": 16763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.076714660602726e-07,
+      "loss": 0.6687,
+      "step": 16764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.069739478373807e-07,
+      "loss": 0.6882,
+      "step": 16765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.062766849968807e-07,
+      "loss": 0.6156,
+      "step": 16766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.055796775583592e-07,
+      "loss": 0.5895,
+      "step": 16767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.048829255413882e-07,
+      "loss": 0.652,
+      "step": 16768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.041864289655411e-07,
+      "loss": 0.6645,
+      "step": 16769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.034901878503766e-07,
+      "loss": 0.6578,
+      "step": 16770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.027942022154512e-07,
+      "loss": 0.64,
+      "step": 16771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.020984720803106e-07,
+      "loss": 0.7219,
+      "step": 16772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.014029974644933e-07,
+      "loss": 0.6177,
+      "step": 16773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.007077783875339e-07,
+      "loss": 0.67,
+      "step": 16774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.000128148689579e-07,
+      "loss": 0.6574,
+      "step": 16775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.993181069282819e-07,
+      "loss": 0.5493,
+      "step": 16776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.986236545850191e-07,
+      "loss": 0.6623,
+      "step": 16777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.979294578586739e-07,
+      "loss": 0.6711,
+      "step": 16778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.972355167687408e-07,
+      "loss": 0.6632,
+      "step": 16779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.965418313347119e-07,
+      "loss": 0.6332,
+      "step": 16780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.958484015760683e-07,
+      "loss": 0.6183,
+      "step": 16781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.951552275122866e-07,
+      "loss": 0.681,
+      "step": 16782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.944623091628335e-07,
+      "loss": 0.6744,
+      "step": 16783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.937696465471712e-07,
+      "loss": 0.6683,
+      "step": 16784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.930772396847542e-07,
+      "loss": 0.6417,
+      "step": 16785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.923850885950259e-07,
+      "loss": 0.6991,
+      "step": 16786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.916931932974282e-07,
+      "loss": 0.6591,
+      "step": 16787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.910015538113903e-07,
+      "loss": 0.6394,
+      "step": 16788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.903101701563422e-07,
+      "loss": 0.6807,
+      "step": 16789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.896190423516993e-07,
+      "loss": 0.7329,
+      "step": 16790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.889281704168739e-07,
+      "loss": 0.6166,
+      "step": 16791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.882375543712663e-07,
+      "loss": 0.6308,
+      "step": 16792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.875471942342739e-07,
+      "loss": 0.6669,
+      "step": 16793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.868570900252871e-07,
+      "loss": 0.5757,
+      "step": 16794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.861672417636869e-07,
+      "loss": 0.6205,
+      "step": 16795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.854776494688489e-07,
+      "loss": 0.64,
+      "step": 16796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.847883131601387e-07,
+      "loss": 0.6204,
+      "step": 16797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.840992328569187e-07,
+      "loss": 0.6266,
+      "step": 16798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.834104085785411e-07,
+      "loss": 0.6328,
+      "step": 16799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.827218403443527e-07,
+      "loss": 0.7014,
+      "step": 16800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.820335281736914e-07,
+      "loss": 0.6541,
+      "step": 16801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.813454720858894e-07,
+      "loss": 0.664,
+      "step": 16802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.806576721002702e-07,
+      "loss": 0.6408,
+      "step": 16803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.79970128236154e-07,
+      "loss": 0.6968,
+      "step": 16804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.792828405128462e-07,
+      "loss": 0.665,
+      "step": 16805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.785958089496527e-07,
+      "loss": 0.5917,
+      "step": 16806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.77909033565868e-07,
+      "loss": 0.6631,
+      "step": 16807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.7722251438078e-07,
+      "loss": 0.6002,
+      "step": 16808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.765362514136721e-07,
+      "loss": 0.6323,
+      "step": 16809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.758502446838169e-07,
+      "loss": 0.6492,
+      "step": 16810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.751644942104831e-07,
+      "loss": 0.6587,
+      "step": 16811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.744790000129277e-07,
+      "loss": 0.6788,
+      "step": 16812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.737937621104042e-07,
+      "loss": 0.6689,
+      "step": 16813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.731087805221572e-07,
+      "loss": 0.6221,
+      "step": 16814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.724240552674257e-07,
+      "loss": 0.6652,
+      "step": 16815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.71739586365441e-07,
+      "loss": 0.6779,
+      "step": 16816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.710553738354254e-07,
+      "loss": 0.6567,
+      "step": 16817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.70371417696596e-07,
+      "loss": 0.6369,
+      "step": 16818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.696877179681628e-07,
+      "loss": 0.6487,
+      "step": 16819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.690042746693272e-07,
+      "loss": 0.6692,
+      "step": 16820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.683210878192838e-07,
+      "loss": 0.6115,
+      "step": 16821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.676381574372206e-07,
+      "loss": 0.5586,
+      "step": 16822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.669554835423177e-07,
+      "loss": 0.6943,
+      "step": 16823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.662730661537499e-07,
+      "loss": 0.6768,
+      "step": 16824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.655909052906808e-07,
+      "loss": 0.6133,
+      "step": 16825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.649090009722705e-07,
+      "loss": 0.6435,
+      "step": 16826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.642273532176715e-07,
+      "loss": 0.5779,
+      "step": 16827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.635459620460262e-07,
+      "loss": 0.6526,
+      "step": 16828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.628648274764706e-07,
+      "loss": 0.6535,
+      "step": 16829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.621839495281414e-07,
+      "loss": 0.6641,
+      "step": 16830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.615033282201556e-07,
+      "loss": 0.599,
+      "step": 16831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.608229635716303e-07,
+      "loss": 0.6594,
+      "step": 16832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.601428556016733e-07,
+      "loss": 0.7177,
+      "step": 16833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.594630043293862e-07,
+      "loss": 0.6621,
+      "step": 16834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.587834097738634e-07,
+      "loss": 0.7014,
+      "step": 16835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.581040719541922e-07,
+      "loss": 0.6956,
+      "step": 16836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.574249908894505e-07,
+      "loss": 0.6374,
+      "step": 16837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.567461665987109e-07,
+      "loss": 0.5873,
+      "step": 16838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.560675991010403e-07,
+      "loss": 0.6199,
+      "step": 16839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.553892884154946e-07,
+      "loss": 0.662,
+      "step": 16840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.547112345611263e-07,
+      "loss": 0.7321,
+      "step": 16841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.540334375569781e-07,
+      "loss": 0.6443,
+      "step": 16842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.533558974220868e-07,
+      "loss": 0.6635,
+      "step": 16843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.526786141754794e-07,
+      "loss": 0.6472,
+      "step": 16844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.520015878361798e-07,
+      "loss": 0.6689,
+      "step": 16845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.513248184232015e-07,
+      "loss": 0.682,
+      "step": 16846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.506483059555526e-07,
+      "loss": 0.6482,
+      "step": 16847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.499720504522335e-07,
+      "loss": 0.6792,
+      "step": 16848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.492960519322347e-07,
+      "loss": 0.6973,
+      "step": 16849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.486203104145475e-07,
+      "loss": 0.5716,
+      "step": 16850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.479448259181455e-07,
+      "loss": 0.6643,
+      "step": 16851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.472695984620016e-07,
+      "loss": 0.6541,
+      "step": 16852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.465946280650794e-07,
+      "loss": 0.6163,
+      "step": 16853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.459199147463371e-07,
+      "loss": 0.5691,
+      "step": 16854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.452454585247227e-07,
+      "loss": 0.6489,
+      "step": 16855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.445712594191824e-07,
+      "loss": 0.6884,
+      "step": 16856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.438973174486442e-07,
+      "loss": 0.6586,
+      "step": 16857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.432236326320431e-07,
+      "loss": 0.293,
+      "step": 16858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.425502049882972e-07,
+      "loss": 0.6476,
+      "step": 16859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.418770345363192e-07,
+      "loss": 0.6918,
+      "step": 16860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.412041212950172e-07,
+      "loss": 0.6535,
+      "step": 16861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.405314652832885e-07,
+      "loss": 0.7439,
+      "step": 16862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.398590665200291e-07,
+      "loss": 0.6203,
+      "step": 16863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.39186925024118e-07,
+      "loss": 0.6852,
+      "step": 16864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.38515040814436e-07,
+      "loss": 0.6331,
+      "step": 16865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.378434139098524e-07,
+      "loss": 0.7457,
+      "step": 16866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.371720443292297e-07,
+      "loss": 0.6823,
+      "step": 16867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.365009320914241e-07,
+      "loss": 0.7661,
+      "step": 16868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.35830077215285e-07,
+      "loss": 0.632,
+      "step": 16869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.351594797196527e-07,
+      "loss": 0.6399,
+      "step": 16870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.344891396233624e-07,
+      "loss": 0.6396,
+      "step": 16871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.338190569452398e-07,
+      "loss": 0.6478,
+      "step": 16872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.331492317041046e-07,
+      "loss": 0.705,
+      "step": 16873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.324796639187693e-07,
+      "loss": 0.6339,
+      "step": 16874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.318103536080402e-07,
+      "loss": 0.678,
+      "step": 16875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.311413007907154e-07,
+      "loss": 0.6725,
+      "step": 16876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.304725054855834e-07,
+      "loss": 0.6121,
+      "step": 16877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.298039677114256e-07,
+      "loss": 0.6901,
+      "step": 16878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.291356874870249e-07,
+      "loss": 0.5813,
+      "step": 16879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.284676648311452e-07,
+      "loss": 0.6324,
+      "step": 16880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.277998997625514e-07,
+      "loss": 0.6768,
+      "step": 16881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.271323922999974e-07,
+      "loss": 0.5834,
+      "step": 16882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.264651424622272e-07,
+      "loss": 0.6481,
+      "step": 16883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.257981502679846e-07,
+      "loss": 0.6462,
+      "step": 16884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.251314157360002e-07,
+      "loss": 0.6077,
+      "step": 16885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.244649388850001e-07,
+      "loss": 0.6245,
+      "step": 16886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.237987197337038e-07,
+      "loss": 0.6652,
+      "step": 16887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.23132758300822e-07,
+      "loss": 0.6306,
+      "step": 16888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.224670546050572e-07,
+      "loss": 0.6136,
+      "step": 16889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.218016086651082e-07,
+      "loss": 0.6783,
+      "step": 16890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.211364204996619e-07,
+      "loss": 0.6143,
+      "step": 16891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.204714901274025e-07,
+      "loss": 0.7163,
+      "step": 16892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198068175670038e-07,
+      "loss": 0.5786,
+      "step": 16893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.191424028371342e-07,
+      "loss": 0.6362,
+      "step": 16894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.184782459564556e-07,
+      "loss": 0.6729,
+      "step": 16895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.178143469436184e-07,
+      "loss": 0.6912,
+      "step": 16896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.17150705817269e-07,
+      "loss": 0.6573,
+      "step": 16897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.164873225960446e-07,
+      "loss": 0.636,
+      "step": 16898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.158241972985803e-07,
+      "loss": 0.6092,
+      "step": 16899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.15161329943499e-07,
+      "loss": 0.688,
+      "step": 16900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.144987205494182e-07,
+      "loss": 0.6007,
+      "step": 16901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.138363691349471e-07,
+      "loss": 0.6735,
+      "step": 16902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.131742757186878e-07,
+      "loss": 0.6548,
+      "step": 16903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.125124403192353e-07,
+      "loss": 0.6255,
+      "step": 16904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.118508629551769e-07,
+      "loss": 0.6621,
+      "step": 16905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.111895436450956e-07,
+      "loss": 0.6515,
+      "step": 16906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.105284824075632e-07,
+      "loss": 0.6089,
+      "step": 16907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.098676792611471e-07,
+      "loss": 0.6049,
+      "step": 16908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.092071342244046e-07,
+      "loss": 0.6458,
+      "step": 16909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.085468473158897e-07,
+      "loss": 0.7112,
+      "step": 16910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.078868185541455e-07,
+      "loss": 0.6791,
+      "step": 16911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.072270479577094e-07,
+      "loss": 0.6643,
+      "step": 16912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.065675355451108e-07,
+      "loss": 0.6971,
+      "step": 16913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.05908281334874e-07,
+      "loss": 0.5899,
+      "step": 16914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.052492853455152e-07,
+      "loss": 0.6742,
+      "step": 16915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.045905475955385e-07,
+      "loss": 0.6562,
+      "step": 16916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.03932068103449e-07,
+      "loss": 0.6263,
+      "step": 16917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.032738468877366e-07,
+      "loss": 0.6141,
+      "step": 16918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.026158839668907e-07,
+      "loss": 0.7144,
+      "step": 16919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.019581793593911e-07,
+      "loss": 0.7075,
+      "step": 16920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.013007330837109e-07,
+      "loss": 0.6453,
+      "step": 16921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.006435451583094e-07,
+      "loss": 0.6847,
+      "step": 16922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.999866156016489e-07,
+      "loss": 0.6442,
+      "step": 16923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.993299444321768e-07,
+      "loss": 0.6433,
+      "step": 16924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.986735316683369e-07,
+      "loss": 0.644,
+      "step": 16925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.980173773285659e-07,
+      "loss": 0.6711,
+      "step": 16926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.973614814312913e-07,
+      "loss": 0.6602,
+      "step": 16927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.967058439949349e-07,
+      "loss": 0.7012,
+      "step": 16928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.960504650379097e-07,
+      "loss": 0.6843,
+      "step": 16929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.953953445786233e-07,
+      "loss": 0.673,
+      "step": 16930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.947404826354743e-07,
+      "loss": 0.6888,
+      "step": 16931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.940858792268547e-07,
+      "loss": 0.6425,
+      "step": 16932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.934315343711507e-07,
+      "loss": 0.6429,
+      "step": 16933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.927774480867411e-07,
+      "loss": 0.6444,
+      "step": 16934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.921236203919913e-07,
+      "loss": 0.6964,
+      "step": 16935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.914700513052686e-07,
+      "loss": 0.6033,
+      "step": 16936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.908167408449274e-07,
+      "loss": 0.6313,
+      "step": 16937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.901636890293141e-07,
+      "loss": 0.7088,
+      "step": 16938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.89510895876775e-07,
+      "loss": 0.6888,
+      "step": 16939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.888583614056411e-07,
+      "loss": 0.6124,
+      "step": 16940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.882060856342411e-07,
+      "loss": 0.636,
+      "step": 16941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.875540685808913e-07,
+      "loss": 0.658,
+      "step": 16942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.86902310263905e-07,
+      "loss": 0.6912,
+      "step": 16943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.862508107015887e-07,
+      "loss": 0.6581,
+      "step": 16944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.855995699122398e-07,
+      "loss": 0.6308,
+      "step": 16945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.849485879141471e-07,
+      "loss": 0.6205,
+      "step": 16946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.84297864725595e-07,
+      "loss": 0.6014,
+      "step": 16947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.836474003648587e-07,
+      "loss": 0.6474,
+      "step": 16948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82997194850208e-07,
+      "loss": 0.6062,
+      "step": 16949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.823472481999039e-07,
+      "loss": 0.6643,
+      "step": 16950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.816975604321997e-07,
+      "loss": 0.6964,
+      "step": 16951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.810481315653428e-07,
+      "loss": 0.6137,
+      "step": 16952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.803989616175723e-07,
+      "loss": 0.6988,
+      "step": 16953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.797500506071231e-07,
+      "loss": 0.6658,
+      "step": 16954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.791013985522166e-07,
+      "loss": 0.7383,
+      "step": 16955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.784530054710715e-07,
+      "loss": 0.5795,
+      "step": 16956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.778048713818975e-07,
+      "loss": 0.6641,
+      "step": 16957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.771569963029002e-07,
+      "loss": 0.5912,
+      "step": 16958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.765093802522727e-07,
+      "loss": 0.6523,
+      "step": 16959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.758620232482083e-07,
+      "loss": 0.5745,
+      "step": 16960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.752149253088826e-07,
+      "loss": 0.6635,
+      "step": 16961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.74568086452474e-07,
+      "loss": 0.6161,
+      "step": 16962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.739215066971461e-07,
+      "loss": 0.5907,
+      "step": 16963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.73275186061061e-07,
+      "loss": 0.6527,
+      "step": 16964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.726291245623696e-07,
+      "loss": 0.6802,
+      "step": 16965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.719833222192163e-07,
+      "loss": 0.6419,
+      "step": 16966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.713377790497411e-07,
+      "loss": 0.7247,
+      "step": 16967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.706924950720718e-07,
+      "loss": 0.6859,
+      "step": 16968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.700474703043337e-07,
+      "loss": 0.7135,
+      "step": 16969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.694027047646413e-07,
+      "loss": 0.6253,
+      "step": 16970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.687581984711035e-07,
+      "loss": 0.6414,
+      "step": 16971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.681139514418212e-07,
+      "loss": 0.6228,
+      "step": 16972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.674699636948913e-07,
+      "loss": 0.5915,
+      "step": 16973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.668262352483957e-07,
+      "loss": 0.6484,
+      "step": 16974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.661827661204169e-07,
+      "loss": 0.6478,
+      "step": 16975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.655395563290269e-07,
+      "loss": 0.6838,
+      "step": 16976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.648966058922891e-07,
+      "loss": 0.7151,
+      "step": 16977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.642539148282624e-07,
+      "loss": 0.6198,
+      "step": 16978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.636114831549956e-07,
+      "loss": 0.6299,
+      "step": 16979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.629693108905356e-07,
+      "loss": 0.3256,
+      "step": 16980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.623273980529144e-07,
+      "loss": 0.6545,
+      "step": 16981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.616857446601622e-07,
+      "loss": 0.653,
+      "step": 16982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.610443507302978e-07,
+      "loss": 0.6684,
+      "step": 16983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.604032162813379e-07,
+      "loss": 0.6605,
+      "step": 16984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.597623413312882e-07,
+      "loss": 0.6543,
+      "step": 16985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.591217258981476e-07,
+      "loss": 0.7028,
+      "step": 16986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.584813699999094e-07,
+      "loss": 0.6589,
+      "step": 16987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.578412736545548e-07,
+      "loss": 0.6806,
+      "step": 16988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.572014368800651e-07,
+      "loss": 0.6166,
+      "step": 16989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.565618596944091e-07,
+      "loss": 0.6223,
+      "step": 16990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.559225421155491e-07,
+      "loss": 0.6954,
+      "step": 16991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.552834841614421e-07,
+      "loss": 0.674,
+      "step": 16992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.546446858500367e-07,
+      "loss": 0.5721,
+      "step": 16993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.54006147199271e-07,
+      "loss": 0.5939,
+      "step": 16994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.533678682270796e-07,
+      "loss": 0.636,
+      "step": 16995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.527298489513901e-07,
+      "loss": 0.5922,
+      "step": 16996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.520920893901218e-07,
+      "loss": 0.664,
+      "step": 16997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.514545895611857e-07,
+      "loss": 0.3005,
+      "step": 16998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.508173494824844e-07,
+      "loss": 0.699,
+      "step": 16999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5018036917192e-07,
+      "loss": 0.644,
+      "step": 17000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.495436486473795e-07,
+      "loss": 0.7067,
+      "step": 17001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.48907187926744e-07,
+      "loss": 0.6853,
+      "step": 17002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.482709870278915e-07,
+      "loss": 0.6497,
+      "step": 17003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.476350459686887e-07,
+      "loss": 0.6653,
+      "step": 17004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46999364766997e-07,
+      "loss": 0.6432,
+      "step": 17005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46363943440671e-07,
+      "loss": 0.6296,
+      "step": 17006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.457287820075531e-07,
+      "loss": 0.6191,
+      "step": 17007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.450938804854835e-07,
+      "loss": 0.6632,
+      "step": 17008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.444592388922955e-07,
+      "loss": 0.649,
+      "step": 17009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.438248572458129e-07,
+      "loss": 0.6467,
+      "step": 17010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.431907355638524e-07,
+      "loss": 0.7068,
+      "step": 17011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.425568738642242e-07,
+      "loss": 0.6502,
+      "step": 17012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.419232721647296e-07,
+      "loss": 0.6526,
+      "step": 17013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.412899304831634e-07,
+      "loss": 0.5913,
+      "step": 17014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.406568488373145e-07,
+      "loss": 0.642,
+      "step": 17015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.40024027244961e-07,
+      "loss": 0.7307,
+      "step": 17016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.393914657238787e-07,
+      "loss": 0.6326,
+      "step": 17017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.387591642918324e-07,
+      "loss": 0.5669,
+      "step": 17018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.38127122966581e-07,
+      "loss": 0.6344,
+      "step": 17019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.374953417658748e-07,
+      "loss": 0.6594,
+      "step": 17020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.368638207074586e-07,
+      "loss": 0.6335,
+      "step": 17021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.362325598090681e-07,
+      "loss": 0.6473,
+      "step": 17022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.356015590884336e-07,
+      "loss": 0.6158,
+      "step": 17023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.349708185632765e-07,
+      "loss": 0.6673,
+      "step": 17024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.343403382513115e-07,
+      "loss": 0.6216,
+      "step": 17025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.337101181702477e-07,
+      "loss": 0.6592,
+      "step": 17026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.330801583377823e-07,
+      "loss": 0.6329,
+      "step": 17027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.324504587716075e-07,
+      "loss": 0.6633,
+      "step": 17028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.318210194894115e-07,
+      "loss": 0.66,
+      "step": 17029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.311918405088724e-07,
+      "loss": 0.6418,
+      "step": 17030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.305629218476595e-07,
+      "loss": 0.7318,
+      "step": 17031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.299342635234386e-07,
+      "loss": 0.6794,
+      "step": 17032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.293058655538631e-07,
+      "loss": 0.5594,
+      "step": 17033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286777279565827e-07,
+      "loss": 0.6403,
+      "step": 17034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.280498507492395e-07,
+      "loss": 0.6608,
+      "step": 17035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.274222339494663e-07,
+      "loss": 0.6984,
+      "step": 17036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.267948775748923e-07,
+      "loss": 0.6576,
+      "step": 17037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.261677816431367e-07,
+      "loss": 0.6608,
+      "step": 17038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.255409461718099e-07,
+      "loss": 0.5845,
+      "step": 17039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.249143711785189e-07,
+      "loss": 0.6133,
+      "step": 17040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.242880566808607e-07,
+      "loss": 0.6474,
+      "step": 17041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.236620026964247e-07,
+      "loss": 0.6839,
+      "step": 17042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.230362092427968e-07,
+      "loss": 0.6595,
+      "step": 17043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.224106763375493e-07,
+      "loss": 0.6353,
+      "step": 17044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.217854039982542e-07,
+      "loss": 0.6588,
+      "step": 17045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.211603922424693e-07,
+      "loss": 0.6229,
+      "step": 17046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.205356410877496e-07,
+      "loss": 0.6526,
+      "step": 17047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.199111505516388e-07,
+      "loss": 0.6071,
+      "step": 17048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.192869206516817e-07,
+      "loss": 0.6926,
+      "step": 17049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.186629514054077e-07,
+      "loss": 0.6446,
+      "step": 17050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.180392428303395e-07,
+      "loss": 0.6524,
+      "step": 17051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.174157949439986e-07,
+      "loss": 0.6394,
+      "step": 17052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.167926077638898e-07,
+      "loss": 0.6882,
+      "step": 17053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.161696813075181e-07,
+      "loss": 0.6577,
+      "step": 17054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.155470155923772e-07,
+      "loss": 0.6065,
+      "step": 17055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.149246106359576e-07,
+      "loss": 0.6519,
+      "step": 17056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.143024664557374e-07,
+      "loss": 0.6532,
+      "step": 17057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.136805830691917e-07,
+      "loss": 0.6741,
+      "step": 17058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.130589604937855e-07,
+      "loss": 0.6627,
+      "step": 17059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.124375987469767e-07,
+      "loss": 0.608,
+      "step": 17060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.118164978462183e-07,
+      "loss": 0.6076,
+      "step": 17061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.11195657808953e-07,
+      "loss": 0.6239,
+      "step": 17062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.105750786526177e-07,
+      "loss": 0.6701,
+      "step": 17063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.099547603946422e-07,
+      "loss": 0.6523,
+      "step": 17064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.093347030524489e-07,
+      "loss": 0.6441,
+      "step": 17065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.087149066434496e-07,
+      "loss": 0.6995,
+      "step": 17066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080953711850547e-07,
+      "loss": 0.6946,
+      "step": 17067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.074760966946625e-07,
+      "loss": 0.6855,
+      "step": 17068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.068570831896637e-07,
+      "loss": 0.6023,
+      "step": 17069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.062383306874487e-07,
+      "loss": 0.6422,
+      "step": 17070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.056198392053948e-07,
+      "loss": 0.6428,
+      "step": 17071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.05001608760868e-07,
+      "loss": 0.6799,
+      "step": 17072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.043836393712355e-07,
+      "loss": 0.7133,
+      "step": 17073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.037659310538514e-07,
+      "loss": 0.628,
+      "step": 17074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.03148483826065e-07,
+      "loss": 0.7385,
+      "step": 17075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.025312977052179e-07,
+      "loss": 0.6823,
+      "step": 17076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.019143727086441e-07,
+      "loss": 0.689,
+      "step": 17077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.012977088536699e-07,
+      "loss": 0.602,
+      "step": 17078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.006813061576145e-07,
+      "loss": 0.6156,
+      "step": 17079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.00065164637791e-07,
+      "loss": 0.6283,
+      "step": 17080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.99449284311503e-07,
+      "loss": 0.6132,
+      "step": 17081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.988336651960481e-07,
+      "loss": 0.5979,
+      "step": 17082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.982183073087156e-07,
+      "loss": 0.6751,
+      "step": 17083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.976032106667919e-07,
+      "loss": 0.7152,
+      "step": 17084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.969883752875461e-07,
+      "loss": 0.6633,
+      "step": 17085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.963738011882504e-07,
+      "loss": 0.623,
+      "step": 17086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.95759488386164e-07,
+      "loss": 0.665,
+      "step": 17087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.9514543689854e-07,
+      "loss": 0.6537,
+      "step": 17088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.945316467426233e-07,
+      "loss": 0.6897,
+      "step": 17089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.939181179356568e-07,
+      "loss": 0.648,
+      "step": 17090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.933048504948692e-07,
+      "loss": 0.6736,
+      "step": 17091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.926918444374842e-07,
+      "loss": 0.6657,
+      "step": 17092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.920790997807169e-07,
+      "loss": 0.6115,
+      "step": 17093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.914666165417794e-07,
+      "loss": 0.6159,
+      "step": 17094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.908543947378721e-07,
+      "loss": 0.6073,
+      "step": 17095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.902424343861902e-07,
+      "loss": 0.6564,
+      "step": 17096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8963073550392e-07,
+      "loss": 0.6042,
+      "step": 17097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.890192981082422e-07,
+      "loss": 0.2929,
+      "step": 17098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.884081222163296e-07,
+      "loss": 0.6826,
+      "step": 17099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.877972078453454e-07,
+      "loss": 0.6113,
+      "step": 17100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.871865550124501e-07,
+      "loss": 0.7171,
+      "step": 17101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.865761637347934e-07,
+      "loss": 0.5975,
+      "step": 17102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.859660340295171e-07,
+      "loss": 0.6847,
+      "step": 17103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.853561659137609e-07,
+      "loss": 0.6505,
+      "step": 17104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.847465594046476e-07,
+      "loss": 0.6616,
+      "step": 17105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.841372145193014e-07,
+      "loss": 0.7109,
+      "step": 17106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.835281312748365e-07,
+      "loss": 0.6382,
+      "step": 17107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.82919309688358e-07,
+      "loss": 0.6477,
+      "step": 17108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.823107497769632e-07,
+      "loss": 0.6477,
+      "step": 17109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.817024515577508e-07,
+      "loss": 0.577,
+      "step": 17110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.810944150477983e-07,
+      "loss": 0.6973,
+      "step": 17111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.804866402641853e-07,
+      "loss": 0.6408,
+      "step": 17112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.798791272239802e-07,
+      "loss": 0.6471,
+      "step": 17113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.792718759442474e-07,
+      "loss": 0.6133,
+      "step": 17114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.786648864420398e-07,
+      "loss": 0.6409,
+      "step": 17115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.780581587344071e-07,
+      "loss": 0.6805,
+      "step": 17116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.774516928383889e-07,
+      "loss": 0.6275,
+      "step": 17117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.768454887710152e-07,
+      "loss": 0.6596,
+      "step": 17118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.762395465493154e-07,
+      "loss": 0.6425,
+      "step": 17119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.756338661903061e-07,
+      "loss": 0.6882,
+      "step": 17120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.750284477109992e-07,
+      "loss": 0.6352,
+      "step": 17121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.744232911283966e-07,
+      "loss": 0.6738,
+      "step": 17122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.738183964594969e-07,
+      "loss": 0.7289,
+      "step": 17123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.732137637212854e-07,
+      "loss": 0.5913,
+      "step": 17124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.726093929307465e-07,
+      "loss": 0.6052,
+      "step": 17125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.720052841048519e-07,
+      "loss": 0.6782,
+      "step": 17126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.714014372605693e-07,
+      "loss": 0.6346,
+      "step": 17127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.707978524148584e-07,
+      "loss": 0.6754,
+      "step": 17128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.701945295846701e-07,
+      "loss": 0.6704,
+      "step": 17129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.695914687869531e-07,
+      "loss": 0.6543,
+      "step": 17130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.689886700386395e-07,
+      "loss": 0.691,
+      "step": 17131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.68386133356661e-07,
+      "loss": 0.648,
+      "step": 17132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.67783858757941e-07,
+      "loss": 0.6157,
+      "step": 17133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.671818462593937e-07,
+      "loss": 0.6143,
+      "step": 17134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.665800958779278e-07,
+      "loss": 0.6199,
+      "step": 17135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.659786076304453e-07,
+      "loss": 0.5929,
+      "step": 17136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.653773815338349e-07,
+      "loss": 0.6434,
+      "step": 17137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.647764176049843e-07,
+      "loss": 0.6482,
+      "step": 17138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.641757158607731e-07,
+      "loss": 0.6252,
+      "step": 17139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.635752763180714e-07,
+      "loss": 0.6679,
+      "step": 17140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.629750989937433e-07,
+      "loss": 0.6216,
+      "step": 17141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.623751839046455e-07,
+      "loss": 0.6295,
+      "step": 17142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.617755310676266e-07,
+      "loss": 0.6586,
+      "step": 17143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.611761404995276e-07,
+      "loss": 0.6179,
+      "step": 17144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.605770122171817e-07,
+      "loss": 0.6864,
+      "step": 17145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.599781462374178e-07,
+      "loss": 0.6629,
+      "step": 17146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.593795425770544e-07,
+      "loss": 0.6893,
+      "step": 17147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.587812012529038e-07,
+      "loss": 0.6628,
+      "step": 17148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.581831222817714e-07,
+      "loss": 0.7231,
+      "step": 17149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.575853056804526e-07,
+      "loss": 0.6068,
+      "step": 17150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.569877514657396e-07,
+      "loss": 0.6635,
+      "step": 17151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.563904596544135e-07,
+      "loss": 0.6607,
+      "step": 17152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.557934302632496e-07,
+      "loss": 0.5816,
+      "step": 17153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.551966633090167e-07,
+      "loss": 0.6711,
+      "step": 17154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.54600158808476e-07,
+      "loss": 0.6421,
+      "step": 17155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.540039167783796e-07,
+      "loss": 0.6295,
+      "step": 17156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.534079372354729e-07,
+      "loss": 0.6663,
+      "step": 17157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.528122201964915e-07,
+      "loss": 0.728,
+      "step": 17158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.52216765678172e-07,
+      "loss": 0.6559,
+      "step": 17159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.516215736972365e-07,
+      "loss": 0.6668,
+      "step": 17160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.510266442703994e-07,
+      "loss": 0.6711,
+      "step": 17161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.504319774143719e-07,
+      "loss": 0.7205,
+      "step": 17162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.498375731458529e-07,
+      "loss": 0.6481,
+      "step": 17163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.492434314815377e-07,
+      "loss": 0.6777,
+      "step": 17164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.486495524381131e-07,
+      "loss": 0.6316,
+      "step": 17165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.480559360322592e-07,
+      "loss": 0.6239,
+      "step": 17166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.474625822806468e-07,
+      "loss": 0.6615,
+      "step": 17167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.468694911999418e-07,
+      "loss": 0.6989,
+      "step": 17168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.462766628068007e-07,
+      "loss": 0.6199,
+      "step": 17169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.456840971178735e-07,
+      "loss": 0.6563,
+      "step": 17170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.450917941498025e-07,
+      "loss": 0.6668,
+      "step": 17171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.444997539192244e-07,
+      "loss": 0.6375,
+      "step": 17172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.439079764427647e-07,
+      "loss": 0.6846,
+      "step": 17173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.433164617370446e-07,
+      "loss": 0.6823,
+      "step": 17174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.427252098186787e-07,
+      "loss": 0.6478,
+      "step": 17175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.421342207042702e-07,
+      "loss": 0.6377,
+      "step": 17176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.415434944104182e-07,
+      "loss": 0.6342,
+      "step": 17177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.409530309537137e-07,
+      "loss": 0.6116,
+      "step": 17178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.403628303507392e-07,
+      "loss": 0.671,
+      "step": 17179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.397728926180725e-07,
+      "loss": 0.641,
+      "step": 17180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.391832177722812e-07,
+      "loss": 0.6865,
+      "step": 17181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.38593805829929e-07,
+      "loss": 0.667,
+      "step": 17182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.380046568075671e-07,
+      "loss": 0.6151,
+      "step": 17183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.374157707217421e-07,
+      "loss": 0.6563,
+      "step": 17184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.368271475889931e-07,
+      "loss": 0.6342,
+      "step": 17185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.362387874258536e-07,
+      "loss": 0.6389,
+      "step": 17186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.356506902488468e-07,
+      "loss": 0.7214,
+      "step": 17187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.350628560744909e-07,
+      "loss": 0.686,
+      "step": 17188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.344752849192937e-07,
+      "loss": 0.7083,
+      "step": 17189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.338879767997585e-07,
+      "loss": 0.608,
+      "step": 17190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.333009317323802e-07,
+      "loss": 0.6408,
+      "step": 17191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.327141497336453e-07,
+      "loss": 0.6208,
+      "step": 17192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.321276308200353e-07,
+      "loss": 0.6353,
+      "step": 17193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.315413750080224e-07,
+      "loss": 0.6088,
+      "step": 17194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.309553823140724e-07,
+      "loss": 0.6842,
+      "step": 17195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.303696527546421e-07,
+      "loss": 0.6307,
+      "step": 17196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.297841863461817e-07,
+      "loss": 0.7073,
+      "step": 17197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.291989831051359e-07,
+      "loss": 0.5918,
+      "step": 17198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.28614043047937e-07,
+      "loss": 0.6201,
+      "step": 17199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.280293661910186e-07,
+      "loss": 0.614,
+      "step": 17200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.274449525508009e-07,
+      "loss": 0.6671,
+      "step": 17201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.268608021436928e-07,
+      "loss": 0.7064,
+      "step": 17202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262769149861036e-07,
+      "loss": 0.6716,
+      "step": 17203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.256932910944324e-07,
+      "loss": 0.6782,
+      "step": 17204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.251099304850694e-07,
+      "loss": 0.6737,
+      "step": 17205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.245268331743993e-07,
+      "loss": 0.6835,
+      "step": 17206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.239439991787988e-07,
+      "loss": 0.665,
+      "step": 17207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.233614285146361e-07,
+      "loss": 0.587,
+      "step": 17208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.227791211982748e-07,
+      "loss": 0.7382,
+      "step": 17209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.221970772460672e-07,
+      "loss": 0.6686,
+      "step": 17210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.216152966743616e-07,
+      "loss": 0.6899,
+      "step": 17211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.210337794994981e-07,
+      "loss": 0.6467,
+      "step": 17212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204525257378069e-07,
+      "loss": 0.6843,
+      "step": 17213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.198715354056162e-07,
+      "loss": 0.641,
+      "step": 17214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.192908085192406e-07,
+      "loss": 0.6091,
+      "step": 17215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.187103450949894e-07,
+      "loss": 0.7036,
+      "step": 17216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.181301451491672e-07,
+      "loss": 0.6649,
+      "step": 17217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.175502086980689e-07,
+      "loss": 0.6463,
+      "step": 17218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.169705357579813e-07,
+      "loss": 0.6217,
+      "step": 17219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.163911263451861e-07,
+      "loss": 0.6436,
+      "step": 17220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.158119804759577e-07,
+      "loss": 0.6505,
+      "step": 17221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.152330981665578e-07,
+      "loss": 0.6685,
+      "step": 17222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.146544794332465e-07,
+      "loss": 0.6826,
+      "step": 17223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.140761242922755e-07,
+      "loss": 0.7515,
+      "step": 17224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.134980327598872e-07,
+      "loss": 0.6457,
+      "step": 17225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.129202048523175e-07,
+      "loss": 0.6696,
+      "step": 17226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.123426405857968e-07,
+      "loss": 0.6342,
+      "step": 17227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.117653399765433e-07,
+      "loss": 0.6677,
+      "step": 17228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.111883030407717e-07,
+      "loss": 0.6305,
+      "step": 17229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.106115297946901e-07,
+      "loss": 0.6424,
+      "step": 17230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.100350202544969e-07,
+      "loss": 0.6389,
+      "step": 17231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.094587744363823e-07,
+      "loss": 0.6365,
+      "step": 17232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.088827923565321e-07,
+      "loss": 0.7249,
+      "step": 17233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.083070740311237e-07,
+      "loss": 0.6308,
+      "step": 17234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.077316194763228e-07,
+      "loss": 0.6466,
+      "step": 17235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.071564287082942e-07,
+      "loss": 0.6537,
+      "step": 17236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.06581501743192e-07,
+      "loss": 0.3066,
+      "step": 17237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.060068385971619e-07,
+      "loss": 0.6495,
+      "step": 17238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.054324392863442e-07,
+      "loss": 0.5779,
+      "step": 17239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.048583038268751e-07,
+      "loss": 0.5839,
+      "step": 17240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.042844322348751e-07,
+      "loss": 0.6403,
+      "step": 17241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.037108245264622e-07,
+      "loss": 0.6232,
+      "step": 17242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.03137480717747e-07,
+      "loss": 0.6708,
+      "step": 17243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.025644008248333e-07,
+      "loss": 0.5928,
+      "step": 17244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.019915848638136e-07,
+      "loss": 0.6675,
+      "step": 17245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.014190328507785e-07,
+      "loss": 0.6624,
+      "step": 17246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.008467448018096e-07,
+      "loss": 0.6514,
+      "step": 17247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.00274720732974e-07,
+      "loss": 0.6194,
+      "step": 17248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.997029606603422e-07,
+      "loss": 0.682,
+      "step": 17249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.991314645999713e-07,
+      "loss": 0.5922,
+      "step": 17250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.985602325679118e-07,
+      "loss": 0.635,
+      "step": 17251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.979892645802077e-07,
+      "loss": 0.6192,
+      "step": 17252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.974185606528959e-07,
+      "loss": 0.7224,
+      "step": 17253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.968481208020016e-07,
+      "loss": 0.6403,
+      "step": 17254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.962779450435474e-07,
+      "loss": 0.617,
+      "step": 17255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.957080333935483e-07,
+      "loss": 0.5942,
+      "step": 17256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.951383858680093e-07,
+      "loss": 0.6093,
+      "step": 17257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.945690024829287e-07,
+      "loss": 0.6871,
+      "step": 17258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.939998832542992e-07,
+      "loss": 0.6639,
+      "step": 17259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.934310281981048e-07,
+      "loss": 0.7007,
+      "step": 17260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.928624373303215e-07,
+      "loss": 0.6166,
+      "step": 17261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.922941106669189e-07,
+      "loss": 0.5997,
+      "step": 17262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.917260482238584e-07,
+      "loss": 0.6915,
+      "step": 17263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.911582500170943e-07,
+      "loss": 0.7287,
+      "step": 17264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.905907160625735e-07,
+      "loss": 0.6118,
+      "step": 17265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.900234463762367e-07,
+      "loss": 0.6282,
+      "step": 17266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.894564409740144e-07,
+      "loss": 0.5758,
+      "step": 17267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.888896998718296e-07,
+      "loss": 0.7066,
+      "step": 17268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.88323223085604e-07,
+      "loss": 0.6954,
+      "step": 17269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.877570106312436e-07,
+      "loss": 0.728,
+      "step": 17270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.871910625246535e-07,
+      "loss": 0.622,
+      "step": 17271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.866253787817267e-07,
+      "loss": 0.6582,
+      "step": 17272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.860599594183536e-07,
+      "loss": 0.7172,
+      "step": 17273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.854948044504094e-07,
+      "loss": 0.6792,
+      "step": 17274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.849299138937703e-07,
+      "loss": 0.6615,
+      "step": 17275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.843652877643014e-07,
+      "loss": 0.67,
+      "step": 17276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.838009260778587e-07,
+      "loss": 0.6316,
+      "step": 17277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.832368288502943e-07,
+      "loss": 0.7046,
+      "step": 17278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.826729960974498e-07,
+      "loss": 0.6569,
+      "step": 17279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.821094278351624e-07,
+      "loss": 0.6277,
+      "step": 17280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.815461240792597e-07,
+      "loss": 0.7043,
+      "step": 17281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.809830848455611e-07,
+      "loss": 0.7069,
+      "step": 17282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.804203101498806e-07,
+      "loss": 0.6324,
+      "step": 17283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.798578000080247e-07,
+      "loss": 0.627,
+      "step": 17284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.792955544357914e-07,
+      "loss": 0.7057,
+      "step": 17285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.78733573448973e-07,
+      "loss": 0.6852,
+      "step": 17286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.781718570633499e-07,
+      "loss": 0.6642,
+      "step": 17287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.776104052946985e-07,
+      "loss": 0.6906,
+      "step": 17288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.770492181587906e-07,
+      "loss": 0.6609,
+      "step": 17289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.764882956713857e-07,
+      "loss": 0.6146,
+      "step": 17290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.759276378482371e-07,
+      "loss": 0.6691,
+      "step": 17291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.753672447050939e-07,
+      "loss": 0.6242,
+      "step": 17292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.748071162576907e-07,
+      "loss": 0.6422,
+      "step": 17293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.742472525217613e-07,
+      "loss": 0.65,
+      "step": 17294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.736876535130298e-07,
+      "loss": 0.6412,
+      "step": 17295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.731283192472114e-07,
+      "loss": 0.6736,
+      "step": 17296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.725692497400181e-07,
+      "loss": 0.6971,
+      "step": 17297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.720104450071484e-07,
+      "loss": 0.6245,
+      "step": 17298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.714519050642997e-07,
+      "loss": 0.7106,
+      "step": 17299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.708936299271573e-07,
+      "loss": 0.6535,
+      "step": 17300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.703356196113996e-07,
+      "loss": 0.6785,
+      "step": 17301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.697778741327009e-07,
+      "loss": 0.6323,
+      "step": 17302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.692203935067242e-07,
+      "loss": 0.597,
+      "step": 17303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.68663177749128e-07,
+      "loss": 0.6237,
+      "step": 17304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.681062268755611e-07,
+      "loss": 0.6485,
+      "step": 17305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.675495409016652e-07,
+      "loss": 0.6155,
+      "step": 17306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.669931198430756e-07,
+      "loss": 0.6932,
+      "step": 17307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.664369637154199e-07,
+      "loss": 0.6873,
+      "step": 17308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.658810725343166e-07,
+      "loss": 0.6447,
+      "step": 17309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.653254463153812e-07,
+      "loss": 0.7112,
+      "step": 17310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.647700850742166e-07,
+      "loss": 0.6795,
+      "step": 17311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.642149888264214e-07,
+      "loss": 0.6036,
+      "step": 17312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.636601575875855e-07,
+      "loss": 0.7044,
+      "step": 17313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.631055913732897e-07,
+      "loss": 0.6277,
+      "step": 17314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.625512901991114e-07,
+      "loss": 0.6522,
+      "step": 17315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.619972540806184e-07,
+      "loss": 0.6657,
+      "step": 17316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.614434830333715e-07,
+      "loss": 0.7434,
+      "step": 17317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.608899770729214e-07,
+      "loss": 0.6831,
+      "step": 17318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.603367362148159e-07,
+      "loss": 0.2954,
+      "step": 17319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.597837604745915e-07,
+      "loss": 0.665,
+      "step": 17320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.5923104986778e-07,
+      "loss": 0.6108,
+      "step": 17321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.586786044099035e-07,
+      "loss": 0.6774,
+      "step": 17322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.581264241164785e-07,
+      "loss": 0.6057,
+      "step": 17323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.575745090030138e-07,
+      "loss": 0.6296,
+      "step": 17324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.570228590850102e-07,
+      "loss": 0.6733,
+      "step": 17325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.564714743779587e-07,
+      "loss": 0.6828,
+      "step": 17326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.55920354897348e-07,
+      "loss": 0.6668,
+      "step": 17327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.553695006586545e-07,
+      "loss": 0.6857,
+      "step": 17328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.548189116773494e-07,
+      "loss": 0.619,
+      "step": 17329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.54268587968898e-07,
+      "loss": 0.6594,
+      "step": 17330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.537185295487568e-07,
+      "loss": 0.6375,
+      "step": 17331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.531687364323724e-07,
+      "loss": 0.7014,
+      "step": 17332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.526192086351856e-07,
+      "loss": 0.6878,
+      "step": 17333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.52069946172632e-07,
+      "loss": 0.6451,
+      "step": 17334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.515209490601381e-07,
+      "loss": 0.6609,
+      "step": 17335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.509722173131215e-07,
+      "loss": 0.6357,
+      "step": 17336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.504237509469934e-07,
+      "loss": 0.6519,
+      "step": 17337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.498755499771591e-07,
+      "loss": 0.6268,
+      "step": 17338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.493276144190151e-07,
+      "loss": 0.6884,
+      "step": 17339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.487799442879482e-07,
+      "loss": 0.5535,
+      "step": 17340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.482325395993426e-07,
+      "loss": 0.6871,
+      "step": 17341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.476854003685716e-07,
+      "loss": 0.6398,
+      "step": 17342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.471385266110018e-07,
+      "loss": 0.6731,
+      "step": 17343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.465919183419921e-07,
+      "loss": 0.703,
+      "step": 17344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.460455755768968e-07,
+      "loss": 0.6883,
+      "step": 17345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.454994983310557e-07,
+      "loss": 0.6133,
+      "step": 17346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44953686619808e-07,
+      "loss": 0.6098,
+      "step": 17347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.444081404584834e-07,
+      "loss": 0.606,
+      "step": 17348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.438628598624019e-07,
+      "loss": 0.6127,
+      "step": 17349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.433178448468812e-07,
+      "loss": 0.7252,
+      "step": 17350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.427730954272281e-07,
+      "loss": 0.6228,
+      "step": 17351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.422286116187391e-07,
+      "loss": 0.7059,
+      "step": 17352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.416843934367089e-07,
+      "loss": 0.6042,
+      "step": 17353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.411404408964205e-07,
+      "loss": 0.6693,
+      "step": 17354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.405967540131518e-07,
+      "loss": 0.6039,
+      "step": 17355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.40053332802174e-07,
+      "loss": 0.6368,
+      "step": 17356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.395101772787459e-07,
+      "loss": 0.5531,
+      "step": 17357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.389672874581264e-07,
+      "loss": 0.6092,
+      "step": 17358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.38424663355559e-07,
+      "loss": 0.5935,
+      "step": 17359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.378823049862869e-07,
+      "loss": 0.621,
+      "step": 17360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.373402123655402e-07,
+      "loss": 0.7296,
+      "step": 17361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.367983855085446e-07,
+      "loss": 0.5986,
+      "step": 17362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.362568244305188e-07,
+      "loss": 0.6183,
+      "step": 17363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.35715529146672e-07,
+      "loss": 0.6138,
+      "step": 17364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.351744996722052e-07,
+      "loss": 0.6683,
+      "step": 17365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.346337360223153e-07,
+      "loss": 0.6568,
+      "step": 17366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.340932382121889e-07,
+      "loss": 0.6564,
+      "step": 17367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.335530062570071e-07,
+      "loss": 0.7195,
+      "step": 17368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.330130401719413e-07,
+      "loss": 0.6019,
+      "step": 17369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.324733399721571e-07,
+      "loss": 0.6585,
+      "step": 17370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.319339056728146e-07,
+      "loss": 0.6479,
+      "step": 17371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.313947372890615e-07,
+      "loss": 0.6763,
+      "step": 17372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.308558348360415e-07,
+      "loss": 0.646,
+      "step": 17373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.303171983288901e-07,
+      "loss": 0.5767,
+      "step": 17374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.297788277827343e-07,
+      "loss": 0.5779,
+      "step": 17375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.29240723212695e-07,
+      "loss": 0.6677,
+      "step": 17376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.287028846338871e-07,
+      "loss": 0.6763,
+      "step": 17377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.281653120614116e-07,
+      "loss": 0.601,
+      "step": 17378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.276280055103699e-07,
+      "loss": 0.6982,
+      "step": 17379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.270909649958522e-07,
+      "loss": 0.6249,
+      "step": 17380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.265541905329408e-07,
+      "loss": 0.6252,
+      "step": 17381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.260176821367125e-07,
+      "loss": 0.6547,
+      "step": 17382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.254814398222341e-07,
+      "loss": 0.7373,
+      "step": 17383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.249454636045681e-07,
+      "loss": 0.7366,
+      "step": 17384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.244097534987647e-07,
+      "loss": 0.6249,
+      "step": 17385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.238743095198717e-07,
+      "loss": 0.6402,
+      "step": 17386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.233391316829261e-07,
+      "loss": 0.6692,
+      "step": 17387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.228042200029593e-07,
+      "loss": 0.6713,
+      "step": 17388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.222695744949957e-07,
+      "loss": 0.6101,
+      "step": 17389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.217351951740491e-07,
+      "loss": 0.6677,
+      "step": 17390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.212010820551294e-07,
+      "loss": 0.6884,
+      "step": 17391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.20667235153236e-07,
+      "loss": 0.6575,
+      "step": 17392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.201336544833635e-07,
+      "loss": 0.5884,
+      "step": 17393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.196003400604977e-07,
+      "loss": 0.6048,
+      "step": 17394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.190672918996165e-07,
+      "loss": 0.6227,
+      "step": 17395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.185345100156913e-07,
+      "loss": 0.6575,
+      "step": 17396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.180019944236859e-07,
+      "loss": 0.6527,
+      "step": 17397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.174697451385535e-07,
+      "loss": 0.6469,
+      "step": 17398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.169377621752459e-07,
+      "loss": 0.6479,
+      "step": 17399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.16406045548703e-07,
+      "loss": 0.6367,
+      "step": 17400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.158745952738586e-07,
+      "loss": 0.6211,
+      "step": 17401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.153434113656386e-07,
+      "loss": 0.6555,
+      "step": 17402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.14812493838962e-07,
+      "loss": 0.3115,
+      "step": 17403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.142818427087392e-07,
+      "loss": 0.6381,
+      "step": 17404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.137514579898728e-07,
+      "loss": 0.593,
+      "step": 17405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.132213396972607e-07,
+      "loss": 0.6221,
+      "step": 17406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.126914878457911e-07,
+      "loss": 0.664,
+      "step": 17407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.121619024503444e-07,
+      "loss": 0.7237,
+      "step": 17408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.116325835257951e-07,
+      "loss": 0.694,
+      "step": 17409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.111035310870094e-07,
+      "loss": 0.6691,
+      "step": 17410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.105747451488452e-07,
+      "loss": 0.657,
+      "step": 17411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.100462257261552e-07,
+      "loss": 0.6807,
+      "step": 17412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.095179728337818e-07,
+      "loss": 0.6538,
+      "step": 17413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.089899864865622e-07,
+      "loss": 0.6329,
+      "step": 17414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.084622666993244e-07,
+      "loss": 0.3153,
+      "step": 17415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.079348134868911e-07,
+      "loss": 0.6106,
+      "step": 17416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.074076268640737e-07,
+      "loss": 0.6424,
+      "step": 17417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.068807068456805e-07,
+      "loss": 0.5599,
+      "step": 17418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.063540534465072e-07,
+      "loss": 0.6492,
+      "step": 17419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0582766668135e-07,
+      "loss": 0.6599,
+      "step": 17420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.053015465649902e-07,
+      "loss": 0.6757,
+      "step": 17421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.047756931122039e-07,
+      "loss": 0.6334,
+      "step": 17422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.042501063377614e-07,
+      "loss": 0.6812,
+      "step": 17423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.03724786256422e-07,
+      "loss": 0.721,
+      "step": 17424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.031997328829408e-07,
+      "loss": 0.6743,
+      "step": 17425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.026749462320635e-07,
+      "loss": 0.6829,
+      "step": 17426
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.021504263185306e-07,
+      "loss": 0.6168,
+      "step": 17427
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.016261731570715e-07,
+      "loss": 0.609,
+      "step": 17428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.01102186762411e-07,
+      "loss": 0.6285,
+      "step": 17429
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.005784671492664e-07,
+      "loss": 0.6516,
+      "step": 17430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.000550143323446e-07,
+      "loss": 0.6656,
+      "step": 17431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.995318283263495e-07,
+      "loss": 0.6609,
+      "step": 17432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.990089091459727e-07,
+      "loss": 0.6495,
+      "step": 17433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984862568059034e-07,
+      "loss": 0.5927,
+      "step": 17434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.979638713208179e-07,
+      "loss": 0.638,
+      "step": 17435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.974417527053898e-07,
+      "loss": 0.6708,
+      "step": 17436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.969199009742808e-07,
+      "loss": 0.6826,
+      "step": 17437
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.963983161421492e-07,
+      "loss": 0.6176,
+      "step": 17438
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958769982236422e-07,
+      "loss": 0.6458,
+      "step": 17439
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.953559472334024e-07,
+      "loss": 0.6523,
+      "step": 17440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.948351631860648e-07,
+      "loss": 0.6235,
+      "step": 17441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.943146460962567e-07,
+      "loss": 0.6149,
+      "step": 17442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.937943959785929e-07,
+      "loss": 0.5476,
+      "step": 17443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.932744128476885e-07,
+      "loss": 0.6771,
+      "step": 17444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.927546967181451e-07,
+      "loss": 0.6643,
+      "step": 17445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.922352476045611e-07,
+      "loss": 0.6392,
+      "step": 17446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.917160655215248e-07,
+      "loss": 0.6871,
+      "step": 17447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.911971504836177e-07,
+      "loss": 0.684,
+      "step": 17448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.906785025054139e-07,
+      "loss": 0.6017,
+      "step": 17449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.901601216014806e-07,
+      "loss": 0.611,
+      "step": 17450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.896420077863751e-07,
+      "loss": 0.6092,
+      "step": 17451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8912416107465e-07,
+      "loss": 0.6985,
+      "step": 17452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.886065814808482e-07,
+      "loss": 0.6523,
+      "step": 17453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.880892690195083e-07,
+      "loss": 0.6093,
+      "step": 17454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875722237051594e-07,
+      "loss": 0.5997,
+      "step": 17455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.870554455523202e-07,
+      "loss": 0.6061,
+      "step": 17456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.865389345755056e-07,
+      "loss": 0.6415,
+      "step": 17457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.86022690789223e-07,
+      "loss": 0.6799,
+      "step": 17458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.855067142079695e-07,
+      "loss": 0.6073,
+      "step": 17459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849910048462403e-07,
+      "loss": 0.6698,
+      "step": 17460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.844755627185161e-07,
+      "loss": 0.65,
+      "step": 17461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.839603878392763e-07,
+      "loss": 0.6436,
+      "step": 17462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.834454802229871e-07,
+      "loss": 0.6254,
+      "step": 17463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829308398841104e-07,
+      "loss": 0.6214,
+      "step": 17464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.824164668371e-07,
+      "loss": 0.6822,
+      "step": 17465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.819023610964036e-07,
+      "loss": 0.7047,
+      "step": 17466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.813885226764592e-07,
+      "loss": 0.6583,
+      "step": 17467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.808749515916988e-07,
+      "loss": 0.6143,
+      "step": 17468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.803616478565453e-07,
+      "loss": 0.6103,
+      "step": 17469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.798486114854173e-07,
+      "loss": 0.6057,
+      "step": 17470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.793358424927208e-07,
+      "loss": 0.6885,
+      "step": 17471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.788233408928588e-07,
+      "loss": 0.6573,
+      "step": 17472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.783111067002244e-07,
+      "loss": 0.5838,
+      "step": 17473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.777991399292059e-07,
+      "loss": 0.67,
+      "step": 17474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.772874405941807e-07,
+      "loss": 0.5741,
+      "step": 17475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7677600870951967e-07,
+      "loss": 0.5936,
+      "step": 17476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7626484428958786e-07,
+      "loss": 0.6663,
+      "step": 17477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.757539473487394e-07,
+      "loss": 0.6656,
+      "step": 17478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.752433179013238e-07,
+      "loss": 0.6278,
+      "step": 17479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7473295596168536e-07,
+      "loss": 0.6597,
+      "step": 17480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7422286154415575e-07,
+      "loss": 0.6207,
+      "step": 17481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7371303466306027e-07,
+      "loss": 0.6559,
+      "step": 17482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.732034753327186e-07,
+      "loss": 0.6418,
+      "step": 17483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.726941835674426e-07,
+      "loss": 0.6217,
+      "step": 17484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.721851593815341e-07,
+      "loss": 0.6388,
+      "step": 17485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7167640278929174e-07,
+      "loss": 0.671,
+      "step": 17486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.711679138050018e-07,
+      "loss": 0.6299,
+      "step": 17487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.706596924429474e-07,
+      "loss": 0.6122,
+      "step": 17488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.701517387174015e-07,
+      "loss": 0.6542,
+      "step": 17489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6964405264263045e-07,
+      "loss": 0.6107,
+      "step": 17490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.691366342328918e-07,
+      "loss": 0.6888,
+      "step": 17491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.686294835024385e-07,
+      "loss": 0.5928,
+      "step": 17492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.681226004655126e-07,
+      "loss": 0.6197,
+      "step": 17493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6761598513635266e-07,
+      "loss": 0.6552,
+      "step": 17494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6710963752918284e-07,
+      "loss": 0.6179,
+      "step": 17495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.666035576582273e-07,
+      "loss": 0.6511,
+      "step": 17496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6609774553769916e-07,
+      "loss": 0.7132,
+      "step": 17497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6559220118180373e-07,
+      "loss": 0.6099,
+      "step": 17498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.650869246047407e-07,
+      "loss": 0.6737,
+      "step": 17499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.645819158206977e-07,
+      "loss": 0.6324,
+      "step": 17500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.640771748438644e-07,
+      "loss": 0.6785,
+      "step": 17501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6357270168841175e-07,
+      "loss": 0.6101,
+      "step": 17502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.630684963685084e-07,
+      "loss": 0.6749,
+      "step": 17503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.625645588983174e-07,
+      "loss": 0.6287,
+      "step": 17504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6206088929198977e-07,
+      "loss": 0.653,
+      "step": 17505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.61557487563673e-07,
+      "loss": 0.7404,
+      "step": 17506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6105435372750694e-07,
+      "loss": 0.6793,
+      "step": 17507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.605514877976169e-07,
+      "loss": 0.6269,
+      "step": 17508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.600488897881306e-07,
+      "loss": 0.6411,
+      "step": 17509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.595465597131632e-07,
+      "loss": 0.7146,
+      "step": 17510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5904449758682243e-07,
+      "loss": 0.6067,
+      "step": 17511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.585427034232093e-07,
+      "loss": 0.595,
+      "step": 17512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.580411772364168e-07,
+      "loss": 0.6014,
+      "step": 17513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.575399190405305e-07,
+      "loss": 0.7252,
+      "step": 17514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.57038928849628e-07,
+      "loss": 0.6223,
+      "step": 17515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5653820667778014e-07,
+      "loss": 0.6436,
+      "step": 17516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5603775253905023e-07,
+      "loss": 0.693,
+      "step": 17517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5553756644749256e-07,
+      "loss": 0.6049,
+      "step": 17518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.55037648417157e-07,
+      "loss": 0.6543,
+      "step": 17519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5453799846208345e-07,
+      "loss": 0.6341,
+      "step": 17520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.54038616596304e-07,
+      "loss": 0.6808,
+      "step": 17521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.535395028338452e-07,
+      "loss": 0.6588,
+      "step": 17522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.530406571887247e-07,
+      "loss": 0.5771,
+      "step": 17523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5254207967495135e-07,
+      "loss": 0.643,
+      "step": 17524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5204377030652945e-07,
+      "loss": 0.6683,
+      "step": 17525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5154572909745454e-07,
+      "loss": 0.6422,
+      "step": 17526
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5104795606171424e-07,
+      "loss": 0.6365,
+      "step": 17527
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.505504512132886e-07,
+      "loss": 0.6588,
+      "step": 17528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5005321456614734e-07,
+      "loss": 0.6769,
+      "step": 17529
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4955624613425954e-07,
+      "loss": 0.5877,
+      "step": 17530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.490595459315816e-07,
+      "loss": 0.649,
+      "step": 17531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4856311397206364e-07,
+      "loss": 0.6907,
+      "step": 17532
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4806695026965e-07,
+      "loss": 0.6733,
+      "step": 17533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.475710548382717e-07,
+      "loss": 0.6096,
+      "step": 17534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.470754276918587e-07,
+      "loss": 0.7032,
+      "step": 17535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.465800688443311e-07,
+      "loss": 0.6516,
+      "step": 17536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4608497830959995e-07,
+      "loss": 0.7061,
+      "step": 17537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4559015610157184e-07,
+      "loss": 0.6455,
+      "step": 17538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4509560223414237e-07,
+      "loss": 0.6463,
+      "step": 17539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4460131672120266e-07,
+      "loss": 0.6568,
+      "step": 17540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.441072995766349e-07,
+      "loss": 0.7111,
+      "step": 17541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4361355081431355e-07,
+      "loss": 0.6135,
+      "step": 17542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4312007044810537e-07,
+      "loss": 0.6327,
+      "step": 17543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4262685849187026e-07,
+      "loss": 0.6507,
+      "step": 17544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4213391495946054e-07,
+      "loss": 0.2944,
+      "step": 17545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.416412398647218e-07,
+      "loss": 0.6341,
+      "step": 17546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.411488332214886e-07,
+      "loss": 0.6315,
+      "step": 17547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4065669504359197e-07,
+      "loss": 0.6159,
+      "step": 17548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4016482534485207e-07,
+      "loss": 0.6538,
+      "step": 17549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3967322413908664e-07,
+      "loss": 0.6138,
+      "step": 17550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3918189144010135e-07,
+      "loss": 0.6155,
+      "step": 17551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.386908272616941e-07,
+      "loss": 0.5821,
+      "step": 17552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3820003161765934e-07,
+      "loss": 0.6685,
+      "step": 17553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.377095045217783e-07,
+      "loss": 0.5854,
+      "step": 17554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3721924598782884e-07,
+      "loss": 0.6651,
+      "step": 17555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3672925602958106e-07,
+      "loss": 0.6807,
+      "step": 17556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3623953466079507e-07,
+      "loss": 0.6252,
+      "step": 17557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3575008189522536e-07,
+      "loss": 0.6335,
+      "step": 17558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3526089774661996e-07,
+      "loss": 0.7119,
+      "step": 17559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3477198222871553e-07,
+      "loss": 0.6848,
+      "step": 17560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.342833353552456e-07,
+      "loss": 0.6259,
+      "step": 17561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.337949571399325e-07,
+      "loss": 0.6998,
+      "step": 17562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.333068475964941e-07,
+      "loss": 0.6653,
+      "step": 17563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.328190067386373e-07,
+      "loss": 0.6469,
+      "step": 17564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3233143458006555e-07,
+      "loss": 0.6419,
+      "step": 17565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3184413113447234e-07,
+      "loss": 0.6231,
+      "step": 17566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3135709641554115e-07,
+      "loss": 0.629,
+      "step": 17567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3087033043695325e-07,
+      "loss": 0.6649,
+      "step": 17568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.303838332123766e-07,
+      "loss": 0.6028,
+      "step": 17569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2989760475547927e-07,
+      "loss": 0.6869,
+      "step": 17570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.294116450799146e-07,
+      "loss": 0.6397,
+      "step": 17571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.289259541993329e-07,
+      "loss": 0.6572,
+      "step": 17572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.284405321273721e-07,
+      "loss": 0.6167,
+      "step": 17573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.279553788776669e-07,
+      "loss": 0.6411,
+      "step": 17574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.27470494463843e-07,
+      "loss": 0.6719,
+      "step": 17575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2698587889951846e-07,
+      "loss": 0.7074,
+      "step": 17576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.265015321983046e-07,
+      "loss": 0.6478,
+      "step": 17577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2601745437380493e-07,
+      "loss": 0.6655,
+      "step": 17578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.255336454396142e-07,
+      "loss": 0.6362,
+      "step": 17579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2505010540932033e-07,
+      "loss": 0.6433,
+      "step": 17580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2456683429650367e-07,
+      "loss": 0.6301,
+      "step": 17581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.240838321147378e-07,
+      "loss": 0.6326,
+      "step": 17582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.236010988775885e-07,
+      "loss": 0.6737,
+      "step": 17583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.231186345986116e-07,
+      "loss": 0.55,
+      "step": 17584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.226364392913607e-07,
+      "loss": 0.6972,
+      "step": 17585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.221545129693749e-07,
+      "loss": 0.701,
+      "step": 17586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.216728556461913e-07,
+      "loss": 0.6024,
+      "step": 17587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.211914673353368e-07,
+      "loss": 0.6414,
+      "step": 17588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.207103480503316e-07,
+      "loss": 0.6452,
+      "step": 17589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2022949780468837e-07,
+      "loss": 0.623,
+      "step": 17590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.197489166119129e-07,
+      "loss": 0.6598,
+      "step": 17591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.192686044855021e-07,
+      "loss": 0.6068,
+      "step": 17592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.187885614389442e-07,
+      "loss": 0.5763,
+      "step": 17593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1830878748572277e-07,
+      "loss": 0.6558,
+      "step": 17594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1782928263931264e-07,
+      "loss": 0.6556,
+      "step": 17595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1735004691318084e-07,
+      "loss": 0.6628,
+      "step": 17596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.168710803207865e-07,
+      "loss": 0.6549,
+      "step": 17597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1639238287558113e-07,
+      "loss": 0.667,
+      "step": 17598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.159139545910107e-07,
+      "loss": 0.6531,
+      "step": 17599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1543579548051217e-07,
+      "loss": 0.6348,
+      "step": 17600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1495790555751367e-07,
+      "loss": 0.6457,
+      "step": 17601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1448028483543676e-07,
+      "loss": 0.6382,
+      "step": 17602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.140029333276974e-07,
+      "loss": 0.7212,
+      "step": 17603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1352585104770136e-07,
+      "loss": 0.6429,
+      "step": 17604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.130490380088481e-07,
+      "loss": 0.6678,
+      "step": 17605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1257249422452794e-07,
+      "loss": 0.697,
+      "step": 17606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.120962197081246e-07,
+      "loss": 0.6728,
+      "step": 17607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1162021447301635e-07,
+      "loss": 0.6115,
+      "step": 17608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1114447853257134e-07,
+      "loss": 0.6693,
+      "step": 17609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1066901190015e-07,
+      "loss": 0.6118,
+      "step": 17610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1019381458910825e-07,
+      "loss": 0.6694,
+      "step": 17611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0971888661278993e-07,
+      "loss": 0.6258,
+      "step": 17612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0924422798453436e-07,
+      "loss": 0.6258,
+      "step": 17613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0876983871767304e-07,
+      "loss": 0.7328,
+      "step": 17614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0829571882552874e-07,
+      "loss": 0.6546,
+      "step": 17615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0782186832141857e-07,
+      "loss": 0.6141,
+      "step": 17616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.073482872186485e-07,
+      "loss": 0.6087,
+      "step": 17617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0687497553052234e-07,
+      "loss": 0.6746,
+      "step": 17618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0640193327033063e-07,
+      "loss": 0.6425,
+      "step": 17619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.059291604513615e-07,
+      "loss": 0.649,
+      "step": 17620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.054566570868912e-07,
+      "loss": 0.6932,
+      "step": 17621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0498442319019005e-07,
+      "loss": 0.6123,
+      "step": 17622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0451245877452193e-07,
+      "loss": 0.6641,
+      "step": 17623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0404076385314405e-07,
+      "loss": 0.5776,
+      "step": 17624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0356933843930023e-07,
+      "loss": 0.6047,
+      "step": 17625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0309818254623214e-07,
+      "loss": 0.6474,
+      "step": 17626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0262729618717357e-07,
+      "loss": 0.6999,
+      "step": 17627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0215667937534954e-07,
+      "loss": 0.693,
+      "step": 17628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.016863321239761e-07,
+      "loss": 0.6446,
+      "step": 17629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0121625444626276e-07,
+      "loss": 0.676,
+      "step": 17630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.007464463554156e-07,
+      "loss": 0.6714,
+      "step": 17631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0027690786462623e-07,
+      "loss": 0.6641,
+      "step": 17632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9980763898708194e-07,
+      "loss": 0.6185,
+      "step": 17633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9933863973596333e-07,
+      "loss": 0.5982,
+      "step": 17634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.988699101244431e-07,
+      "loss": 0.6688,
+      "step": 17635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.984014501656841e-07,
+      "loss": 0.5909,
+      "step": 17636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9793325987284583e-07,
+      "loss": 0.6728,
+      "step": 17637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.974653392590733e-07,
+      "loss": 0.6541,
+      "step": 17638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.969976883375126e-07,
+      "loss": 0.6641,
+      "step": 17639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.965303071212956e-07,
+      "loss": 0.6406,
+      "step": 17640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9606319562354946e-07,
+      "loss": 0.6483,
+      "step": 17641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.955963538573937e-07,
+      "loss": 0.6438,
+      "step": 17642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9512978183594006e-07,
+      "loss": 0.6368,
+      "step": 17643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9466347957229366e-07,
+      "loss": 0.5761,
+      "step": 17644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.941974470795462e-07,
+      "loss": 0.679,
+      "step": 17645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9373168437079057e-07,
+      "loss": 0.644,
+      "step": 17646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.932661914591074e-07,
+      "loss": 0.6585,
+      "step": 17647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9280096835756843e-07,
+      "loss": 0.6929,
+      "step": 17648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.92336015079241e-07,
+      "loss": 0.6493,
+      "step": 17649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.918713316371847e-07,
+      "loss": 0.6796,
+      "step": 17650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.91406918044448e-07,
+      "loss": 0.5357,
+      "step": 17651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909427743140759e-07,
+      "loss": 0.592,
+      "step": 17652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9047890045910363e-07,
+      "loss": 0.5921,
+      "step": 17653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9001529649255853e-07,
+      "loss": 0.7422,
+      "step": 17654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.895519624274635e-07,
+      "loss": 0.5939,
+      "step": 17655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.890888982768293e-07,
+      "loss": 0.6648,
+      "step": 17656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.886261040536632e-07,
+      "loss": 0.6286,
+      "step": 17657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.881635797709604e-07,
+      "loss": 0.6172,
+      "step": 17658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.877013254417117e-07,
+      "loss": 0.6849,
+      "step": 17659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.87239341078901e-07,
+      "loss": 0.6481,
+      "step": 17660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8677762669550367e-07,
+      "loss": 0.6373,
+      "step": 17661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8631618230448696e-07,
+      "loss": 0.6473,
+      "step": 17662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.858550079188106e-07,
+      "loss": 0.6756,
+      "step": 17663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.853941035514264e-07,
+      "loss": 0.5971,
+      "step": 17664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8493346921527843e-07,
+      "loss": 0.6375,
+      "step": 17665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8447310492330523e-07,
+      "loss": 0.6326,
+      "step": 17666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.840130106884354e-07,
+      "loss": 0.6577,
+      "step": 17667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8355318652359197e-07,
+      "loss": 0.6593,
+      "step": 17668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8309363244168895e-07,
+      "loss": 0.6133,
+      "step": 17669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8263434845563276e-07,
+      "loss": 0.6774,
+      "step": 17670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8217533457832303e-07,
+      "loss": 0.6821,
+      "step": 17671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8171659082265166e-07,
+      "loss": 0.6918,
+      "step": 17672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8125811720150174e-07,
+      "loss": 0.6639,
+      "step": 17673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.807999137277507e-07,
+      "loss": 0.5951,
+      "step": 17674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8034198041426715e-07,
+      "loss": 0.6699,
+      "step": 17675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7988431727391307e-07,
+      "loss": 0.5602,
+      "step": 17676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7942692431954034e-07,
+      "loss": 0.6501,
+      "step": 17677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7896980156399533e-07,
+      "loss": 0.6625,
+      "step": 17678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.785129490201178e-07,
+      "loss": 0.6434,
+      "step": 17679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7805636670073753e-07,
+      "loss": 0.7183,
+      "step": 17680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.776000546186798e-07,
+      "loss": 0.6646,
+      "step": 17681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.771440127867587e-07,
+      "loss": 0.6371,
+      "step": 17682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7668824121778524e-07,
+      "loss": 0.6746,
+      "step": 17683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.762327399245547e-07,
+      "loss": 0.6412,
+      "step": 17684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7577750891986455e-07,
+      "loss": 0.6425,
+      "step": 17685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7532254821649795e-07,
+      "loss": 0.5629,
+      "step": 17686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.748678578272336e-07,
+      "loss": 0.5677,
+      "step": 17687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7441343776484116e-07,
+      "loss": 0.6105,
+      "step": 17688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7395928804208394e-07,
+      "loss": 0.5459,
+      "step": 17689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.73505408671716e-07,
+      "loss": 0.5811,
+      "step": 17690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.730517996664862e-07,
+      "loss": 0.6865,
+      "step": 17691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.725984610391331e-07,
+      "loss": 0.6357,
+      "step": 17692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7214539280238994e-07,
+      "loss": 0.6865,
+      "step": 17693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.716925949689809e-07,
+      "loss": 0.663,
+      "step": 17694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7124006755162254e-07,
+      "loss": 0.6907,
+      "step": 17695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.707878105630258e-07,
+      "loss": 0.7123,
+      "step": 17696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.703358240158916e-07,
+      "loss": 0.6169,
+      "step": 17697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6988410792291317e-07,
+      "loss": 0.6768,
+      "step": 17698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.694326622967781e-07,
+      "loss": 0.5879,
+      "step": 17699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6898148715016624e-07,
+      "loss": 0.6197,
+      "step": 17700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6853058249574745e-07,
+      "loss": 0.6203,
+      "step": 17701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6807994834618943e-07,
+      "loss": 0.6444,
+      "step": 17702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.676295847141431e-07,
+      "loss": 0.6001,
+      "step": 17703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6717949161226066e-07,
+      "loss": 0.7089,
+      "step": 17704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.667296690531819e-07,
+      "loss": 0.6801,
+      "step": 17705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6628011704954113e-07,
+      "loss": 0.692,
+      "step": 17706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6583083561396394e-07,
+      "loss": 0.6174,
+      "step": 17707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.653818247590679e-07,
+      "loss": 0.6891,
+      "step": 17708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6493308449746525e-07,
+      "loss": 0.6082,
+      "step": 17709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.644846148417569e-07,
+      "loss": 0.6887,
+      "step": 17710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6403641580454063e-07,
+      "loss": 0.621,
+      "step": 17711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.63588487398403e-07,
+      "loss": 0.602,
+      "step": 17712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.63140829635924e-07,
+      "loss": 0.6342,
+      "step": 17713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.62693442529678e-07,
+      "loss": 0.6048,
+      "step": 17714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.622463260922293e-07,
+      "loss": 0.5934,
+      "step": 17715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6179948033613575e-07,
+      "loss": 0.6195,
+      "step": 17716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6135290527394506e-07,
+      "loss": 0.5884,
+      "step": 17717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6090660091820163e-07,
+      "loss": 0.6494,
+      "step": 17718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6046056728143986e-07,
+      "loss": 0.6639,
+      "step": 17719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.600148043761853e-07,
+      "loss": 0.6432,
+      "step": 17720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5956931221496016e-07,
+      "loss": 0.6224,
+      "step": 17721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5912409081027554e-07,
+      "loss": 0.6584,
+      "step": 17722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.586791401746348e-07,
+      "loss": 0.6273,
+      "step": 17723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5823446032053456e-07,
+      "loss": 0.6394,
+      "step": 17724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5779005126046486e-07,
+      "loss": 0.6531,
+      "step": 17725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.573459130069057e-07,
+      "loss": 0.6987,
+      "step": 17726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5690204557233157e-07,
+      "loss": 0.6371,
+      "step": 17727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5645844896921024e-07,
+      "loss": 0.6674,
+      "step": 17728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5601512320999844e-07,
+      "loss": 0.6168,
+      "step": 17729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.555720683071473e-07,
+      "loss": 0.63,
+      "step": 17730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5512928427310134e-07,
+      "loss": 0.6714,
+      "step": 17731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5468677112029503e-07,
+      "loss": 0.66,
+      "step": 17732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.542445288611585e-07,
+      "loss": 0.6323,
+      "step": 17733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5380255750811057e-07,
+      "loss": 0.6118,
+      "step": 17734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5336085707356584e-07,
+      "loss": 0.5963,
+      "step": 17735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.529194275699288e-07,
+      "loss": 0.6757,
+      "step": 17736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5247826900959626e-07,
+      "loss": 0.6794,
+      "step": 17737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5203738140495935e-07,
+      "loss": 0.6215,
+      "step": 17738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5159676476840046e-07,
+      "loss": 0.5742,
+      "step": 17739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.51156419112294e-07,
+      "loss": 0.7236,
+      "step": 17740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5071634444900913e-07,
+      "loss": 0.6399,
+      "step": 17741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5027654079090477e-07,
+      "loss": 0.603,
+      "step": 17742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4983700815033216e-07,
+      "loss": 0.3267,
+      "step": 17743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4939774653963587e-07,
+      "loss": 0.6705,
+      "step": 17744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.489587559711538e-07,
+      "loss": 0.6453,
+      "step": 17745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.48520036457215e-07,
+      "loss": 0.6617,
+      "step": 17746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.480815880101396e-07,
+      "loss": 0.7185,
+      "step": 17747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476434106422444e-07,
+      "loss": 0.6622,
+      "step": 17748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472055043658329e-07,
+      "loss": 0.6516,
+      "step": 17749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467678691932064e-07,
+      "loss": 0.6891,
+      "step": 17750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46330505136655e-07,
+      "loss": 0.6646,
+      "step": 17751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.458934122084612e-07,
+      "loss": 0.6675,
+      "step": 17752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4545659042090284e-07,
+      "loss": 0.669,
+      "step": 17753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.45020039786248e-07,
+      "loss": 0.569,
+      "step": 17754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.445837603167579e-07,
+      "loss": 0.6189,
+      "step": 17755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.441477520246839e-07,
+      "loss": 0.675,
+      "step": 17756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.437120149222717e-07,
+      "loss": 0.6528,
+      "step": 17757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4327654902176045e-07,
+      "loss": 0.5876,
+      "step": 17758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4284135433538037e-07,
+      "loss": 0.5992,
+      "step": 17759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.424064308753516e-07,
+      "loss": 0.6049,
+      "step": 17760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4197177865389453e-07,
+      "loss": 0.7227,
+      "step": 17761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.415373976832115e-07,
+      "loss": 0.6729,
+      "step": 17762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4110328797550383e-07,
+      "loss": 0.6672,
+      "step": 17763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.406694495429652e-07,
+      "loss": 0.6569,
+      "step": 17764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.40235882397778e-07,
+      "loss": 0.6667,
+      "step": 17765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.398025865521204e-07,
+      "loss": 0.6182,
+      "step": 17766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.393695620181614e-07,
+      "loss": 0.6351,
+      "step": 17767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3893680880806467e-07,
+      "loss": 0.643,
+      "step": 17768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.385043269339805e-07,
+      "loss": 0.6971,
+      "step": 17769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.380721164080569e-07,
+      "loss": 0.6645,
+      "step": 17770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3764017724243534e-07,
+      "loss": 0.6919,
+      "step": 17771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3720850944924386e-07,
+      "loss": 0.7176,
+      "step": 17772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.367771130406072e-07,
+      "loss": 0.603,
+      "step": 17773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.363459880286424e-07,
+      "loss": 0.7423,
+      "step": 17774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.359151344254552e-07,
+      "loss": 0.6577,
+      "step": 17775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3548455224314825e-07,
+      "loss": 0.6786,
+      "step": 17776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3505424149381407e-07,
+      "loss": 0.6292,
+      "step": 17777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.346242021895396e-07,
+      "loss": 0.5729,
+      "step": 17778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3419443434240083e-07,
+      "loss": 0.6668,
+      "step": 17779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.33764937964467e-07,
+      "loss": 0.6528,
+      "step": 17780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.33335713067805e-07,
+      "loss": 0.6473,
+      "step": 17781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.329067596644664e-07,
+      "loss": 0.6498,
+      "step": 17782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.324780777664993e-07,
+      "loss": 0.6445,
+      "step": 17783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.320496673859441e-07,
+      "loss": 0.6564,
+      "step": 17784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.316215285348323e-07,
+      "loss": 0.6333,
+      "step": 17785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.311936612251876e-07,
+      "loss": 0.7045,
+      "step": 17786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.307660654690303e-07,
+      "loss": 0.6818,
+      "step": 17787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.303387412783654e-07,
+      "loss": 0.6608,
+      "step": 17788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.299116886651954e-07,
+      "loss": 0.6897,
+      "step": 17789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.294849076415163e-07,
+      "loss": 0.673,
+      "step": 17790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2905839821931295e-07,
+      "loss": 0.698,
+      "step": 17791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2863216041056465e-07,
+      "loss": 0.7094,
+      "step": 17792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2820619422724296e-07,
+      "loss": 0.6794,
+      "step": 17793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2778049968131164e-07,
+      "loss": 0.6398,
+      "step": 17794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2735507678472443e-07,
+      "loss": 0.6436,
+      "step": 17795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.269299255494307e-07,
+      "loss": 0.6156,
+      "step": 17796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2650504598737085e-07,
+      "loss": 0.6757,
+      "step": 17797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2608043811047874e-07,
+      "loss": 0.622,
+      "step": 17798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.256561019306792e-07,
+      "loss": 0.6944,
+      "step": 17799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2523203745988943e-07,
+      "loss": 0.6935,
+      "step": 17800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.248082447100198e-07,
+      "loss": 0.6299,
+      "step": 17801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2438472369297314e-07,
+      "loss": 0.6311,
+      "step": 17802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.239614744206443e-07,
+      "loss": 0.5859,
+      "step": 17803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2353849690492044e-07,
+      "loss": 0.6587,
+      "step": 17804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2311579115767987e-07,
+      "loss": 0.616,
+      "step": 17805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2269335719079644e-07,
+      "loss": 0.6942,
+      "step": 17806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.222711950161339e-07,
+      "loss": 0.6985,
+      "step": 17807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.218493046455473e-07,
+      "loss": 0.6485,
+      "step": 17808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2142768609088495e-07,
+      "loss": 0.6218,
+      "step": 17809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.210063393639928e-07,
+      "loss": 0.6859,
+      "step": 17810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2058526447670047e-07,
+      "loss": 0.6959,
+      "step": 17811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2016446144083614e-07,
+      "loss": 0.6144,
+      "step": 17812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1974393026821816e-07,
+      "loss": 0.5773,
+      "step": 17813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.19323670970656e-07,
+      "loss": 0.662,
+      "step": 17814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.189036835599546e-07,
+      "loss": 0.6008,
+      "step": 17815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1848396804790683e-07,
+      "loss": 0.6932,
+      "step": 17816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1806452444630434e-07,
+      "loss": 0.6738,
+      "step": 17817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1764535276692435e-07,
+      "loss": 0.6343,
+      "step": 17818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.172264530215408e-07,
+      "loss": 0.6271,
+      "step": 17819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1680782522191754e-07,
+      "loss": 0.6512,
+      "step": 17820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.163894693798142e-07,
+      "loss": 0.6407,
+      "step": 17821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.15971385506979e-07,
+      "loss": 0.6418,
+      "step": 17822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.155535736151538e-07,
+      "loss": 0.6725,
+      "step": 17823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1513603371607473e-07,
+      "loss": 0.6667,
+      "step": 17824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1471876582146676e-07,
+      "loss": 0.6646,
+      "step": 17825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.143017699430506e-07,
+      "loss": 0.5813,
+      "step": 17826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.138850460925358e-07,
+      "loss": 0.6247,
+      "step": 17827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1346859428162845e-07,
+      "loss": 0.6871,
+      "step": 17828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.130524145220215e-07,
+      "loss": 0.6634,
+      "step": 17829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1263650682540783e-07,
+      "loss": 0.6614,
+      "step": 17830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.122208712034669e-07,
+      "loss": 0.6356,
+      "step": 17831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1180550766787054e-07,
+      "loss": 0.5707,
+      "step": 17832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.113904162302872e-07,
+      "loss": 0.6524,
+      "step": 17833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.10975596902372e-07,
+      "loss": 0.6232,
+      "step": 17834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1056104969577673e-07,
+      "loss": 0.6414,
+      "step": 17835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.101467746221443e-07,
+      "loss": 0.6436,
+      "step": 17836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.097327716931098e-07,
+      "loss": 0.6216,
+      "step": 17837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0931904092029953e-07,
+      "loss": 0.6694,
+      "step": 17838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0890558231533527e-07,
+      "loss": 0.7086,
+      "step": 17839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0849239588982784e-07,
+      "loss": 0.6717,
+      "step": 17840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.080794816553834e-07,
+      "loss": 0.3405,
+      "step": 17841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.07666839623596e-07,
+      "loss": 0.6773,
+      "step": 17842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0725446980605753e-07,
+      "loss": 0.6866,
+      "step": 17843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0684237221434877e-07,
+      "loss": 0.6577,
+      "step": 17844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.064305468600437e-07,
+      "loss": 0.6629,
+      "step": 17845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.060189937547098e-07,
+      "loss": 0.6007,
+      "step": 17846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0560771290990334e-07,
+      "loss": 0.6709,
+      "step": 17847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.051967043371762e-07,
+      "loss": 0.6611,
+      "step": 17848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0478596804807246e-07,
+      "loss": 0.6605,
+      "step": 17849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.043755040541263e-07,
+      "loss": 0.6213,
+      "step": 17850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.039653123668673e-07,
+      "loss": 0.6748,
+      "step": 17851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.035553929978174e-07,
+      "loss": 0.6328,
+      "step": 17852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0314574595848743e-07,
+      "loss": 0.6469,
+      "step": 17853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.027363712603826e-07,
+      "loss": 0.6348,
+      "step": 17854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.023272689149992e-07,
+      "loss": 0.3108,
+      "step": 17855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.019184389338303e-07,
+      "loss": 0.637,
+      "step": 17856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.015098813283546e-07,
+      "loss": 0.6296,
+      "step": 17857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.011015961100494e-07,
+      "loss": 0.6815,
+      "step": 17858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.006935832903801e-07,
+      "loss": 0.5789,
+      "step": 17859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0028584288080643e-07,
+      "loss": 0.666,
+      "step": 17860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.998783748927803e-07,
+      "loss": 0.7383,
+      "step": 17861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.994711793377447e-07,
+      "loss": 0.6607,
+      "step": 17862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.990642562271373e-07,
+      "loss": 0.6199,
+      "step": 17863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9865760557238553e-07,
+      "loss": 0.7302,
+      "step": 17864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9825122738491143e-07,
+      "loss": 0.5747,
+      "step": 17865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.97845121676128e-07,
+      "loss": 0.6664,
+      "step": 17866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.974392884574395e-07,
+      "loss": 0.6803,
+      "step": 17867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9703372774024573e-07,
+      "loss": 0.6289,
+      "step": 17868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9662843953593647e-07,
+      "loss": 0.6452,
+      "step": 17869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9622342385589256e-07,
+      "loss": 0.726,
+      "step": 17870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.958186807114938e-07,
+      "loss": 0.6038,
+      "step": 17871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9541421011410444e-07,
+      "loss": 0.5483,
+      "step": 17872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9501001207508427e-07,
+      "loss": 0.7068,
+      "step": 17873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9460608660578537e-07,
+      "loss": 0.6941,
+      "step": 17874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9420243371755196e-07,
+      "loss": 0.6422,
+      "step": 17875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.937990534217228e-07,
+      "loss": 0.6292,
+      "step": 17876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9339594572962537e-07,
+      "loss": 0.618,
+      "step": 17877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9299311065258185e-07,
+      "loss": 0.6409,
+      "step": 17878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9259054820190426e-07,
+      "loss": 0.6659,
+      "step": 17879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.921882583889013e-07,
+      "loss": 0.6915,
+      "step": 17880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9178624122486953e-07,
+      "loss": 0.6974,
+      "step": 17881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9138449672110213e-07,
+      "loss": 0.6501,
+      "step": 17882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9098302488888007e-07,
+      "loss": 0.63,
+      "step": 17883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 0.6877,
+      "step": 17884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.901808992841704e-07,
+      "loss": 0.6593,
+      "step": 17885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8978024553420915e-07,
+      "loss": 0.653,
+      "step": 17886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.893798645008505e-07,
+      "loss": 0.665,
+      "step": 17887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.889797561953389e-07,
+      "loss": 0.6296,
+      "step": 17888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8857992062891305e-07,
+      "loss": 0.6129,
+      "step": 17889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8818035781279953e-07,
+      "loss": 0.6227,
+      "step": 17890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8778106775822377e-07,
+      "loss": 0.5913,
+      "step": 17891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.87382050476398e-07,
+      "loss": 0.6393,
+      "step": 17892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8698330597852874e-07,
+      "loss": 0.6376,
+      "step": 17893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8658483427581487e-07,
+      "loss": 0.6349,
+      "step": 17894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8618663537944955e-07,
+      "loss": 0.6359,
+      "step": 17895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.85788709300614e-07,
+      "loss": 0.7292,
+      "step": 17896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8539105605048577e-07,
+      "loss": 0.7333,
+      "step": 17897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.849936756402327e-07,
+      "loss": 0.6365,
+      "step": 17898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8459656808101477e-07,
+      "loss": 0.6289,
+      "step": 17899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8419973338398524e-07,
+      "loss": 0.7079,
+      "step": 17900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838031715602907e-07,
+      "loss": 0.6272,
+      "step": 17901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.834068826210667e-07,
+      "loss": 0.6092,
+      "step": 17902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.830108665774456e-07,
+      "loss": 0.6187,
+      "step": 17903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.826151234405483e-07,
+      "loss": 0.6508,
+      "step": 17904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.822196532214894e-07,
+      "loss": 0.5909,
+      "step": 17905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8182445593137544e-07,
+      "loss": 0.6073,
+      "step": 17906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.814295315813065e-07,
+      "loss": 0.2811,
+      "step": 17907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.810348801823748e-07,
+      "loss": 0.62,
+      "step": 17908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.806405017456626e-07,
+      "loss": 0.676,
+      "step": 17909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.802463962822477e-07,
+      "loss": 0.6682,
+      "step": 17910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7985256380319903e-07,
+      "loss": 0.6126,
+      "step": 17911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.794590043195766e-07,
+      "loss": 0.6258,
+      "step": 17912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.790657178424339e-07,
+      "loss": 0.6534,
+      "step": 17913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7867270438281634e-07,
+      "loss": 0.6045,
+      "step": 17914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7827996395176195e-07,
+      "loss": 0.6682,
+      "step": 17915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.778874965603018e-07,
+      "loss": 0.6882,
+      "step": 17916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.774953022194593e-07,
+      "loss": 0.6536,
+      "step": 17917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.771033809402468e-07,
+      "loss": 0.673,
+      "step": 17918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7671173273367323e-07,
+      "loss": 0.6608,
+      "step": 17919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.763203576107376e-07,
+      "loss": 0.6948,
+      "step": 17920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.759292555824333e-07,
+      "loss": 0.6605,
+      "step": 17921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7553842665974385e-07,
+      "loss": 0.6106,
+      "step": 17922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.751478708536448e-07,
+      "loss": 0.6164,
+      "step": 17923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7475758817510743e-07,
+      "loss": 0.6083,
+      "step": 17924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.743675786350908e-07,
+      "loss": 0.6369,
+      "step": 17925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.739778422445494e-07,
+      "loss": 0.6882,
+      "step": 17926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7358837901442894e-07,
+      "loss": 0.6197,
+      "step": 17927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.731991889556684e-07,
+      "loss": 0.6963,
+      "step": 17928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7281027207919695e-07,
+      "loss": 0.6606,
+      "step": 17929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.724216283959391e-07,
+      "loss": 0.5803,
+      "step": 17930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7203325791680945e-07,
+      "loss": 0.6336,
+      "step": 17931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7164516065271596e-07,
+      "loss": 0.6063,
+      "step": 17932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7125733661455764e-07,
+      "loss": 0.5903,
+      "step": 17933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7086978581322697e-07,
+      "loss": 0.6989,
+      "step": 17934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7048250825960963e-07,
+      "loss": 0.6604,
+      "step": 17935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.700955039645814e-07,
+      "loss": 0.7339,
+      "step": 17936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6970877293901246e-07,
+      "loss": 0.715,
+      "step": 17937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6932231519376404e-07,
+      "loss": 0.5935,
+      "step": 17938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6893613073968763e-07,
+      "loss": 0.6782,
+      "step": 17939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.685502195876322e-07,
+      "loss": 0.6628,
+      "step": 17940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6816458174843575e-07,
+      "loss": 0.7158,
+      "step": 17941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6777921723292966e-07,
+      "loss": 0.7025,
+      "step": 17942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6739412605193635e-07,
+      "loss": 0.6586,
+      "step": 17943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.670093082162706e-07,
+      "loss": 0.6517,
+      "step": 17944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.666247637367414e-07,
+      "loss": 0.5834,
+      "step": 17945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.66240492624148e-07,
+      "loss": 0.647,
+      "step": 17946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6585649488928386e-07,
+      "loss": 0.6484,
+      "step": 17947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6547277054293276e-07,
+      "loss": 0.6376,
+      "step": 17948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.650893195958715e-07,
+      "loss": 0.6126,
+      "step": 17949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6470614205887035e-07,
+      "loss": 0.6855,
+      "step": 17950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.643232379426919e-07,
+      "loss": 0.6982,
+      "step": 17951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6394060725808855e-07,
+      "loss": 0.5888,
+      "step": 17952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6355825001580735e-07,
+      "loss": 0.6082,
+      "step": 17953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.631761662265875e-07,
+      "loss": 0.6494,
+      "step": 17954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.627943559011592e-07,
+      "loss": 0.6702,
+      "step": 17955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.624128190502473e-07,
+      "loss": 0.6027,
+      "step": 17956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.620315556845643e-07,
+      "loss": 0.6392,
+      "step": 17957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.616505658148216e-07,
+      "loss": 0.6393,
+      "step": 17958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.612698494517174e-07,
+      "loss": 0.5962,
+      "step": 17959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.608894066059431e-07,
+      "loss": 0.6201,
+      "step": 17960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.605092372881879e-07,
+      "loss": 0.6245,
+      "step": 17961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.601293415091255e-07,
+      "loss": 0.5991,
+      "step": 17962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.597497192794285e-07,
+      "loss": 0.6392,
+      "step": 17963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5937037060975614e-07,
+      "loss": 0.6203,
+      "step": 17964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.589912955107632e-07,
+      "loss": 0.6497,
+      "step": 17965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5861249399309563e-07,
+      "loss": 0.6773,
+      "step": 17966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.582339660673938e-07,
+      "loss": 0.6179,
+      "step": 17967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5785571174428813e-07,
+      "loss": 0.6817,
+      "step": 17968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.574777310344012e-07,
+      "loss": 0.6468,
+      "step": 17969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.571000239483512e-07,
+      "loss": 0.5787,
+      "step": 17970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5672259049674407e-07,
+      "loss": 0.6495,
+      "step": 17971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5634543069018027e-07,
+      "loss": 0.6064,
+      "step": 17972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5596854453925346e-07,
+      "loss": 0.6688,
+      "step": 17973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.555919320545486e-07,
+      "loss": 0.6134,
+      "step": 17974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.552155932466427e-07,
+      "loss": 0.6324,
+      "step": 17975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.548395281261062e-07,
+      "loss": 0.6407,
+      "step": 17976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.544637367034997e-07,
+      "loss": 0.6282,
+      "step": 17977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5408821898937786e-07,
+      "loss": 0.6268,
+      "step": 17978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5371297499428793e-07,
+      "loss": 0.6477,
+      "step": 17979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.533380047287681e-07,
+      "loss": 0.6576,
+      "step": 17980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5296330820335e-07,
+      "loss": 0.6726,
+      "step": 17981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5258888542855965e-07,
+      "loss": 0.6364,
+      "step": 17982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.522147364149086e-07,
+      "loss": 0.6662,
+      "step": 17983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5184086117290617e-07,
+      "loss": 0.6096,
+      "step": 17984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.514672597130552e-07,
+      "loss": 0.7135,
+      "step": 17985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.510939320458461e-07,
+      "loss": 0.6143,
+      "step": 17986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.5961,
+      "step": 17987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.503480981312878e-07,
+      "loss": 0.6578,
+      "step": 17988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.499755919048863e-07,
+      "loss": 0.6103,
+      "step": 17989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4960335951302204e-07,
+      "loss": 0.613,
+      "step": 17990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.492314009661489e-07,
+      "loss": 0.6833,
+      "step": 17991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4885971627471306e-07,
+      "loss": 0.6765,
+      "step": 17992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.484883054491549e-07,
+      "loss": 0.6322,
+      "step": 17993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.481171684999051e-07,
+      "loss": 0.5599,
+      "step": 17994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.477463054373874e-07,
+      "loss": 0.6835,
+      "step": 17995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4737571627201585e-07,
+      "loss": 0.675,
+      "step": 17996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.470054010142009e-07,
+      "loss": 0.6494,
+      "step": 17997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4663535967434203e-07,
+      "loss": 0.6324,
+      "step": 17998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.462655922628321e-07,
+      "loss": 0.6321,
+      "step": 17999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.458960987900549e-07,
+      "loss": 0.6215,
+      "step": 18000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.455268792663912e-07,
+      "loss": 0.5997,
+      "step": 18001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.451579337022092e-07,
+      "loss": 0.6241,
+      "step": 18002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4478926210786957e-07,
+      "loss": 0.7121,
+      "step": 18003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.444208644937285e-07,
+      "loss": 0.5966,
+      "step": 18004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4405274087013096e-07,
+      "loss": 0.7177,
+      "step": 18005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4368489124741655e-07,
+      "loss": 0.5924,
+      "step": 18006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4331731563591696e-07,
+      "loss": 0.6091,
+      "step": 18007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.429500140459573e-07,
+      "loss": 0.7209,
+      "step": 18008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.425829864878482e-07,
+      "loss": 0.6207,
+      "step": 18009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4221623297190355e-07,
+      "loss": 0.6514,
+      "step": 18010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.418497535084197e-07,
+      "loss": 0.6306,
+      "step": 18011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.414835481076927e-07,
+      "loss": 0.6192,
+      "step": 18012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4111761678000555e-07,
+      "loss": 0.6528,
+      "step": 18013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.407519595356367e-07,
+      "loss": 0.6741,
+      "step": 18014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4038657638485563e-07,
+      "loss": 0.652,
+      "step": 18015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.400214673379231e-07,
+      "loss": 0.6681,
+      "step": 18016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3965663240509417e-07,
+      "loss": 0.6703,
+      "step": 18017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3929207159661514e-07,
+      "loss": 0.6352,
+      "step": 18018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.389277849227267e-07,
+      "loss": 0.6482,
+      "step": 18019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3856377239365624e-07,
+      "loss": 0.6365,
+      "step": 18020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.382000340196322e-07,
+      "loss": 0.6737,
+      "step": 18021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3783656981086645e-07,
+      "loss": 0.6462,
+      "step": 18022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3747337977756857e-07,
+      "loss": 0.627,
+      "step": 18023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.371104639299393e-07,
+      "loss": 0.7115,
+      "step": 18024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.367478222781705e-07,
+      "loss": 0.6172,
+      "step": 18025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3638545483244735e-07,
+      "loss": 0.5998,
+      "step": 18026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3602336160294726e-07,
+      "loss": 0.6349,
+      "step": 18027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3566154259983987e-07,
+      "loss": 0.6598,
+      "step": 18028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3529999783328593e-07,
+      "loss": 0.6606,
+      "step": 18029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3493872731344182e-07,
+      "loss": 0.6428,
+      "step": 18030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3457773105045267e-07,
+      "loss": 0.6775,
+      "step": 18031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.342170090544571e-07,
+      "loss": 0.6864,
+      "step": 18032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3385656133558699e-07,
+      "loss": 0.643,
+      "step": 18033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3349638790396533e-07,
+      "loss": 0.6131,
+      "step": 18034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3313648876970742e-07,
+      "loss": 0.6387,
+      "step": 18035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3277686394292065e-07,
+      "loss": 0.6434,
+      "step": 18036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3241751343370699e-07,
+      "loss": 0.6418,
+      "step": 18037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3205843725215726e-07,
+      "loss": 0.6078,
+      "step": 18038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3169963540835783e-07,
+      "loss": 0.66,
+      "step": 18039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3134110791238395e-07,
+      "loss": 0.6729,
+      "step": 18040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.309828547743076e-07,
+      "loss": 0.6955,
+      "step": 18041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.306248760041885e-07,
+      "loss": 0.619,
+      "step": 18042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.302671716120819e-07,
+      "loss": 0.6632,
+      "step": 18043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2990974160803203e-07,
+      "loss": 0.7003,
+      "step": 18044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2955258600207975e-07,
+      "loss": 0.7077,
+      "step": 18045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.291957048042559e-07,
+      "loss": 0.6974,
+      "step": 18046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2883909802458358e-07,
+      "loss": 0.6319,
+      "step": 18047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2848276567307593e-07,
+      "loss": 0.59,
+      "step": 18048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2812670775974267e-07,
+      "loss": 0.6404,
+      "step": 18049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2777092429458358e-07,
+      "loss": 0.7054,
+      "step": 18050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2741541528759182e-07,
+      "loss": 0.7399,
+      "step": 18051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2706018074875046e-07,
+      "loss": 0.6334,
+      "step": 18052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2670522068803824e-07,
+      "loss": 0.7263,
+      "step": 18053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2635053511542382e-07,
+      "loss": 0.6191,
+      "step": 18054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2599612404086813e-07,
+      "loss": 0.6413,
+      "step": 18055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.256419874743243e-07,
+      "loss": 0.6823,
+      "step": 18056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2528812542573996e-07,
+      "loss": 0.6397,
+      "step": 18057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2493453790505381e-07,
+      "loss": 0.6994,
+      "step": 18058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2458122492219458e-07,
+      "loss": 0.5919,
+      "step": 18059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2422818648708656e-07,
+      "loss": 0.6349,
+      "step": 18060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23875422609644e-07,
+      "loss": 0.6149,
+      "step": 18061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2352293329977682e-07,
+      "loss": 0.6669,
+      "step": 18062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.231707185673826e-07,
+      "loss": 0.6331,
+      "step": 18063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2281877842235344e-07,
+      "loss": 0.6354,
+      "step": 18064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.224671128745759e-07,
+      "loss": 0.6665,
+      "step": 18065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2211572193392428e-07,
+      "loss": 0.6609,
+      "step": 18066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2176460561026958e-07,
+      "loss": 0.6056,
+      "step": 18067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2141376391347168e-07,
+      "loss": 0.5664,
+      "step": 18068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2106319685338384e-07,
+      "loss": 0.6155,
+      "step": 18069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.207129044398526e-07,
+      "loss": 0.6691,
+      "step": 18070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2036288668271677e-07,
+      "loss": 0.5895,
+      "step": 18071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2001314359180627e-07,
+      "loss": 0.6709,
+      "step": 18072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196636751769432e-07,
+      "loss": 0.6882,
+      "step": 18073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1931448144794421e-07,
+      "loss": 0.6209,
+      "step": 18074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1896556241461474e-07,
+      "loss": 0.6654,
+      "step": 18075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186169180867559e-07,
+      "loss": 0.6883,
+      "step": 18076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1826854847415756e-07,
+      "loss": 0.6489,
+      "step": 18077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.179204535866053e-07,
+      "loss": 0.6387,
+      "step": 18078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.175726334338757e-07,
+      "loss": 0.662,
+      "step": 18079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.172250880257365e-07,
+      "loss": 0.606,
+      "step": 18080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.168778173719488e-07,
+      "loss": 0.6935,
+      "step": 18081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.16530821482267e-07,
+      "loss": 0.6886,
+      "step": 18082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1618410036643666e-07,
+      "loss": 0.5798,
+      "step": 18083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.158376540341933e-07,
+      "loss": 0.5791,
+      "step": 18084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.154914824952703e-07,
+      "loss": 0.6487,
+      "step": 18085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.151455857593876e-07,
+      "loss": 0.6564,
+      "step": 18086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1479996383626188e-07,
+      "loss": 0.6933,
+      "step": 18087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1445461673559764e-07,
+      "loss": 0.6266,
+      "step": 18088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1410954446709596e-07,
+      "loss": 0.6483,
+      "step": 18089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1376474704044693e-07,
+      "loss": 0.7053,
+      "step": 18090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1342022446533606e-07,
+      "loss": 0.6136,
+      "step": 18091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.13075976751439e-07,
+      "loss": 0.5778,
+      "step": 18092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1273200390842464e-07,
+      "loss": 0.6313,
+      "step": 18093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1238830594595195e-07,
+      "loss": 0.7053,
+      "step": 18094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1204488287367543e-07,
+      "loss": 0.6326,
+      "step": 18095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1170173470123845e-07,
+      "loss": 0.6837,
+      "step": 18096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.113588614382811e-07,
+      "loss": 0.626,
+      "step": 18097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1101626309443013e-07,
+      "loss": 0.7477,
+      "step": 18098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1067393967931004e-07,
+      "loss": 0.654,
+      "step": 18099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1033189120253427e-07,
+      "loss": 0.7408,
+      "step": 18100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.099901176737096e-07,
+      "loss": 0.6216,
+      "step": 18101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0964861910243494e-07,
+      "loss": 0.6218,
+      "step": 18102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0930739549830047e-07,
+      "loss": 0.6711,
+      "step": 18103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0896644687089074e-07,
+      "loss": 0.623,
+      "step": 18104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.086257732297825e-07,
+      "loss": 0.7751,
+      "step": 18105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0828537458454146e-07,
+      "loss": 0.6714,
+      "step": 18106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0794525094472883e-07,
+      "loss": 0.3155,
+      "step": 18107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0760540231989813e-07,
+      "loss": 0.6334,
+      "step": 18108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0726582871959166e-07,
+      "loss": 0.6788,
+      "step": 18109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.069265301533485e-07,
+      "loss": 0.6823,
+      "step": 18110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0658750663069772e-07,
+      "loss": 0.5974,
+      "step": 18111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0624875816116163e-07,
+      "loss": 0.6652,
+      "step": 18112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.059102847542538e-07,
+      "loss": 0.6438,
+      "step": 18113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.055720864194799e-07,
+      "loss": 0.6629,
+      "step": 18114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0523416316633793e-07,
+      "loss": 0.6172,
+      "step": 18115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.048965150043203e-07,
+      "loss": 0.6362,
+      "step": 18116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0455914194290827e-07,
+      "loss": 0.6809,
+      "step": 18117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0422204399157762e-07,
+      "loss": 0.6539,
+      "step": 18118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0388522115979638e-07,
+      "loss": 0.6312,
+      "step": 18119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0354867345702467e-07,
+      "loss": 0.6606,
+      "step": 18120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0321240089271389e-07,
+      "loss": 0.6241,
+      "step": 18121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0287640347630865e-07,
+      "loss": 0.6498,
+      "step": 18122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0254068121724478e-07,
+      "loss": 0.5886,
+      "step": 18123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0220523412495364e-07,
+      "loss": 0.6778,
+      "step": 18124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.018700622088543e-07,
+      "loss": 0.6786,
+      "step": 18125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.015351654783615e-07,
+      "loss": 0.6526,
+      "step": 18126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.012005439428799e-07,
+      "loss": 0.6735,
+      "step": 18127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0086619761180758e-07,
+      "loss": 0.5551,
+      "step": 18128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.005321264945348e-07,
+      "loss": 0.6216,
+      "step": 18129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0019833060044512e-07,
+      "loss": 0.6369,
+      "step": 18130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9986480993891222e-07,
+      "loss": 0.6673,
+      "step": 18131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9953156451930523e-07,
+      "loss": 0.6785,
+      "step": 18132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.991985943509811e-07,
+      "loss": 0.6465,
+      "step": 18133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9886589944329237e-07,
+      "loss": 0.582,
+      "step": 18134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.985334798055827e-07,
+      "loss": 0.7112,
+      "step": 18135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.98201335447189e-07,
+      "loss": 0.5622,
+      "step": 18136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9786946637743942e-07,
+      "loss": 0.6538,
+      "step": 18137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9753787260565428e-07,
+      "loss": 0.6655,
+      "step": 18138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.972065541411472e-07,
+      "loss": 0.6844,
+      "step": 18139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.96875510993223e-07,
+      "loss": 0.6588,
+      "step": 18140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9654474317117867e-07,
+      "loss": 0.6624,
+      "step": 18141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9621425068430454e-07,
+      "loss": 0.6848,
+      "step": 18142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9588403354188324e-07,
+      "loss": 0.6684,
+      "step": 18143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.955540917531884e-07,
+      "loss": 0.6708,
+      "step": 18144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.952244253274882e-07,
+      "loss": 0.7411,
+      "step": 18145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9489503427403855e-07,
+      "loss": 0.5754,
+      "step": 18146
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9456591860209205e-07,
+      "loss": 0.6068,
+      "step": 18147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9423707832089244e-07,
+      "loss": 0.6199,
+      "step": 18148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9390851343967564e-07,
+      "loss": 0.6753,
+      "step": 18149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9358022396766762e-07,
+      "loss": 0.6309,
+      "step": 18150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9325220991408988e-07,
+      "loss": 0.6856,
+      "step": 18151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9292447128815616e-07,
+      "loss": 0.6904,
+      "step": 18152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9259700809907023e-07,
+      "loss": 0.6479,
+      "step": 18153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9226982035602803e-07,
+      "loss": 0.7044,
+      "step": 18154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9194290806821892e-07,
+      "loss": 0.6015,
+      "step": 18155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9161627124482552e-07,
+      "loss": 0.7055,
+      "step": 18156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.912899098950216e-07,
+      "loss": 0.5806,
+      "step": 18157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9096382402797208e-07,
+      "loss": 0.6698,
+      "step": 18158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.906380136528352e-07,
+      "loss": 0.6577,
+      "step": 18159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9031247877876358e-07,
+      "loss": 0.6314,
+      "step": 18160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8998721941489772e-07,
+      "loss": 0.6572,
+      "step": 18161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8966223557037367e-07,
+      "loss": 0.6619,
+      "step": 18162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8933752725431852e-07,
+      "loss": 0.7179,
+      "step": 18163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.890130944758528e-07,
+      "loss": 0.6217,
+      "step": 18164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.886889372440881e-07,
+      "loss": 0.6863,
+      "step": 18165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8836505556812712e-07,
+      "loss": 0.6361,
+      "step": 18166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8804144945706705e-07,
+      "loss": 0.6906,
+      "step": 18167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8771811891999726e-07,
+      "loss": 0.6332,
+      "step": 18168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8739506396599717e-07,
+      "loss": 0.5992,
+      "step": 18169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8707228460414172e-07,
+      "loss": 0.599,
+      "step": 18170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.867497808434948e-07,
+      "loss": 0.6046,
+      "step": 18171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8642755269311586e-07,
+      "loss": 0.6372,
+      "step": 18172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8610560016205314e-07,
+      "loss": 0.6931,
+      "step": 18173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8578392325934947e-07,
+      "loss": 0.6193,
+      "step": 18174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.854625219940387e-07,
+      "loss": 0.6859,
+      "step": 18175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8514139637514916e-07,
+      "loss": 0.6022,
+      "step": 18176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8482054641169812e-07,
+      "loss": 0.6796,
+      "step": 18177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.844999721126983e-07,
+      "loss": 0.6055,
+      "step": 18178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.841796734871515e-07,
+      "loss": 0.5826,
+      "step": 18179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8385965054405485e-07,
+      "loss": 0.6585,
+      "step": 18180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8353990329239567e-07,
+      "loss": 0.6956,
+      "step": 18181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8322043174115457e-07,
+      "loss": 0.5756,
+      "step": 18182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.829012358993043e-07,
+      "loss": 0.6655,
+      "step": 18183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8258231577581e-07,
+      "loss": 0.6437,
+      "step": 18184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8226367137962664e-07,
+      "loss": 0.5783,
+      "step": 18185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8194530271970602e-07,
+      "loss": 0.3142,
+      "step": 18186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8162720980498761e-07,
+      "loss": 0.6254,
+      "step": 18187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.813093926444065e-07,
+      "loss": 0.6623,
+      "step": 18188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.809918512468889e-07,
+      "loss": 0.6346,
+      "step": 18189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8067458562135208e-07,
+      "loss": 0.6931,
+      "step": 18190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8035759577670676e-07,
+      "loss": 0.5844,
+      "step": 18191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8004088172185687e-07,
+      "loss": 0.3115,
+      "step": 18192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7972444346569752e-07,
+      "loss": 0.6386,
+      "step": 18193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7940828101711383e-07,
+      "loss": 0.6577,
+      "step": 18194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.790923943849876e-07,
+      "loss": 0.6556,
+      "step": 18195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.787767835781906e-07,
+      "loss": 0.5663,
+      "step": 18196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.784614486055858e-07,
+      "loss": 0.6411,
+      "step": 18197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7814638947603047e-07,
+      "loss": 0.678,
+      "step": 18198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7783160619837202e-07,
+      "loss": 0.2949,
+      "step": 18199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7751709878145118e-07,
+      "loss": 0.6094,
+      "step": 18200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7720286723410307e-07,
+      "loss": 0.6115,
+      "step": 18201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7688891156515286e-07,
+      "loss": 0.6275,
+      "step": 18202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7657523178341574e-07,
+      "loss": 0.6309,
+      "step": 18203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7626182789770463e-07,
+      "loss": 0.6307,
+      "step": 18204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7594869991681917e-07,
+      "loss": 0.6822,
+      "step": 18205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7563584784955456e-07,
+      "loss": 0.6405,
+      "step": 18206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.753232717046971e-07,
+      "loss": 0.6644,
+      "step": 18207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.750109714910264e-07,
+      "loss": 0.6369,
+      "step": 18208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7469894721731328e-07,
+      "loss": 0.647,
+      "step": 18209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.743871988923207e-07,
+      "loss": 0.7394,
+      "step": 18210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7407572652480498e-07,
+      "loss": 0.7022,
+      "step": 18211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7376453012351248e-07,
+      "loss": 0.6597,
+      "step": 18212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7345360969718617e-07,
+      "loss": 0.6752,
+      "step": 18213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7314296525455575e-07,
+      "loss": 0.6893,
+      "step": 18214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.728325968043465e-07,
+      "loss": 0.6018,
+      "step": 18215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7252250435527584e-07,
+      "loss": 0.6091,
+      "step": 18216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.722126879160535e-07,
+      "loss": 0.6251,
+      "step": 18217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7190314749537918e-07,
+      "loss": 0.6395,
+      "step": 18218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7159388310194703e-07,
+      "loss": 0.645,
+      "step": 18219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7128489474444233e-07,
+      "loss": 0.6051,
+      "step": 18220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.709761824315448e-07,
+      "loss": 0.6138,
+      "step": 18221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7066774617192417e-07,
+      "loss": 0.6614,
+      "step": 18222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.703595859742424e-07,
+      "loss": 0.5929,
+      "step": 18223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7005170184715592e-07,
+      "loss": 0.6762,
+      "step": 18224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.697440937993089e-07,
+      "loss": 0.6282,
+      "step": 18225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6943676183934333e-07,
+      "loss": 0.6787,
+      "step": 18226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6912970597588897e-07,
+      "loss": 0.6621,
+      "step": 18227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6882292621757113e-07,
+      "loss": 0.6792,
+      "step": 18228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6851642257300516e-07,
+      "loss": 0.7044,
+      "step": 18229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6821019505079973e-07,
+      "loss": 0.6929,
+      "step": 18230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.679042436595557e-07,
+      "loss": 0.6277,
+      "step": 18231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6759856840786403e-07,
+      "loss": 0.685,
+      "step": 18232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6729316930431228e-07,
+      "loss": 0.6795,
+      "step": 18233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.669880463574758e-07,
+      "loss": 0.6744,
+      "step": 18234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6668319957592549e-07,
+      "loss": 0.6055,
+      "step": 18235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6637862896822233e-07,
+      "loss": 0.6161,
+      "step": 18236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6607433454292054e-07,
+      "loss": 0.6748,
+      "step": 18237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6577031630856554e-07,
+      "loss": 0.7124,
+      "step": 18238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6546657427369718e-07,
+      "loss": 0.628,
+      "step": 18239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6516310844684525e-07,
+      "loss": 0.7143,
+      "step": 18240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6485991883653408e-07,
+      "loss": 0.6439,
+      "step": 18241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6455700545127684e-07,
+      "loss": 0.5783,
+      "step": 18242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.642543682995845e-07,
+      "loss": 0.6501,
+      "step": 18243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6395200738995254e-07,
+      "loss": 0.5767,
+      "step": 18244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.636499227308752e-07,
+      "loss": 0.7252,
+      "step": 18245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6334811433083686e-07,
+      "loss": 0.6666,
+      "step": 18246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6304658219831292e-07,
+      "loss": 0.6739,
+      "step": 18247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.627453263417733e-07,
+      "loss": 0.6477,
+      "step": 18248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.624443467696779e-07,
+      "loss": 0.6885,
+      "step": 18249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6214364349047996e-07,
+      "loss": 0.6494,
+      "step": 18250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6184321651262491e-07,
+      "loss": 0.5764,
+      "step": 18251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6154306584455048e-07,
+      "loss": 0.5894,
+      "step": 18252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.612431914946877e-07,
+      "loss": 0.6237,
+      "step": 18253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609435934714576e-07,
+      "loss": 0.6812,
+      "step": 18254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6064427178327458e-07,
+      "loss": 0.6288,
+      "step": 18255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6034522643854522e-07,
+      "loss": 0.6825,
+      "step": 18256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.600464574456684e-07,
+      "loss": 0.631,
+      "step": 18257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5974796481303512e-07,
+      "loss": 0.5975,
+      "step": 18258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5944974854902873e-07,
+      "loss": 0.6395,
+      "step": 18259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5915180866202585e-07,
+      "loss": 0.5971,
+      "step": 18260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5885414516039199e-07,
+      "loss": 0.6044,
+      "step": 18261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.585567580524894e-07,
+      "loss": 0.6364,
+      "step": 18262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.582596473466702e-07,
+      "loss": 0.6109,
+      "step": 18263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.579628130512767e-07,
+      "loss": 0.676,
+      "step": 18264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5766625517464774e-07,
+      "loss": 0.6383,
+      "step": 18265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.573699737251122e-07,
+      "loss": 0.6177,
+      "step": 18266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5707396871099012e-07,
+      "loss": 0.6385,
+      "step": 18267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5677824014059595e-07,
+      "loss": 0.6397,
+      "step": 18268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5648278802223526e-07,
+      "loss": 0.6677,
+      "step": 18269
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5618761236420587e-07,
+      "loss": 0.6208,
+      "step": 18270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.558927131747967e-07,
+      "loss": 0.6779,
+      "step": 18271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5559809046229224e-07,
+      "loss": 0.5696,
+      "step": 18272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5530374423496696e-07,
+      "loss": 0.6723,
+      "step": 18273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.550096745010854e-07,
+      "loss": 0.7029,
+      "step": 18274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5471588126890979e-07,
+      "loss": 0.659,
+      "step": 18275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5442236454668912e-07,
+      "loss": 0.5859,
+      "step": 18276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.541291243426679e-07,
+      "loss": 0.6002,
+      "step": 18277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5383616066508177e-07,
+      "loss": 0.637,
+      "step": 18278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.535434735221586e-07,
+      "loss": 0.5994,
+      "step": 18279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5325106292211843e-07,
+      "loss": 0.6771,
+      "step": 18280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5295892887317476e-07,
+      "loss": 0.5767,
+      "step": 18281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.526670713835321e-07,
+      "loss": 0.6485,
+      "step": 18282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5237549046138612e-07,
+      "loss": 0.6566,
+      "step": 18283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5208418611492804e-07,
+      "loss": 0.6455,
+      "step": 18284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5179315835233689e-07,
+      "loss": 0.6406,
+      "step": 18285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5150240718178833e-07,
+      "loss": 0.6454,
+      "step": 18286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5121193261144805e-07,
+      "loss": 0.6699,
+      "step": 18287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5092173464947402e-07,
+      "loss": 0.6255,
+      "step": 18288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5063181330401522e-07,
+      "loss": 0.6173,
+      "step": 18289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5034216858321515e-07,
+      "loss": 0.6387,
+      "step": 18290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5005280049520954e-07,
+      "loss": 0.6023,
+      "step": 18291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.497637090481241e-07,
+      "loss": 0.5587,
+      "step": 18292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4947489425008012e-07,
+      "loss": 0.6529,
+      "step": 18293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4918635610918664e-07,
+      "loss": 0.6204,
+      "step": 18294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4889809463354942e-07,
+      "loss": 0.6664,
+      "step": 18295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4861010983126202e-07,
+      "loss": 0.6223,
+      "step": 18296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4832240171041569e-07,
+      "loss": 0.6064,
+      "step": 18297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4803497027908842e-07,
+      "loss": 0.6317,
+      "step": 18298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4774781554535377e-07,
+      "loss": 0.6213,
+      "step": 18299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.474609375172764e-07,
+      "loss": 0.6606,
+      "step": 18300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4717433620291322e-07,
+      "loss": 0.6544,
+      "step": 18301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.468880116103155e-07,
+      "loss": 0.6339,
+      "step": 18302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4660196374752243e-07,
+      "loss": 0.6058,
+      "step": 18303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.463161926225687e-07,
+      "loss": 0.6207,
+      "step": 18304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4603069824348005e-07,
+      "loss": 0.6095,
+      "step": 18305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4574548061827564e-07,
+      "loss": 0.6759,
+      "step": 18306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4546053975496465e-07,
+      "loss": 0.6166,
+      "step": 18307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.451758756615518e-07,
+      "loss": 0.6555,
+      "step": 18308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.448914883460306e-07,
+      "loss": 0.6319,
+      "step": 18309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4460737781638701e-07,
+      "loss": 0.6388,
+      "step": 18310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4432354408060233e-07,
+      "loss": 0.6825,
+      "step": 18311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.440399871466469e-07,
+      "loss": 0.6471,
+      "step": 18312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4375670702248656e-07,
+      "loss": 0.6917,
+      "step": 18313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4347370371607605e-07,
+      "loss": 0.6464,
+      "step": 18314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4319097723536345e-07,
+      "loss": 0.6786,
+      "step": 18315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.429085275882891e-07,
+      "loss": 0.611,
+      "step": 18316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4262635478278665e-07,
+      "loss": 0.6676,
+      "step": 18317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4234445882677972e-07,
+      "loss": 0.6996,
+      "step": 18318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4206283972818757e-07,
+      "loss": 0.5947,
+      "step": 18319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4178149749491722e-07,
+      "loss": 0.6719,
+      "step": 18320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.415004321348723e-07,
+      "loss": 0.6135,
+      "step": 18321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4121964365594654e-07,
+      "loss": 0.6558,
+      "step": 18322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4093913206602473e-07,
+      "loss": 0.6584,
+      "step": 18323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.40658897372985e-07,
+      "loss": 0.6211,
+      "step": 18324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4037893958469994e-07,
+      "loss": 0.5592,
+      "step": 18325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4009925870903108e-07,
+      "loss": 0.678,
+      "step": 18326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.398198547538332e-07,
+      "loss": 0.6775,
+      "step": 18327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3954072772695338e-07,
+      "loss": 0.6226,
+      "step": 18328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3926187763623201e-07,
+      "loss": 0.7136,
+      "step": 18329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3898330448949837e-07,
+      "loss": 0.6597,
+      "step": 18330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3870500829457956e-07,
+      "loss": 0.6619,
+      "step": 18331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.384269890592904e-07,
+      "loss": 0.708,
+      "step": 18332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3814924679143915e-07,
+      "loss": 0.5867,
+      "step": 18333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3787178149882618e-07,
+      "loss": 0.6362,
+      "step": 18334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3759459318924417e-07,
+      "loss": 0.6469,
+      "step": 18335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.37317681870478e-07,
+      "loss": 0.3086,
+      "step": 18336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.370410475503048e-07,
+      "loss": 0.6328,
+      "step": 18337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3676469023649497e-07,
+      "loss": 0.6301,
+      "step": 18338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3648860993680903e-07,
+      "loss": 0.6464,
+      "step": 18339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3621280665900184e-07,
+      "loss": 0.6515,
+      "step": 18340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3593728041081833e-07,
+      "loss": 0.7554,
+      "step": 18341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3566203119999678e-07,
+      "loss": 0.2823,
+      "step": 18342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3538705903426875e-07,
+      "loss": 0.6379,
+      "step": 18343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3511236392135696e-07,
+      "loss": 0.672,
+      "step": 18344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3483794586897636e-07,
+      "loss": 0.7148,
+      "step": 18345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.34563804884833e-07,
+      "loss": 0.6141,
+      "step": 18346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.342899409766274e-07,
+      "loss": 0.6565,
+      "step": 18347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3401635415205005e-07,
+      "loss": 0.6819,
+      "step": 18348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3374304441878594e-07,
+      "loss": 0.6854,
+      "step": 18349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3347001178451003e-07,
+      "loss": 0.6196,
+      "step": 18350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3319725625689173e-07,
+      "loss": 0.6131,
+      "step": 18351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.329247778435905e-07,
+      "loss": 0.5935,
+      "step": 18352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3265257655226126e-07,
+      "loss": 0.5958,
+      "step": 18353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3238065239054575e-07,
+      "loss": 0.6616,
+      "step": 18354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.321090053660823e-07,
+      "loss": 0.6409,
+      "step": 18355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.318376354865014e-07,
+      "loss": 0.6769,
+      "step": 18356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.315665427594237e-07,
+      "loss": 0.6039,
+      "step": 18357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3129572719246198e-07,
+      "loss": 0.6251,
+      "step": 18358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3102518879322456e-07,
+      "loss": 0.6791,
+      "step": 18359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3075492756930763e-07,
+      "loss": 0.3099,
+      "step": 18360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3048494352830175e-07,
+      "loss": 0.7153,
+      "step": 18361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.302152366777909e-07,
+      "loss": 0.5918,
+      "step": 18362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2994580702534898e-07,
+      "loss": 0.6676,
+      "step": 18363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2967665457854438e-07,
+      "loss": 0.6726,
+      "step": 18364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2940777934493332e-07,
+      "loss": 0.6579,
+      "step": 18365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2913918133207193e-07,
+      "loss": 0.6783,
+      "step": 18366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2887086054749866e-07,
+      "loss": 0.6643,
+      "step": 18367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2860281699875298e-07,
+      "loss": 0.6526,
+      "step": 18368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2833505069336228e-07,
+      "loss": 0.6955,
+      "step": 18369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2806756163884604e-07,
+      "loss": 0.6767,
+      "step": 18370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2780034984271828e-07,
+      "loss": 0.6284,
+      "step": 18371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2753341531248298e-07,
+      "loss": 0.6335,
+      "step": 18372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2726675805563748e-07,
+      "loss": 0.6059,
+      "step": 18373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2700037807967026e-07,
+      "loss": 0.66,
+      "step": 18374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2673427539206306e-07,
+      "loss": 0.6116,
+      "step": 18375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2646845000028996e-07,
+      "loss": 0.6487,
+      "step": 18376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2620290191181606e-07,
+      "loss": 0.6377,
+      "step": 18377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2593763113409986e-07,
+      "loss": 0.6454,
+      "step": 18378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2567263767459204e-07,
+      "loss": 0.6534,
+      "step": 18379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2540792154073444e-07,
+      "loss": 0.6229,
+      "step": 18380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2514348273996113e-07,
+      "loss": 0.561,
+      "step": 18381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.248793212797006e-07,
+      "loss": 0.7023,
+      "step": 18382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2461543716737133e-07,
+      "loss": 0.7166,
+      "step": 18383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2435183041038412e-07,
+      "loss": 0.6509,
+      "step": 18384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.24088501016143e-07,
+      "loss": 0.5602,
+      "step": 18385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.238254489920454e-07,
+      "loss": 0.6577,
+      "step": 18386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2356267434547543e-07,
+      "loss": 0.631,
+      "step": 18387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2330017708381603e-07,
+      "loss": 0.6678,
+      "step": 18388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.23037957214438e-07,
+      "loss": 0.5788,
+      "step": 18389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2277601474470767e-07,
+      "loss": 0.6597,
+      "step": 18390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2251434968197918e-07,
+      "loss": 0.6367,
+      "step": 18391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2225296203360547e-07,
+      "loss": 0.6185,
+      "step": 18392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2199185180692407e-07,
+      "loss": 0.6523,
+      "step": 18393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2173101900927131e-07,
+      "loss": 0.5717,
+      "step": 18394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2147046364797023e-07,
+      "loss": 0.6288,
+      "step": 18395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.212101857303405e-07,
+      "loss": 0.6419,
+      "step": 18396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2095018526369074e-07,
+      "loss": 0.6334,
+      "step": 18397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2069046225532399e-07,
+      "loss": 0.6346,
+      "step": 18398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2043101671253553e-07,
+      "loss": 0.6778,
+      "step": 18399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2017184864261065e-07,
+      "loss": 0.7333,
+      "step": 18400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.199129580528291e-07,
+      "loss": 0.5587,
+      "step": 18401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1965434495046056e-07,
+      "loss": 0.6198,
+      "step": 18402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.193960093427704e-07,
+      "loss": 0.6448,
+      "step": 18403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1913795123701277e-07,
+      "loss": 0.7498,
+      "step": 18404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1888017064043633e-07,
+      "loss": 0.6674,
+      "step": 18405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1862266756027862e-07,
+      "loss": 0.5823,
+      "step": 18406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1836544200377498e-07,
+      "loss": 0.7009,
+      "step": 18407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1810849397814739e-07,
+      "loss": 0.6837,
+      "step": 18408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1785182349061342e-07,
+      "loss": 0.6177,
+      "step": 18409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1759543054838063e-07,
+      "loss": 0.6673,
+      "step": 18410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1733931515865105e-07,
+      "loss": 0.6614,
+      "step": 18411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1708347732861891e-07,
+      "loss": 0.6354,
+      "step": 18412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1682791706546737e-07,
+      "loss": 0.6785,
+      "step": 18413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1657263437637511e-07,
+      "loss": 0.6172,
+      "step": 18414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1631762926851197e-07,
+      "loss": 0.6327,
+      "step": 18415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.160629017490389e-07,
+      "loss": 0.6121,
+      "step": 18416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1580845182511124e-07,
+      "loss": 0.6137,
+      "step": 18417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1555427950387443e-07,
+      "loss": 0.6495,
+      "step": 18418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1530038479246719e-07,
+      "loss": 0.6974,
+      "step": 18419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1504676769802048e-07,
+      "loss": 0.6084,
+      "step": 18420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1479342822765749e-07,
+      "loss": 0.6545,
+      "step": 18421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1454036638849364e-07,
+      "loss": 0.6104,
+      "step": 18422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1428758218763547e-07,
+      "loss": 0.6834,
+      "step": 18423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1403507563218285e-07,
+      "loss": 0.6752,
+      "step": 18424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1378284672922901e-07,
+      "loss": 0.6343,
+      "step": 18425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1353089548585494e-07,
+      "loss": 0.6175,
+      "step": 18426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1327922190913943e-07,
+      "loss": 0.6114,
+      "step": 18427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1302782600614904e-07,
+      "loss": 0.6447,
+      "step": 18428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.127767077839459e-07,
+      "loss": 0.6077,
+      "step": 18429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1252586724958214e-07,
+      "loss": 0.6291,
+      "step": 18430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1227530441010104e-07,
+      "loss": 0.7053,
+      "step": 18431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1202501927254361e-07,
+      "loss": 0.6603,
+      "step": 18432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1177501184393647e-07,
+      "loss": 0.6357,
+      "step": 18433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1152528213130176e-07,
+      "loss": 0.6345,
+      "step": 18434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1127583014165278e-07,
+      "loss": 0.6064,
+      "step": 18435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1102665588199723e-07,
+      "loss": 0.6761,
+      "step": 18436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1077775935933177e-07,
+      "loss": 0.6451,
+      "step": 18437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1052914058064745e-07,
+      "loss": 0.6054,
+      "step": 18438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1028079955292537e-07,
+      "loss": 0.6704,
+      "step": 18439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1003273628314215e-07,
+      "loss": 0.7351,
+      "step": 18440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0978495077826445e-07,
+      "loss": 0.6155,
+      "step": 18441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0953744304525116e-07,
+      "loss": 0.6841,
+      "step": 18442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0929021309105337e-07,
+      "loss": 0.662,
+      "step": 18443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0904326092261441e-07,
+      "loss": 0.6989,
+      "step": 18444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0879658654687209e-07,
+      "loss": 0.582,
+      "step": 18445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0855018997075196e-07,
+      "loss": 0.6459,
+      "step": 18446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0830407120117625e-07,
+      "loss": 0.5834,
+      "step": 18447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0805823024505502e-07,
+      "loss": 0.726,
+      "step": 18448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0781266710929384e-07,
+      "loss": 0.6396,
+      "step": 18449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0756738180079051e-07,
+      "loss": 0.5819,
+      "step": 18450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0732237432643289e-07,
+      "loss": 0.6243,
+      "step": 18451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.070776446931021e-07,
+      "loss": 0.6571,
+      "step": 18452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.068331929076727e-07,
+      "loss": 0.6418,
+      "step": 18453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0658901897700912e-07,
+      "loss": 0.6653,
+      "step": 18454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0634512290796816e-07,
+      "loss": 0.6009,
+      "step": 18455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.061015047074021e-07,
+      "loss": 0.5915,
+      "step": 18456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0585816438215213e-07,
+      "loss": 0.6393,
+      "step": 18457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0561510193905277e-07,
+      "loss": 0.6427,
+      "step": 18458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.053723173849297e-07,
+      "loss": 0.6371,
+      "step": 18459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0512981072660077e-07,
+      "loss": 0.6344,
+      "step": 18460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0488758197087944e-07,
+      "loss": 0.6652,
+      "step": 18461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0464563112456805e-07,
+      "loss": 0.6849,
+      "step": 18462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0440395819446114e-07,
+      "loss": 0.6411,
+      "step": 18463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0416256318734775e-07,
+      "loss": 0.6749,
+      "step": 18464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0392144611000577e-07,
+      "loss": 0.6697,
+      "step": 18465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0368060696920756e-07,
+      "loss": 0.6293,
+      "step": 18466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0344004577171663e-07,
+      "loss": 0.3192,
+      "step": 18467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0319976252429086e-07,
+      "loss": 0.715,
+      "step": 18468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0295975723367824e-07,
+      "loss": 0.6208,
+      "step": 18469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0272002990661889e-07,
+      "loss": 0.6316,
+      "step": 18470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0248058054984633e-07,
+      "loss": 0.6881,
+      "step": 18471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0224140917008518e-07,
+      "loss": 0.6153,
+      "step": 18472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0200251577405229e-07,
+      "loss": 0.6185,
+      "step": 18473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0176390036845784e-07,
+      "loss": 0.6656,
+      "step": 18474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0152556296000315e-07,
+      "loss": 0.6608,
+      "step": 18475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0128750355538175e-07,
+      "loss": 0.6569,
+      "step": 18476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0104972216128162e-07,
+      "loss": 0.6239,
+      "step": 18477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0081221878437742e-07,
+      "loss": 0.6901,
+      "step": 18478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0057499343134269e-07,
+      "loss": 0.5607,
+      "step": 18479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0033804610883769e-07,
+      "loss": 0.6894,
+      "step": 18480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0010137682351928e-07,
+      "loss": 0.6618,
+      "step": 18481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.986498558203328e-08,
+      "loss": 0.7393,
+      "step": 18482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.962887239101882e-08,
+      "loss": 0.6479,
+      "step": 18483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.939303725710836e-08,
+      "loss": 0.661,
+      "step": 18484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.91574801869244e-08,
+      "loss": 0.6481,
+      "step": 18485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.892220118708273e-08,
+      "loss": 0.6531,
+      "step": 18486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.868720026419143e-08,
+      "loss": 0.6724,
+      "step": 18487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.845247742485076e-08,
+      "loss": 0.6437,
+      "step": 18488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.821803267565322e-08,
+      "loss": 0.5939,
+      "step": 18489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.798386602318243e-08,
+      "loss": 0.6823,
+      "step": 18490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.774997747401538e-08,
+      "loss": 0.674,
+      "step": 18491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.751636703472233e-08,
+      "loss": 0.602,
+      "step": 18492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.72830347118625e-08,
+      "loss": 0.602,
+      "step": 18493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.704998051199066e-08,
+      "loss": 0.6066,
+      "step": 18494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.681720444165155e-08,
+      "loss": 0.6022,
+      "step": 18495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.658470650738217e-08,
+      "loss": 0.6392,
+      "step": 18496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.635248671571395e-08,
+      "loss": 0.6633,
+      "step": 18497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.612054507316726e-08,
+      "loss": 0.6468,
+      "step": 18498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.588888158625687e-08,
+      "loss": 0.3121,
+      "step": 18499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.565749626148869e-08,
+      "loss": 0.2851,
+      "step": 18500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.542638910536194e-08,
+      "loss": 0.6624,
+      "step": 18501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.519556012436815e-08,
+      "loss": 0.7164,
+      "step": 18502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.496500932498986e-08,
+      "loss": 0.6017,
+      "step": 18503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.473473671370082e-08,
+      "loss": 0.6209,
+      "step": 18504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.450474229697027e-08,
+      "loss": 0.6267,
+      "step": 18505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.427502608125527e-08,
+      "loss": 0.7405,
+      "step": 18506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.404558807301067e-08,
+      "loss": 0.5825,
+      "step": 18507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.381642827867798e-08,
+      "loss": 0.5933,
+      "step": 18508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.358754670469316e-08,
+      "loss": 0.5955,
+      "step": 18509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.335894335748663e-08,
+      "loss": 0.6379,
+      "step": 18510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.313061824347658e-08,
+      "loss": 0.631,
+      "step": 18511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.290257136907788e-08,
+      "loss": 0.6885,
+      "step": 18512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.267480274069319e-08,
+      "loss": 0.6043,
+      "step": 18513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.24473123647196e-08,
+      "loss": 0.6488,
+      "step": 18514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.222010024754868e-08,
+      "loss": 0.658,
+      "step": 18515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.199316639555977e-08,
+      "loss": 0.675,
+      "step": 18516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.176651081512666e-08,
+      "loss": 0.6482,
+      "step": 18517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.154013351261537e-08,
+      "loss": 0.658,
+      "step": 18518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.131403449438303e-08,
+      "loss": 0.6674,
+      "step": 18519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.108821376678123e-08,
+      "loss": 0.6265,
+      "step": 18520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.086267133615046e-08,
+      "loss": 0.5984,
+      "step": 18521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.063740720882675e-08,
+      "loss": 0.6985,
+      "step": 18522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.041242139113726e-08,
+      "loss": 0.622,
+      "step": 18523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.018771388939917e-08,
+      "loss": 0.6637,
+      "step": 18524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.99632847099241e-08,
+      "loss": 0.6071,
+      "step": 18525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.973913385901478e-08,
+      "loss": 0.7027,
+      "step": 18526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.951526134296729e-08,
+      "loss": 0.7001,
+      "step": 18527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.929166716806992e-08,
+      "loss": 0.6475,
+      "step": 18528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.906835134060099e-08,
+      "loss": 0.6574,
+      "step": 18529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.884531386683215e-08,
+      "loss": 0.6379,
+      "step": 18530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.86225547530295e-08,
+      "loss": 0.6535,
+      "step": 18531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.840007400544804e-08,
+      "loss": 0.7794,
+      "step": 18532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.817787163033609e-08,
+      "loss": 0.7086,
+      "step": 18533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.795594763393533e-08,
+      "loss": 0.6432,
+      "step": 18534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.773430202247746e-08,
+      "loss": 0.6328,
+      "step": 18535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.751293480218858e-08,
+      "loss": 0.6318,
+      "step": 18536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.729184597928487e-08,
+      "loss": 0.6791,
+      "step": 18537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.707103555997576e-08,
+      "loss": 0.6642,
+      "step": 18538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.685050355046409e-08,
+      "loss": 0.704,
+      "step": 18539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.663024995694269e-08,
+      "loss": 0.6245,
+      "step": 18540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.641027478559661e-08,
+      "loss": 0.2912,
+      "step": 18541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.619057804260644e-08,
+      "loss": 0.6439,
+      "step": 18542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.59711597341395e-08,
+      "loss": 0.6278,
+      "step": 18543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.575201986636084e-08,
+      "loss": 0.6947,
+      "step": 18544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.553315844542332e-08,
+      "loss": 0.6617,
+      "step": 18545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.531457547747534e-08,
+      "loss": 0.672,
+      "step": 18546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.509627096865425e-08,
+      "loss": 0.6575,
+      "step": 18547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.48782449250929e-08,
+      "loss": 0.5978,
+      "step": 18548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.466049735291415e-08,
+      "loss": 0.6129,
+      "step": 18549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.444302825823092e-08,
+      "loss": 0.6515,
+      "step": 18550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.422583764715498e-08,
+      "loss": 0.6788,
+      "step": 18551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.400892552578477e-08,
+      "loss": 0.6769,
+      "step": 18552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.379229190021098e-08,
+      "loss": 0.5636,
+      "step": 18553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.357593677651988e-08,
+      "loss": 0.6252,
+      "step": 18554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.335986016078768e-08,
+      "loss": 0.292,
+      "step": 18555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.314406205908176e-08,
+      "loss": 0.6087,
+      "step": 18556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.292854247746284e-08,
+      "loss": 0.6672,
+      "step": 18557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.271330142198608e-08,
+      "loss": 0.6837,
+      "step": 18558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.24983388986933e-08,
+      "loss": 0.7064,
+      "step": 18559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.228365491362522e-08,
+      "loss": 0.7036,
+      "step": 18560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.206924947280815e-08,
+      "loss": 0.722,
+      "step": 18561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.185512258226725e-08,
+      "loss": 0.6379,
+      "step": 18562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.164127424801438e-08,
+      "loss": 0.6426,
+      "step": 18563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.142770447605585e-08,
+      "loss": 0.6649,
+      "step": 18564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.12144132723891e-08,
+      "loss": 0.62,
+      "step": 18565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.100140064300598e-08,
+      "loss": 0.6081,
+      "step": 18566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.078866659388729e-08,
+      "loss": 0.6176,
+      "step": 18567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.057621113101044e-08,
+      "loss": 0.6189,
+      "step": 18568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.036403426033957e-08,
+      "loss": 0.7159,
+      "step": 18569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.015213598783434e-08,
+      "loss": 0.6632,
+      "step": 18570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.994051631944777e-08,
+      "loss": 0.6185,
+      "step": 18571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.972917526112178e-08,
+      "loss": 0.6485,
+      "step": 18572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.951811281879273e-08,
+      "loss": 0.6947,
+      "step": 18573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.9307328998387e-08,
+      "loss": 0.7015,
+      "step": 18574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.909682380582762e-08,
+      "loss": 0.5608,
+      "step": 18575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.888659724702319e-08,
+      "loss": 0.6187,
+      "step": 18576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.867664932788011e-08,
+      "loss": 0.6638,
+      "step": 18577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.846698005429476e-08,
+      "loss": 0.5979,
+      "step": 18578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.825758943215466e-08,
+      "loss": 0.7366,
+      "step": 18579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.804847746734178e-08,
+      "loss": 0.6619,
+      "step": 18580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.783964416572809e-08,
+      "loss": 0.7029,
+      "step": 18581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.763108953318e-08,
+      "loss": 0.2852,
+      "step": 18582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.742281357555504e-08,
+      "loss": 0.6733,
+      "step": 18583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.721481629870076e-08,
+      "loss": 0.5961,
+      "step": 18584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.700709770846138e-08,
+      "loss": 0.6512,
+      "step": 18585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.67996578106689e-08,
+      "loss": 0.5827,
+      "step": 18586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.659249661114976e-08,
+      "loss": 0.6713,
+      "step": 18587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.638561411572375e-08,
+      "loss": 0.5973,
+      "step": 18588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.617901033019959e-08,
+      "loss": 0.6989,
+      "step": 18589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.597268526037926e-08,
+      "loss": 0.6427,
+      "step": 18590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.576663891205926e-08,
+      "loss": 0.6516,
+      "step": 18591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.556087129102496e-08,
+      "loss": 0.645,
+      "step": 18592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.535538240305729e-08,
+      "loss": 0.624,
+      "step": 18593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.515017225392607e-08,
+      "loss": 0.7027,
+      "step": 18594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.494524084939558e-08,
+      "loss": 0.7208,
+      "step": 18595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.474058819522013e-08,
+      "loss": 0.6667,
+      "step": 18596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.453621429714953e-08,
+      "loss": 0.6747,
+      "step": 18597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.433211916092143e-08,
+      "loss": 0.6283,
+      "step": 18598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.412830279226901e-08,
+      "loss": 0.6112,
+      "step": 18599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.392476519691772e-08,
+      "loss": 0.6605,
+      "step": 18600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.372150638058184e-08,
+      "loss": 0.6547,
+      "step": 18601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.351852634897127e-08,
+      "loss": 0.6342,
+      "step": 18602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.331582510778701e-08,
+      "loss": 0.6443,
+      "step": 18603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.311340266272116e-08,
+      "loss": 0.6852,
+      "step": 18604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.291125901946027e-08,
+      "loss": 0.6229,
+      "step": 18605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.270939418367984e-08,
+      "loss": 0.656,
+      "step": 18606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.250780816105086e-08,
+      "loss": 0.7405,
+      "step": 18607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.230650095723436e-08,
+      "loss": 0.7009,
+      "step": 18608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.210547257788358e-08,
+      "loss": 0.6945,
+      "step": 18609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.190472302864515e-08,
+      "loss": 0.5903,
+      "step": 18610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.170425231515676e-08,
+      "loss": 0.6178,
+      "step": 18611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.150406044304947e-08,
+      "loss": 0.6381,
+      "step": 18612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.130414741794656e-08,
+      "loss": 0.5903,
+      "step": 18613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.11045132454613e-08,
+      "loss": 0.7025,
+      "step": 18614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.090515793119924e-08,
+      "loss": 0.6424,
+      "step": 18615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.070608148076252e-08,
+      "loss": 0.6116,
+      "step": 18616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.050728389974004e-08,
+      "loss": 0.6116,
+      "step": 18617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.03087651937151e-08,
+      "loss": 0.6653,
+      "step": 18618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.011052536826435e-08,
+      "loss": 0.6353,
+      "step": 18619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.991256442895556e-08,
+      "loss": 0.6756,
+      "step": 18620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.971488238134761e-08,
+      "loss": 0.6324,
+      "step": 18621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.951747923099272e-08,
+      "loss": 0.6548,
+      "step": 18622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.932035498343647e-08,
+      "loss": 0.6343,
+      "step": 18623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.912350964421221e-08,
+      "loss": 0.6888,
+      "step": 18624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.892694321885108e-08,
+      "loss": 0.5926,
+      "step": 18625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.873065571287307e-08,
+      "loss": 0.6485,
+      "step": 18626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.853464713179159e-08,
+      "loss": 0.6922,
+      "step": 18627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.833891748110999e-08,
+      "loss": 0.667,
+      "step": 18628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.81434667663261e-08,
+      "loss": 0.6768,
+      "step": 18629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.794829499292998e-08,
+      "loss": 0.6564,
+      "step": 18630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.775340216640169e-08,
+      "loss": 0.635,
+      "step": 18631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.755878829221574e-08,
+      "loss": 0.6054,
+      "step": 18632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.736445337583996e-08,
+      "loss": 0.576,
+      "step": 18633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.71703974227289e-08,
+      "loss": 0.6534,
+      "step": 18634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.697662043833485e-08,
+      "loss": 0.6635,
+      "step": 18635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.678312242809903e-08,
+      "loss": 0.6325,
+      "step": 18636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.658990339745708e-08,
+      "loss": 0.6095,
+      "step": 18637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.639696335183465e-08,
+      "loss": 0.6633,
+      "step": 18638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.620430229665186e-08,
+      "loss": 0.6271,
+      "step": 18639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.601192023731773e-08,
+      "loss": 0.6757,
+      "step": 18640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.58198171792368e-08,
+      "loss": 0.6798,
+      "step": 18641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.562799312780476e-08,
+      "loss": 0.6933,
+      "step": 18642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.543644808840732e-08,
+      "loss": 0.5951,
+      "step": 18643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.524518206642571e-08,
+      "loss": 0.6119,
+      "step": 18644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.50541950672312e-08,
+      "loss": 0.6244,
+      "step": 18645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.48634870961884e-08,
+      "loss": 0.619,
+      "step": 18646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.467305815865188e-08,
+      "loss": 0.618,
+      "step": 18647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.448290825997072e-08,
+      "loss": 0.6962,
+      "step": 18648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.429303740548621e-08,
+      "loss": 0.7006,
+      "step": 18649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.410344560053072e-08,
+      "loss": 0.6343,
+      "step": 18650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.391413285042669e-08,
+      "loss": 0.6096,
+      "step": 18651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.37250991604954e-08,
+      "loss": 0.6475,
+      "step": 18652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.353634453604263e-08,
+      "loss": 0.7157,
+      "step": 18653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.334786898237078e-08,
+      "loss": 0.5856,
+      "step": 18654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.315967250477228e-08,
+      "loss": 0.6568,
+      "step": 18655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.297175510853404e-08,
+      "loss": 0.6202,
+      "step": 18656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.278411679893292e-08,
+      "loss": 0.6577,
+      "step": 18657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.259675758124029e-08,
+      "loss": 0.6206,
+      "step": 18658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.240967746071635e-08,
+      "loss": 0.2969,
+      "step": 18659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.222287644261582e-08,
+      "loss": 0.6182,
+      "step": 18660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.20363545321856e-08,
+      "loss": 0.7021,
+      "step": 18661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.185011173466482e-08,
+      "loss": 0.3214,
+      "step": 18662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.166414805528265e-08,
+      "loss": 0.6537,
+      "step": 18663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.147846349926379e-08,
+      "loss": 0.6418,
+      "step": 18664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.129305807182185e-08,
+      "loss": 0.6803,
+      "step": 18665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.110793177816488e-08,
+      "loss": 0.589,
+      "step": 18666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.092308462349094e-08,
+      "loss": 0.6141,
+      "step": 18667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.073851661299368e-08,
+      "loss": 0.6839,
+      "step": 18668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.055422775185449e-08,
+      "loss": 0.6569,
+      "step": 18669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.037021804525145e-08,
+      "loss": 0.6341,
+      "step": 18670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.018648749834932e-08,
+      "loss": 0.6302,
+      "step": 18671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.000303611631287e-08,
+      "loss": 0.6896,
+      "step": 18672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.981986390429128e-08,
+      "loss": 0.6027,
+      "step": 18673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.963697086742937e-08,
+      "loss": 0.6646,
+      "step": 18674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9454357010864106e-08,
+      "loss": 0.646,
+      "step": 18675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9272022339723624e-08,
+      "loss": 0.6558,
+      "step": 18676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.908996685913049e-08,
+      "loss": 0.6923,
+      "step": 18677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.890819057419728e-08,
+      "loss": 0.7334,
+      "step": 18678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8726693490028795e-08,
+      "loss": 0.6483,
+      "step": 18679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.854547561172097e-08,
+      "loss": 0.6562,
+      "step": 18680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.836453694436639e-08,
+      "loss": 0.6895,
+      "step": 18681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8183877493045435e-08,
+      "loss": 0.6359,
+      "step": 18682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.800349726283183e-08,
+      "loss": 0.6365,
+      "step": 18683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.782339625879152e-08,
+      "loss": 0.7253,
+      "step": 18684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.76435744859849e-08,
+      "loss": 0.6175,
+      "step": 18685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.746403194945904e-08,
+      "loss": 0.6241,
+      "step": 18686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.72847686542577e-08,
+      "loss": 0.5999,
+      "step": 18687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.710578460541571e-08,
+      "loss": 0.6987,
+      "step": 18688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.69270798079613e-08,
+      "loss": 0.7081,
+      "step": 18689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.674865426691156e-08,
+      "loss": 0.3243,
+      "step": 18690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6570507987278034e-08,
+      "loss": 0.6232,
+      "step": 18691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6392640974064495e-08,
+      "loss": 0.5689,
+      "step": 18692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.621505323226695e-08,
+      "loss": 0.6272,
+      "step": 18693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6037744766872515e-08,
+      "loss": 0.6453,
+      "step": 18694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5860715582859436e-08,
+      "loss": 0.6774,
+      "step": 18695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.568396568520262e-08,
+      "loss": 0.6597,
+      "step": 18696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5507495078864774e-08,
+      "loss": 0.6488,
+      "step": 18697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.533130376880191e-08,
+      "loss": 0.5826,
+      "step": 18698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5155391759962317e-08,
+      "loss": 0.6316,
+      "step": 18699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.497975905728537e-08,
+      "loss": 0.6208,
+      "step": 18700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.480440566570711e-08,
+      "loss": 0.6601,
+      "step": 18701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.462933159015027e-08,
+      "loss": 0.687,
+      "step": 18702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.445453683553204e-08,
+      "loss": 0.6195,
+      "step": 18703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.428002140676292e-08,
+      "loss": 0.6148,
+      "step": 18704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4105785308742334e-08,
+      "loss": 0.6698,
+      "step": 18705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.393182854636525e-08,
+      "loss": 0.6107,
+      "step": 18706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3758151124515546e-08,
+      "loss": 0.6372,
+      "step": 18707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3584753048073756e-08,
+      "loss": 0.6604,
+      "step": 18708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.34116343219071e-08,
+      "loss": 0.6587,
+      "step": 18709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.323879495087947e-08,
+      "loss": 0.6639,
+      "step": 18710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.306623493984475e-08,
+      "loss": 0.6684,
+      "step": 18711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2893954293649073e-08,
+      "loss": 0.624,
+      "step": 18712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.272195301713079e-08,
+      "loss": 0.6275,
+      "step": 18713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2550231115121586e-08,
+      "loss": 0.6505,
+      "step": 18714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2378788592443167e-08,
+      "loss": 0.682,
+      "step": 18715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.220762545391167e-08,
+      "loss": 0.6429,
+      "step": 18716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.203674170433437e-08,
+      "loss": 0.6587,
+      "step": 18717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1866137348510757e-08,
+      "loss": 0.6735,
+      "step": 18718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.169581239123034e-08,
+      "loss": 0.6337,
+      "step": 18719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.152576683727817e-08,
+      "loss": 0.6567,
+      "step": 18720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.135600069143043e-08,
+      "loss": 0.6786,
+      "step": 18721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.118651395845553e-08,
+      "loss": 0.6652,
+      "step": 18722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.101730664311189e-08,
+      "loss": 0.6498,
+      "step": 18723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.084837875015347e-08,
+      "loss": 0.6772,
+      "step": 18724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.067973028432316e-08,
+      "loss": 0.6808,
+      "step": 18725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.051136125035827e-08,
+      "loss": 0.6865,
+      "step": 18726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.034327165298836e-08,
+      "loss": 0.6587,
+      "step": 18727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.017546149693298e-08,
+      "loss": 0.6923,
+      "step": 18728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.000793078690613e-08,
+      "loss": 0.5389,
+      "step": 18729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.984067952761184e-08,
+      "loss": 0.6059,
+      "step": 18730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9673707723748574e-08,
+      "loss": 0.6455,
+      "step": 18731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.950701538000591e-08,
+      "loss": 0.6606,
+      "step": 18732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.934060250106454e-08,
+      "loss": 0.6789,
+      "step": 18733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.917446909159851e-08,
+      "loss": 0.7057,
+      "step": 18734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9008615156274083e-08,
+      "loss": 0.6096,
+      "step": 18735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.884304069974977e-08,
+      "loss": 0.2961,
+      "step": 18736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8677745726675166e-08,
+      "loss": 0.6272,
+      "step": 18737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8512730241692115e-08,
+      "loss": 0.6555,
+      "step": 18738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8347994249436924e-08,
+      "loss": 0.6006,
+      "step": 18739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.818353775453477e-08,
+      "loss": 0.6479,
+      "step": 18740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8019360761604186e-08,
+      "loss": 0.66,
+      "step": 18741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.785546327525703e-08,
+      "loss": 0.6448,
+      "step": 18742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.769184530009741e-08,
+      "loss": 0.6349,
+      "step": 18743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.752850684071941e-08,
+      "loss": 0.617,
+      "step": 18744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.736544790170938e-08,
+      "loss": 0.629,
+      "step": 18745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7202668487649205e-08,
+      "loss": 0.6373,
+      "step": 18746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.704016860310967e-08,
+      "loss": 0.6109,
+      "step": 18747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.68779482526549e-08,
+      "loss": 0.6828,
+      "step": 18748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.671600744083904e-08,
+      "loss": 0.6612,
+      "step": 18749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655434617221288e-08,
+      "loss": 0.644,
+      "step": 18750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6392964451316134e-08,
+      "loss": 0.5901,
+      "step": 18751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.623186228268073e-08,
+      "loss": 0.6091,
+      "step": 18752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6071039670830816e-08,
+      "loss": 0.6406,
+      "step": 18753
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.591049662028502e-08,
+      "loss": 0.6529,
+      "step": 18754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.575023313554972e-08,
+      "loss": 0.7604,
+      "step": 18755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.559024922112798e-08,
+      "loss": 0.6843,
+      "step": 18756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5430544881513996e-08,
+      "loss": 0.6655,
+      "step": 18757
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.527112012118973e-08,
+      "loss": 0.6703,
+      "step": 18758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.511197494463493e-08,
+      "loss": 0.6823,
+      "step": 18759
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4953109356319355e-08,
+      "loss": 0.5951,
+      "step": 18760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.479452336070278e-08,
+      "loss": 0.6395,
+      "step": 18761
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.463621696224274e-08,
+      "loss": 0.6053,
+      "step": 18762
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.447819016538235e-08,
+      "loss": 0.6397,
+      "step": 18763
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.43204429745614e-08,
+      "loss": 0.6716,
+      "step": 18764
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.416297539420966e-08,
+      "loss": 0.6945,
+      "step": 18765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.400578742875028e-08,
+      "loss": 0.6319,
+      "step": 18766
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.384887908259638e-08,
+      "loss": 0.6484,
+      "step": 18767
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3692250360156674e-08,
+      "loss": 0.6815,
+      "step": 18768
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.353590126582985e-08,
+      "loss": 0.6229,
+      "step": 18769
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3379831804005736e-08,
+      "loss": 0.6534,
+      "step": 18770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.322404197906971e-08,
+      "loss": 0.5604,
+      "step": 18771
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.306853179539494e-08,
+      "loss": 0.6964,
+      "step": 18772
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.291330125735127e-08,
+      "loss": 0.6172,
+      "step": 18773
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.275835036929632e-08,
+      "loss": 0.6668,
+      "step": 18774
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2603679135583276e-08,
+      "loss": 0.6636,
+      "step": 18775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.244928756055644e-08,
+      "loss": 0.7328,
+      "step": 18776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.229517564855012e-08,
+      "loss": 0.6174,
+      "step": 18777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2141343403894195e-08,
+      "loss": 0.639,
+      "step": 18778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1987790830909646e-08,
+      "loss": 0.6726,
+      "step": 18779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.183451793390747e-08,
+      "loss": 0.6509,
+      "step": 18780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.168152471719311e-08,
+      "loss": 0.6462,
+      "step": 18781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1528811185063137e-08,
+      "loss": 0.7031,
+      "step": 18782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.137637734180855e-08,
+      "loss": 0.5968,
+      "step": 18783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.122422319170705e-08,
+      "loss": 0.6449,
+      "step": 18784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1072348739035205e-08,
+      "loss": 0.6457,
+      "step": 18785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.092075398805628e-08,
+      "loss": 0.627,
+      "step": 18786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.07694389430302e-08,
+      "loss": 0.6281,
+      "step": 18787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.061840360820357e-08,
+      "loss": 0.6366,
+      "step": 18788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.046764798782077e-08,
+      "loss": 0.6944,
+      "step": 18789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.031717208611507e-08,
+      "loss": 0.6751,
+      "step": 18790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.016697590731311e-08,
+      "loss": 0.6754,
+      "step": 18791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.001705945563261e-08,
+      "loss": 0.6653,
+      "step": 18792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9867422735283544e-08,
+      "loss": 0.536,
+      "step": 18793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.971806575047033e-08,
+      "loss": 0.683,
+      "step": 18794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.956898850538626e-08,
+      "loss": 0.6081,
+      "step": 18795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.942019100421912e-08,
+      "loss": 0.6034,
+      "step": 18796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9271673251146666e-08,
+      "loss": 0.6768,
+      "step": 18797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9123435250341126e-08,
+      "loss": 0.7404,
+      "step": 18798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8975477005965825e-08,
+      "loss": 0.6742,
+      "step": 18799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.882779852217522e-08,
+      "loss": 0.6605,
+      "step": 18800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.868039980311933e-08,
+      "loss": 0.2963,
+      "step": 18801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.853328085293484e-08,
+      "loss": 0.6227,
+      "step": 18802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8386441675756224e-08,
+      "loss": 0.6453,
+      "step": 18803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.823988227570574e-08,
+      "loss": 0.6178,
+      "step": 18804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8093602656901204e-08,
+      "loss": 0.5712,
+      "step": 18805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7947602823449335e-08,
+      "loss": 0.6219,
+      "step": 18806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.78018827794524e-08,
+      "loss": 0.6444,
+      "step": 18807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.765644252900047e-08,
+      "loss": 0.6183,
+      "step": 18808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.751128207618138e-08,
+      "loss": 0.6251,
+      "step": 18809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7366401425069645e-08,
+      "loss": 0.6118,
+      "step": 18810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7221800579735346e-08,
+      "loss": 0.6576,
+      "step": 18811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.707747954423968e-08,
+      "loss": 0.7462,
+      "step": 18812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.693343832263607e-08,
+      "loss": 0.5564,
+      "step": 18813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.678967691896906e-08,
+      "loss": 0.6492,
+      "step": 18814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.664619533727654e-08,
+      "loss": 0.6619,
+      "step": 18815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.650299358158971e-08,
+      "loss": 0.6777,
+      "step": 18816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.636007165592981e-08,
+      "loss": 0.679,
+      "step": 18817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.621742956430918e-08,
+      "loss": 0.6413,
+      "step": 18818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.607506731073573e-08,
+      "loss": 0.6446,
+      "step": 18819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.593298489920738e-08,
+      "loss": 0.665,
+      "step": 18820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.579118233371426e-08,
+      "loss": 0.6662,
+      "step": 18821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.564965961823985e-08,
+      "loss": 0.6469,
+      "step": 18822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.550841675675765e-08,
+      "loss": 0.3137,
+      "step": 18823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.536745375323558e-08,
+      "loss": 0.6751,
+      "step": 18824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.522677061163271e-08,
+      "loss": 0.6866,
+      "step": 18825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.508636733589921e-08,
+      "loss": 0.6974,
+      "step": 18826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4946243929978585e-08,
+      "loss": 0.6087,
+      "step": 18827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4806400397807696e-08,
+      "loss": 0.7357,
+      "step": 18828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.466683674331228e-08,
+      "loss": 0.6374,
+      "step": 18829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4527552970412547e-08,
+      "loss": 0.6415,
+      "step": 18830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.438854908302092e-08,
+      "loss": 0.6484,
+      "step": 18831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4249825085040936e-08,
+      "loss": 0.7089,
+      "step": 18832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.411138098036948e-08,
+      "loss": 0.6265,
+      "step": 18833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3973216772894554e-08,
+      "loss": 0.6363,
+      "step": 18834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.383533246649529e-08,
+      "loss": 0.6356,
+      "step": 18835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.369772806504523e-08,
+      "loss": 0.6637,
+      "step": 18836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.356040357241019e-08,
+      "loss": 0.741,
+      "step": 18837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.342335899244486e-08,
+      "loss": 0.662,
+      "step": 18838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.32865943289995e-08,
+      "loss": 0.6812,
+      "step": 18839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3150109585914356e-08,
+      "loss": 0.5918,
+      "step": 18840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.301390476702415e-08,
+      "loss": 0.6635,
+      "step": 18841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2877979876152486e-08,
+      "loss": 0.6564,
+      "step": 18842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2742334917118535e-08,
+      "loss": 0.6937,
+      "step": 18843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.260696989373036e-08,
+      "loss": 0.6062,
+      "step": 18844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.247188480979158e-08,
+      "loss": 0.6906,
+      "step": 18845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.233707966909361e-08,
+      "loss": 0.643,
+      "step": 18846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.220255447542564e-08,
+      "loss": 0.6574,
+      "step": 18847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2068309232563545e-08,
+      "loss": 0.6095,
+      "step": 18848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.193434394427874e-08,
+      "loss": 0.6207,
+      "step": 18849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.180065861433268e-08,
+      "loss": 0.6802,
+      "step": 18850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1667253246481234e-08,
+      "loss": 0.6529,
+      "step": 18851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1534127844470296e-08,
+      "loss": 0.624,
+      "step": 18852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.14012824120391e-08,
+      "loss": 0.6495,
+      "step": 18853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.126871695291911e-08,
+      "loss": 0.6863,
+      "step": 18854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1136431470832896e-08,
+      "loss": 0.671,
+      "step": 18855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.100442596949638e-08,
+      "loss": 0.6876,
+      "step": 18856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.087270045261548e-08,
+      "loss": 0.6084,
+      "step": 18857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.074125492389168e-08,
+      "loss": 0.6719,
+      "step": 18858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0610089387015375e-08,
+      "loss": 0.6631,
+      "step": 18859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.047920384567138e-08,
+      "loss": 0.7068,
+      "step": 18860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0348598303534535e-08,
+      "loss": 0.6839,
+      "step": 18861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0218272764274136e-08,
+      "loss": 0.6806,
+      "step": 18862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0088227231549473e-08,
+      "loss": 0.6333,
+      "step": 18863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.995846170901428e-08,
+      "loss": 0.641,
+      "step": 18864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.982897620031122e-08,
+      "loss": 0.599,
+      "step": 18865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.969977070907848e-08,
+      "loss": 0.6307,
+      "step": 18866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.957084523894427e-08,
+      "loss": 0.6339,
+      "step": 18867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9442199793530135e-08,
+      "loss": 0.6411,
+      "step": 18868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9313834376447637e-08,
+      "loss": 0.5848,
+      "step": 18869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9185748991302775e-08,
+      "loss": 0.6768,
+      "step": 18870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9057943641693787e-08,
+      "loss": 0.7073,
+      "step": 18871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.893041833120891e-08,
+      "loss": 0.5286,
+      "step": 18872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8803173063430835e-08,
+      "loss": 0.6765,
+      "step": 18873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8676207841933367e-08,
+      "loss": 0.7466,
+      "step": 18874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8549522670281438e-08,
+      "loss": 0.6236,
+      "step": 18875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8423117552032197e-08,
+      "loss": 0.6931,
+      "step": 18876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.829699249073836e-08,
+      "loss": 0.7253,
+      "step": 18877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8171147489940432e-08,
+      "loss": 0.577,
+      "step": 18878
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8045582553173356e-08,
+      "loss": 0.6057,
+      "step": 18879
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7920297683963203e-08,
+      "loss": 0.5832,
+      "step": 18880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7795292885829384e-08,
+      "loss": 0.6706,
+      "step": 18881
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.767056816228242e-08,
+      "loss": 0.6276,
+      "step": 18882
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7546123516826172e-08,
+      "loss": 0.6632,
+      "step": 18883
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.742195895295341e-08,
+      "loss": 0.6835,
+      "step": 18884
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7298074474153557e-08,
+      "loss": 0.6522,
+      "step": 18885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7174470083904946e-08,
+      "loss": 0.5723,
+      "step": 18886
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7051145785680354e-08,
+      "loss": 0.6767,
+      "step": 18887
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.692810158294146e-08,
+      "loss": 0.6658,
+      "step": 18888
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.680533747914438e-08,
+      "loss": 0.6331,
+      "step": 18889
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.668285347773858e-08,
+      "loss": 0.6135,
+      "step": 18890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6560649582161314e-08,
+      "loss": 0.658,
+      "step": 18891
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.643872579584872e-08,
+      "loss": 0.546,
+      "step": 18892
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6317082122221395e-08,
+      "loss": 0.6658,
+      "step": 18893
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6195718564698825e-08,
+      "loss": 0.6423,
+      "step": 18894
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6074635126686066e-08,
+      "loss": 0.6763,
+      "step": 18895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.595383181158706e-08,
+      "loss": 0.6,
+      "step": 18896
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.583330862279243e-08,
+      "loss": 0.724,
+      "step": 18897
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5713065563689465e-08,
+      "loss": 0.6922,
+      "step": 18898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5593102637652136e-08,
+      "loss": 0.6694,
+      "step": 18899
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5473419848052183e-08,
+      "loss": 0.7067,
+      "step": 18900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5354017198250257e-08,
+      "loss": 0.6092,
+      "step": 18901
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.523489469159923e-08,
+      "loss": 0.6776,
+      "step": 18902
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5116052331445317e-08,
+      "loss": 0.6688,
+      "step": 18903
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4997490121125843e-08,
+      "loss": 0.6299,
+      "step": 18904
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4879208063971483e-08,
+      "loss": 0.6704,
+      "step": 18905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.476120616330402e-08,
+      "loss": 0.6696,
+      "step": 18906
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4643484422436358e-08,
+      "loss": 0.6787,
+      "step": 18907
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4526042844676966e-08,
+      "loss": 0.6316,
+      "step": 18908
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4408881433322096e-08,
+      "loss": 0.6516,
+      "step": 18909
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4292000191663555e-08,
+      "loss": 0.6265,
+      "step": 18910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4175399122983167e-08,
+      "loss": 0.6534,
+      "step": 18911
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.40590782305572e-08,
+      "loss": 0.6831,
+      "step": 18912
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3943037517650812e-08,
+      "loss": 0.6254,
+      "step": 18913
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3827276987524738e-08,
+      "loss": 0.6326,
+      "step": 18914
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3711796643428596e-08,
+      "loss": 0.6895,
+      "step": 18915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.359659648860646e-08,
+      "loss": 0.6441,
+      "step": 18916
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3481676526293517e-08,
+      "loss": 0.7156,
+      "step": 18917
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3367036759718298e-08,
+      "loss": 0.6645,
+      "step": 18918
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3252677192099336e-08,
+      "loss": 0.6208,
+      "step": 18919
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3138597826649622e-08,
+      "loss": 0.6462,
+      "step": 18920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3024798666572146e-08,
+      "loss": 0.6505,
+      "step": 18921
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.291127971506324e-08,
+      "loss": 0.6648,
+      "step": 18922
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2798040975310353e-08,
+      "loss": 0.6653,
+      "step": 18923
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.26850824504965e-08,
+      "loss": 0.6446,
+      "step": 18924
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.257240414379136e-08,
+      "loss": 0.6411,
+      "step": 18925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2460006058361294e-08,
+      "loss": 0.5951,
+      "step": 18926
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2347888197361555e-08,
+      "loss": 0.6727,
+      "step": 18927
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2236050563942957e-08,
+      "loss": 0.6432,
+      "step": 18928
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.21244931612441e-08,
+      "loss": 0.6929,
+      "step": 18929
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.201321599240025e-08,
+      "loss": 0.6617,
+      "step": 18930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.190221906053558e-08,
+      "loss": 0.6998,
+      "step": 18931
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1791502368768702e-08,
+      "loss": 0.6787,
+      "step": 18932
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1681065920207133e-08,
+      "loss": 0.6028,
+      "step": 18933
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1570909717955058e-08,
+      "loss": 0.6483,
+      "step": 18934
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1461033765103335e-08,
+      "loss": 0.6494,
+      "step": 18935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1351438064740605e-08,
+      "loss": 0.6891,
+      "step": 18936
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1242122619943297e-08,
+      "loss": 0.6644,
+      "step": 18937
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1133087433782284e-08,
+      "loss": 0.5819,
+      "step": 18938
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1024332509319568e-08,
+      "loss": 0.6108,
+      "step": 18939
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0915857849609368e-08,
+      "loss": 0.6373,
+      "step": 18940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0807663457699244e-08,
+      "loss": 0.6706,
+      "step": 18941
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0699749336626775e-08,
+      "loss": 0.666,
+      "step": 18942
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.059211548942286e-08,
+      "loss": 0.6767,
+      "step": 18943
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.048476191911064e-08,
+      "loss": 0.663,
+      "step": 18944
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0377688628706594e-08,
+      "loss": 0.6877,
+      "step": 18945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027089562121609e-08,
+      "loss": 0.7108,
+      "step": 18946
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0164382899638958e-08,
+      "loss": 0.6106,
+      "step": 18947
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0058150466966132e-08,
+      "loss": 0.6899,
+      "step": 18948
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9952198326183004e-08,
+      "loss": 0.6135,
+      "step": 18949
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9846526480263863e-08,
+      "loss": 0.6463,
+      "step": 18950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9741134932176333e-08,
+      "loss": 0.6532,
+      "step": 18951
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.963602368488138e-08,
+      "loss": 0.7067,
+      "step": 18952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9531192741329975e-08,
+      "loss": 0.6678,
+      "step": 18953
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9426642104467542e-08,
+      "loss": 0.632,
+      "step": 18954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9322371777230618e-08,
+      "loss": 0.6119,
+      "step": 18955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9218381762545758e-08,
+      "loss": 0.6301,
+      "step": 18956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.911467206333506e-08,
+      "loss": 0.6844,
+      "step": 18957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9011242682511754e-08,
+      "loss": 0.5936,
+      "step": 18958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8908093622979074e-08,
+      "loss": 0.5759,
+      "step": 18959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8805224887634698e-08,
+      "loss": 0.6635,
+      "step": 18960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.870263647936854e-08,
+      "loss": 0.7051,
+      "step": 18961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860032840106163e-08,
+      "loss": 0.6299,
+      "step": 18962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.849830065558611e-08,
+      "loss": 0.6712,
+      "step": 18963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.839655324580858e-08,
+      "loss": 0.6973,
+      "step": 18964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.829508617458786e-08,
+      "loss": 0.6438,
+      "step": 18965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.819389944477057e-08,
+      "loss": 0.6045,
+      "step": 18966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.809299305920109e-08,
+      "loss": 0.6546,
+      "step": 18967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.799236702071161e-08,
+      "loss": 0.6826,
+      "step": 18968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.789202133212986e-08,
+      "loss": 0.6768,
+      "step": 18969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7791955996273592e-08,
+      "loss": 0.6536,
+      "step": 18970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7692171015953884e-08,
+      "loss": 0.704,
+      "step": 18971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7592666393971836e-08,
+      "loss": 0.5975,
+      "step": 18972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7493442133121874e-08,
+      "loss": 0.6721,
+      "step": 18973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.739449823619288e-08,
+      "loss": 0.636,
+      "step": 18974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7295834705961524e-08,
+      "loss": 0.6909,
+      "step": 18975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7197451545201137e-08,
+      "loss": 0.7113,
+      "step": 18976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.709934875667285e-08,
+      "loss": 0.6727,
+      "step": 18977
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7001526343132236e-08,
+      "loss": 0.6296,
+      "step": 18978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6903984307327093e-08,
+      "loss": 0.6618,
+      "step": 18979
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6806722651997454e-08,
+      "loss": 0.6835,
+      "step": 18980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6709741379873358e-08,
+      "loss": 0.6073,
+      "step": 18981
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.66130404936804e-08,
+      "loss": 0.7058,
+      "step": 18982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6516619996134188e-08,
+      "loss": 0.6652,
+      "step": 18983
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6420479889941444e-08,
+      "loss": 0.6427,
+      "step": 18984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.632462017780445e-08,
+      "loss": 0.6551,
+      "step": 18985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.622904086241328e-08,
+      "loss": 0.6563,
+      "step": 18986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.613374194645245e-08,
+      "loss": 0.6796,
+      "step": 18987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6038723432599823e-08,
+      "loss": 0.6297,
+      "step": 18988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.594398532352326e-08,
+      "loss": 0.6661,
+      "step": 18989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5849527621883964e-08,
+      "loss": 0.6859,
+      "step": 18990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5755350330335372e-08,
+      "loss": 0.5919,
+      "step": 18991
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5661453451520924e-08,
+      "loss": 0.6572,
+      "step": 18992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.556783698807851e-08,
+      "loss": 0.6573,
+      "step": 18993
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.547450094263714e-08,
+      "loss": 0.3345,
+      "step": 18994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5381445317819153e-08,
+      "loss": 0.5895,
+      "step": 18995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5288670116236915e-08,
+      "loss": 0.6136,
+      "step": 18996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.519617534049722e-08,
+      "loss": 0.6216,
+      "step": 18997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5103960993196888e-08,
+      "loss": 0.6493,
+      "step": 18998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.501202707692606e-08,
+      "loss": 0.6777,
+      "step": 18999
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4920373594267123e-08,
+      "loss": 0.6011,
+      "step": 19000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4829000547793571e-08,
+      "loss": 0.6721,
+      "step": 19001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4737907940071128e-08,
+      "loss": 0.6563,
+      "step": 19002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.464709577365997e-08,
+      "loss": 0.6064,
+      "step": 19003
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4556564051110278e-08,
+      "loss": 0.6708,
+      "step": 19004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4466312774963353e-08,
+      "loss": 0.6844,
+      "step": 19005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4376341947754946e-08,
+      "loss": 0.7144,
+      "step": 19006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4286651572010813e-08,
+      "loss": 0.6176,
+      "step": 19007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4197241650251158e-08,
+      "loss": 0.6392,
+      "step": 19008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4108112184987311e-08,
+      "loss": 0.641,
+      "step": 19009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.401926317872171e-08,
+      "loss": 0.6519,
+      "step": 19010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3930694633950137e-08,
+      "loss": 0.6387,
+      "step": 19011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3842406553159493e-08,
+      "loss": 0.6364,
+      "step": 19012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3754398938830015e-08,
+      "loss": 0.6805,
+      "step": 19013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3666671793433062e-08,
+      "loss": 0.5589,
+      "step": 19014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3579225119433326e-08,
+      "loss": 0.6629,
+      "step": 19015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.349205891928551e-08,
+      "loss": 0.6471,
+      "step": 19016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.340517319543877e-08,
+      "loss": 0.6785,
+      "step": 19017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3318567950332262e-08,
+      "loss": 0.6973,
+      "step": 19018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3232243186399596e-08,
+      "loss": 0.7304,
+      "step": 19019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.314619890606439e-08,
+      "loss": 0.6151,
+      "step": 19020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3060435111743596e-08,
+      "loss": 0.6409,
+      "step": 19021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2974951805845293e-08,
+      "loss": 0.6725,
+      "step": 19022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2889748990771999e-08,
+      "loss": 0.6948,
+      "step": 19023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2804826668915139e-08,
+      "loss": 0.65,
+      "step": 19024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2720184842659466e-08,
+      "loss": 0.5703,
+      "step": 19025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2635823514384194e-08,
+      "loss": 0.7117,
+      "step": 19026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2551742686456313e-08,
+      "loss": 0.6473,
+      "step": 19027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2467942361239494e-08,
+      "loss": 0.6924,
+      "step": 19028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2384422541085183e-08,
+      "loss": 0.5766,
+      "step": 19029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2301183228340397e-08,
+      "loss": 0.6533,
+      "step": 19030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2218224425343262e-08,
+      "loss": 0.6292,
+      "step": 19031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2135546134421916e-08,
+      "loss": 0.6826,
+      "step": 19032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2053148357900057e-08,
+      "loss": 0.5955,
+      "step": 19033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.197103109809139e-08,
+      "loss": 0.6573,
+      "step": 19034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1889194357302957e-08,
+      "loss": 0.641,
+      "step": 19035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1807638137831811e-08,
+      "loss": 0.6728,
+      "step": 19036
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1726362441969452e-08,
+      "loss": 0.6282,
+      "step": 19037
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1645367271998498e-08,
+      "loss": 0.7104,
+      "step": 19038
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1564652630192686e-08,
+      "loss": 0.6318,
+      "step": 19039
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1484218518820201e-08,
+      "loss": 0.5829,
+      "step": 19040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1404064940140347e-08,
+      "loss": 0.7032,
+      "step": 19041
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1324191896402437e-08,
+      "loss": 0.6653,
+      "step": 19042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1244599389851341e-08,
+      "loss": 0.6656,
+      "step": 19043
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1165287422721938e-08,
+      "loss": 0.7064,
+      "step": 19044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1086255997242446e-08,
+      "loss": 0.6263,
+      "step": 19045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.100750511563109e-08,
+      "loss": 0.6633,
+      "step": 19046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0929034780100545e-08,
+      "loss": 0.6694,
+      "step": 19047
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0850844992854603e-08,
+      "loss": 0.2937,
+      "step": 19048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0772935756089287e-08,
+      "loss": 0.6448,
+      "step": 19049
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0695307071991733e-08,
+      "loss": 0.6817,
+      "step": 19050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0617958942744644e-08,
+      "loss": 0.6261,
+      "step": 19051
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0540891370517392e-08,
+      "loss": 0.6548,
+      "step": 19052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0464104357477133e-08,
+      "loss": 0.6581,
+      "step": 19053
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0387597905778812e-08,
+      "loss": 0.6913,
+      "step": 19054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0311372017570709e-08,
+      "loss": 0.6666,
+      "step": 19055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0235426694995555e-08,
+      "loss": 0.676,
+      "step": 19056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0159761940184975e-08,
+      "loss": 0.6478,
+      "step": 19057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0084377755265051e-08,
+      "loss": 0.6299,
+      "step": 19058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0009274142350756e-08,
+      "loss": 0.6273,
+      "step": 19059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.934451103553733e-09,
+      "loss": 0.662,
+      "step": 19060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.859908640975635e-09,
+      "loss": 0.6726,
+      "step": 19061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.785646756707012e-09,
+      "loss": 0.277,
+      "step": 19062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.711665452837305e-09,
+      "loss": 0.5369,
+      "step": 19063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.637964731441518e-09,
+      "loss": 0.6868,
+      "step": 19064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.56454459459133e-09,
+      "loss": 0.6414,
+      "step": 19065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.491405044348422e-09,
+      "loss": 0.5926,
+      "step": 19066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.4185460827656e-09,
+      "loss": 0.6674,
+      "step": 19067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.345967711891224e-09,
+      "loss": 0.6629,
+      "step": 19068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.273669933762553e-09,
+      "loss": 0.6306,
+      "step": 19069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.201652750409073e-09,
+      "loss": 0.6434,
+      "step": 19070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.129916163855835e-09,
+      "loss": 0.6133,
+      "step": 19071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.058460176115668e-09,
+      "loss": 0.7184,
+      "step": 19072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.987284789195861e-09,
+      "loss": 0.6816,
+      "step": 19073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.916390005095921e-09,
+      "loss": 0.6945,
+      "step": 19074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.84577582580648e-09,
+      "loss": 0.6133,
+      "step": 19075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.775442253311505e-09,
+      "loss": 0.657,
+      "step": 19076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.705389289586086e-09,
+      "loss": 0.7064,
+      "step": 19077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.635616936597535e-09,
+      "loss": 0.7167,
+      "step": 19078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.566125196304287e-09,
+      "loss": 0.6755,
+      "step": 19079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.496914070660334e-09,
+      "loss": 0.5562,
+      "step": 19080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.427983561607455e-09,
+      "loss": 0.6436,
+      "step": 19081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.359333671081881e-09,
+      "loss": 0.7135,
+      "step": 19082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.29096440101318e-09,
+      "loss": 0.591,
+      "step": 19083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.222875753320925e-09,
+      "loss": 0.6675,
+      "step": 19084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.155067729916922e-09,
+      "loss": 0.6255,
+      "step": 19085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.087540332705201e-09,
+      "loss": 0.7012,
+      "step": 19086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.020293563583136e-09,
+      "loss": 0.6024,
+      "step": 19087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.953327424439217e-09,
+      "loss": 0.6193,
+      "step": 19088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.886641917154159e-09,
+      "loss": 0.7278,
+      "step": 19089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.82023704360091e-09,
+      "loss": 0.3066,
+      "step": 19090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.754112805644643e-09,
+      "loss": 0.6347,
+      "step": 19091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.688269205141653e-09,
+      "loss": 0.6295,
+      "step": 19092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.622706243941568e-09,
+      "loss": 0.6793,
+      "step": 19093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.557423923886254e-09,
+      "loss": 0.6721,
+      "step": 19094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.492422246809794e-09,
+      "loss": 0.6902,
+      "step": 19095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.427701214536287e-09,
+      "loss": 0.6561,
+      "step": 19096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3632608288842775e-09,
+      "loss": 0.589,
+      "step": 19097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.2991010916634294e-09,
+      "loss": 0.6345,
+      "step": 19098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.235222004675635e-09,
+      "loss": 0.6435,
+      "step": 19099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.171623569715014e-09,
+      "loss": 0.6933,
+      "step": 19100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.108305788567915e-09,
+      "loss": 0.6455,
+      "step": 19101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.045268663011806e-09,
+      "loss": 0.6889,
+      "step": 19102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.982512194818602e-09,
+      "loss": 0.6328,
+      "step": 19103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.920036385750228e-09,
+      "loss": 0.6261,
+      "step": 19104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.857841237560836e-09,
+      "loss": 0.6328,
+      "step": 19105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7959267519968065e-09,
+      "loss": 0.6245,
+      "step": 19106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.734292930798969e-09,
+      "loss": 0.6935,
+      "step": 19107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.672939775697052e-09,
+      "loss": 0.6452,
+      "step": 19108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.61186728841412e-09,
+      "loss": 0.6453,
+      "step": 19109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.551075470665469e-09,
+      "loss": 0.6232,
+      "step": 19110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.490564324158621e-09,
+      "loss": 0.6379,
+      "step": 19111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.4303338505933284e-09,
+      "loss": 0.6466,
+      "step": 19112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.370384051661571e-09,
+      "loss": 0.6464,
+      "step": 19113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.310714929045336e-09,
+      "loss": 0.6661,
+      "step": 19114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2513264844221715e-09,
+      "loss": 0.6086,
+      "step": 19115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.19221871945963e-09,
+      "loss": 0.7233,
+      "step": 19116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.133391635817498e-09,
+      "loss": 0.6199,
+      "step": 19117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.074845235147786e-09,
+      "loss": 0.6144,
+      "step": 19118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.016579519094734e-09,
+      "loss": 0.6841,
+      "step": 19119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.958594489295921e-09,
+      "loss": 0.6227,
+      "step": 19120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.900890147377825e-09,
+      "loss": 0.5938,
+      "step": 19121
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.84346649496359e-09,
+      "loss": 0.6258,
+      "step": 19122
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.786323533664151e-09,
+      "loss": 0.6369,
+      "step": 19123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.72946126508489e-09,
+      "loss": 0.6252,
+      "step": 19124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.672879690823418e-09,
+      "loss": 0.6348,
+      "step": 19125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.616578812468465e-09,
+      "loss": 0.6418,
+      "step": 19126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.560558631599877e-09,
+      "loss": 0.6641,
+      "step": 19127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5048191497941716e-09,
+      "loss": 0.6921,
+      "step": 19128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.449360368613432e-09,
+      "loss": 0.5924,
+      "step": 19129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.394182289617522e-09,
+      "loss": 0.6409,
+      "step": 19130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.339284914355203e-09,
+      "loss": 0.657,
+      "step": 19131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.284668244368573e-09,
+      "loss": 0.6176,
+      "step": 19132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.230332281190853e-09,
+      "loss": 0.5995,
+      "step": 19133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.176277026349708e-09,
+      "loss": 0.607,
+      "step": 19134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.122502481361702e-09,
+      "loss": 0.6834,
+      "step": 19135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.06900864773785e-09,
+      "loss": 0.6579,
+      "step": 19136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.015795526980283e-09,
+      "loss": 0.6268,
+      "step": 19137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.962863120583361e-09,
+      "loss": 0.6801,
+      "step": 19138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9102114300347835e-09,
+      "loss": 0.7186,
+      "step": 19139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8578404568122575e-09,
+      "loss": 0.6608,
+      "step": 19140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.805750202386827e-09,
+      "loss": 0.6005,
+      "step": 19141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.753940668221768e-09,
+      "loss": 0.6014,
+      "step": 19142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.702411855771472e-09,
+      "loss": 0.6892,
+      "step": 19143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.651163766484779e-09,
+      "loss": 0.6122,
+      "step": 19144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6001964017994284e-09,
+      "loss": 0.6469,
+      "step": 19145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5495097631464976e-09,
+      "loss": 0.6112,
+      "step": 19146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.499103851951514e-09,
+      "loss": 0.6989,
+      "step": 19147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4489786696289005e-09,
+      "loss": 0.7164,
+      "step": 19148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.399134217586421e-09,
+      "loss": 0.6846,
+      "step": 19149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.349570497224065e-09,
+      "loss": 0.658,
+      "step": 19150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.300287509934053e-09,
+      "loss": 0.673,
+      "step": 19151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2512852570997245e-09,
+      "loss": 0.7258,
+      "step": 19152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2025637400988635e-09,
+      "loss": 0.609,
+      "step": 19153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.154122960299267e-09,
+      "loss": 0.6239,
+      "step": 19154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.105962919059847e-09,
+      "loss": 0.6973,
+      "step": 19155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.058083617736186e-09,
+      "loss": 0.6902,
+      "step": 19156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.010485057669433e-09,
+      "loss": 0.5578,
+      "step": 19157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.963167240199628e-09,
+      "loss": 0.5166,
+      "step": 19158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.916130166654597e-09,
+      "loss": 0.5915,
+      "step": 19159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.869373838354395e-09,
+      "loss": 0.6241,
+      "step": 19160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.822898256613528e-09,
+      "loss": 0.6114,
+      "step": 19161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.776703422736505e-09,
+      "loss": 0.6596,
+      "step": 19162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.730789338021179e-09,
+      "loss": 0.7009,
+      "step": 19163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.685156003756518e-09,
+      "loss": 0.618,
+      "step": 19164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.63980342122483e-09,
+      "loss": 0.601,
+      "step": 19165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5947315916995405e-09,
+      "loss": 0.5823,
+      "step": 19166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.549940516446304e-09,
+      "loss": 0.6524,
+      "step": 19167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5054301967241133e-09,
+      "loss": 0.5922,
+      "step": 19168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.461200633780859e-09,
+      "loss": 0.6593,
+      "step": 19169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4172518288611014e-09,
+      "loss": 0.7022,
+      "step": 19170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3735837831982977e-09,
+      "loss": 0.6252,
+      "step": 19171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3301964980192445e-09,
+      "loss": 0.586,
+      "step": 19172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.287089974540747e-09,
+      "loss": 0.6745,
+      "step": 19173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.244264213976278e-09,
+      "loss": 0.6902,
+      "step": 19174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.201719217525989e-09,
+      "loss": 0.6024,
+      "step": 19175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.159454986386701e-09,
+      "loss": 0.6509,
+      "step": 19176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1174715217441307e-09,
+      "loss": 0.609,
+      "step": 19177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0757688247784468e-09,
+      "loss": 0.6739,
+      "step": 19178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0343468966598234e-09,
+      "loss": 0.6165,
+      "step": 19179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9932057385528846e-09,
+      "loss": 0.6231,
+      "step": 19180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9523453516122625e-09,
+      "loss": 0.6229,
+      "step": 19181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9117657369848173e-09,
+      "loss": 0.6035,
+      "step": 19182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8714668958118585e-09,
+      "loss": 0.6432,
+      "step": 19183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.831448829224703e-09,
+      "loss": 0.647,
+      "step": 19184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.791711538345787e-09,
+      "loss": 0.6679,
+      "step": 19185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7522550242931045e-09,
+      "loss": 0.6626,
+      "step": 19186
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7130792881746583e-09,
+      "loss": 0.6432,
+      "step": 19187
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.674184331089569e-09,
+      "loss": 0.6306,
+      "step": 19188
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.635570154131406e-09,
+      "loss": 0.6285,
+      "step": 19189
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.597236758383748e-09,
+      "loss": 0.6984,
+      "step": 19190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5591841449235098e-09,
+      "loss": 0.6757,
+      "step": 19191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5214123148187275e-09,
+      "loss": 0.6536,
+      "step": 19192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4839212691318838e-09,
+      "loss": 0.6661,
+      "step": 19193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.44671100891547e-09,
+      "loss": 0.6751,
+      "step": 19194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.409781535213096e-09,
+      "loss": 0.603,
+      "step": 19195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3731328490628202e-09,
+      "loss": 0.6724,
+      "step": 19196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3367649514949296e-09,
+      "loss": 0.628,
+      "step": 19197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3006778435297193e-09,
+      "loss": 0.6366,
+      "step": 19198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.264871526180823e-09,
+      "loss": 0.6507,
+      "step": 19199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2293460004541022e-09,
+      "loss": 0.6996,
+      "step": 19200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.194101267346538e-09,
+      "loss": 0.6332,
+      "step": 19201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1591373278495587e-09,
+      "loss": 0.6613,
+      "step": 19202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1244541829434917e-09,
+      "loss": 0.6531,
+      "step": 19203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.090051833604223e-09,
+      "loss": 0.6649,
+      "step": 19204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0559302807965364e-09,
+      "loss": 0.6142,
+      "step": 19205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0220895254785544e-09,
+      "loss": 0.6787,
+      "step": 19206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.988529568601738e-09,
+      "loss": 0.5905,
+      "step": 19207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9552504111086667e-09,
+      "loss": 0.5882,
+      "step": 19208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.922252053933038e-09,
+      "loss": 0.6291,
+      "step": 19209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.889534498002998e-09,
+      "loss": 0.6728,
+      "step": 19210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.857097744235592e-09,
+      "loss": 0.6506,
+      "step": 19211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8249417935434221e-09,
+      "loss": 0.6177,
+      "step": 19212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7930666468291002e-09,
+      "loss": 0.6366,
+      "step": 19213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.761472304987466e-09,
+      "loss": 0.6738,
+      "step": 19214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7301587689055876e-09,
+      "loss": 0.7418,
+      "step": 19215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.699126039463872e-09,
+      "loss": 0.5872,
+      "step": 19216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6683741175338442e-09,
+      "loss": 0.6249,
+      "step": 19217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6379030039792576e-09,
+      "loss": 0.602,
+      "step": 19218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.607712699654984e-09,
+      "loss": 0.6389,
+      "step": 19219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.577803205409234e-09,
+      "loss": 0.6573,
+      "step": 19220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5481745220824461e-09,
+      "loss": 0.6275,
+      "step": 19221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5188266505061776e-09,
+      "loss": 0.6914,
+      "step": 19222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4897595915053242e-09,
+      "loss": 0.5989,
+      "step": 19223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4609733458970098e-09,
+      "loss": 0.6349,
+      "step": 19224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4324679144872566e-09,
+      "loss": 0.6998,
+      "step": 19225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4042432980787557e-09,
+      "loss": 0.6405,
+      "step": 19226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3762994974642064e-09,
+      "loss": 0.697,
+      "step": 19227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3486365134263157e-09,
+      "loss": 0.6481,
+      "step": 19228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3212543467444606e-09,
+      "loss": 0.6628,
+      "step": 19229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.294152998186915e-09,
+      "loss": 0.335,
+      "step": 19230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2673324685141818e-09,
+      "loss": 0.6801,
+      "step": 19231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2407927584789926e-09,
+      "loss": 0.6242,
+      "step": 19232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2145338688285268e-09,
+      "loss": 0.6539,
+      "step": 19233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.188555800299973e-09,
+      "loss": 0.5902,
+      "step": 19234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1628585536216374e-09,
+      "loss": 0.6213,
+      "step": 19235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1374421295162753e-09,
+      "loss": 0.649,
+      "step": 19236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1123065286977597e-09,
+      "loss": 0.7137,
+      "step": 19237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0874517518710825e-09,
+      "loss": 0.6825,
+      "step": 19238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0628777997356842e-09,
+      "loss": 0.6582,
+      "step": 19239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0385846729810134e-09,
+      "loss": 0.6793,
+      "step": 19240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0145723722887469e-09,
+      "loss": 0.6079,
+      "step": 19241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.908408983339002e-10,
+      "loss": 0.5839,
+      "step": 19242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.673902517837176e-10,
+      "loss": 0.6438,
+      "step": 19243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.442204332954507e-10,
+      "loss": 0.6093,
+      "step": 19244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.21331443520801e-10,
+      "loss": 0.6885,
+      "step": 19245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.987232831014769e-10,
+      "loss": 0.6587,
+      "step": 19246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.763959526747468e-10,
+      "loss": 0.5798,
+      "step": 19247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.543494528645557e-10,
+      "loss": 0.5642,
+      "step": 19248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.325837842926288e-10,
+      "loss": 0.6133,
+      "step": 19249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.110989475695885e-10,
+      "loss": 0.6686,
+      "step": 19250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.898949432971759e-10,
+      "loss": 0.6995,
+      "step": 19251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.68971772073801e-10,
+      "loss": 0.6917,
+      "step": 19252
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.483294344856618e-10,
+      "loss": 0.6776,
+      "step": 19253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.279679311111843e-10,
+      "loss": 0.606,
+      "step": 19254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.078872625243538e-10,
+      "loss": 0.6902,
+      "step": 19255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.880874292891638e-10,
+      "loss": 0.6153,
+      "step": 19256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.685684319596153e-10,
+      "loss": 0.6645,
+      "step": 19257
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.493302710863791e-10,
+      "loss": 0.6638,
+      "step": 19258
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.303729472079134e-10,
+      "loss": 0.5617,
+      "step": 19259
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.116964608571252e-10,
+      "loss": 0.6295,
+      "step": 19260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.933008125580397e-10,
+      "loss": 0.7255,
+      "step": 19261
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.751860028291312e-10,
+      "loss": 0.6852,
+      "step": 19262
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.573520321777714e-10,
+      "loss": 0.6105,
+      "step": 19263
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.397989011046712e-10,
+      "loss": 0.6659,
+      "step": 19264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.225266101038795e-10,
+      "loss": 0.5821,
+      "step": 19265
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.055351596594538e-10,
+      "loss": 0.6931,
+      "step": 19266
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.888245502487898e-10,
+      "loss": 0.5889,
+      "step": 19267
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.723947823426223e-10,
+      "loss": 0.6129,
+      "step": 19268
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5624585639947316e-10,
+      "loss": 0.6284,
+      "step": 19269
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.403777728756442e-10,
+      "loss": 0.6323,
+      "step": 19270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2479053221633483e-10,
+      "loss": 0.5968,
+      "step": 19271
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0948413485786264e-10,
+      "loss": 0.6411,
+      "step": 19272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9445858123099425e-10,
+      "loss": 0.6367,
+      "step": 19273
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.797138717576143e-10,
+      "loss": 0.6356,
+      "step": 19274
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6525000685294633e-10,
+      "loss": 0.6718,
+      "step": 19275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5106698692111143e-10,
+      "loss": 0.6251,
+      "step": 19276
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.371648123617899e-10,
+      "loss": 0.6557,
+      "step": 19277
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2354348356578025e-10,
+      "loss": 0.67,
+      "step": 19278
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.102030009149992e-10,
+      "loss": 0.5668,
+      "step": 19279
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9714336478359194e-10,
+      "loss": 0.5785,
+      "step": 19280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.843645755401525e-10,
+      "loss": 0.6523,
+      "step": 19281
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.718666335410625e-10,
+      "loss": 0.6147,
+      "step": 19282
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5964953913937275e-10,
+      "loss": 0.6364,
+      "step": 19283
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4771329267703206e-10,
+      "loss": 0.7373,
+      "step": 19284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.360578944904379e-10,
+      "loss": 0.6767,
+      "step": 19285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2468334490599597e-10,
+      "loss": 0.6897,
+      "step": 19286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.135896442434504e-10,
+      "loss": 0.6449,
+      "step": 19287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0277679281366369e-10,
+      "loss": 0.649,
+      "step": 19288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9224479092194714e-10,
+      "loss": 0.6845,
+      "step": 19289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8199363886250986e-10,
+      "loss": 0.6645,
+      "step": 19290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7202333692400986e-10,
+      "loss": 0.5774,
+      "step": 19291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.623338853862233e-10,
+      "loss": 0.6461,
+      "step": 19292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5292528452115486e-10,
+      "loss": 0.6937,
+      "step": 19293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.437975345941478e-10,
+      "loss": 0.6732,
+      "step": 19294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3495063585944324e-10,
+      "loss": 0.6098,
+      "step": 19295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2638458856795156e-10,
+      "loss": 0.66,
+      "step": 19296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.180993929583707e-10,
+      "loss": 0.6628,
+      "step": 19297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1009504926495773e-10,
+      "loss": 0.6568,
+      "step": 19298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0237155771086748e-10,
+      "loss": 0.6597,
+      "step": 19299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.49289185148139e-11,
+      "loss": 0.628,
+      "step": 19300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.776713188329843e-11,
+      "loss": 0.643,
+      "step": 19301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.088619802060216e-11,
+      "loss": 0.6004,
+      "step": 19302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.428611711768341e-11,
+      "loss": 0.663,
+      "step": 19303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.796688935994943e-11,
+      "loss": 0.7202,
+      "step": 19304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.192851492614615e-11,
+      "loss": 0.6722,
+      "step": 19305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.617099398502745e-11,
+      "loss": 0.2941,
+      "step": 19306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.06943266986859e-11,
+      "loss": 0.5777,
+      "step": 19307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.54985132214425e-11,
+      "loss": 0.6602,
+      "step": 19308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0583553697626234e-11,
+      "loss": 0.3133,
+      "step": 19309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.594944826712521e-11,
+      "loss": 0.6232,
+      "step": 19310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1596197057615076e-11,
+      "loss": 0.688,
+      "step": 19311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.752380019455103e-11,
+      "loss": 0.2974,
+      "step": 19312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3732257790065606e-11,
+      "loss": 0.6094,
+      "step": 19313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0221569950740206e-11,
+      "loss": 0.6638,
+      "step": 19314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6991736776494905e-11,
+      "loss": 0.6991,
+      "step": 19315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4042758356147545e-11,
+      "loss": 0.6013,
+      "step": 19316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1374634774075077e-11,
+      "loss": 0.6566,
+      "step": 19317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.987366105772667e-12,
+      "loss": 0.663,
+      "step": 19318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.880952415633246e-12,
+      "loss": 0.6398,
+      "step": 19319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.055393765829309e-12,
+      "loss": 0.6782,
+      "step": 19320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.510690205210665e-12,
+      "loss": 0.6695,
+      "step": 19321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.246841778186237e-12,
+      "loss": 0.6641,
+      "step": 19322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.263848521393385e-12,
+      "loss": 0.6079,
+      "step": 19323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.617104603672374e-13,
+      "loss": 0.581,
+      "step": 19324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.404276162020324e-13,
+      "loss": 0.6214,
+      "step": 19325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.4588,
+      "step": 19326
+    },
+    {
+      "epoch": 1.0,
+      "step": 19326,
+      "total_flos": 8754991215181824.0,
+      "train_loss": 0.7098402904696831,
+      "train_runtime": 114662.5659,
+      "train_samples_per_second": 21.573,
+      "train_steps_per_second": 0.169
+    }
+  ],
+  "max_steps": 19326,
+  "num_train_epochs": 1,
+  "total_flos": 8754991215181824.0,
+  "trial_name": null,
+  "trial_params": null
+}