diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,30949 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 10308,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.6597,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.8832,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.7804,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.9388,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.8721,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.733,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.9589,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.8646,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 0.7823,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 0.8528,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.8967,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999998141556122e-05,
+      "loss": 0.8121,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999992566225167e-05,
+      "loss": 0.648,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999983274009217e-05,
+      "loss": 0.6722,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999970264911722e-05,
+      "loss": 0.6083,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999953538937513e-05,
+      "loss": 0.6039,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999933096092813e-05,
+      "loss": 0.6011,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999990893638522e-05,
+      "loss": 0.6769,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999988105982371e-05,
+      "loss": 0.6146,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999849466418647e-05,
+      "loss": 0.5803,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999814156181775e-05,
+      "loss": 0.5976,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999977512912622e-05,
+      "loss": 0.6241,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999732385266477e-05,
+      "loss": 0.701,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999685924618444e-05,
+      "loss": 0.5838,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999963574719939e-05,
+      "loss": 0.6776,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999581853027964e-05,
+      "loss": 0.5736,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999952424212419e-05,
+      "loss": 0.5138,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999462914509492e-05,
+      "loss": 0.5569,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999939787020666e-05,
+      "loss": 0.5926,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999932910923987e-05,
+      "loss": 0.5259,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999925663163468e-05,
+      "loss": 0.6545,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999180437418034e-05,
+      "loss": 0.6051,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999100526618245e-05,
+      "loss": 0.6621,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9999016899265017e-05,
+      "loss": 0.626,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998929555389436e-05,
+      "loss": 0.5555,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998838495023964e-05,
+      "loss": 0.624,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998743718202448e-05,
+      "loss": 0.5187,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998645224960118e-05,
+      "loss": 0.7195,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999854301533358e-05,
+      "loss": 0.6118,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998437089360823e-05,
+      "loss": 0.607,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999832744708122e-05,
+      "loss": 0.557,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999821408853553e-05,
+      "loss": 0.6589,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999809701376587e-05,
+      "loss": 0.693,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9997976222815776e-05,
+      "loss": 0.6252,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999785171573013e-05,
+      "loss": 0.6195,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999772349255522e-05,
+      "loss": 0.6387,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9997591553338695e-05,
+      "loss": 0.6132,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9997455898129603e-05,
+      "loss": 0.6487,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9997316526978366e-05,
+      "loss": 0.5981,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9997173439936778e-05,
+      "loss": 0.596,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999702663705803e-05,
+      "loss": 0.6272,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999687611839669e-05,
+      "loss": 0.6333,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9996721884008695e-05,
+      "loss": 0.5929,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999656393395138e-05,
+      "loss": 0.5339,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9996402268283445e-05,
+      "loss": 0.5592,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9996236887064987e-05,
+      "loss": 0.5905,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9996067790357478e-05,
+      "loss": 0.6146,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999589497822376e-05,
+      "loss": 0.5901,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9995718450728073e-05,
+      "loss": 0.6077,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9995538207936024e-05,
+      "loss": 0.6125,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995354249914617e-05,
+      "loss": 0.5612,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995166576732216e-05,
+      "loss": 0.6058,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994975188458582e-05,
+      "loss": 0.65,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994780085164856e-05,
+      "loss": 0.5877,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994581266923546e-05,
+      "loss": 0.656,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994378733808562e-05,
+      "loss": 0.5937,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994172485895174e-05,
+      "loss": 0.6006,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993962523260046e-05,
+      "loss": 0.6482,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999374884598122e-05,
+      "loss": 0.6662,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993531454138116e-05,
+      "loss": 0.6414,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993310347811533e-05,
+      "loss": 0.5541,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999308552708366e-05,
+      "loss": 0.6667,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999285699203805e-05,
+      "loss": 0.5885,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992624742759662e-05,
+      "loss": 0.5312,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992388779334807e-05,
+      "loss": 0.699,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992149101851197e-05,
+      "loss": 0.5835,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991905710397918e-05,
+      "loss": 0.6572,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9991658605065428e-05,
+      "loss": 0.6011,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9991407785945584e-05,
+      "loss": 0.6404,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9991153253131603e-05,
+      "loss": 0.6216,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99908950067181e-05,
+      "loss": 0.6268,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999063304680106e-05,
+      "loss": 0.6016,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9990367373477847e-05,
+      "loss": 0.5634,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999009798684721e-05,
+      "loss": 0.6172,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989824887009276e-05,
+      "loss": 0.6384,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989548074065557e-05,
+      "loss": 0.5853,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989267548118945e-05,
+      "loss": 0.5738,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988983309273696e-05,
+      "loss": 0.6077,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988695357635468e-05,
+      "loss": 0.6555,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988403693311285e-05,
+      "loss": 0.5848,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988108316409562e-05,
+      "loss": 0.589,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987809227040077e-05,
+      "loss": 0.5408,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987506425314004e-05,
+      "loss": 0.6158,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998719991134389e-05,
+      "loss": 0.5704,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9986889685243664e-05,
+      "loss": 0.6461,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9986575747128634e-05,
+      "loss": 0.59,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9986258097115486e-05,
+      "loss": 0.5979,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9985936735322284e-05,
+      "loss": 0.6094,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998561166186848e-05,
+      "loss": 0.6529,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9985282876874898e-05,
+      "loss": 0.7103,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9984950380463743e-05,
+      "loss": 0.6967,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99846141727586e-05,
+      "loss": 0.5704,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9984274253884434e-05,
+      "loss": 0.6534,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9983930623967592e-05,
+      "loss": 0.5817,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9983583283135792e-05,
+      "loss": 0.5577,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9983232231518136e-05,
+      "loss": 0.607,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9982877469245118e-05,
+      "loss": 0.628,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9982518996448583e-05,
+      "loss": 0.6843,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998215681326178e-05,
+      "loss": 0.5788,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981790919819327e-05,
+      "loss": 0.5538,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981421316257226e-05,
+      "loss": 0.5746,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9981048002712852e-05,
+      "loss": 0.5493,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9980670979324953e-05,
+      "loss": 0.6736,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.998029024623368e-05,
+      "loss": 0.5851,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9979905803580535e-05,
+      "loss": 0.6372,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9979517651508416e-05,
+      "loss": 0.6034,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997912579016159e-05,
+      "loss": 0.5776,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9978730219685717e-05,
+      "loss": 0.6312,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9978330940227822e-05,
+      "loss": 0.6013,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977927951936305e-05,
+      "loss": 0.5865,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997752125496096e-05,
+      "loss": 0.7048,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9977110849452954e-05,
+      "loss": 0.6035,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976696735564825e-05,
+      "loss": 0.7057,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997627891345049e-05,
+      "loss": 0.5722,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9975857383265263e-05,
+      "loss": 0.5649,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997543214516581e-05,
+      "loss": 0.6012,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997500319931019e-05,
+      "loss": 0.5912,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9974570545857838e-05,
+      "loss": 0.7004,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997413418496957e-05,
+      "loss": 0.6295,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973694116807566e-05,
+      "loss": 0.6313,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9973250341535408e-05,
+      "loss": 0.5892,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9972802859318034e-05,
+      "loss": 0.6397,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997235167032177e-05,
+      "loss": 0.6387,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9971896774714318e-05,
+      "loss": 0.5865,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9971438172664753e-05,
+      "loss": 0.6836,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9970975864343543e-05,
+      "loss": 0.6184,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.997050984992251e-05,
+      "loss": 0.5886,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9970040129574878e-05,
+      "loss": 0.5975,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9969566703475228e-05,
+      "loss": 0.5215,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9969089571799533e-05,
+      "loss": 0.6835,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9968608734725136e-05,
+      "loss": 0.6087,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9968124192430756e-05,
+      "loss": 0.5671,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9967635945096493e-05,
+      "loss": 0.6142,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9967143992903827e-05,
+      "loss": 0.5846,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9966648336035606e-05,
+      "loss": 0.724,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9966148974676062e-05,
+      "loss": 0.5357,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99656459090108e-05,
+      "loss": 0.6059,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996513913922681e-05,
+      "loss": 0.5883,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9964628665512443e-05,
+      "loss": 0.619,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996411448805745e-05,
+      "loss": 0.6403,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996359660705293e-05,
+      "loss": 0.576,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9963075022691383e-05,
+      "loss": 0.6133,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9962549735166673e-05,
+      "loss": 0.528,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9962020744674045e-05,
+      "loss": 0.6013,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9961488051410115e-05,
+      "loss": 0.5911,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9960951655572886e-05,
+      "loss": 0.5813,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9960411557361726e-05,
+      "loss": 0.6289,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.995986775697738e-05,
+      "loss": 0.6237,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9959320254621982e-05,
+      "loss": 0.5999,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.995876905049902e-05,
+      "loss": 0.6176,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.995821414481338e-05,
+      "loss": 0.5831,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9957655537771312e-05,
+      "loss": 0.6295,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9957093229580444e-05,
+      "loss": 0.5838,
+      "step": 326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9956527220449776e-05,
+      "loss": 0.6431,
+      "step": 328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.995595751058969e-05,
+      "loss": 0.6713,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9955384100211944e-05,
+      "loss": 0.5981,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9954806989529666e-05,
+      "loss": 0.5623,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9954226178757356e-05,
+      "loss": 0.5457,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9953641668110905e-05,
+      "loss": 0.7338,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9953053457807557e-05,
+      "loss": 0.579,
+      "step": 340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9952461548065956e-05,
+      "loss": 0.6099,
+      "step": 342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9951865939106098e-05,
+      "loss": 0.6504,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.995126663114937e-05,
+      "loss": 0.6337,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9950663624418524e-05,
+      "loss": 0.6062,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9950056919137694e-05,
+      "loss": 0.5435,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.994944651553239e-05,
+      "loss": 0.5927,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9948832413829478e-05,
+      "loss": 0.6174,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9948214614257226e-05,
+      "loss": 0.5378,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9947593117045255e-05,
+      "loss": 0.5508,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9946967922424575e-05,
+      "loss": 0.6309,
+      "step": 360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9946339030627562e-05,
+      "loss": 0.5845,
+      "step": 362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9945706441887965e-05,
+      "loss": 0.585,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9945070156440915e-05,
+      "loss": 0.5974,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9944430174522903e-05,
+      "loss": 0.606,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9943786496371816e-05,
+      "loss": 0.5039,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.994313912222689e-05,
+      "loss": 0.6011,
+      "step": 372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9942488052328755e-05,
+      "loss": 0.5854,
+      "step": 374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9941833286919404e-05,
+      "loss": 0.649,
+      "step": 376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9941174826242205e-05,
+      "loss": 0.5963,
+      "step": 378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99405126705419e-05,
+      "loss": 0.5945,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9939846820064605e-05,
+      "loss": 0.5049,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.993917727505781e-05,
+      "loss": 0.6082,
+      "step": 384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9938504035770377e-05,
+      "loss": 0.6567,
+      "step": 386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9937827102452543e-05,
+      "loss": 0.5815,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.993714647535591e-05,
+      "loss": 0.5957,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9936462154733472e-05,
+      "loss": 0.671,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9935774140839572e-05,
+      "loss": 0.5912,
+      "step": 394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9935082433929943e-05,
+      "loss": 0.6308,
+      "step": 396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.993438703426168e-05,
+      "loss": 0.6841,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.993368794209326e-05,
+      "loss": 0.5567,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9932985157684524e-05,
+      "loss": 0.6086,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.993227868129669e-05,
+      "loss": 0.5515,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9931568513192346e-05,
+      "loss": 0.5975,
+      "step": 406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.993085465363546e-05,
+      "loss": 0.5389,
+      "step": 408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.993013710289136e-05,
+      "loss": 0.597,
+      "step": 410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.992941586122675e-05,
+      "loss": 0.5618,
+      "step": 412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9928690928909714e-05,
+      "loss": 0.705,
+      "step": 414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9927962306209697e-05,
+      "loss": 0.6393,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9927229993397517e-05,
+      "loss": 0.5656,
+      "step": 418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9926493990745372e-05,
+      "loss": 0.6071,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9925754298526824e-05,
+      "loss": 0.5931,
+      "step": 422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9925010917016805e-05,
+      "loss": 0.6623,
+      "step": 424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9924263846491626e-05,
+      "loss": 0.6146,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9923513087228966e-05,
+      "loss": 0.657,
+      "step": 428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9922758639507863e-05,
+      "loss": 0.5798,
+      "step": 430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992200050360875e-05,
+      "loss": 0.717,
+      "step": 432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9921238679813416e-05,
+      "loss": 0.5658,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9920473168405013e-05,
+      "loss": 0.5894,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.991970396966808e-05,
+      "loss": 0.4908,
+      "step": 438
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9918931083888518e-05,
+      "loss": 0.6269,
+      "step": 440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9918154511353603e-05,
+      "loss": 0.5881,
+      "step": 442
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.991737425235197e-05,
+      "loss": 0.5812,
+      "step": 444
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9916590307173643e-05,
+      "loss": 0.6113,
+      "step": 446
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9915802676109998e-05,
+      "loss": 0.5629,
+      "step": 448
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.991501135945379e-05,
+      "loss": 0.5377,
+      "step": 450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9914216357499146e-05,
+      "loss": 0.6187,
+      "step": 452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.991341767054156e-05,
+      "loss": 0.706,
+      "step": 454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9912615298877887e-05,
+      "loss": 0.5596,
+      "step": 456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9911809242806364e-05,
+      "loss": 0.5179,
+      "step": 458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9910999502626597e-05,
+      "loss": 0.6913,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.991018607863955e-05,
+      "loss": 0.5975,
+      "step": 462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990936897114757e-05,
+      "loss": 0.6167,
+      "step": 464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908548180454364e-05,
+      "loss": 0.7039,
+      "step": 466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990772370686501e-05,
+      "loss": 0.5998,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990689555068596e-05,
+      "loss": 0.5641,
+      "step": 470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9906063712225022e-05,
+      "loss": 0.6397,
+      "step": 472
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9905228191791387e-05,
+      "loss": 0.5995,
+      "step": 474
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990438898969561e-05,
+      "loss": 0.6245,
+      "step": 476
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9903546106249606e-05,
+      "loss": 0.5652,
+      "step": 478
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9902699541766673e-05,
+      "loss": 0.6412,
+      "step": 480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9901849296561465e-05,
+      "loss": 0.5865,
+      "step": 482
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.990099537095001e-05,
+      "loss": 0.6723,
+      "step": 484
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99001377652497e-05,
+      "loss": 0.5782,
+      "step": 486
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9899276479779304e-05,
+      "loss": 0.5837,
+      "step": 488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9898411514858944e-05,
+      "loss": 0.6459,
+      "step": 490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9897542870810123e-05,
+      "loss": 0.6209,
+      "step": 492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9896670547955707e-05,
+      "loss": 0.6379,
+      "step": 494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9895794546619923e-05,
+      "loss": 0.5919,
+      "step": 496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9894914867128373e-05,
+      "loss": 0.5868,
+      "step": 498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989403150980803e-05,
+      "loss": 0.5646,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989314447498722e-05,
+      "loss": 0.5743,
+      "step": 502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989225376299565e-05,
+      "loss": 0.6689,
+      "step": 504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9891359374164385e-05,
+      "loss": 0.588,
+      "step": 506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989046130882586e-05,
+      "loss": 0.5972,
+      "step": 508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988955956731387e-05,
+      "loss": 0.6529,
+      "step": 510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988865414996359e-05,
+      "loss": 0.5865,
+      "step": 512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988774505711155e-05,
+      "loss": 0.5106,
+      "step": 514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9886832289095653e-05,
+      "loss": 0.5481,
+      "step": 516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988591584625516e-05,
+      "loss": 0.7072,
+      "step": 518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884995728930706e-05,
+      "loss": 0.6704,
+      "step": 520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884071937464284e-05,
+      "loss": 0.623,
+      "step": 522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9883144472199262e-05,
+      "loss": 0.6492,
+      "step": 524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988221333348036e-05,
+      "loss": 0.5402,
+      "step": 526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9881278521653682e-05,
+      "loss": 0.5853,
+      "step": 528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9880340037066686e-05,
+      "loss": 0.6022,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.987939788006819e-05,
+      "loss": 0.5108,
+      "step": 532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878452051008387e-05,
+      "loss": 0.6065,
+      "step": 534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987750255023883e-05,
+      "loss": 0.6014,
+      "step": 536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9876549378112437e-05,
+      "loss": 0.5495,
+      "step": 538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9875592534983493e-05,
+      "loss": 0.6278,
+      "step": 540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9874632021207647e-05,
+      "loss": 0.6758,
+      "step": 542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987366783714191e-05,
+      "loss": 0.5396,
+      "step": 544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9872699983144655e-05,
+      "loss": 0.5975,
+      "step": 546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9871728459575626e-05,
+      "loss": 0.6104,
+      "step": 548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9870753266795928e-05,
+      "loss": 0.5966,
+      "step": 550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986977440516803e-05,
+      "loss": 0.6736,
+      "step": 552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986879187505576e-05,
+      "loss": 0.5919,
+      "step": 554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9867805676824313e-05,
+      "loss": 0.6201,
+      "step": 556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986681581084025e-05,
+      "loss": 0.56,
+      "step": 558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9865822277471496e-05,
+      "loss": 0.6552,
+      "step": 560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9864825077087332e-05,
+      "loss": 0.5916,
+      "step": 562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9863824210058408e-05,
+      "loss": 0.5601,
+      "step": 564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862819676756734e-05,
+      "loss": 0.5427,
+      "step": 566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9861811477555685e-05,
+      "loss": 0.5575,
+      "step": 568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9860799612829996e-05,
+      "loss": 0.5704,
+      "step": 570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985978408295577e-05,
+      "loss": 0.5246,
+      "step": 572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9858764888310464e-05,
+      "loss": 0.662,
+      "step": 574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9857742029272896e-05,
+      "loss": 0.5669,
+      "step": 576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985671550622326e-05,
+      "loss": 0.5319,
+      "step": 578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.98556853195431e-05,
+      "loss": 0.6246,
+      "step": 580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9854651469615324e-05,
+      "loss": 0.5794,
+      "step": 582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9853613956824207e-05,
+      "loss": 0.6718,
+      "step": 584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9852572781555376e-05,
+      "loss": 0.5818,
+      "step": 586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9851527944195824e-05,
+      "loss": 0.6007,
+      "step": 588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9850479445133905e-05,
+      "loss": 0.5835,
+      "step": 590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9849427284759338e-05,
+      "loss": 0.6021,
+      "step": 592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9848371463463196e-05,
+      "loss": 0.7067,
+      "step": 594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9847311981637916e-05,
+      "loss": 0.6103,
+      "step": 596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.98462488396773e-05,
+      "loss": 0.5426,
+      "step": 598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845182037976504e-05,
+      "loss": 0.5415,
+      "step": 600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9844111576932042e-05,
+      "loss": 0.5677,
+      "step": 602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.984303745694179e-05,
+      "loss": 0.6421,
+      "step": 604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9841959678405e-05,
+      "loss": 0.5529,
+      "step": 606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.984087824172226e-05,
+      "loss": 0.609,
+      "step": 608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9839793147295526e-05,
+      "loss": 0.6079,
+      "step": 610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9838704395528118e-05,
+      "loss": 0.567,
+      "step": 612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.983761198682472e-05,
+      "loss": 0.5238,
+      "step": 614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.983651592159136e-05,
+      "loss": 0.667,
+      "step": 616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9835416200235428e-05,
+      "loss": 0.5493,
+      "step": 618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.983431282316569e-05,
+      "loss": 0.6042,
+      "step": 620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9833205790792255e-05,
+      "loss": 0.6681,
+      "step": 622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9832095103526592e-05,
+      "loss": 0.6463,
+      "step": 624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9830980761781533e-05,
+      "loss": 0.668,
+      "step": 626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9829862765971265e-05,
+      "loss": 0.652,
+      "step": 628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9828741116511338e-05,
+      "loss": 0.6443,
+      "step": 630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9827615813818648e-05,
+      "loss": 0.5897,
+      "step": 632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826486858311468e-05,
+      "loss": 0.5956,
+      "step": 634
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9825354250409413e-05,
+      "loss": 0.5513,
+      "step": 636
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.982421799053346e-05,
+      "loss": 0.5358,
+      "step": 638
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.982307807910594e-05,
+      "loss": 0.5963,
+      "step": 640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9821934516550555e-05,
+      "loss": 0.5834,
+      "step": 642
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.982078730329235e-05,
+      "loss": 0.6317,
+      "step": 644
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.981963643975773e-05,
+      "loss": 0.6703,
+      "step": 646
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.981848192637446e-05,
+      "loss": 0.5913,
+      "step": 648
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9817323763571657e-05,
+      "loss": 0.6404,
+      "step": 650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9816161951779798e-05,
+      "loss": 0.5518,
+      "step": 652
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9814996491430716e-05,
+      "loss": 0.4916,
+      "step": 654
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9813827382957602e-05,
+      "loss": 0.6268,
+      "step": 656
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9812654626794998e-05,
+      "loss": 0.5484,
+      "step": 658
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9811478223378802e-05,
+      "loss": 0.6208,
+      "step": 660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9810298173146274e-05,
+      "loss": 0.5052,
+      "step": 662
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.980911447653602e-05,
+      "loss": 0.5982,
+      "step": 664
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9807927133988016e-05,
+      "loss": 0.5512,
+      "step": 666
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9806736145943577e-05,
+      "loss": 0.5623,
+      "step": 668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9805541512845382e-05,
+      "loss": 0.5513,
+      "step": 670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.980434323513746e-05,
+      "loss": 0.6353,
+      "step": 672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.98031413132652e-05,
+      "loss": 0.6235,
+      "step": 674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.980193574767534e-05,
+      "loss": 0.6249,
+      "step": 676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9800726538815984e-05,
+      "loss": 0.5755,
+      "step": 678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.979951368713657e-05,
+      "loss": 0.5976,
+      "step": 680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9798297193087904e-05,
+      "loss": 0.5217,
+      "step": 682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9797077057122148e-05,
+      "loss": 0.5534,
+      "step": 684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.979585327969281e-05,
+      "loss": 0.644,
+      "step": 686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9794625861254756e-05,
+      "loss": 0.6215,
+      "step": 688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97933948022642e-05,
+      "loss": 0.6073,
+      "step": 690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9792160103178717e-05,
+      "loss": 0.5889,
+      "step": 692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9790921764457228e-05,
+      "loss": 0.5585,
+      "step": 694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.978967978656001e-05,
+      "loss": 0.5917,
+      "step": 696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9788434169948692e-05,
+      "loss": 0.6309,
+      "step": 698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9787184915086256e-05,
+      "loss": 0.6091,
+      "step": 700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.978593202243704e-05,
+      "loss": 0.5434,
+      "step": 702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9784675492466726e-05,
+      "loss": 0.5728,
+      "step": 704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.978341532564235e-05,
+      "loss": 0.5821,
+      "step": 706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9782151522432305e-05,
+      "loss": 0.617,
+      "step": 708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9780884083306333e-05,
+      "loss": 0.5761,
+      "step": 710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9779613008735524e-05,
+      "loss": 0.643,
+      "step": 712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9778338299192326e-05,
+      "loss": 0.5426,
+      "step": 714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9777059955150532e-05,
+      "loss": 0.5398,
+      "step": 716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9775777977085287e-05,
+      "loss": 0.5163,
+      "step": 718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.977449236547309e-05,
+      "loss": 0.5157,
+      "step": 720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9773203120791788e-05,
+      "loss": 0.5897,
+      "step": 722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9771910243520577e-05,
+      "loss": 0.5745,
+      "step": 724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.977061373414001e-05,
+      "loss": 0.6795,
+      "step": 726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9769313593131975e-05,
+      "loss": 0.6414,
+      "step": 728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.976800982097973e-05,
+      "loss": 0.5852,
+      "step": 730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9766702418167868e-05,
+      "loss": 0.5946,
+      "step": 732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.976539138518234e-05,
+      "loss": 0.5733,
+      "step": 734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.976407672251043e-05,
+      "loss": 0.5686,
+      "step": 736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9762758430640802e-05,
+      "loss": 0.5856,
+      "step": 738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9761436510063436e-05,
+      "loss": 0.6486,
+      "step": 740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.976011096126968e-05,
+      "loss": 0.6043,
+      "step": 742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9758781784752227e-05,
+      "loss": 0.5916,
+      "step": 744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9757448981005112e-05,
+      "loss": 0.629,
+      "step": 746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9756112550523726e-05,
+      "loss": 0.5821,
+      "step": 748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.975477249380481e-05,
+      "loss": 0.5839,
+      "step": 750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.975342881134644e-05,
+      "loss": 0.5953,
+      "step": 752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9752081503648052e-05,
+      "loss": 0.6484,
+      "step": 754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9750730571210423e-05,
+      "loss": 0.6353,
+      "step": 756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.974937601453568e-05,
+      "loss": 0.5043,
+      "step": 758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9748017834127304e-05,
+      "loss": 0.6073,
+      "step": 760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9746656030490104e-05,
+      "loss": 0.5672,
+      "step": 762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.974529060413025e-05,
+      "loss": 0.6209,
+      "step": 764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.974392155555526e-05,
+      "loss": 0.5633,
+      "step": 766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9742548885273992e-05,
+      "loss": 0.5987,
+      "step": 768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9741172593796652e-05,
+      "loss": 0.58,
+      "step": 770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.973979268163479e-05,
+      "loss": 0.6605,
+      "step": 772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9738409149301306e-05,
+      "loss": 0.6149,
+      "step": 774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9737021997310445e-05,
+      "loss": 0.5547,
+      "step": 776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9735631226177793e-05,
+      "loss": 0.4604,
+      "step": 778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9734236836420282e-05,
+      "loss": 0.5432,
+      "step": 780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9732838828556197e-05,
+      "loss": 0.6387,
+      "step": 782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.973143720310516e-05,
+      "loss": 0.625,
+      "step": 784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9730031960588133e-05,
+      "loss": 0.6566,
+      "step": 786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9728623101527438e-05,
+      "loss": 0.6232,
+      "step": 788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9727210626446727e-05,
+      "loss": 0.5889,
+      "step": 790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9725794535871003e-05,
+      "loss": 0.495,
+      "step": 792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.972437483032661e-05,
+      "loss": 0.5968,
+      "step": 794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.972295151034124e-05,
+      "loss": 0.6065,
+      "step": 796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9721524576443918e-05,
+      "loss": 0.6008,
+      "step": 798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9720094029165023e-05,
+      "loss": 0.5565,
+      "step": 800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9718659869036277e-05,
+      "loss": 0.6049,
+      "step": 802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9717222096590736e-05,
+      "loss": 0.6257,
+      "step": 804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9715780712362803e-05,
+      "loss": 0.6057,
+      "step": 806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.971433571688823e-05,
+      "loss": 0.6086,
+      "step": 808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9712887110704103e-05,
+      "loss": 0.4638,
+      "step": 810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9711434894348854e-05,
+      "loss": 0.5538,
+      "step": 812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970997906836225e-05,
+      "loss": 0.6489,
+      "step": 814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970851963328541e-05,
+      "loss": 0.6019,
+      "step": 816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970705658966079e-05,
+      "loss": 0.6224,
+      "step": 818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9705589938032187e-05,
+      "loss": 0.6233,
+      "step": 820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9704119678944736e-05,
+      "loss": 0.5651,
+      "step": 822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9702645812944916e-05,
+      "loss": 0.5437,
+      "step": 824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9701168340580547e-05,
+      "loss": 0.5387,
+      "step": 826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9699687262400793e-05,
+      "loss": 0.6158,
+      "step": 828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969820257895615e-05,
+      "loss": 0.4762,
+      "step": 830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9696714290798458e-05,
+      "loss": 0.6192,
+      "step": 832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9695222398480898e-05,
+      "loss": 0.5504,
+      "step": 834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9693726902557992e-05,
+      "loss": 0.5936,
+      "step": 836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9692227803585593e-05,
+      "loss": 0.5941,
+      "step": 838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9690725102120904e-05,
+      "loss": 0.6025,
+      "step": 840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9689218798722464e-05,
+      "loss": 0.4879,
+      "step": 842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9687708893950143e-05,
+      "loss": 0.5821,
+      "step": 844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.968619538836516e-05,
+      "loss": 0.6153,
+      "step": 846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9684678282530067e-05,
+      "loss": 0.6596,
+      "step": 848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9683157577008755e-05,
+      "loss": 0.5923,
+      "step": 850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.968163327236645e-05,
+      "loss": 0.5981,
+      "step": 852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9680105369169727e-05,
+      "loss": 0.6273,
+      "step": 854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9678573867986487e-05,
+      "loss": 0.568,
+      "step": 856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9677038769385968e-05,
+      "loss": 0.5801,
+      "step": 858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.967550007393875e-05,
+      "loss": 0.5841,
+      "step": 860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9673957782216754e-05,
+      "loss": 0.5137,
+      "step": 862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9672411894793227e-05,
+      "loss": 0.5025,
+      "step": 864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.967086241224276e-05,
+      "loss": 0.568,
+      "step": 866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9669309335141277e-05,
+      "loss": 0.6577,
+      "step": 868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966775266406604e-05,
+      "loss": 0.7151,
+      "step": 870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966619239959565e-05,
+      "loss": 0.5669,
+      "step": 872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9664628542310034e-05,
+      "loss": 0.5384,
+      "step": 874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9663061092790462e-05,
+      "loss": 0.6176,
+      "step": 876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9661490051619537e-05,
+      "loss": 0.5476,
+      "step": 878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.96599154193812e-05,
+      "loss": 0.688,
+      "step": 880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9658337196660724e-05,
+      "loss": 0.5946,
+      "step": 882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9656755384044716e-05,
+      "loss": 0.5228,
+      "step": 884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9655169982121116e-05,
+      "loss": 0.55,
+      "step": 886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9653580991479197e-05,
+      "loss": 0.5406,
+      "step": 888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9651988412709575e-05,
+      "loss": 0.6088,
+      "step": 890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9650392246404194e-05,
+      "loss": 0.6333,
+      "step": 892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.964879249315633e-05,
+      "loss": 0.5343,
+      "step": 894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9647189153560592e-05,
+      "loss": 0.6145,
+      "step": 896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.964558222821292e-05,
+      "loss": 0.5997,
+      "step": 898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9643971717710597e-05,
+      "loss": 0.5252,
+      "step": 900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9642357622652225e-05,
+      "loss": 0.5751,
+      "step": 902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9640739943637753e-05,
+      "loss": 0.5853,
+      "step": 904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9639118681268446e-05,
+      "loss": 0.6531,
+      "step": 906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9637493836146918e-05,
+      "loss": 0.5948,
+      "step": 908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9635865408877092e-05,
+      "loss": 0.6614,
+      "step": 910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9634233400064252e-05,
+      "loss": 0.7063,
+      "step": 912
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9632597810314985e-05,
+      "loss": 0.6144,
+      "step": 914
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.963095864023723e-05,
+      "loss": 0.5594,
+      "step": 916
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.962931589044024e-05,
+      "loss": 0.6239,
+      "step": 918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9627669561534614e-05,
+      "loss": 0.588,
+      "step": 920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9626019654132273e-05,
+      "loss": 0.4905,
+      "step": 922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9624366168846462e-05,
+      "loss": 0.5833,
+      "step": 924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9622709106291768e-05,
+      "loss": 0.4711,
+      "step": 926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9621048467084107e-05,
+      "loss": 0.7316,
+      "step": 928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.961938425184071e-05,
+      "loss": 0.5823,
+      "step": 930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9617716461180157e-05,
+      "loss": 0.531,
+      "step": 932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9616045095722343e-05,
+      "loss": 0.6414,
+      "step": 934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.961437015608849e-05,
+      "loss": 0.552,
+      "step": 936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9612691642901166e-05,
+      "loss": 0.655,
+      "step": 938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9611009556784247e-05,
+      "loss": 0.5614,
+      "step": 940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9609323898362946e-05,
+      "loss": 0.5865,
+      "step": 942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9607634668263806e-05,
+      "loss": 0.5598,
+      "step": 944
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9605941867114696e-05,
+      "loss": 0.6202,
+      "step": 946
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9604245495544805e-05,
+      "loss": 0.6514,
+      "step": 948
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9602545554184662e-05,
+      "loss": 0.4977,
+      "step": 950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9600842043666112e-05,
+      "loss": 0.6059,
+      "step": 952
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9599134964622337e-05,
+      "loss": 0.6796,
+      "step": 954
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9597424317687827e-05,
+      "loss": 0.6111,
+      "step": 956
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.959571010349842e-05,
+      "loss": 0.5642,
+      "step": 958
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.959399232269127e-05,
+      "loss": 0.6166,
+      "step": 960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9592270975904853e-05,
+      "loss": 0.5971,
+      "step": 962
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9590546063778976e-05,
+      "loss": 0.5239,
+      "step": 964
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9588817586954772e-05,
+      "loss": 0.5714,
+      "step": 966
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.958708554607469e-05,
+      "loss": 0.5439,
+      "step": 968
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9585349941782518e-05,
+      "loss": 0.65,
+      "step": 970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.958361077472335e-05,
+      "loss": 0.5621,
+      "step": 972
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.958186804554363e-05,
+      "loss": 0.5545,
+      "step": 974
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9580121754891094e-05,
+      "loss": 0.6022,
+      "step": 976
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.957837190341483e-05,
+      "loss": 0.6764,
+      "step": 978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.957661849176524e-05,
+      "loss": 0.6177,
+      "step": 980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9574861520594037e-05,
+      "loss": 0.5638,
+      "step": 982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.957310099055427e-05,
+      "loss": 0.6482,
+      "step": 984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.957133690230032e-05,
+      "loss": 0.6454,
+      "step": 986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9569569256487866e-05,
+      "loss": 0.6079,
+      "step": 988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9567798053773924e-05,
+      "loss": 0.6061,
+      "step": 990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9566023294816836e-05,
+      "loss": 0.5909,
+      "step": 992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9564244980276256e-05,
+      "loss": 0.6213,
+      "step": 994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.956246311081316e-05,
+      "loss": 0.5524,
+      "step": 996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9560677687089858e-05,
+      "loss": 0.5497,
+      "step": 998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9558888709769964e-05,
+      "loss": 0.5445,
+      "step": 1000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9557096179518423e-05,
+      "loss": 0.7148,
+      "step": 1002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.95553000970015e-05,
+      "loss": 0.6641,
+      "step": 1004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.955350046288678e-05,
+      "loss": 0.6558,
+      "step": 1006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.955169727784316e-05,
+      "loss": 0.5617,
+      "step": 1008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.954989054254087e-05,
+      "loss": 0.5761,
+      "step": 1010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9548080257651453e-05,
+      "loss": 0.6106,
+      "step": 1012
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.954626642384777e-05,
+      "loss": 0.5274,
+      "step": 1014
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9544449041804e-05,
+      "loss": 0.6526,
+      "step": 1016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9542628112195644e-05,
+      "loss": 0.6327,
+      "step": 1018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9540803635699527e-05,
+      "loss": 0.4861,
+      "step": 1020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9538975612993782e-05,
+      "loss": 0.6781,
+      "step": 1022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9537144044757865e-05,
+      "loss": 0.6084,
+      "step": 1024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.953530893167255e-05,
+      "loss": 0.6102,
+      "step": 1026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9533470274419928e-05,
+      "loss": 0.5916,
+      "step": 1028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9531628073683404e-05,
+      "loss": 0.5629,
+      "step": 1030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9529782330147708e-05,
+      "loss": 0.6212,
+      "step": 1032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.952793304449888e-05,
+      "loss": 0.6145,
+      "step": 1034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.952608021742428e-05,
+      "loss": 0.6259,
+      "step": 1036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9524223849612577e-05,
+      "loss": 0.5013,
+      "step": 1038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.952236394175377e-05,
+      "loss": 0.535,
+      "step": 1040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9520500494539164e-05,
+      "loss": 0.5421,
+      "step": 1042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.951863350866138e-05,
+      "loss": 0.5772,
+      "step": 1044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9516762984814357e-05,
+      "loss": 0.5532,
+      "step": 1046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9514888923693345e-05,
+      "loss": 0.5715,
+      "step": 1048
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9513011325994912e-05,
+      "loss": 0.5877,
+      "step": 1050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9511130192416943e-05,
+      "loss": 0.6468,
+      "step": 1052
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.950924552365863e-05,
+      "loss": 0.5482,
+      "step": 1054
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9507357320420487e-05,
+      "loss": 0.6255,
+      "step": 1056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.950546558340434e-05,
+      "loss": 0.6123,
+      "step": 1058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.950357031331332e-05,
+      "loss": 0.6309,
+      "step": 1060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.950167151085188e-05,
+      "loss": 0.5981,
+      "step": 1062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9499769176725785e-05,
+      "loss": 0.653,
+      "step": 1064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.949786331164211e-05,
+      "loss": 0.5706,
+      "step": 1066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9495953916309247e-05,
+      "loss": 0.5031,
+      "step": 1068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.949404099143689e-05,
+      "loss": 0.581,
+      "step": 1070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9492124537736056e-05,
+      "loss": 0.6317,
+      "step": 1072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.949020455591907e-05,
+      "loss": 0.6181,
+      "step": 1074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9488281046699565e-05,
+      "loss": 0.6569,
+      "step": 1076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9486354010792495e-05,
+      "loss": 0.6451,
+      "step": 1078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.948442344891411e-05,
+      "loss": 0.6046,
+      "step": 1080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9482489361781977e-05,
+      "loss": 0.5443,
+      "step": 1082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9480551750114977e-05,
+      "loss": 0.7137,
+      "step": 1084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.94786106146333e-05,
+      "loss": 0.5656,
+      "step": 1086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9476665956058448e-05,
+      "loss": 0.6597,
+      "step": 1088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9474717775113218e-05,
+      "loss": 0.6356,
+      "step": 1090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.947276607252174e-05,
+      "loss": 0.5186,
+      "step": 1092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9470810849009428e-05,
+      "loss": 0.4941,
+      "step": 1094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9468852105303022e-05,
+      "loss": 0.6725,
+      "step": 1096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9466889842130566e-05,
+      "loss": 0.5597,
+      "step": 1098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.946492406022141e-05,
+      "loss": 0.6739,
+      "step": 1100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9462954760306213e-05,
+      "loss": 0.5792,
+      "step": 1102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9460981943116943e-05,
+      "loss": 0.5691,
+      "step": 1104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945900560938687e-05,
+      "loss": 0.5493,
+      "step": 1106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9457025759850582e-05,
+      "loss": 0.628,
+      "step": 1108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945504239524396e-05,
+      "loss": 0.5978,
+      "step": 1110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9453055516304202e-05,
+      "loss": 0.6084,
+      "step": 1112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9451065123769808e-05,
+      "loss": 0.5876,
+      "step": 1114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9449071218380585e-05,
+      "loss": 0.6838,
+      "step": 1116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9447073800877643e-05,
+      "loss": 0.5132,
+      "step": 1118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9445072872003402e-05,
+      "loss": 0.5561,
+      "step": 1120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.944306843250158e-05,
+      "loss": 0.6176,
+      "step": 1122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9441060483117215e-05,
+      "loss": 0.6251,
+      "step": 1124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.943904902459663e-05,
+      "loss": 0.6201,
+      "step": 1126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9437034057687464e-05,
+      "loss": 0.6082,
+      "step": 1128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9435015583138657e-05,
+      "loss": 0.6236,
+      "step": 1130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9432993601700455e-05,
+      "loss": 0.6357,
+      "step": 1132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9430968114124408e-05,
+      "loss": 0.5078,
+      "step": 1134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942893912116336e-05,
+      "loss": 0.6138,
+      "step": 1136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9426906623571475e-05,
+      "loss": 0.6766,
+      "step": 1138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.94248706221042e-05,
+      "loss": 0.5546,
+      "step": 1140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9422831117518298e-05,
+      "loss": 0.6393,
+      "step": 1142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942078811057183e-05,
+      "loss": 0.5881,
+      "step": 1144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9418741602024157e-05,
+      "loss": 0.6799,
+      "step": 1146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9416691592635945e-05,
+      "loss": 0.5532,
+      "step": 1148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9414638083169158e-05,
+      "loss": 0.6789,
+      "step": 1150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9412581074387066e-05,
+      "loss": 0.4965,
+      "step": 1152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.941052056705423e-05,
+      "loss": 0.538,
+      "step": 1154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9408456561936523e-05,
+      "loss": 0.5856,
+      "step": 1156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.940638905980111e-05,
+      "loss": 0.5881,
+      "step": 1158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9404318061416457e-05,
+      "loss": 0.5802,
+      "step": 1160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9402243567552335e-05,
+      "loss": 0.5579,
+      "step": 1162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.940016557897981e-05,
+      "loss": 0.5414,
+      "step": 1164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.939808409647124e-05,
+      "loss": 0.4745,
+      "step": 1166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9395999120800298e-05,
+      "loss": 0.5447,
+      "step": 1168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9393910652741937e-05,
+      "loss": 0.6156,
+      "step": 1170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9391818693072423e-05,
+      "loss": 0.6465,
+      "step": 1172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9389723242569316e-05,
+      "loss": 0.6854,
+      "step": 1174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9387624302011464e-05,
+      "loss": 0.6886,
+      "step": 1176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9385521872179028e-05,
+      "loss": 0.639,
+      "step": 1178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.938341595385345e-05,
+      "loss": 0.5662,
+      "step": 1180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9381306547817478e-05,
+      "loss": 0.4757,
+      "step": 1182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.937919365485516e-05,
+      "loss": 0.5197,
+      "step": 1184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.937707727575183e-05,
+      "loss": 0.5676,
+      "step": 1186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.937495741129412e-05,
+      "loss": 0.546,
+      "step": 1188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9372834062269964e-05,
+      "loss": 0.5532,
+      "step": 1190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9370707229468584e-05,
+      "loss": 0.6595,
+      "step": 1192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9368576913680502e-05,
+      "loss": 0.6766,
+      "step": 1194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9366443115697537e-05,
+      "loss": 0.6226,
+      "step": 1196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.936430583631279e-05,
+      "loss": 0.5202,
+      "step": 1198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9362165076320667e-05,
+      "loss": 0.5807,
+      "step": 1200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9360020836516863e-05,
+      "loss": 0.5398,
+      "step": 1202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9357873117698368e-05,
+      "loss": 0.5116,
+      "step": 1204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9355721920663467e-05,
+      "loss": 0.5631,
+      "step": 1206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9353567246211736e-05,
+      "loss": 0.5819,
+      "step": 1208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9351409095144043e-05,
+      "loss": 0.5225,
+      "step": 1210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9349247468262544e-05,
+      "loss": 0.5086,
+      "step": 1212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.93470823663707e-05,
+      "loss": 0.4927,
+      "step": 1214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9344913790273246e-05,
+      "loss": 0.6887,
+      "step": 1216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.934274174077622e-05,
+      "loss": 0.5887,
+      "step": 1218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9340566218686952e-05,
+      "loss": 0.5794,
+      "step": 1220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9338387224814063e-05,
+      "loss": 0.6053,
+      "step": 1222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9336204759967448e-05,
+      "loss": 0.584,
+      "step": 1224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.933401882495831e-05,
+      "loss": 0.6498,
+      "step": 1226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9331829420599143e-05,
+      "loss": 0.5892,
+      "step": 1228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9329636547703715e-05,
+      "loss": 0.5942,
+      "step": 1230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9327440207087098e-05,
+      "loss": 0.6046,
+      "step": 1232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9325240399565643e-05,
+      "loss": 0.6261,
+      "step": 1234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9323037125956997e-05,
+      "loss": 0.6278,
+      "step": 1236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9320830387080093e-05,
+      "loss": 0.5263,
+      "step": 1238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9318620183755148e-05,
+      "loss": 0.5537,
+      "step": 1240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9316406516803668e-05,
+      "loss": 0.6146,
+      "step": 1242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9314189387048458e-05,
+      "loss": 0.6696,
+      "step": 1244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.931196879531359e-05,
+      "loss": 0.6757,
+      "step": 1246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9309744742424432e-05,
+      "loss": 0.7317,
+      "step": 1248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9307517229207652e-05,
+      "loss": 0.56,
+      "step": 1250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9305286256491176e-05,
+      "loss": 0.5865,
+      "step": 1252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9303051825104244e-05,
+      "loss": 0.6094,
+      "step": 1254
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.930081393587736e-05,
+      "loss": 0.5544,
+      "step": 1256
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.929857258964233e-05,
+      "loss": 0.5458,
+      "step": 1258
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9296327787232232e-05,
+      "loss": 0.5444,
+      "step": 1260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9294079529481435e-05,
+      "loss": 0.549,
+      "step": 1262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.929182781722559e-05,
+      "loss": 0.5879,
+      "step": 1264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9289572651301644e-05,
+      "loss": 0.7001,
+      "step": 1266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.92873140325478e-05,
+      "loss": 0.5152,
+      "step": 1268
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.928505196180357e-05,
+      "loss": 0.6534,
+      "step": 1270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.928278643990974e-05,
+      "loss": 0.548,
+      "step": 1272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9280517467708377e-05,
+      "loss": 0.5577,
+      "step": 1274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9278245046042835e-05,
+      "loss": 0.5684,
+      "step": 1276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9275969175757746e-05,
+      "loss": 0.5923,
+      "step": 1278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.927368985769902e-05,
+      "loss": 0.7258,
+      "step": 1280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.927140709271387e-05,
+      "loss": 0.5459,
+      "step": 1282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.926912088165076e-05,
+      "loss": 0.6263,
+      "step": 1284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9266831225359453e-05,
+      "loss": 0.5043,
+      "step": 1286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9264538124690987e-05,
+      "loss": 0.5612,
+      "step": 1288
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9262241580497685e-05,
+      "loss": 0.639,
+      "step": 1290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9259941593633145e-05,
+      "loss": 0.5485,
+      "step": 1292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9257638164952247e-05,
+      "loss": 0.5658,
+      "step": 1294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9255331295311146e-05,
+      "loss": 0.5069,
+      "step": 1296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9253020985567285e-05,
+      "loss": 0.5919,
+      "step": 1298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9250707236579376e-05,
+      "loss": 0.6436,
+      "step": 1300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9248390049207416e-05,
+      "loss": 0.5472,
+      "step": 1302
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9246069424312678e-05,
+      "loss": 0.6824,
+      "step": 1304
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.924374536275771e-05,
+      "loss": 0.5441,
+      "step": 1306
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.924141786540634e-05,
+      "loss": 0.6617,
+      "step": 1308
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.923908693312367e-05,
+      "loss": 0.6006,
+      "step": 1310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.923675256677609e-05,
+      "loss": 0.6422,
+      "step": 1312
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.923441476723125e-05,
+      "loss": 0.533,
+      "step": 1314
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.923207353535809e-05,
+      "loss": 0.6135,
+      "step": 1316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.922972887202681e-05,
+      "loss": 0.5906,
+      "step": 1318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.92273807781089e-05,
+      "loss": 0.4476,
+      "step": 1320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9225029254477126e-05,
+      "loss": 0.5984,
+      "step": 1322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.922267430200552e-05,
+      "loss": 0.463,
+      "step": 1324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9220315921569384e-05,
+      "loss": 0.5288,
+      "step": 1326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9217954114045304e-05,
+      "loss": 0.6625,
+      "step": 1328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9215588880311147e-05,
+      "loss": 0.5577,
+      "step": 1330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9213220221246035e-05,
+      "loss": 0.6646,
+      "step": 1332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.921084813773037e-05,
+      "loss": 0.5474,
+      "step": 1334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.920847263064584e-05,
+      "loss": 0.515,
+      "step": 1336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9206093700875383e-05,
+      "loss": 0.5399,
+      "step": 1338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9203711349303225e-05,
+      "loss": 0.4982,
+      "step": 1340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9201325576814858e-05,
+      "loss": 0.6383,
+      "step": 1342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9198936384297048e-05,
+      "loss": 0.6108,
+      "step": 1344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.919654377263783e-05,
+      "loss": 0.5554,
+      "step": 1346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9194147742726517e-05,
+      "loss": 0.5153,
+      "step": 1348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9191748295453677e-05,
+      "loss": 0.5027,
+      "step": 1350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.918934543171116e-05,
+      "loss": 0.5686,
+      "step": 1352
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9186939152392087e-05,
+      "loss": 0.6614,
+      "step": 1354
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9184529458390845e-05,
+      "loss": 0.4884,
+      "step": 1356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.918211635060309e-05,
+      "loss": 0.6031,
+      "step": 1358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.917969982992574e-05,
+      "loss": 0.5216,
+      "step": 1360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9177279897257e-05,
+      "loss": 0.644,
+      "step": 1362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9174856553496323e-05,
+      "loss": 0.5776,
+      "step": 1364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9172429799544444e-05,
+      "loss": 0.6399,
+      "step": 1366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9169999636303356e-05,
+      "loss": 0.5529,
+      "step": 1368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9167566064676325e-05,
+      "loss": 0.6757,
+      "step": 1370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9165129085567885e-05,
+      "loss": 0.5543,
+      "step": 1372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9162688699883828e-05,
+      "loss": 0.5279,
+      "step": 1374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9160244908531226e-05,
+      "loss": 0.6148,
+      "step": 1376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.91577977124184e-05,
+      "loss": 0.4855,
+      "step": 1378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9155347112454948e-05,
+      "loss": 0.531,
+      "step": 1380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9152893109551736e-05,
+      "loss": 0.621,
+      "step": 1382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9150435704620886e-05,
+      "loss": 0.5847,
+      "step": 1384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9147974898575782e-05,
+      "loss": 0.4894,
+      "step": 1386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9145510692331086e-05,
+      "loss": 0.5849,
+      "step": 1388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9143043086802714e-05,
+      "loss": 0.5384,
+      "step": 1390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9140572082907847e-05,
+      "loss": 0.5552,
+      "step": 1392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9138097681564926e-05,
+      "loss": 0.5894,
+      "step": 1394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.913561988369366e-05,
+      "loss": 0.5431,
+      "step": 1396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9133138690215025e-05,
+      "loss": 0.561,
+      "step": 1398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.913065410205124e-05,
+      "loss": 0.5032,
+      "step": 1400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9128166120125814e-05,
+      "loss": 0.5467,
+      "step": 1402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.912567474536349e-05,
+      "loss": 0.534,
+      "step": 1404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9123179978690286e-05,
+      "loss": 0.5117,
+      "step": 1406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9120681821033484e-05,
+      "loss": 0.5814,
+      "step": 1408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9118180273321613e-05,
+      "loss": 0.5293,
+      "step": 1410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9115675336484478e-05,
+      "loss": 0.4986,
+      "step": 1412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9113167011453134e-05,
+      "loss": 0.6765,
+      "step": 1414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9110655299159894e-05,
+      "loss": 0.6499,
+      "step": 1416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9108140200538337e-05,
+      "loss": 0.6838,
+      "step": 1418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9105621716523293e-05,
+      "loss": 0.5712,
+      "step": 1420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9103099848050858e-05,
+      "loss": 0.4988,
+      "step": 1422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.910057459605838e-05,
+      "loss": 0.5992,
+      "step": 1424
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9098045961484468e-05,
+      "loss": 0.5766,
+      "step": 1426
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.909551394526899e-05,
+      "loss": 0.5595,
+      "step": 1428
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9092978548353058e-05,
+      "loss": 0.6893,
+      "step": 1430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9090439771679063e-05,
+      "loss": 0.5048,
+      "step": 1432
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.908789761619063e-05,
+      "loss": 0.6083,
+      "step": 1434
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9085352082832657e-05,
+      "loss": 0.5672,
+      "step": 1436
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9082803172551285e-05,
+      "loss": 0.617,
+      "step": 1438
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9080250886293918e-05,
+      "loss": 0.5596,
+      "step": 1440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9077695225009212e-05,
+      "loss": 0.5568,
+      "step": 1442
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9075136189647073e-05,
+      "loss": 0.5571,
+      "step": 1444
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9072573781158677e-05,
+      "loss": 0.6135,
+      "step": 1446
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9070008000496433e-05,
+      "loss": 0.5251,
+      "step": 1448
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.906743884861401e-05,
+      "loss": 0.5848,
+      "step": 1450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.906486632646634e-05,
+      "loss": 0.6429,
+      "step": 1452
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.90622904350096e-05,
+      "loss": 0.6232,
+      "step": 1454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.905971117520122e-05,
+      "loss": 0.6064,
+      "step": 1456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.905712854799988e-05,
+      "loss": 0.551,
+      "step": 1458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9054542554365514e-05,
+      "loss": 0.4942,
+      "step": 1460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9051953195259303e-05,
+      "loss": 0.605,
+      "step": 1462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.904936047164369e-05,
+      "loss": 0.4846,
+      "step": 1464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9046764384482356e-05,
+      "loss": 0.5796,
+      "step": 1466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9044164934740242e-05,
+      "loss": 0.5281,
+      "step": 1468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.904156212338353e-05,
+      "loss": 0.5493,
+      "step": 1470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9038955951379656e-05,
+      "loss": 0.5251,
+      "step": 1472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.903634641969731e-05,
+      "loss": 0.5468,
+      "step": 1474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.903373352930642e-05,
+      "loss": 0.5835,
+      "step": 1476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.903111728117817e-05,
+      "loss": 0.6492,
+      "step": 1478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9028497676284987e-05,
+      "loss": 0.5439,
+      "step": 1480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9025874715600555e-05,
+      "loss": 0.7175,
+      "step": 1482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9023248400099793e-05,
+      "loss": 0.5514,
+      "step": 1484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.902061873075888e-05,
+      "loss": 0.61,
+      "step": 1486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9017985708555226e-05,
+      "loss": 0.5736,
+      "step": 1488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.90153493344675e-05,
+      "loss": 0.6859,
+      "step": 1490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9012709609475618e-05,
+      "loss": 0.5755,
+      "step": 1492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9010066534560728e-05,
+      "loss": 0.5433,
+      "step": 1494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9007420110705233e-05,
+      "loss": 0.5988,
+      "step": 1496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9004770338892782e-05,
+      "loss": 0.5627,
+      "step": 1498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.900211722010826e-05,
+      "loss": 0.5707,
+      "step": 1500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.899946075533781e-05,
+      "loss": 0.5523,
+      "step": 1502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8996800945568802e-05,
+      "loss": 0.5343,
+      "step": 1504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8994137791789865e-05,
+      "loss": 0.497,
+      "step": 1506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8991471294990855e-05,
+      "loss": 0.5255,
+      "step": 1508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.898880145616288e-05,
+      "loss": 0.6533,
+      "step": 1510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8986128276298303e-05,
+      "loss": 0.5818,
+      "step": 1512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8983451756390693e-05,
+      "loss": 0.5483,
+      "step": 1514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8980771897434898e-05,
+      "loss": 0.6135,
+      "step": 1516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8978088700426988e-05,
+      "loss": 0.5934,
+      "step": 1518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8975402166364274e-05,
+      "loss": 0.5212,
+      "step": 1520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8972712296245316e-05,
+      "loss": 0.5999,
+      "step": 1522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.89700190910699e-05,
+      "loss": 0.5798,
+      "step": 1524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8967322551839072e-05,
+      "loss": 0.591,
+      "step": 1526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8964622679555097e-05,
+      "loss": 0.5338,
+      "step": 1528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8961919475221487e-05,
+      "loss": 0.5769,
+      "step": 1530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8959212939842994e-05,
+      "loss": 0.4957,
+      "step": 1532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.895650307442561e-05,
+      "loss": 0.5037,
+      "step": 1534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8953789879976557e-05,
+      "loss": 0.61,
+      "step": 1536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8951073357504304e-05,
+      "loss": 0.5521,
+      "step": 1538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8948353508018545e-05,
+      "loss": 0.6333,
+      "step": 1540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8945630332530226e-05,
+      "loss": 0.6029,
+      "step": 1542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8942903832051516e-05,
+      "loss": 0.6066,
+      "step": 1544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.894017400759582e-05,
+      "loss": 0.5514,
+      "step": 1546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8937440860177793e-05,
+      "loss": 0.611,
+      "step": 1548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.893470439081331e-05,
+      "loss": 0.5613,
+      "step": 1550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8931964600519484e-05,
+      "loss": 0.5775,
+      "step": 1552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8929221490314663e-05,
+      "loss": 0.5648,
+      "step": 1554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.892647506121844e-05,
+      "loss": 0.5802,
+      "step": 1556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.892372531425162e-05,
+      "loss": 0.5806,
+      "step": 1558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8920972250436266e-05,
+      "loss": 0.5508,
+      "step": 1560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.891821587079565e-05,
+      "loss": 0.5495,
+      "step": 1562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.891545617635429e-05,
+      "loss": 0.5946,
+      "step": 1564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8912693168137934e-05,
+      "loss": 0.5634,
+      "step": 1566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8909926847173566e-05,
+      "loss": 0.5874,
+      "step": 1568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8907157214489386e-05,
+      "loss": 0.6399,
+      "step": 1570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8904384271114848e-05,
+      "loss": 0.5244,
+      "step": 1572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8901608018080613e-05,
+      "loss": 0.5942,
+      "step": 1574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8898828456418587e-05,
+      "loss": 0.622,
+      "step": 1576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.88960455871619e-05,
+      "loss": 0.5134,
+      "step": 1578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.889325941134492e-05,
+      "loss": 0.5779,
+      "step": 1580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8890469930003233e-05,
+      "loss": 0.5726,
+      "step": 1582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.888767714417366e-05,
+      "loss": 0.5765,
+      "step": 1584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.888488105489424e-05,
+      "loss": 0.6367,
+      "step": 1586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8882081663204258e-05,
+      "loss": 0.5699,
+      "step": 1588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.887927897014421e-05,
+      "loss": 0.5613,
+      "step": 1590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.887647297675583e-05,
+      "loss": 0.5637,
+      "step": 1592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8873663684082074e-05,
+      "loss": 0.6095,
+      "step": 1594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8870851093167117e-05,
+      "loss": 0.5233,
+      "step": 1596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.886803520505638e-05,
+      "loss": 0.5782,
+      "step": 1598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8865216020796486e-05,
+      "loss": 0.6292,
+      "step": 1600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8862393541435296e-05,
+      "loss": 0.5586,
+      "step": 1602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.88595677680219e-05,
+      "loss": 0.6182,
+      "step": 1604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8856738701606602e-05,
+      "loss": 0.5966,
+      "step": 1606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8853906343240934e-05,
+      "loss": 0.6814,
+      "step": 1608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8851070693977656e-05,
+      "loss": 0.5717,
+      "step": 1610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.884823175487074e-05,
+      "loss": 0.6391,
+      "step": 1612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.884538952697539e-05,
+      "loss": 0.5036,
+      "step": 1614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8842544011348037e-05,
+      "loss": 0.6018,
+      "step": 1616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8839695209046317e-05,
+      "loss": 0.6042,
+      "step": 1618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8836843121129106e-05,
+      "loss": 0.6586,
+      "step": 1620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8833987748656488e-05,
+      "loss": 0.6025,
+      "step": 1622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.883112909268977e-05,
+      "loss": 0.6254,
+      "step": 1624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8828267154291494e-05,
+      "loss": 0.6459,
+      "step": 1626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8825401934525403e-05,
+      "loss": 0.6183,
+      "step": 1628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.882253343445647e-05,
+      "loss": 0.5933,
+      "step": 1630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8819661655150873e-05,
+      "loss": 0.5052,
+      "step": 1632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8816786597676033e-05,
+      "loss": 0.5698,
+      "step": 1634
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8813908263100575e-05,
+      "loss": 0.5631,
+      "step": 1636
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.881102665249434e-05,
+      "loss": 0.5997,
+      "step": 1638
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8808141766928387e-05,
+      "loss": 0.5284,
+      "step": 1640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8805253607475004e-05,
+      "loss": 0.5682,
+      "step": 1642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8802362175207682e-05,
+      "loss": 0.5742,
+      "step": 1644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8799467471201133e-05,
+      "loss": 0.5089,
+      "step": 1646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8796569496531294e-05,
+      "loss": 0.6551,
+      "step": 1648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.87936682522753e-05,
+      "loss": 0.6407,
+      "step": 1650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8790763739511515e-05,
+      "loss": 0.5345,
+      "step": 1652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8787855959319518e-05,
+      "loss": 0.5495,
+      "step": 1654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.878494491278009e-05,
+      "loss": 0.5653,
+      "step": 1656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8782030600975236e-05,
+      "loss": 0.5109,
+      "step": 1658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.877911302498818e-05,
+      "loss": 0.5484,
+      "step": 1660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8776192185903345e-05,
+      "loss": 0.5834,
+      "step": 1662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8773268084806376e-05,
+      "loss": 0.6688,
+      "step": 1664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8770340722784132e-05,
+      "loss": 0.5444,
+      "step": 1666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.876741010092468e-05,
+      "loss": 0.5842,
+      "step": 1668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8764476220317292e-05,
+      "loss": 0.5023,
+      "step": 1670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8761539082052464e-05,
+      "loss": 0.5929,
+      "step": 1672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8758598687221897e-05,
+      "loss": 0.5675,
+      "step": 1674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8755655036918505e-05,
+      "loss": 0.5941,
+      "step": 1676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8752708132236405e-05,
+      "loss": 0.5411,
+      "step": 1678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8749757974270933e-05,
+      "loss": 0.5635,
+      "step": 1680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.874680456411863e-05,
+      "loss": 0.5019,
+      "step": 1682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8743847902877233e-05,
+      "loss": 0.5631,
+      "step": 1684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.874088799164572e-05,
+      "loss": 0.6827,
+      "step": 1686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8737924831524243e-05,
+      "loss": 0.4641,
+      "step": 1688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8734958423614175e-05,
+      "loss": 0.6909,
+      "step": 1690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8731988769018103e-05,
+      "loss": 0.5763,
+      "step": 1692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.872901586883981e-05,
+      "loss": 0.5694,
+      "step": 1694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8726039724184296e-05,
+      "loss": 0.5469,
+      "step": 1696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8723060336157752e-05,
+      "loss": 0.5729,
+      "step": 1698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.872007770586759e-05,
+      "loss": 0.4758,
+      "step": 1700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8717091834422415e-05,
+      "loss": 0.5997,
+      "step": 1702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8714102722932042e-05,
+      "loss": 0.6396,
+      "step": 1704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8711110372507494e-05,
+      "loss": 0.6007,
+      "step": 1706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8708114784260992e-05,
+      "loss": 0.5736,
+      "step": 1708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.870511595930596e-05,
+      "loss": 0.6317,
+      "step": 1710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.870211389875704e-05,
+      "loss": 0.5945,
+      "step": 1712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8699108603730045e-05,
+      "loss": 0.5703,
+      "step": 1714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.869610007534202e-05,
+      "loss": 0.6212,
+      "step": 1716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.86930883147112e-05,
+      "loss": 0.5416,
+      "step": 1718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8690073322957026e-05,
+      "loss": 0.5961,
+      "step": 1720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8687055101200132e-05,
+      "loss": 0.5614,
+      "step": 1722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8684033650562357e-05,
+      "loss": 0.5421,
+      "step": 1724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8681008972166746e-05,
+      "loss": 0.5238,
+      "step": 1726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8677981067137533e-05,
+      "loss": 0.5389,
+      "step": 1728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8674949936600156e-05,
+      "loss": 0.5831,
+      "step": 1730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8671915581681253e-05,
+      "loss": 0.5482,
+      "step": 1732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8668878003508658e-05,
+      "loss": 0.5049,
+      "step": 1734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.866583720321141e-05,
+      "loss": 0.5048,
+      "step": 1736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8662793181919734e-05,
+      "loss": 0.6042,
+      "step": 1738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8659745940765064e-05,
+      "loss": 0.5505,
+      "step": 1740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.865669548088002e-05,
+      "loss": 0.5386,
+      "step": 1742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8653641803398434e-05,
+      "loss": 0.4597,
+      "step": 1744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8650584909455308e-05,
+      "loss": 0.5598,
+      "step": 1746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8647524800186868e-05,
+      "loss": 0.5611,
+      "step": 1748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8644461476730516e-05,
+      "loss": 0.476,
+      "step": 1750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8641394940224858e-05,
+      "loss": 0.4891,
+      "step": 1752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8638325191809692e-05,
+      "loss": 0.5815,
+      "step": 1754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8635252232626004e-05,
+      "loss": 0.6075,
+      "step": 1756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8632176063815983e-05,
+      "loss": 0.5341,
+      "step": 1758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8629096686523002e-05,
+      "loss": 0.5832,
+      "step": 1760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8626014101891635e-05,
+      "loss": 0.5788,
+      "step": 1762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8622928311067647e-05,
+      "loss": 0.5739,
+      "step": 1764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8619839315197983e-05,
+      "loss": 0.6352,
+      "step": 1766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8616747115430793e-05,
+      "loss": 0.5641,
+      "step": 1768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8613651712915414e-05,
+      "loss": 0.5412,
+      "step": 1770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.861055310880237e-05,
+      "loss": 0.6429,
+      "step": 1772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.860745130424338e-05,
+      "loss": 0.5023,
+      "step": 1774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.860434630039135e-05,
+      "loss": 0.6452,
+      "step": 1776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8601238098400368e-05,
+      "loss": 0.5796,
+      "step": 1778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.859812669942573e-05,
+      "loss": 0.507,
+      "step": 1780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8595012104623897e-05,
+      "loss": 0.604,
+      "step": 1782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8591894315152534e-05,
+      "loss": 0.5975,
+      "step": 1784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.858877333217049e-05,
+      "loss": 0.577,
+      "step": 1786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8585649156837794e-05,
+      "loss": 0.568,
+      "step": 1788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.858252179031567e-05,
+      "loss": 0.5464,
+      "step": 1790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8579391233766527e-05,
+      "loss": 0.5616,
+      "step": 1792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.857625748835395e-05,
+      "loss": 0.5645,
+      "step": 1794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8573120555242733e-05,
+      "loss": 0.5609,
+      "step": 1796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.856998043559882e-05,
+      "loss": 0.533,
+      "step": 1798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.856683713058937e-05,
+      "loss": 0.5889,
+      "step": 1800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.856369064138271e-05,
+      "loss": 0.4551,
+      "step": 1802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8560540969148358e-05,
+      "loss": 0.5304,
+      "step": 1804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.855738811505701e-05,
+      "loss": 0.6365,
+      "step": 1806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.855423208028054e-05,
+      "loss": 0.4874,
+      "step": 1808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.855107286599202e-05,
+      "loss": 0.5416,
+      "step": 1810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8547910473365697e-05,
+      "loss": 0.5518,
+      "step": 1812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8544744903576983e-05,
+      "loss": 0.5058,
+      "step": 1814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8541576157802495e-05,
+      "loss": 0.5838,
+      "step": 1816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8538404237220016e-05,
+      "loss": 0.586,
+      "step": 1818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.853522914300852e-05,
+      "loss": 0.5475,
+      "step": 1820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8532050876348146e-05,
+      "loss": 0.5092,
+      "step": 1822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8528869438420225e-05,
+      "loss": 0.6056,
+      "step": 1824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.852568483040726e-05,
+      "loss": 0.527,
+      "step": 1826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8522497053492935e-05,
+      "loss": 0.5974,
+      "step": 1828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8519306108862105e-05,
+      "loss": 0.5405,
+      "step": 1830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8516111997700817e-05,
+      "loss": 0.6404,
+      "step": 1832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.851291472119628e-05,
+      "loss": 0.5159,
+      "step": 1834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8509714280536894e-05,
+      "loss": 0.5354,
+      "step": 1836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8506510676912215e-05,
+      "loss": 0.563,
+      "step": 1838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8503303911512996e-05,
+      "loss": 0.5687,
+      "step": 1840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8500093985531152e-05,
+      "loss": 0.5313,
+      "step": 1842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8496880900159776e-05,
+      "loss": 0.6295,
+      "step": 1844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8493664656593134e-05,
+      "loss": 0.5125,
+      "step": 1846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8490445256026674e-05,
+      "loss": 0.484,
+      "step": 1848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.848722269965701e-05,
+      "loss": 0.5819,
+      "step": 1850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8483996988681922e-05,
+      "loss": 0.5551,
+      "step": 1852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8480768124300378e-05,
+      "loss": 0.6238,
+      "step": 1854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8477536107712507e-05,
+      "loss": 0.5759,
+      "step": 1856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8474300940119615e-05,
+      "loss": 0.6275,
+      "step": 1858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8471062622724182e-05,
+      "loss": 0.54,
+      "step": 1860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8467821156729844e-05,
+      "loss": 0.4778,
+      "step": 1862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8464576543341427e-05,
+      "loss": 0.6007,
+      "step": 1864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.846132878376491e-05,
+      "loss": 0.6636,
+      "step": 1866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8458077879207456e-05,
+      "loss": 0.5114,
+      "step": 1868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.845482383087738e-05,
+      "loss": 0.5008,
+      "step": 1870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8451566639984188e-05,
+      "loss": 0.5377,
+      "step": 1872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8448306307738532e-05,
+      "loss": 0.4504,
+      "step": 1874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8445042835352243e-05,
+      "loss": 0.6138,
+      "step": 1876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.844177622403832e-05,
+      "loss": 0.5235,
+      "step": 1878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8438506475010916e-05,
+      "loss": 0.6279,
+      "step": 1880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8435233589485376e-05,
+      "loss": 0.4175,
+      "step": 1882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.843195756867818e-05,
+      "loss": 0.6113,
+      "step": 1884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8428678413806998e-05,
+      "loss": 0.5809,
+      "step": 1886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.842539612609065e-05,
+      "loss": 0.6177,
+      "step": 1888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8422110706749124e-05,
+      "loss": 0.5803,
+      "step": 1890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.841882215700358e-05,
+      "loss": 0.5488,
+      "step": 1892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.841553047807633e-05,
+      "loss": 0.6073,
+      "step": 1894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8412235671190858e-05,
+      "loss": 0.7104,
+      "step": 1896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.84089377375718e-05,
+      "loss": 0.5501,
+      "step": 1898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8405636678444968e-05,
+      "loss": 0.6196,
+      "step": 1900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8402332495037326e-05,
+      "loss": 0.5739,
+      "step": 1902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8399025188577e-05,
+      "loss": 0.602,
+      "step": 1904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.839571476029328e-05,
+      "loss": 0.5666,
+      "step": 1906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8392401211416615e-05,
+      "loss": 0.6013,
+      "step": 1908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8389084543178617e-05,
+      "loss": 0.4678,
+      "step": 1910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.838576475681205e-05,
+      "loss": 0.548,
+      "step": 1912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8382441853550843e-05,
+      "loss": 0.5047,
+      "step": 1914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.837911583463008e-05,
+      "loss": 0.6213,
+      "step": 1916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.837578670128601e-05,
+      "loss": 0.5974,
+      "step": 1918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.837245445475603e-05,
+      "loss": 0.5006,
+      "step": 1920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8369119096278697e-05,
+      "loss": 0.5643,
+      "step": 1922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8365780627093732e-05,
+      "loss": 0.6016,
+      "step": 1924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8362439048442003e-05,
+      "loss": 0.6074,
+      "step": 1926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8359094361565543e-05,
+      "loss": 0.5666,
+      "step": 1928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.835574656770752e-05,
+      "loss": 0.5527,
+      "step": 1930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8352395668112288e-05,
+      "loss": 0.6252,
+      "step": 1932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.834904166402533e-05,
+      "loss": 0.5819,
+      "step": 1934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.834568455669329e-05,
+      "loss": 0.5198,
+      "step": 1936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8342324347363974e-05,
+      "loss": 0.7239,
+      "step": 1938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.833896103728633e-05,
+      "loss": 0.5517,
+      "step": 1940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8335594627710464e-05,
+      "loss": 0.63,
+      "step": 1942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8332225119887633e-05,
+      "loss": 0.6421,
+      "step": 1944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8328852515070237e-05,
+      "loss": 0.6463,
+      "step": 1946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.832547681451185e-05,
+      "loss": 0.5381,
+      "step": 1948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8322098019467175e-05,
+      "loss": 0.5316,
+      "step": 1950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8318716131192066e-05,
+      "loss": 0.5564,
+      "step": 1952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.831533115094354e-05,
+      "loss": 0.5744,
+      "step": 1954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8311943079979755e-05,
+      "loss": 0.5931,
+      "step": 1956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.830855191956002e-05,
+      "loss": 0.5253,
+      "step": 1958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8305157670944794e-05,
+      "loss": 0.6657,
+      "step": 1960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.830176033539567e-05,
+      "loss": 0.51,
+      "step": 1962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.829835991417541e-05,
+      "loss": 0.5039,
+      "step": 1964
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8294956408547907e-05,
+      "loss": 0.4866,
+      "step": 1966
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.829154981977821e-05,
+      "loss": 0.5277,
+      "step": 1968
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8288140149132508e-05,
+      "loss": 0.6111,
+      "step": 1970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8284727397878133e-05,
+      "loss": 0.5179,
+      "step": 1972
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8281311567283573e-05,
+      "loss": 0.5829,
+      "step": 1974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8277892658618452e-05,
+      "loss": 0.6009,
+      "step": 1976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8274470673153537e-05,
+      "loss": 0.6488,
+      "step": 1978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8271045612160745e-05,
+      "loss": 0.6409,
+      "step": 1980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.826761747691313e-05,
+      "loss": 0.5746,
+      "step": 1982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8264186268684897e-05,
+      "loss": 0.4902,
+      "step": 1984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8260751988751377e-05,
+      "loss": 0.5835,
+      "step": 1986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8257314638389065e-05,
+      "loss": 0.5387,
+      "step": 1988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8253874218875577e-05,
+      "loss": 0.5853,
+      "step": 1990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8250430731489684e-05,
+      "loss": 0.5594,
+      "step": 1992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8246984177511288e-05,
+      "loss": 0.5633,
+      "step": 1994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8243534558221437e-05,
+      "loss": 0.5923,
+      "step": 1996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8240081874902313e-05,
+      "loss": 0.491,
+      "step": 1998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.823662612883724e-05,
+      "loss": 0.5538,
+      "step": 2000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8233167321310686e-05,
+      "loss": 0.5021,
+      "step": 2002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.822970545360824e-05,
+      "loss": 0.5196,
+      "step": 2004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8226240527016647e-05,
+      "loss": 0.5776,
+      "step": 2006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8222772542823783e-05,
+      "loss": 0.5344,
+      "step": 2008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8219301502318652e-05,
+      "loss": 0.6396,
+      "step": 2010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8215827406791408e-05,
+      "loss": 0.4815,
+      "step": 2012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8212350257533326e-05,
+      "loss": 0.5536,
+      "step": 2014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8208870055836828e-05,
+      "loss": 0.4872,
+      "step": 2016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8205386802995462e-05,
+      "loss": 0.5763,
+      "step": 2018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8201900500303923e-05,
+      "loss": 0.6757,
+      "step": 2020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.819841114905802e-05,
+      "loss": 0.6157,
+      "step": 2022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8194918750554712e-05,
+      "loss": 0.64,
+      "step": 2024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.819142330609208e-05,
+      "loss": 0.5319,
+      "step": 2026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8187924816969347e-05,
+      "loss": 0.5554,
+      "step": 2028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.818442328448686e-05,
+      "loss": 0.5217,
+      "step": 2030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8180918709946092e-05,
+      "loss": 0.5755,
+      "step": 2032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8177411094649667e-05,
+      "loss": 0.5471,
+      "step": 2034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.817390043990132e-05,
+      "loss": 0.5968,
+      "step": 2036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8170386747005918e-05,
+      "loss": 0.5892,
+      "step": 2038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8166870017269465e-05,
+      "loss": 0.6117,
+      "step": 2040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.816335025199909e-05,
+      "loss": 0.5562,
+      "step": 2042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.815982745250305e-05,
+      "loss": 0.5277,
+      "step": 2044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.815630162009073e-05,
+      "loss": 0.6796,
+      "step": 2046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8152772756072644e-05,
+      "loss": 0.5723,
+      "step": 2048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.814924086176043e-05,
+      "loss": 0.5251,
+      "step": 2050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8145705938466852e-05,
+      "loss": 0.6545,
+      "step": 2052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.81421679875058e-05,
+      "loss": 0.6039,
+      "step": 2054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8138627010192297e-05,
+      "loss": 0.598,
+      "step": 2056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.813508300784248e-05,
+      "loss": 0.5938,
+      "step": 2058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8131535981773612e-05,
+      "loss": 0.5759,
+      "step": 2060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8127985933304092e-05,
+      "loss": 0.6019,
+      "step": 2062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.812443286375342e-05,
+      "loss": 0.6979,
+      "step": 2064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8120876774442243e-05,
+      "loss": 0.5374,
+      "step": 2066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.811731766669232e-05,
+      "loss": 0.57,
+      "step": 2068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8113755541826522e-05,
+      "loss": 0.5927,
+      "step": 2070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.811019040116886e-05,
+      "loss": 0.593,
+      "step": 2072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.810662224604445e-05,
+      "loss": 0.5451,
+      "step": 2074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8103051077779538e-05,
+      "loss": 0.5662,
+      "step": 2076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.809947689770149e-05,
+      "loss": 0.5816,
+      "step": 2078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8095899707138786e-05,
+      "loss": 0.636,
+      "step": 2080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8092319507421025e-05,
+      "loss": 0.5456,
+      "step": 2082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.808873629987893e-05,
+      "loss": 0.491,
+      "step": 2084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.808515008584434e-05,
+      "loss": 0.5161,
+      "step": 2086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.808156086665021e-05,
+      "loss": 0.6286,
+      "step": 2088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8077968643630607e-05,
+      "loss": 0.5275,
+      "step": 2090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.807437341812073e-05,
+      "loss": 0.5263,
+      "step": 2092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8070775191456874e-05,
+      "loss": 0.5194,
+      "step": 2094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.806717396497647e-05,
+      "loss": 0.4929,
+      "step": 2096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8063569740018042e-05,
+      "loss": 0.5456,
+      "step": 2098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8059962517921246e-05,
+      "loss": 0.553,
+      "step": 2100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8056352300026847e-05,
+      "loss": 0.588,
+      "step": 2102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.805273908767672e-05,
+      "loss": 0.5228,
+      "step": 2104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8049122882213858e-05,
+      "loss": 0.5016,
+      "step": 2106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8045503684982358e-05,
+      "loss": 0.53,
+      "step": 2108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8041881497327445e-05,
+      "loss": 0.5985,
+      "step": 2110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8038256320595438e-05,
+      "loss": 0.5078,
+      "step": 2112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8034628156133775e-05,
+      "loss": 0.5018,
+      "step": 2114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8030997005291007e-05,
+      "loss": 0.574,
+      "step": 2116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8027362869416788e-05,
+      "loss": 0.5763,
+      "step": 2118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.802372574986189e-05,
+      "loss": 0.5546,
+      "step": 2120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8020085647978188e-05,
+      "loss": 0.5674,
+      "step": 2122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8016442565118665e-05,
+      "loss": 0.5177,
+      "step": 2124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8012796502637412e-05,
+      "loss": 0.6155,
+      "step": 2126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8009147461889636e-05,
+      "loss": 0.5172,
+      "step": 2128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8005495444231646e-05,
+      "loss": 0.4809,
+      "step": 2130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8001840451020847e-05,
+      "loss": 0.5629,
+      "step": 2132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7998182483615762e-05,
+      "loss": 0.5234,
+      "step": 2134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.799452154337602e-05,
+      "loss": 0.5034,
+      "step": 2136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.799085763166235e-05,
+      "loss": 0.505,
+      "step": 2138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7987190749836583e-05,
+      "loss": 0.6071,
+      "step": 2140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7983520899261662e-05,
+      "loss": 0.5128,
+      "step": 2142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.797984808130163e-05,
+      "loss": 0.6191,
+      "step": 2144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7976172297321628e-05,
+      "loss": 0.5116,
+      "step": 2146
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7972493548687908e-05,
+      "loss": 0.5986,
+      "step": 2148
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7968811836767816e-05,
+      "loss": 0.529,
+      "step": 2150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7965127162929805e-05,
+      "loss": 0.5928,
+      "step": 2152
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7961439528543427e-05,
+      "loss": 0.5424,
+      "step": 2154
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7957748934979332e-05,
+      "loss": 0.6024,
+      "step": 2156
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7954055383609274e-05,
+      "loss": 0.6566,
+      "step": 2158
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7950358875806106e-05,
+      "loss": 0.5408,
+      "step": 2160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7946659412943778e-05,
+      "loss": 0.548,
+      "step": 2162
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7942956996397335e-05,
+      "loss": 0.4887,
+      "step": 2164
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7939251627542928e-05,
+      "loss": 0.5758,
+      "step": 2166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7935543307757802e-05,
+      "loss": 0.5825,
+      "step": 2168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.793183203842029e-05,
+      "loss": 0.6202,
+      "step": 2170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.792811782090984e-05,
+      "loss": 0.5014,
+      "step": 2172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7924400656606975e-05,
+      "loss": 0.6063,
+      "step": 2174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7920680546893332e-05,
+      "loss": 0.5543,
+      "step": 2176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.791695749315163e-05,
+      "loss": 0.6721,
+      "step": 2178
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.791323149676568e-05,
+      "loss": 0.65,
+      "step": 2180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.79095025591204e-05,
+      "loss": 0.662,
+      "step": 2182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.79057706816018e-05,
+      "loss": 0.4885,
+      "step": 2184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7902035865596967e-05,
+      "loss": 0.5634,
+      "step": 2186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7898298112494092e-05,
+      "loss": 0.6104,
+      "step": 2188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.789455742368246e-05,
+      "loss": 0.5981,
+      "step": 2190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7890813800552438e-05,
+      "loss": 0.6237,
+      "step": 2192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7887067244495494e-05,
+      "loss": 0.5892,
+      "step": 2194
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.788331775690417e-05,
+      "loss": 0.5046,
+      "step": 2196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7879565339172126e-05,
+      "loss": 0.5579,
+      "step": 2198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7875809992694083e-05,
+      "loss": 0.5894,
+      "step": 2200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7872051718865858e-05,
+      "loss": 0.5858,
+      "step": 2202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7868290519084364e-05,
+      "loss": 0.533,
+      "step": 2204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.78645263947476e-05,
+      "loss": 0.4956,
+      "step": 2206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7860759347254643e-05,
+      "loss": 0.4863,
+      "step": 2208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7856989378005663e-05,
+      "loss": 0.5842,
+      "step": 2210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7853216488401916e-05,
+      "loss": 0.5055,
+      "step": 2212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7849440679845745e-05,
+      "loss": 0.5733,
+      "step": 2214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7845661953740573e-05,
+      "loss": 0.6332,
+      "step": 2216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.784188031149091e-05,
+      "loss": 0.6171,
+      "step": 2218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.783809575450235e-05,
+      "loss": 0.51,
+      "step": 2220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7834308284181575e-05,
+      "loss": 0.6306,
+      "step": 2222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7830517901936337e-05,
+      "loss": 0.5811,
+      "step": 2224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7826724609175486e-05,
+      "loss": 0.5886,
+      "step": 2226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.782292840730894e-05,
+      "loss": 0.527,
+      "step": 2228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7819129297747713e-05,
+      "loss": 0.6287,
+      "step": 2230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7815327281903884e-05,
+      "loss": 0.5381,
+      "step": 2232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7811522361190618e-05,
+      "loss": 0.531,
+      "step": 2234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.780771453702217e-05,
+      "loss": 0.5255,
+      "step": 2236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7803903810813858e-05,
+      "loss": 0.5768,
+      "step": 2238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.780009018398209e-05,
+      "loss": 0.5018,
+      "step": 2240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.779627365794435e-05,
+      "loss": 0.5504,
+      "step": 2242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.779245423411919e-05,
+      "loss": 0.4966,
+      "step": 2244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.778863191392626e-05,
+      "loss": 0.6283,
+      "step": 2246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7784806698786257e-05,
+      "loss": 0.5834,
+      "step": 2248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7780978590120984e-05,
+      "loss": 0.5821,
+      "step": 2250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.77771475893533e-05,
+      "loss": 0.6467,
+      "step": 2252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7773313697907142e-05,
+      "loss": 0.5688,
+      "step": 2254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.776947691720753e-05,
+      "loss": 0.527,
+      "step": 2256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7765637248680555e-05,
+      "loss": 0.5958,
+      "step": 2258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7761794693753374e-05,
+      "loss": 0.523,
+      "step": 2260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7757949253854218e-05,
+      "loss": 0.5664,
+      "step": 2262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.77541009304124e-05,
+      "loss": 0.6129,
+      "step": 2264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.77502497248583e-05,
+      "loss": 0.5078,
+      "step": 2266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.774639563862336e-05,
+      "loss": 0.6069,
+      "step": 2268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7742538673140108e-05,
+      "loss": 0.5343,
+      "step": 2270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7738678829842132e-05,
+      "loss": 0.5743,
+      "step": 2272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7734816110164087e-05,
+      "loss": 0.5326,
+      "step": 2274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7730950515541714e-05,
+      "loss": 0.6408,
+      "step": 2276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.77270820474118e-05,
+      "loss": 0.5542,
+      "step": 2278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.772321070721222e-05,
+      "loss": 0.5961,
+      "step": 2280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7719336496381904e-05,
+      "loss": 0.588,
+      "step": 2282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.771545941636085e-05,
+      "loss": 0.6319,
+      "step": 2284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7711579468590127e-05,
+      "loss": 0.6345,
+      "step": 2286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.770769665451187e-05,
+      "loss": 0.6483,
+      "step": 2288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7703810975569274e-05,
+      "loss": 0.6486,
+      "step": 2290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.769992243320661e-05,
+      "loss": 0.62,
+      "step": 2292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7696031028869196e-05,
+      "loss": 0.5699,
+      "step": 2294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7692136764003428e-05,
+      "loss": 0.5598,
+      "step": 2296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.768823964005676e-05,
+      "loss": 0.6331,
+      "step": 2298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7684339658477707e-05,
+      "loss": 0.5165,
+      "step": 2300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7680436820715853e-05,
+      "loss": 0.6119,
+      "step": 2302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7676531128221836e-05,
+      "loss": 0.6511,
+      "step": 2304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7672622582447357e-05,
+      "loss": 0.6098,
+      "step": 2306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.766871118484518e-05,
+      "loss": 0.7256,
+      "step": 2308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.766479693686913e-05,
+      "loss": 0.6297,
+      "step": 2310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7660879839974084e-05,
+      "loss": 0.5432,
+      "step": 2312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7656959895615987e-05,
+      "loss": 0.6177,
+      "step": 2314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7653037105251836e-05,
+      "loss": 0.5428,
+      "step": 2316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.764911147033969e-05,
+      "loss": 0.584,
+      "step": 2318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.764518299233866e-05,
+      "loss": 0.5421,
+      "step": 2320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7641251672708924e-05,
+      "loss": 0.5118,
+      "step": 2322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7637317512911702e-05,
+      "loss": 0.6065,
+      "step": 2324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.763338051440928e-05,
+      "loss": 0.519,
+      "step": 2326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7629440678664997e-05,
+      "loss": 0.6035,
+      "step": 2328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7625498007143246e-05,
+      "loss": 0.5894,
+      "step": 2330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.762155250130947e-05,
+      "loss": 0.5347,
+      "step": 2332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7617604162630177e-05,
+      "loss": 0.6336,
+      "step": 2334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7613652992572912e-05,
+      "loss": 0.5247,
+      "step": 2336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7609698992606282e-05,
+      "loss": 0.5598,
+      "step": 2338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7605742164199947e-05,
+      "loss": 0.6029,
+      "step": 2340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7601782508824616e-05,
+      "loss": 0.6136,
+      "step": 2342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7597820027952047e-05,
+      "loss": 0.5878,
+      "step": 2344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.759385472305505e-05,
+      "loss": 0.5473,
+      "step": 2346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7589886595607484e-05,
+      "loss": 0.5272,
+      "step": 2348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.758591564708426e-05,
+      "loss": 0.5141,
+      "step": 2350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.758194187896133e-05,
+      "loss": 0.5927,
+      "step": 2352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7577965292715702e-05,
+      "loss": 0.4981,
+      "step": 2354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7573985889825428e-05,
+      "loss": 0.5968,
+      "step": 2356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7570003671769607e-05,
+      "loss": 0.5756,
+      "step": 2358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7566018640028385e-05,
+      "loss": 0.6089,
+      "step": 2360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7562030796082958e-05,
+      "loss": 0.5324,
+      "step": 2362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7558040141415555e-05,
+      "loss": 0.4789,
+      "step": 2364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.755404667750946e-05,
+      "loss": 0.5179,
+      "step": 2366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7550050405849006e-05,
+      "loss": 0.5497,
+      "step": 2368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.754605132791955e-05,
+      "loss": 0.5821,
+      "step": 2370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7542049445207517e-05,
+      "loss": 0.4316,
+      "step": 2372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.753804475920035e-05,
+      "loss": 0.5617,
+      "step": 2374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7534037271386554e-05,
+      "loss": 0.4803,
+      "step": 2376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7530026983255663e-05,
+      "loss": 0.5941,
+      "step": 2378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.752601389629826e-05,
+      "loss": 0.6316,
+      "step": 2380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7521998012005964e-05,
+      "loss": 0.5545,
+      "step": 2382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7517979331871425e-05,
+      "loss": 0.6069,
+      "step": 2384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7513957857388356e-05,
+      "loss": 0.5508,
+      "step": 2386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7509933590051483e-05,
+      "loss": 0.5332,
+      "step": 2388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7505906531356586e-05,
+      "loss": 0.5626,
+      "step": 2390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7501876682800472e-05,
+      "loss": 0.5225,
+      "step": 2392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7497844045881e-05,
+      "loss": 0.5194,
+      "step": 2394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7493808622097047e-05,
+      "loss": 0.5896,
+      "step": 2396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.748977041294854e-05,
+      "loss": 0.4918,
+      "step": 2398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.748572941993643e-05,
+      "loss": 0.6066,
+      "step": 2400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.748168564456272e-05,
+      "loss": 0.628,
+      "step": 2402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7477639088330422e-05,
+      "loss": 0.6119,
+      "step": 2404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7473589752743606e-05,
+      "loss": 0.5331,
+      "step": 2406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.746953763930736e-05,
+      "loss": 0.4788,
+      "step": 2408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7465482749527808e-05,
+      "loss": 0.558,
+      "step": 2410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7461425084912112e-05,
+      "loss": 0.5227,
+      "step": 2412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7457364646968456e-05,
+      "loss": 0.5657,
+      "step": 2414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.745330143720606e-05,
+      "loss": 0.5465,
+      "step": 2416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7449235457135175e-05,
+      "loss": 0.5513,
+      "step": 2418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.744516670826708e-05,
+      "loss": 0.5946,
+      "step": 2420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7441095192114084e-05,
+      "loss": 0.5357,
+      "step": 2422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.743702091018952e-05,
+      "loss": 0.5227,
+      "step": 2424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7432943864007757e-05,
+      "loss": 0.6091,
+      "step": 2426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7428864055084182e-05,
+      "loss": 0.555,
+      "step": 2428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.742478148493522e-05,
+      "loss": 0.5069,
+      "step": 2430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7420696155078315e-05,
+      "loss": 0.6085,
+      "step": 2432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7416608067031938e-05,
+      "loss": 0.5501,
+      "step": 2434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.741251722231558e-05,
+      "loss": 0.5717,
+      "step": 2436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.740842362244977e-05,
+      "loss": 0.5889,
+      "step": 2438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7404327268956053e-05,
+      "loss": 0.5116,
+      "step": 2440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7400228163356992e-05,
+      "loss": 0.5634,
+      "step": 2442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7396126307176182e-05,
+      "loss": 0.5428,
+      "step": 2444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7392021701938236e-05,
+      "loss": 0.5642,
+      "step": 2446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7387914349168792e-05,
+      "loss": 0.5729,
+      "step": 2448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.73838042503945e-05,
+      "loss": 0.6389,
+      "step": 2450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7379691407143043e-05,
+      "loss": 0.5452,
+      "step": 2452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.737557582094312e-05,
+      "loss": 0.5386,
+      "step": 2454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7371457493324447e-05,
+      "loss": 0.515,
+      "step": 2456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.736733642581776e-05,
+      "loss": 0.4562,
+      "step": 2458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.736321261995481e-05,
+      "loss": 0.5097,
+      "step": 2460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7359086077268378e-05,
+      "loss": 0.5447,
+      "step": 2462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7354956799292245e-05,
+      "loss": 0.5844,
+      "step": 2464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7350824787561218e-05,
+      "loss": 0.5377,
+      "step": 2466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.734669004361112e-05,
+      "loss": 0.619,
+      "step": 2468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7342552568978798e-05,
+      "loss": 0.6356,
+      "step": 2470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.733841236520209e-05,
+      "loss": 0.5711,
+      "step": 2472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7334269433819873e-05,
+      "loss": 0.5097,
+      "step": 2474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7330123776372024e-05,
+      "loss": 0.5114,
+      "step": 2476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7325975394399443e-05,
+      "loss": 0.5909,
+      "step": 2478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7321824289444033e-05,
+      "loss": 0.5952,
+      "step": 2480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.731767046304871e-05,
+      "loss": 0.4905,
+      "step": 2482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7313513916757408e-05,
+      "loss": 0.5588,
+      "step": 2484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7309354652115072e-05,
+      "loss": 0.6102,
+      "step": 2486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.730519267066765e-05,
+      "loss": 0.5365,
+      "step": 2488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.73010279739621e-05,
+      "loss": 0.5428,
+      "step": 2490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.72968605635464e-05,
+      "loss": 0.6014,
+      "step": 2492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7292690440969527e-05,
+      "loss": 0.5275,
+      "step": 2494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.728851760778147e-05,
+      "loss": 0.5448,
+      "step": 2496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7284342065533223e-05,
+      "loss": 0.6619,
+      "step": 2498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7280163815776788e-05,
+      "loss": 0.4987,
+      "step": 2500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.727598286006517e-05,
+      "loss": 0.6311,
+      "step": 2502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7271799199952387e-05,
+      "loss": 0.6357,
+      "step": 2504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7267612836993462e-05,
+      "loss": 0.4785,
+      "step": 2506
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7263423772744415e-05,
+      "loss": 0.586,
+      "step": 2508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7259232008762278e-05,
+      "loss": 0.4984,
+      "step": 2510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7255037546605073e-05,
+      "loss": 0.6751,
+      "step": 2512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7250840387831843e-05,
+      "loss": 0.5023,
+      "step": 2514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7246640534002626e-05,
+      "loss": 0.5851,
+      "step": 2516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7242437986678455e-05,
+      "loss": 0.5404,
+      "step": 2518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.723823274742137e-05,
+      "loss": 0.5923,
+      "step": 2520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7234024817794417e-05,
+      "loss": 0.5184,
+      "step": 2522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7229814199361632e-05,
+      "loss": 0.4897,
+      "step": 2524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7225600893688054e-05,
+      "loss": 0.5106,
+      "step": 2526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.722138490233972e-05,
+      "loss": 0.7481,
+      "step": 2528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.721716622688367e-05,
+      "loss": 0.6403,
+      "step": 2530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7212944868887942e-05,
+      "loss": 0.6112,
+      "step": 2532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.720872082992156e-05,
+      "loss": 0.5151,
+      "step": 2534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7204494111554552e-05,
+      "loss": 0.4741,
+      "step": 2536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7200264715357945e-05,
+      "loss": 0.5324,
+      "step": 2538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.719603264290376e-05,
+      "loss": 0.6655,
+      "step": 2540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7191797895765008e-05,
+      "loss": 0.5092,
+      "step": 2542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7187560475515696e-05,
+      "loss": 0.5833,
+      "step": 2544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7183320383730828e-05,
+      "loss": 0.5638,
+      "step": 2546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7179077621986397e-05,
+      "loss": 0.6093,
+      "step": 2548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.717483219185939e-05,
+      "loss": 0.5449,
+      "step": 2550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.717058409492779e-05,
+      "loss": 0.5195,
+      "step": 2552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7166333332770557e-05,
+      "loss": 0.5534,
+      "step": 2554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7162079906967662e-05,
+      "loss": 0.4846,
+      "step": 2556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.715782381910005e-05,
+      "loss": 0.5679,
+      "step": 2558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7153565070749662e-05,
+      "loss": 0.6229,
+      "step": 2560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7149303663499427e-05,
+      "loss": 0.6794,
+      "step": 2562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7145039598933263e-05,
+      "loss": 0.498,
+      "step": 2564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7140772878636076e-05,
+      "loss": 0.4375,
+      "step": 2566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7136503504193752e-05,
+      "loss": 0.529,
+      "step": 2568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.713223147719318e-05,
+      "loss": 0.5448,
+      "step": 2570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7127956799222216e-05,
+      "loss": 0.6281,
+      "step": 2572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.712367947186971e-05,
+      "loss": 0.5408,
+      "step": 2574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7119399496725502e-05,
+      "loss": 0.5849,
+      "step": 2576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7115116875380408e-05,
+      "loss": 0.5618,
+      "step": 2578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.711083160942623e-05,
+      "loss": 0.6042,
+      "step": 2580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.710654370045575e-05,
+      "loss": 0.6101,
+      "step": 2582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.710225315006274e-05,
+      "loss": 0.4839,
+      "step": 2584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.709795995984195e-05,
+      "loss": 0.5336,
+      "step": 2586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7093664131389108e-05,
+      "loss": 0.458,
+      "step": 2588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7089365666300928e-05,
+      "loss": 0.5902,
+      "step": 2590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.70850645661751e-05,
+      "loss": 0.5566,
+      "step": 2592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7080760832610292e-05,
+      "loss": 0.5572,
+      "step": 2594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7076454467206158e-05,
+      "loss": 0.644,
+      "step": 2596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7072145471563315e-05,
+      "loss": 0.4844,
+      "step": 2598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7067833847283384e-05,
+      "loss": 0.5722,
+      "step": 2600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7063519595968938e-05,
+      "loss": 0.6707,
+      "step": 2602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7059202719223536e-05,
+      "loss": 0.5349,
+      "step": 2604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7054883218651718e-05,
+      "loss": 0.583,
+      "step": 2606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7050561095858985e-05,
+      "loss": 0.5157,
+      "step": 2608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7046236352451834e-05,
+      "loss": 0.5426,
+      "step": 2610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7041908990037708e-05,
+      "loss": 0.5554,
+      "step": 2612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7037579010225053e-05,
+      "loss": 0.525,
+      "step": 2614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.703324641462327e-05,
+      "loss": 0.5648,
+      "step": 2616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.702891120484273e-05,
+      "loss": 0.5294,
+      "step": 2618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7024573382494786e-05,
+      "loss": 0.5423,
+      "step": 2620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.702023294919176e-05,
+      "loss": 0.5582,
+      "step": 2622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.701588990654694e-05,
+      "loss": 0.5847,
+      "step": 2624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7011544256174586e-05,
+      "loss": 0.5496,
+      "step": 2626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7007195999689926e-05,
+      "loss": 0.5945,
+      "step": 2628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7002845138709166e-05,
+      "loss": 0.598,
+      "step": 2630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.699849167484946e-05,
+      "loss": 0.4376,
+      "step": 2632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6994135609728952e-05,
+      "loss": 0.5129,
+      "step": 2634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.698977694496674e-05,
+      "loss": 0.5556,
+      "step": 2636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6985415682182883e-05,
+      "loss": 0.635,
+      "step": 2638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6981051822998423e-05,
+      "loss": 0.5208,
+      "step": 2640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6976685369035357e-05,
+      "loss": 0.602,
+      "step": 2642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.697231632191664e-05,
+      "loss": 0.5493,
+      "step": 2644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6967944683266206e-05,
+      "loss": 0.5013,
+      "step": 2646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.696357045470894e-05,
+      "loss": 0.5331,
+      "step": 2648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.695919363787069e-05,
+      "loss": 0.5651,
+      "step": 2650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6954814234378276e-05,
+      "loss": 0.627,
+      "step": 2652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.695043224585947e-05,
+      "loss": 0.4464,
+      "step": 2654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6946047673943007e-05,
+      "loss": 0.5691,
+      "step": 2656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6941660520258588e-05,
+      "loss": 0.5855,
+      "step": 2658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.693727078643686e-05,
+      "loss": 0.6107,
+      "step": 2660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6932878474109445e-05,
+      "loss": 0.5593,
+      "step": 2662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6928483584908918e-05,
+      "loss": 0.5037,
+      "step": 2664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6924086120468805e-05,
+      "loss": 0.4844,
+      "step": 2666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6919686082423595e-05,
+      "loss": 0.6387,
+      "step": 2668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.691528347240873e-05,
+      "loss": 0.5156,
+      "step": 2670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6910878292060617e-05,
+      "loss": 0.5117,
+      "step": 2672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.690647054301661e-05,
+      "loss": 0.5131,
+      "step": 2674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6902060226915017e-05,
+      "loss": 0.6115,
+      "step": 2676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6897647345395104e-05,
+      "loss": 0.5453,
+      "step": 2678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6893231900097092e-05,
+      "loss": 0.5303,
+      "step": 2680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6888813892662154e-05,
+      "loss": 0.5707,
+      "step": 2682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6884393324732403e-05,
+      "loss": 0.5669,
+      "step": 2684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.687997019795093e-05,
+      "loss": 0.657,
+      "step": 2686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.687554451396175e-05,
+      "loss": 0.5286,
+      "step": 2688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6871116274409844e-05,
+      "loss": 0.5214,
+      "step": 2690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.686668548094114e-05,
+      "loss": 0.5765,
+      "step": 2692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6862252135202513e-05,
+      "loss": 0.5586,
+      "step": 2694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6857816238841787e-05,
+      "loss": 0.5378,
+      "step": 2696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6853377793507733e-05,
+      "loss": 0.5681,
+      "step": 2698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6848936800850077e-05,
+      "loss": 0.4886,
+      "step": 2700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6844493262519487e-05,
+      "loss": 0.4995,
+      "step": 2702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6840047180167567e-05,
+      "loss": 0.5009,
+      "step": 2704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6835598555446883e-05,
+      "loss": 0.5088,
+      "step": 2706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6831147390010934e-05,
+      "loss": 0.5016,
+      "step": 2708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6826693685514176e-05,
+      "loss": 0.617,
+      "step": 2710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6822237443611996e-05,
+      "loss": 0.5326,
+      "step": 2712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6817778665960732e-05,
+      "loss": 0.6026,
+      "step": 2714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6813317354217652e-05,
+      "loss": 0.5703,
+      "step": 2716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6808853510040986e-05,
+      "loss": 0.5011,
+      "step": 2718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6804387135089896e-05,
+      "loss": 0.5455,
+      "step": 2720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6799918231024473e-05,
+      "loss": 0.5864,
+      "step": 2722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6795446799505767e-05,
+      "loss": 0.6228,
+      "step": 2724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6790972842195755e-05,
+      "loss": 0.6523,
+      "step": 2726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6786496360757358e-05,
+      "loss": 0.6272,
+      "step": 2728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6782017356854435e-05,
+      "loss": 0.5137,
+      "step": 2730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.677753583215178e-05,
+      "loss": 0.5812,
+      "step": 2732
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6773051788315125e-05,
+      "loss": 0.6064,
+      "step": 2734
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.676856522701114e-05,
+      "loss": 0.5748,
+      "step": 2736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6764076149907432e-05,
+      "loss": 0.5104,
+      "step": 2738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6759584558672535e-05,
+      "loss": 0.6183,
+      "step": 2740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6755090454975925e-05,
+      "loss": 0.5982,
+      "step": 2742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.675059384048801e-05,
+      "loss": 0.5746,
+      "step": 2744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6746094716880134e-05,
+      "loss": 0.5371,
+      "step": 2746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6741593085824566e-05,
+      "loss": 0.4939,
+      "step": 2748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6737088948994518e-05,
+      "loss": 0.5926,
+      "step": 2750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.673258230806412e-05,
+      "loss": 0.5837,
+      "step": 2752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6728073164708446e-05,
+      "loss": 0.482,
+      "step": 2754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.672356152060349e-05,
+      "loss": 0.5982,
+      "step": 2756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6719047377426178e-05,
+      "loss": 0.4479,
+      "step": 2758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.671453073685437e-05,
+      "loss": 0.5444,
+      "step": 2760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6710011600566852e-05,
+      "loss": 0.58,
+      "step": 2762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6705489970243333e-05,
+      "loss": 0.6107,
+      "step": 2764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6700965847564453e-05,
+      "loss": 0.5362,
+      "step": 2766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6696439234211777e-05,
+      "loss": 0.6747,
+      "step": 2768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.66919101318678e-05,
+      "loss": 0.5847,
+      "step": 2770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6687378542215932e-05,
+      "loss": 0.5891,
+      "step": 2772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6682844466940518e-05,
+      "loss": 0.5013,
+      "step": 2774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6678307907726824e-05,
+      "loss": 0.5956,
+      "step": 2776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6673768866261038e-05,
+      "loss": 0.5182,
+      "step": 2778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6669227344230266e-05,
+      "loss": 0.6132,
+      "step": 2780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6664683343322545e-05,
+      "loss": 0.5266,
+      "step": 2782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6660136865226827e-05,
+      "loss": 0.5911,
+      "step": 2784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.665558791163299e-05,
+      "loss": 0.5299,
+      "step": 2786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6651036484231825e-05,
+      "loss": 0.6064,
+      "step": 2788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6646482584715054e-05,
+      "loss": 0.4642,
+      "step": 2790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6641926214775302e-05,
+      "loss": 0.5897,
+      "step": 2792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6637367376106123e-05,
+      "loss": 0.5176,
+      "step": 2794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6632806070401984e-05,
+      "loss": 0.4505,
+      "step": 2796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.662824229935828e-05,
+      "loss": 0.544,
+      "step": 2798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6623676064671303e-05,
+      "loss": 0.4875,
+      "step": 2800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6619107368038276e-05,
+      "loss": 0.6371,
+      "step": 2802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661453621115733e-05,
+      "loss": 0.6036,
+      "step": 2804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660996259572752e-05,
+      "loss": 0.5809,
+      "step": 2806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6605386523448798e-05,
+      "loss": 0.5684,
+      "step": 2808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6600807996022046e-05,
+      "loss": 0.5597,
+      "step": 2810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.659622701514904e-05,
+      "loss": 0.5787,
+      "step": 2812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6593935605202262e-05,
+      "loss": 0.5396,
+      "step": 2814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6589350947352735e-05,
+      "loss": 0.5655,
+      "step": 2816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6584763840315427e-05,
+      "loss": 0.6118,
+      "step": 2818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6580174285795293e-05,
+      "loss": 0.5562,
+      "step": 2820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.657558228549824e-05,
+      "loss": 0.4713,
+      "step": 2822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6570987841131058e-05,
+      "loss": 0.5338,
+      "step": 2824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.656639095440144e-05,
+      "loss": 0.4904,
+      "step": 2826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6561791627018006e-05,
+      "loss": 0.6086,
+      "step": 2828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6557189860690274e-05,
+      "loss": 0.6543,
+      "step": 2830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6552585657128662e-05,
+      "loss": 0.5292,
+      "step": 2832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.654797901804451e-05,
+      "loss": 0.603,
+      "step": 2834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6543369945150052e-05,
+      "loss": 0.6011,
+      "step": 2836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6538758440158425e-05,
+      "loss": 0.4711,
+      "step": 2838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6534144504783676e-05,
+      "loss": 0.5962,
+      "step": 2840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6529528140740753e-05,
+      "loss": 0.5626,
+      "step": 2842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.65249093497455e-05,
+      "loss": 0.5889,
+      "step": 2844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6520288133514685e-05,
+      "loss": 0.6033,
+      "step": 2846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6515664493765947e-05,
+      "loss": 0.5475,
+      "step": 2848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.651103843221785e-05,
+      "loss": 0.6136,
+      "step": 2850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6506409950589843e-05,
+      "loss": 0.4828,
+      "step": 2852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6501779050602288e-05,
+      "loss": 0.5646,
+      "step": 2854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6497145733976432e-05,
+      "loss": 0.4854,
+      "step": 2856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6492510002434433e-05,
+      "loss": 0.6016,
+      "step": 2858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.648787185769933e-05,
+      "loss": 0.6697,
+      "step": 2860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6483231301495082e-05,
+      "loss": 0.5015,
+      "step": 2862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6478588335546524e-05,
+      "loss": 0.5851,
+      "step": 2864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6473942961579394e-05,
+      "loss": 0.5265,
+      "step": 2866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.646929518132033e-05,
+      "loss": 0.44,
+      "step": 2868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6464644996496856e-05,
+      "loss": 0.51,
+      "step": 2870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6459992408837396e-05,
+      "loss": 0.5525,
+      "step": 2872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.645533742007126e-05,
+      "loss": 0.52,
+      "step": 2874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.645068003192866e-05,
+      "loss": 0.6273,
+      "step": 2876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6446020246140693e-05,
+      "loss": 0.5639,
+      "step": 2878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.644135806443935e-05,
+      "loss": 0.557,
+      "step": 2880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.643669348855751e-05,
+      "loss": 0.6167,
+      "step": 2882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6432026520228946e-05,
+      "loss": 0.6703,
+      "step": 2884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6427357161188317e-05,
+      "loss": 0.5337,
+      "step": 2886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6422685413171166e-05,
+      "loss": 0.5687,
+      "step": 2888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6418011277913937e-05,
+      "loss": 0.5662,
+      "step": 2890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6413334757153946e-05,
+      "loss": 0.5209,
+      "step": 2892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6408655852629414e-05,
+      "loss": 0.5438,
+      "step": 2894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6403974566079427e-05,
+      "loss": 0.6638,
+      "step": 2896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6399290899243973e-05,
+      "loss": 0.5219,
+      "step": 2898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6394604853863917e-05,
+      "loss": 0.603,
+      "step": 2900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6389916431681005e-05,
+      "loss": 0.6254,
+      "step": 2902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6385225634437878e-05,
+      "loss": 0.5161,
+      "step": 2904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.638053246387805e-05,
+      "loss": 0.6342,
+      "step": 2906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6375836921745914e-05,
+      "loss": 0.5355,
+      "step": 2908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6371139009786757e-05,
+      "loss": 0.5213,
+      "step": 2910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6366438729746738e-05,
+      "loss": 0.573,
+      "step": 2912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.63617360833729e-05,
+      "loss": 0.6097,
+      "step": 2914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.635703107241316e-05,
+      "loss": 0.6028,
+      "step": 2916
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6352323698616324e-05,
+      "loss": 0.4915,
+      "step": 2918
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6347613963732066e-05,
+      "loss": 0.5643,
+      "step": 2920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.634290186951094e-05,
+      "loss": 0.5172,
+      "step": 2922
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.633818741770438e-05,
+      "loss": 0.5497,
+      "step": 2924
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6333470610064696e-05,
+      "loss": 0.5352,
+      "step": 2926
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6328751448345074e-05,
+      "loss": 0.5345,
+      "step": 2928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.632402993429957e-05,
+      "loss": 0.5227,
+      "step": 2930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6319306069683118e-05,
+      "loss": 0.4792,
+      "step": 2932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.631457985625153e-05,
+      "loss": 0.5311,
+      "step": 2934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.630985129576148e-05,
+      "loss": 0.5215,
+      "step": 2936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6305120389970522e-05,
+      "loss": 0.5684,
+      "step": 2938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6300387140637083e-05,
+      "loss": 0.6854,
+      "step": 2940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6295651549520458e-05,
+      "loss": 0.578,
+      "step": 2942
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6290913618380817e-05,
+      "loss": 0.5984,
+      "step": 2944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.628617334897919e-05,
+      "loss": 0.5568,
+      "step": 2946
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6281430743077483e-05,
+      "loss": 0.5587,
+      "step": 2948
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6276685802438474e-05,
+      "loss": 0.5411,
+      "step": 2950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6271938528825796e-05,
+      "loss": 0.6343,
+      "step": 2952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6267188924003962e-05,
+      "loss": 0.537,
+      "step": 2954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6262436989738345e-05,
+      "loss": 0.6037,
+      "step": 2956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6257682727795194e-05,
+      "loss": 0.5262,
+      "step": 2958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.62529261399416e-05,
+      "loss": 0.5785,
+      "step": 2960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6248167227945546e-05,
+      "loss": 0.5607,
+      "step": 2962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.624340599357586e-05,
+      "loss": 0.5235,
+      "step": 2964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.623864243860224e-05,
+      "loss": 0.6049,
+      "step": 2966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.623387656479525e-05,
+      "loss": 0.5583,
+      "step": 2968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.622910837392631e-05,
+      "loss": 0.6093,
+      "step": 2970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6224337867767695e-05,
+      "loss": 0.4959,
+      "step": 2972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.621956504809256e-05,
+      "loss": 0.524,
+      "step": 2974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.62147899166749e-05,
+      "loss": 0.5868,
+      "step": 2976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6210012475289586e-05,
+      "loss": 0.4837,
+      "step": 2978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6205232725712333e-05,
+      "loss": 0.5862,
+      "step": 2980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.620045066971972e-05,
+      "loss": 0.542,
+      "step": 2982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.619566630908919e-05,
+      "loss": 0.5829,
+      "step": 2984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6190879645599026e-05,
+      "loss": 0.5641,
+      "step": 2986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6186090681028384e-05,
+      "loss": 0.4789,
+      "step": 2988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.618129941715727e-05,
+      "loss": 0.5066,
+      "step": 2990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6176505855766542e-05,
+      "loss": 0.4842,
+      "step": 2992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6171709998637907e-05,
+      "loss": 0.4327,
+      "step": 2994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6166911847553936e-05,
+      "loss": 0.6017,
+      "step": 2996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.616211140429805e-05,
+      "loss": 0.5272,
+      "step": 2998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6157308670654518e-05,
+      "loss": 0.573,
+      "step": 3000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6152503648408458e-05,
+      "loss": 0.5709,
+      "step": 3002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6147696339345848e-05,
+      "loss": 0.5206,
+      "step": 3004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6142886745253508e-05,
+      "loss": 0.6002,
+      "step": 3006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6138074867919112e-05,
+      "loss": 0.627,
+      "step": 3008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.613326070913118e-05,
+      "loss": 0.4945,
+      "step": 3010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.612844427067908e-05,
+      "loss": 0.5713,
+      "step": 3012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.612362555435303e-05,
+      "loss": 0.5547,
+      "step": 3014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6118804561944093e-05,
+      "loss": 0.6353,
+      "step": 3016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6113981295244176e-05,
+      "loss": 0.5453,
+      "step": 3018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6109155756046032e-05,
+      "loss": 0.5699,
+      "step": 3020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.610432794614326e-05,
+      "loss": 0.4485,
+      "step": 3022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.609949786733031e-05,
+      "loss": 0.5259,
+      "step": 3024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6094665521402455e-05,
+      "loss": 0.5422,
+      "step": 3026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6089830910155833e-05,
+      "loss": 0.5249,
+      "step": 3028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6084994035387417e-05,
+      "loss": 0.606,
+      "step": 3030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.608015489889501e-05,
+      "loss": 0.5106,
+      "step": 3032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6075313502477272e-05,
+      "loss": 0.6021,
+      "step": 3034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6070469847933692e-05,
+      "loss": 0.4508,
+      "step": 3036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.60656239370646e-05,
+      "loss": 0.5428,
+      "step": 3038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6060775771671178e-05,
+      "loss": 0.5722,
+      "step": 3040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6055925353555423e-05,
+      "loss": 0.4873,
+      "step": 3042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6051072684520182e-05,
+      "loss": 0.5103,
+      "step": 3044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6046217766369145e-05,
+      "loss": 0.4856,
+      "step": 3046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.604136060090682e-05,
+      "loss": 0.6346,
+      "step": 3048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.603650118993857e-05,
+      "loss": 0.622,
+      "step": 3050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.603163953527058e-05,
+      "loss": 0.5801,
+      "step": 3052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6026775638709868e-05,
+      "loss": 0.574,
+      "step": 3054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.60219095020643e-05,
+      "loss": 0.556,
+      "step": 3056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6017041127142557e-05,
+      "loss": 0.5283,
+      "step": 3058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.601217051575416e-05,
+      "loss": 0.5215,
+      "step": 3060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6007297669709463e-05,
+      "loss": 0.5475,
+      "step": 3062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6002422590819648e-05,
+      "loss": 0.5861,
+      "step": 3064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5997545280896725e-05,
+      "loss": 0.5829,
+      "step": 3066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5992665741753533e-05,
+      "loss": 0.487,
+      "step": 3068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5987783975203747e-05,
+      "loss": 0.5427,
+      "step": 3070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5982899983061864e-05,
+      "loss": 0.6022,
+      "step": 3072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.597801376714321e-05,
+      "loss": 0.5261,
+      "step": 3074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5973125329263932e-05,
+      "loss": 0.5079,
+      "step": 3076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.596823467124101e-05,
+      "loss": 0.6917,
+      "step": 3078
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.596334179489225e-05,
+      "loss": 0.5338,
+      "step": 3080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5958446702036274e-05,
+      "loss": 0.5915,
+      "step": 3082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5953549394492534e-05,
+      "loss": 0.5066,
+      "step": 3084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5948649874081307e-05,
+      "loss": 0.5343,
+      "step": 3086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5943748142623688e-05,
+      "loss": 0.5217,
+      "step": 3088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5938844201941597e-05,
+      "loss": 0.5594,
+      "step": 3090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.593393805385777e-05,
+      "loss": 0.5912,
+      "step": 3092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.592902970019577e-05,
+      "loss": 0.6508,
+      "step": 3094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5924119142779977e-05,
+      "loss": 0.643,
+      "step": 3096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.591920638343559e-05,
+      "loss": 0.5345,
+      "step": 3098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5914291423988627e-05,
+      "loss": 0.5843,
+      "step": 3100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.590937426626592e-05,
+      "loss": 0.5041,
+      "step": 3102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5904454912095128e-05,
+      "loss": 0.5033,
+      "step": 3104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5899533363304716e-05,
+      "loss": 0.544,
+      "step": 3106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5894609621723962e-05,
+      "loss": 0.525,
+      "step": 3108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5889683689182975e-05,
+      "loss": 0.5434,
+      "step": 3110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5884755567512665e-05,
+      "loss": 0.5119,
+      "step": 3112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5879825258544757e-05,
+      "loss": 0.5305,
+      "step": 3114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.58748927641118e-05,
+      "loss": 0.4997,
+      "step": 3116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.586995808604714e-05,
+      "loss": 0.5677,
+      "step": 3118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.586502122618494e-05,
+      "loss": 0.556,
+      "step": 3120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.586008218636018e-05,
+      "loss": 0.552,
+      "step": 3122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.585514096840864e-05,
+      "loss": 0.5077,
+      "step": 3124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5850197574166922e-05,
+      "loss": 0.5467,
+      "step": 3126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5845252005472427e-05,
+      "loss": 0.5115,
+      "step": 3128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5840304264163368e-05,
+      "loss": 0.5486,
+      "step": 3130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.583535435207876e-05,
+      "loss": 0.4655,
+      "step": 3132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.583040227105844e-05,
+      "loss": 0.556,
+      "step": 3134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.582544802294303e-05,
+      "loss": 0.5877,
+      "step": 3136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5820491609573977e-05,
+      "loss": 0.5044,
+      "step": 3138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.581553303279352e-05,
+      "loss": 0.4986,
+      "step": 3140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5810572294444706e-05,
+      "loss": 0.6685,
+      "step": 3142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.580560939637139e-05,
+      "loss": 0.54,
+      "step": 3144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5800644340418218e-05,
+      "loss": 0.5799,
+      "step": 3146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.579567712843065e-05,
+      "loss": 0.5236,
+      "step": 3148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5790707762254948e-05,
+      "loss": 0.5323,
+      "step": 3150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5785736243738158e-05,
+      "loss": 0.5652,
+      "step": 3152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.578076257472815e-05,
+      "loss": 0.5737,
+      "step": 3154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5775786757073573e-05,
+      "loss": 0.5263,
+      "step": 3156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.577080879262388e-05,
+      "loss": 0.6274,
+      "step": 3158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5765828683229336e-05,
+      "loss": 0.5519,
+      "step": 3160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5760846430740983e-05,
+      "loss": 0.589,
+      "step": 3162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5755862037010666e-05,
+      "loss": 0.5027,
+      "step": 3164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5750875503891037e-05,
+      "loss": 0.6471,
+      "step": 3166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.574588683323553e-05,
+      "loss": 0.5605,
+      "step": 3168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5740896026898374e-05,
+      "loss": 0.5076,
+      "step": 3170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.57359030867346e-05,
+      "loss": 0.4882,
+      "step": 3172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.573090801460003e-05,
+      "loss": 0.535,
+      "step": 3174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5725910812351273e-05,
+      "loss": 0.567,
+      "step": 3176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.572091148184573e-05,
+      "loss": 0.6027,
+      "step": 3178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5715910024941604e-05,
+      "loss": 0.5879,
+      "step": 3180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5710906443497875e-05,
+      "loss": 0.6317,
+      "step": 3182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.570590073937432e-05,
+      "loss": 0.6201,
+      "step": 3184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5700892914431497e-05,
+      "loss": 0.5697,
+      "step": 3186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.569588297053077e-05,
+      "loss": 0.6411,
+      "step": 3188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5690870909534265e-05,
+      "loss": 0.5455,
+      "step": 3190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.568585673330492e-05,
+      "loss": 0.493,
+      "step": 3192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5680840443706446e-05,
+      "loss": 0.4848,
+      "step": 3194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5675822042603335e-05,
+      "loss": 0.6846,
+      "step": 3196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5670801531860877e-05,
+      "loss": 0.5578,
+      "step": 3198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.566577891334514e-05,
+      "loss": 0.5031,
+      "step": 3200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5660754188922972e-05,
+      "loss": 0.4508,
+      "step": 3202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5655727360462003e-05,
+      "loss": 0.5529,
+      "step": 3204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5650698429830654e-05,
+      "loss": 0.5757,
+      "step": 3206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5645667398898123e-05,
+      "loss": 0.5419,
+      "step": 3208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5640634269534385e-05,
+      "loss": 0.5668,
+      "step": 3210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5635599043610196e-05,
+      "loss": 0.5024,
+      "step": 3212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.56305617229971e-05,
+      "loss": 0.499,
+      "step": 3214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5625522309567406e-05,
+      "loss": 0.5661,
+      "step": 3216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5620480805194208e-05,
+      "loss": 0.5033,
+      "step": 3218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.561543721175138e-05,
+      "loss": 0.5465,
+      "step": 3220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5610391531113562e-05,
+      "loss": 0.4546,
+      "step": 3222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5605343765156187e-05,
+      "loss": 0.6687,
+      "step": 3224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5600293915755445e-05,
+      "loss": 0.5287,
+      "step": 3226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5595241984788312e-05,
+      "loss": 0.5531,
+      "step": 3228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5590187974132534e-05,
+      "loss": 0.5163,
+      "step": 3230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5585131885666627e-05,
+      "loss": 0.612,
+      "step": 3232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5580073721269882e-05,
+      "loss": 0.5451,
+      "step": 3234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.557501348282237e-05,
+      "loss": 0.5929,
+      "step": 3236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5569951172204916e-05,
+      "loss": 0.5991,
+      "step": 3238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.556488679129913e-05,
+      "loss": 0.6053,
+      "step": 3240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.555982034198738e-05,
+      "loss": 0.5239,
+      "step": 3242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.555475182615281e-05,
+      "loss": 0.4703,
+      "step": 3244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5549681245679334e-05,
+      "loss": 0.5255,
+      "step": 3246
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.554460860245163e-05,
+      "loss": 0.5382,
+      "step": 3248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5539533898355137e-05,
+      "loss": 0.5163,
+      "step": 3250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.553445713527607e-05,
+      "loss": 0.4922,
+      "step": 3252
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.55293783151014e-05,
+      "loss": 0.4372,
+      "step": 3254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5524297439718874e-05,
+      "loss": 0.5477,
+      "step": 3256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5519214511016993e-05,
+      "loss": 0.566,
+      "step": 3258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5514129530885022e-05,
+      "loss": 0.5144,
+      "step": 3260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5509042501212995e-05,
+      "loss": 0.5091,
+      "step": 3262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5503953423891703e-05,
+      "loss": 0.4589,
+      "step": 3264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.54988623008127e-05,
+      "loss": 0.5096,
+      "step": 3266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5493769133868295e-05,
+      "loss": 0.6897,
+      "step": 3268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5488673924951568e-05,
+      "loss": 0.59,
+      "step": 3270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5483576675956344e-05,
+      "loss": 0.4883,
+      "step": 3272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5478477388777218e-05,
+      "loss": 0.5793,
+      "step": 3274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5473376065309534e-05,
+      "loss": 0.4725,
+      "step": 3276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5468272707449398e-05,
+      "loss": 0.6198,
+      "step": 3278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5463167317093676e-05,
+      "loss": 0.6104,
+      "step": 3280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5458059896139975e-05,
+      "loss": 0.4928,
+      "step": 3282
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.545295044648667e-05,
+      "loss": 0.55,
+      "step": 3284
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.544783897003289e-05,
+      "loss": 0.5673,
+      "step": 3286
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5442725468678507e-05,
+      "loss": 0.5661,
+      "step": 3288
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.543760994432416e-05,
+      "loss": 0.5008,
+      "step": 3290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5432492398871223e-05,
+      "loss": 0.6061,
+      "step": 3292
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5427372834221836e-05,
+      "loss": 0.5799,
+      "step": 3294
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5422251252278883e-05,
+      "loss": 0.5099,
+      "step": 3296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5417127654945996e-05,
+      "loss": 0.5391,
+      "step": 3298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.541200204412756e-05,
+      "loss": 0.5891,
+      "step": 3300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.540687442172871e-05,
+      "loss": 0.5051,
+      "step": 3302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.540174478965532e-05,
+      "loss": 0.5755,
+      "step": 3304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5396613149814023e-05,
+      "loss": 0.5888,
+      "step": 3306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5391479504112186e-05,
+      "loss": 0.6019,
+      "step": 3308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5386343854457933e-05,
+      "loss": 0.5617,
+      "step": 3310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5381206202760122e-05,
+      "loss": 0.5276,
+      "step": 3312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5376066550928363e-05,
+      "loss": 0.5705,
+      "step": 3314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.537092490087301e-05,
+      "loss": 0.5068,
+      "step": 3316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.536578125450515e-05,
+      "loss": 0.5742,
+      "step": 3318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5360635613736623e-05,
+      "loss": 0.5751,
+      "step": 3320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5355487980480007e-05,
+      "loss": 0.5061,
+      "step": 3322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5350338356648615e-05,
+      "loss": 0.5067,
+      "step": 3324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5345186744156505e-05,
+      "loss": 0.4982,
+      "step": 3326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.534003314491848e-05,
+      "loss": 0.5898,
+      "step": 3328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5334877560850067e-05,
+      "loss": 0.5558,
+      "step": 3330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5329719993867544e-05,
+      "loss": 0.5025,
+      "step": 3332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5324560445887918e-05,
+      "loss": 0.5324,
+      "step": 3334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.531939891882894e-05,
+      "loss": 0.5521,
+      "step": 3336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.531423541460908e-05,
+      "loss": 0.5538,
+      "step": 3338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5309069935147566e-05,
+      "loss": 0.5478,
+      "step": 3340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5303902482364348e-05,
+      "loss": 0.5315,
+      "step": 3342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5298733058180108e-05,
+      "loss": 0.5632,
+      "step": 3344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.529356166451626e-05,
+      "loss": 0.647,
+      "step": 3346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.528838830329495e-05,
+      "loss": 0.5868,
+      "step": 3348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.528321297643907e-05,
+      "loss": 0.5042,
+      "step": 3350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5278035685872218e-05,
+      "loss": 0.6287,
+      "step": 3352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.527285643351875e-05,
+      "loss": 0.528,
+      "step": 3354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.526767522130372e-05,
+      "loss": 0.532,
+      "step": 3356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5262492051152936e-05,
+      "loss": 0.5105,
+      "step": 3358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5257306924992921e-05,
+      "loss": 0.4832,
+      "step": 3360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.525211984475093e-05,
+      "loss": 0.6274,
+      "step": 3362
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.524693081235494e-05,
+      "loss": 0.5725,
+      "step": 3364
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5241739829733656e-05,
+      "loss": 0.4787,
+      "step": 3366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5236546898816516e-05,
+      "loss": 0.4233,
+      "step": 3368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5231352021533661e-05,
+      "loss": 0.6013,
+      "step": 3370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5226155199815977e-05,
+      "loss": 0.5123,
+      "step": 3372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5220956435595065e-05,
+      "loss": 0.5534,
+      "step": 3374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5215755730803242e-05,
+      "loss": 0.5253,
+      "step": 3376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5210553087373554e-05,
+      "loss": 0.4766,
+      "step": 3378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5205348507239768e-05,
+      "loss": 0.6017,
+      "step": 3380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5200141992336361e-05,
+      "loss": 0.6631,
+      "step": 3382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5194933544598547e-05,
+      "loss": 0.519,
+      "step": 3384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5189723165962238e-05,
+      "loss": 0.6671,
+      "step": 3386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5184510858364077e-05,
+      "loss": 0.4805,
+      "step": 3388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.517929662374142e-05,
+      "loss": 0.5282,
+      "step": 3390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.517408046403234e-05,
+      "loss": 0.5938,
+      "step": 3392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5168862381175623e-05,
+      "loss": 0.471,
+      "step": 3394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5163642377110772e-05,
+      "loss": 0.5653,
+      "step": 3396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5158420453778008e-05,
+      "loss": 0.4917,
+      "step": 3398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5153196613118258e-05,
+      "loss": 0.5849,
+      "step": 3400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5147970857073167e-05,
+      "loss": 0.5,
+      "step": 3402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5142743187585081e-05,
+      "loss": 0.5666,
+      "step": 3404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5137513606597081e-05,
+      "loss": 0.5639,
+      "step": 3406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.513228211605293e-05,
+      "loss": 0.587,
+      "step": 3408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5127048717897124e-05,
+      "loss": 0.5464,
+      "step": 3410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5121813414074854e-05,
+      "loss": 0.6027,
+      "step": 3412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5116576206532023e-05,
+      "loss": 0.4892,
+      "step": 3414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5111337097215244e-05,
+      "loss": 0.5435,
+      "step": 3416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5106096088071833e-05,
+      "loss": 0.6106,
+      "step": 3418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5100853181049817e-05,
+      "loss": 0.5009,
+      "step": 3420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5095608378097923e-05,
+      "loss": 0.589,
+      "step": 3422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.509036168116559e-05,
+      "loss": 0.5169,
+      "step": 3424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5085113092202953e-05,
+      "loss": 0.5591,
+      "step": 3426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5079862613160852e-05,
+      "loss": 0.5847,
+      "step": 3428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5074610245990833e-05,
+      "loss": 0.5502,
+      "step": 3430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5069355992645145e-05,
+      "loss": 0.5788,
+      "step": 3432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5064099855076729e-05,
+      "loss": 0.5029,
+      "step": 3434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5058841835239233e-05,
+      "loss": 0.5885,
+      "step": 3436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5053581935087012e-05,
+      "loss": 0.2914,
+      "step": 3438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5048320156575102e-05,
+      "loss": 0.3394,
+      "step": 3440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5043056501659252e-05,
+      "loss": 0.3261,
+      "step": 3442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5037790972295903e-05,
+      "loss": 0.3153,
+      "step": 3444
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5032523570442192e-05,
+      "loss": 0.2759,
+      "step": 3446
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5027254298055956e-05,
+      "loss": 0.2843,
+      "step": 3448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.502198315709572e-05,
+      "loss": 0.3589,
+      "step": 3450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5016710149520708e-05,
+      "loss": 0.2735,
+      "step": 3452
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5011435277290844e-05,
+      "loss": 0.29,
+      "step": 3454
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5006158542366733e-05,
+      "loss": 0.3255,
+      "step": 3456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.500087994670968e-05,
+      "loss": 0.304,
+      "step": 3458
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4995599492281683e-05,
+      "loss": 0.2821,
+      "step": 3460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.499031718104542e-05,
+      "loss": 0.2675,
+      "step": 3462
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4985033014964274e-05,
+      "loss": 0.2794,
+      "step": 3464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4979746996002304e-05,
+      "loss": 0.2863,
+      "step": 3466
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.497445912612427e-05,
+      "loss": 0.2588,
+      "step": 3468
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4969169407295607e-05,
+      "loss": 0.2867,
+      "step": 3470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4963877841482455e-05,
+      "loss": 0.3346,
+      "step": 3472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4958584430651617e-05,
+      "loss": 0.2879,
+      "step": 3474
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4953289176770602e-05,
+      "loss": 0.2632,
+      "step": 3476
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4947992081807594e-05,
+      "loss": 0.2605,
+      "step": 3478
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4942693147731468e-05,
+      "loss": 0.2859,
+      "step": 3480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.493739237651177e-05,
+      "loss": 0.2637,
+      "step": 3482
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4932089770118744e-05,
+      "loss": 0.2884,
+      "step": 3484
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.492678533052331e-05,
+      "loss": 0.3153,
+      "step": 3486
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4921479059697061e-05,
+      "loss": 0.2844,
+      "step": 3488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4916170959612287e-05,
+      "loss": 0.2673,
+      "step": 3490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4910861032241941e-05,
+      "loss": 0.2758,
+      "step": 3492
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4905549279559672e-05,
+      "loss": 0.2502,
+      "step": 3494
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4900235703539792e-05,
+      "loss": 0.3163,
+      "step": 3496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4894920306157303e-05,
+      "loss": 0.2549,
+      "step": 3498
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4889603089387872e-05,
+      "loss": 0.258,
+      "step": 3500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4884284055207856e-05,
+      "loss": 0.2471,
+      "step": 3502
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4878963205594275e-05,
+      "loss": 0.213,
+      "step": 3504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4873640542524832e-05,
+      "loss": 0.2395,
+      "step": 3506
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4868316067977899e-05,
+      "loss": 0.2486,
+      "step": 3508
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4862989783932525e-05,
+      "loss": 0.2774,
+      "step": 3510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4857661692368428e-05,
+      "loss": 0.2735,
+      "step": 3512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4852331795266003e-05,
+      "loss": 0.2924,
+      "step": 3514
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4847000094606311e-05,
+      "loss": 0.2826,
+      "step": 3516
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4841666592371086e-05,
+      "loss": 0.2854,
+      "step": 3518
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4836331290542732e-05,
+      "loss": 0.2107,
+      "step": 3520
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4830994191104317e-05,
+      "loss": 0.2728,
+      "step": 3522
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4825655296039583e-05,
+      "loss": 0.2774,
+      "step": 3524
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4820314607332938e-05,
+      "loss": 0.2728,
+      "step": 3526
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4814972126969456e-05,
+      "loss": 0.2896,
+      "step": 3528
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4809627856934876e-05,
+      "loss": 0.2962,
+      "step": 3530
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4804281799215606e-05,
+      "loss": 0.2884,
+      "step": 3532
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4798933955798708e-05,
+      "loss": 0.2284,
+      "step": 3534
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4793584328671928e-05,
+      "loss": 0.2519,
+      "step": 3536
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4788232919823651e-05,
+      "loss": 0.2757,
+      "step": 3538
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.478287973124294e-05,
+      "loss": 0.269,
+      "step": 3540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4777524764919518e-05,
+      "loss": 0.2414,
+      "step": 3542
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4772168022843758e-05,
+      "loss": 0.2861,
+      "step": 3544
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4766809507006712e-05,
+      "loss": 0.3715,
+      "step": 3546
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4761449219400066e-05,
+      "loss": 0.3125,
+      "step": 3548
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4756087162016193e-05,
+      "loss": 0.3207,
+      "step": 3550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.47507233368481e-05,
+      "loss": 0.2288,
+      "step": 3552
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4745357745889467e-05,
+      "loss": 0.2811,
+      "step": 3554
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4739990391134615e-05,
+      "loss": 0.329,
+      "step": 3556
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.473462127457854e-05,
+      "loss": 0.2312,
+      "step": 3558
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4729250398216876e-05,
+      "loss": 0.2757,
+      "step": 3560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4723877764045921e-05,
+      "loss": 0.2191,
+      "step": 3562
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.471850337406262e-05,
+      "loss": 0.2638,
+      "step": 3564
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4713127230264572e-05,
+      "loss": 0.2746,
+      "step": 3566
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4707749334650036e-05,
+      "loss": 0.2947,
+      "step": 3568
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4702369689217907e-05,
+      "loss": 0.2501,
+      "step": 3570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4696988295967744e-05,
+      "loss": 0.2354,
+      "step": 3572
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.469160515689975e-05,
+      "loss": 0.2855,
+      "step": 3574
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.468622027401478e-05,
+      "loss": 0.2863,
+      "step": 3576
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4680833649314326e-05,
+      "loss": 0.2733,
+      "step": 3578
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4675445284800542e-05,
+      "loss": 0.2817,
+      "step": 3580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4670055182476224e-05,
+      "loss": 0.243,
+      "step": 3582
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4664663344344811e-05,
+      "loss": 0.272,
+      "step": 3584
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4659269772410385e-05,
+      "loss": 0.278,
+      "step": 3586
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.465387446867768e-05,
+      "loss": 0.3142,
+      "step": 3588
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4648477435152073e-05,
+      "loss": 0.237,
+      "step": 3590
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4643078673839572e-05,
+      "loss": 0.2547,
+      "step": 3592
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4637678186746845e-05,
+      "loss": 0.234,
+      "step": 3594
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4632275975881186e-05,
+      "loss": 0.2979,
+      "step": 3596
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4626872043250538e-05,
+      "loss": 0.2968,
+      "step": 3598
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4621466390863481e-05,
+      "loss": 0.2935,
+      "step": 3600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4616059020729241e-05,
+      "loss": 0.249,
+      "step": 3602
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.461064993485767e-05,
+      "loss": 0.2356,
+      "step": 3604
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4605239135259268e-05,
+      "loss": 0.2674,
+      "step": 3606
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4599826623945169e-05,
+      "loss": 0.2733,
+      "step": 3608
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4594412402927139e-05,
+      "loss": 0.2634,
+      "step": 3610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4588996474217587e-05,
+      "loss": 0.247,
+      "step": 3612
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4583578839829549e-05,
+      "loss": 0.2572,
+      "step": 3614
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4578159501776705e-05,
+      "loss": 0.2965,
+      "step": 3616
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4572738462073357e-05,
+      "loss": 0.309,
+      "step": 3618
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4567315722734445e-05,
+      "loss": 0.2715,
+      "step": 3620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4561891285775542e-05,
+      "loss": 0.2659,
+      "step": 3622
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.455646515321285e-05,
+      "loss": 0.2497,
+      "step": 3624
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4551037327063202e-05,
+      "loss": 0.276,
+      "step": 3626
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.454560780934406e-05,
+      "loss": 0.261,
+      "step": 3628
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4540176602073512e-05,
+      "loss": 0.2547,
+      "step": 3630
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4534743707270282e-05,
+      "loss": 0.3118,
+      "step": 3632
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.452930912695371e-05,
+      "loss": 0.339,
+      "step": 3634
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4523872863143773e-05,
+      "loss": 0.3441,
+      "step": 3636
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4518434917861068e-05,
+      "loss": 0.2476,
+      "step": 3638
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4512995293126816e-05,
+      "loss": 0.297,
+      "step": 3640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.450755399096287e-05,
+      "loss": 0.2692,
+      "step": 3642
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4502111013391694e-05,
+      "loss": 0.272,
+      "step": 3644
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4496666362436387e-05,
+      "loss": 0.2675,
+      "step": 3646
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4491220040120662e-05,
+      "loss": 0.2988,
+      "step": 3648
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4485772048468858e-05,
+      "loss": 0.3195,
+      "step": 3650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4480322389505927e-05,
+      "loss": 0.3067,
+      "step": 3652
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4474871065257454e-05,
+      "loss": 0.2509,
+      "step": 3654
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.446941807774963e-05,
+      "loss": 0.239,
+      "step": 3656
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4463963429009272e-05,
+      "loss": 0.2834,
+      "step": 3658
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4458507121063807e-05,
+      "loss": 0.2901,
+      "step": 3660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4453049155941287e-05,
+      "loss": 0.2537,
+      "step": 3662
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4447589535670379e-05,
+      "loss": 0.2855,
+      "step": 3664
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4442128262280356e-05,
+      "loss": 0.2575,
+      "step": 3666
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4436665337801115e-05,
+      "loss": 0.2733,
+      "step": 3668
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4431200764263164e-05,
+      "loss": 0.2892,
+      "step": 3670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4425734543697628e-05,
+      "loss": 0.2628,
+      "step": 3672
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4420266678136233e-05,
+      "loss": 0.2719,
+      "step": 3674
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4414797169611326e-05,
+      "loss": 0.2532,
+      "step": 3676
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4409326020155861e-05,
+      "loss": 0.2456,
+      "step": 3678
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4403853231803405e-05,
+      "loss": 0.2795,
+      "step": 3680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.439837880658813e-05,
+      "loss": 0.2598,
+      "step": 3682
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4392902746544817e-05,
+      "loss": 0.3466,
+      "step": 3684
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.438742505370886e-05,
+      "loss": 0.2731,
+      "step": 3686
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4381945730116256e-05,
+      "loss": 0.3081,
+      "step": 3688
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4376464777803605e-05,
+      "loss": 0.2854,
+      "step": 3690
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4370982198808112e-05,
+      "loss": 0.308,
+      "step": 3692
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.43654979951676e-05,
+      "loss": 0.2769,
+      "step": 3694
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4360012168920482e-05,
+      "loss": 0.261,
+      "step": 3696
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4354524722105774e-05,
+      "loss": 0.2908,
+      "step": 3698
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4349035656763098e-05,
+      "loss": 0.298,
+      "step": 3700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4343544974932684e-05,
+      "loss": 0.2812,
+      "step": 3702
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4338052678655355e-05,
+      "loss": 0.2941,
+      "step": 3704
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4332558769972535e-05,
+      "loss": 0.2861,
+      "step": 3706
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4327063250926246e-05,
+      "loss": 0.2769,
+      "step": 3708
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4321566123559112e-05,
+      "loss": 0.2991,
+      "step": 3710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4316067389914356e-05,
+      "loss": 0.2588,
+      "step": 3712
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4310567052035795e-05,
+      "loss": 0.2576,
+      "step": 3714
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4305065111967839e-05,
+      "loss": 0.2586,
+      "step": 3716
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.42995615717555e-05,
+      "loss": 0.223,
+      "step": 3718
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4294056433444383e-05,
+      "loss": 0.2853,
+      "step": 3720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4288549699080687e-05,
+      "loss": 0.2419,
+      "step": 3722
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4283041370711199e-05,
+      "loss": 0.318,
+      "step": 3724
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4277531450383307e-05,
+      "loss": 0.2735,
+      "step": 3726
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4272019940144986e-05,
+      "loss": 0.2844,
+      "step": 3728
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4266506842044798e-05,
+      "loss": 0.2603,
+      "step": 3730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4260992158131906e-05,
+      "loss": 0.2472,
+      "step": 3732
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4255475890456052e-05,
+      "loss": 0.2903,
+      "step": 3734
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4249958041067573e-05,
+      "loss": 0.3285,
+      "step": 3736
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.424443861201739e-05,
+      "loss": 0.262,
+      "step": 3738
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4238917605357014e-05,
+      "loss": 0.2377,
+      "step": 3740
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4233395023138541e-05,
+      "loss": 0.2432,
+      "step": 3742
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4227870867414653e-05,
+      "loss": 0.2469,
+      "step": 3744
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4222345140238618e-05,
+      "loss": 0.261,
+      "step": 3746
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.421681784366428e-05,
+      "loss": 0.2513,
+      "step": 3748
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.421128897974608e-05,
+      "loss": 0.2719,
+      "step": 3750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4205758550539034e-05,
+      "loss": 0.2486,
+      "step": 3752
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4200226558098741e-05,
+      "loss": 0.2707,
+      "step": 3754
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4194693004481378e-05,
+      "loss": 0.2699,
+      "step": 3756
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4189157891743704e-05,
+      "loss": 0.3271,
+      "step": 3758
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4183621221943061e-05,
+      "loss": 0.3203,
+      "step": 3760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4178082997137365e-05,
+      "loss": 0.3192,
+      "step": 3762
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4172543219385112e-05,
+      "loss": 0.3133,
+      "step": 3764
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4167001890745378e-05,
+      "loss": 0.244,
+      "step": 3766
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4161459013277812e-05,
+      "loss": 0.2829,
+      "step": 3768
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4155914589042637e-05,
+      "loss": 0.265,
+      "step": 3770
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4150368620100652e-05,
+      "loss": 0.309,
+      "step": 3772
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4144821108513233e-05,
+      "loss": 0.2731,
+      "step": 3774
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4139272056342335e-05,
+      "loss": 0.2772,
+      "step": 3776
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4133721465650468e-05,
+      "loss": 0.2818,
+      "step": 3778
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4128169338500729e-05,
+      "loss": 0.2733,
+      "step": 3780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4122615676956777e-05,
+      "loss": 0.2601,
+      "step": 3782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4117060483082854e-05,
+      "loss": 0.2786,
+      "step": 3784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4111503758943754e-05,
+      "loss": 0.2759,
+      "step": 3786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4105945506604857e-05,
+      "loss": 0.2624,
+      "step": 3788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.41003857281321e-05,
+      "loss": 0.2191,
+      "step": 3790
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4094824425591989e-05,
+      "loss": 0.305,
+      "step": 3792
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4089261601051599e-05,
+      "loss": 0.2872,
+      "step": 3794
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4083697256578564e-05,
+      "loss": 0.2552,
+      "step": 3796
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4078131394241097e-05,
+      "loss": 0.2785,
+      "step": 3798
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4072564016107964e-05,
+      "loss": 0.2543,
+      "step": 3800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4066995124248496e-05,
+      "loss": 0.2409,
+      "step": 3802
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4061424720732584e-05,
+      "loss": 0.3273,
+      "step": 3804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.405585280763069e-05,
+      "loss": 0.2536,
+      "step": 3806
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.405027938701383e-05,
+      "loss": 0.2968,
+      "step": 3808
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.404470446095358e-05,
+      "loss": 0.2872,
+      "step": 3810
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.403912803152208e-05,
+      "loss": 0.2497,
+      "step": 3812
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4036339253689806e-05,
+      "loss": 0.2314,
+      "step": 3814
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4033550100792024e-05,
+      "loss": 0.3078,
+      "step": 3816
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4027970670836669e-05,
+      "loss": 0.2895,
+      "step": 3818
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4022389743729823e-05,
+      "loss": 0.2896,
+      "step": 3820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4016807321545855e-05,
+      "loss": 0.2494,
+      "step": 3822
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4011223406359691e-05,
+      "loss": 0.2647,
+      "step": 3824
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4005638000246808e-05,
+      "loss": 0.2918,
+      "step": 3826
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4000051105283238e-05,
+      "loss": 0.2683,
+      "step": 3828
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3994462723545568e-05,
+      "loss": 0.2971,
+      "step": 3830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3988872857110936e-05,
+      "loss": 0.2775,
+      "step": 3832
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3983281508057037e-05,
+      "loss": 0.2532,
+      "step": 3834
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3977688678462105e-05,
+      "loss": 0.2797,
+      "step": 3836
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3972094370404936e-05,
+      "loss": 0.2511,
+      "step": 3838
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3966498585964874e-05,
+      "loss": 0.2776,
+      "step": 3840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3960901327221806e-05,
+      "loss": 0.2795,
+      "step": 3842
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.395530259625617e-05,
+      "loss": 0.2838,
+      "step": 3844
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3949702395148949e-05,
+      "loss": 0.2921,
+      "step": 3846
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3944100725981686e-05,
+      "loss": 0.2364,
+      "step": 3848
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3938497590836445e-05,
+      "loss": 0.2634,
+      "step": 3850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3932892991795858e-05,
+      "loss": 0.2566,
+      "step": 3852
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3927286930943085e-05,
+      "loss": 0.2862,
+      "step": 3854
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3921679410361842e-05,
+      "loss": 0.2754,
+      "step": 3856
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3916070432136376e-05,
+      "loss": 0.2672,
+      "step": 3858
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3910459998351485e-05,
+      "loss": 0.2716,
+      "step": 3860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3904848111092502e-05,
+      "loss": 0.2726,
+      "step": 3862
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3899234772445303e-05,
+      "loss": 0.2737,
+      "step": 3864
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3893619984496302e-05,
+      "loss": 0.2671,
+      "step": 3866
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3888003749332457e-05,
+      "loss": 0.2824,
+      "step": 3868
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3882386069041253e-05,
+      "loss": 0.2769,
+      "step": 3870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3876766945710725e-05,
+      "loss": 0.3045,
+      "step": 3872
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3871146381429435e-05,
+      "loss": 0.2779,
+      "step": 3874
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3865524378286482e-05,
+      "loss": 0.2701,
+      "step": 3876
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3859900938371504e-05,
+      "loss": 0.256,
+      "step": 3878
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.385427606377467e-05,
+      "loss": 0.2502,
+      "step": 3880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3848649756586685e-05,
+      "loss": 0.2463,
+      "step": 3882
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.384302201889878e-05,
+      "loss": 0.2982,
+      "step": 3884
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3837392852802723e-05,
+      "loss": 0.3125,
+      "step": 3886
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3831762260390816e-05,
+      "loss": 0.2673,
+      "step": 3888
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.382613024375588e-05,
+      "loss": 0.3083,
+      "step": 3890
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3820496804991282e-05,
+      "loss": 0.299,
+      "step": 3892
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3814861946190892e-05,
+      "loss": 0.2761,
+      "step": 3894
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3809225669449142e-05,
+      "loss": 0.2828,
+      "step": 3896
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3803587976860962e-05,
+      "loss": 0.2563,
+      "step": 3898
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3797948870521822e-05,
+      "loss": 0.2506,
+      "step": 3900
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3792308352527712e-05,
+      "loss": 0.2698,
+      "step": 3902
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3786666424975154e-05,
+      "loss": 0.2831,
+      "step": 3904
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3781023089961188e-05,
+      "loss": 0.2375,
+      "step": 3906
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3775378349583374e-05,
+      "loss": 0.2714,
+      "step": 3908
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3769732205939804e-05,
+      "loss": 0.3235,
+      "step": 3910
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.376408466112908e-05,
+      "loss": 0.2843,
+      "step": 3912
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3758435717250338e-05,
+      "loss": 0.2514,
+      "step": 3914
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3752785376403224e-05,
+      "loss": 0.2414,
+      "step": 3916
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3747133640687903e-05,
+      "loss": 0.3135,
+      "step": 3918
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3741480512205066e-05,
+      "loss": 0.2375,
+      "step": 3920
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3735825993055918e-05,
+      "loss": 0.2478,
+      "step": 3922
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3730170085342179e-05,
+      "loss": 0.2559,
+      "step": 3924
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.372451279116608e-05,
+      "loss": 0.2682,
+      "step": 3926
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3718854112630384e-05,
+      "loss": 0.268,
+      "step": 3928
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3713194051838357e-05,
+      "loss": 0.2464,
+      "step": 3930
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3707532610893773e-05,
+      "loss": 0.3108,
+      "step": 3932
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.370186979190093e-05,
+      "loss": 0.2723,
+      "step": 3934
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3696205596964638e-05,
+      "loss": 0.2887,
+      "step": 3936
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3690540028190207e-05,
+      "loss": 0.261,
+      "step": 3938
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3684873087683473e-05,
+      "loss": 0.263,
+      "step": 3940
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3679204777550765e-05,
+      "loss": 0.2536,
+      "step": 3942
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.367353509989894e-05,
+      "loss": 0.2872,
+      "step": 3944
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3667864056835347e-05,
+      "loss": 0.3554,
+      "step": 3946
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3662191650467855e-05,
+      "loss": 0.3762,
+      "step": 3948
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3656517882904825e-05,
+      "loss": 0.2952,
+      "step": 3950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3650842756255145e-05,
+      "loss": 0.2709,
+      "step": 3952
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3645166272628182e-05,
+      "loss": 0.2201,
+      "step": 3954
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3639488434133834e-05,
+      "loss": 0.3008,
+      "step": 3956
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3633809242882479e-05,
+      "loss": 0.359,
+      "step": 3958
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3628128700985015e-05,
+      "loss": 0.2826,
+      "step": 3960
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3622446810552837e-05,
+      "loss": 0.2171,
+      "step": 3962
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3616763573697833e-05,
+      "loss": 0.2391,
+      "step": 3964
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3611078992532405e-05,
+      "loss": 0.2947,
+      "step": 3966
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3608236198493538e-05,
+      "loss": 0.2946,
+      "step": 3968
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3602549604824304e-05,
+      "loss": 0.2332,
+      "step": 3970
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3596861672127817e-05,
+      "loss": 0.2575,
+      "step": 3972
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3591172402518216e-05,
+      "loss": 0.308,
+      "step": 3974
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3585481798110135e-05,
+      "loss": 0.3067,
+      "step": 3976
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.357978986101872e-05,
+      "loss": 0.2649,
+      "step": 3978
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3574096593359588e-05,
+      "loss": 0.3048,
+      "step": 3980
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.356840199724887e-05,
+      "loss": 0.3111,
+      "step": 3982
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3562706074803177e-05,
+      "loss": 0.2849,
+      "step": 3984
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3559857616866294e-05,
+      "loss": 0.267,
+      "step": 3986
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3554159708887882e-05,
+      "loss": 0.3064,
+      "step": 3988
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3548460479868195e-05,
+      "loss": 0.3034,
+      "step": 3990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3542759931925577e-05,
+      "loss": 0.2655,
+      "step": 3992
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3537058067178857e-05,
+      "loss": 0.2778,
+      "step": 3994
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3531354887747349e-05,
+      "loss": 0.2837,
+      "step": 3996
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3525650395750869e-05,
+      "loss": 0.2445,
+      "step": 3998
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3519944593309703e-05,
+      "loss": 0.2457,
+      "step": 4000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3514237482544641e-05,
+      "loss": 0.2726,
+      "step": 4002
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3508529065576949e-05,
+      "loss": 0.3047,
+      "step": 4004
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3502819344528386e-05,
+      "loss": 0.2725,
+      "step": 4006
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3497108321521184e-05,
+      "loss": 0.2699,
+      "step": 4008
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3491395998678068e-05,
+      "loss": 0.278,
+      "step": 4010
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.348568237812225e-05,
+      "loss": 0.3209,
+      "step": 4012
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3479967461977412e-05,
+      "loss": 0.2661,
+      "step": 4014
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3474251252367727e-05,
+      "loss": 0.245,
+      "step": 4016
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3468533751417847e-05,
+      "loss": 0.2456,
+      "step": 4018
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3462814961252902e-05,
+      "loss": 0.2545,
+      "step": 4020
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.34570948839985e-05,
+      "loss": 0.2792,
+      "step": 4022
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3451373521780733e-05,
+      "loss": 0.2912,
+      "step": 4024
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3445650876726164e-05,
+      "loss": 0.3052,
+      "step": 4026
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.343992695096184e-05,
+      "loss": 0.2762,
+      "step": 4028
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3434201746615274e-05,
+      "loss": 0.2923,
+      "step": 4030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3428475265814465e-05,
+      "loss": 0.2675,
+      "step": 4032
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.342274751068788e-05,
+      "loss": 0.2439,
+      "step": 4034
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3417018483364464e-05,
+      "loss": 0.2818,
+      "step": 4036
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3411288185973626e-05,
+      "loss": 0.3018,
+      "step": 4038
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3405556620645258e-05,
+      "loss": 0.3128,
+      "step": 4040
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3399823789509718e-05,
+      "loss": 0.341,
+      "step": 4042
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3394089694697833e-05,
+      "loss": 0.3085,
+      "step": 4044
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3388354338340905e-05,
+      "loss": 0.2768,
+      "step": 4046
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3382617722570696e-05,
+      "loss": 0.2765,
+      "step": 4048
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3376879849519451e-05,
+      "loss": 0.2968,
+      "step": 4050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3371140721319864e-05,
+      "loss": 0.2849,
+      "step": 4052
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3365400340105108e-05,
+      "loss": 0.2762,
+      "step": 4054
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3359658708008817e-05,
+      "loss": 0.2982,
+      "step": 4056
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3353915827165094e-05,
+      "loss": 0.2723,
+      "step": 4058
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3348171699708501e-05,
+      "loss": 0.3018,
+      "step": 4060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3342426327774067e-05,
+      "loss": 0.2726,
+      "step": 4062
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3336679713497283e-05,
+      "loss": 0.3069,
+      "step": 4064
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3330931859014097e-05,
+      "loss": 0.2864,
+      "step": 4066
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3325182766460927e-05,
+      "loss": 0.2877,
+      "step": 4068
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3319432437974642e-05,
+      "loss": 0.2928,
+      "step": 4070
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3313680875692576e-05,
+      "loss": 0.2726,
+      "step": 4072
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3307928081752522e-05,
+      "loss": 0.2419,
+      "step": 4074
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3302174058292728e-05,
+      "loss": 0.2372,
+      "step": 4076
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3296418807451898e-05,
+      "loss": 0.2527,
+      "step": 4078
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3290662331369195e-05,
+      "loss": 0.3089,
+      "step": 4080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.328490463218424e-05,
+      "loss": 0.2939,
+      "step": 4082
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3279145712037101e-05,
+      "loss": 0.2783,
+      "step": 4084
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3273385573068302e-05,
+      "loss": 0.2689,
+      "step": 4086
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3267624217418828e-05,
+      "loss": 0.32,
+      "step": 4088
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.326186164723011e-05,
+      "loss": 0.2502,
+      "step": 4090
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3256097864644026e-05,
+      "loss": 0.2666,
+      "step": 4092
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3250332871802911e-05,
+      "loss": 0.256,
+      "step": 4094
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3244566670849548e-05,
+      "loss": 0.2713,
+      "step": 4096
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.323879926392717e-05,
+      "loss": 0.3407,
+      "step": 4098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.323303065317946e-05,
+      "loss": 0.2758,
+      "step": 4100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3227260840750541e-05,
+      "loss": 0.2753,
+      "step": 4102
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3221489828784986e-05,
+      "loss": 0.2798,
+      "step": 4104
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3215717619427826e-05,
+      "loss": 0.3016,
+      "step": 4106
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3209944214824515e-05,
+      "loss": 0.256,
+      "step": 4108
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.320416961712097e-05,
+      "loss": 0.2564,
+      "step": 4110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3198393828463539e-05,
+      "loss": 0.2704,
+      "step": 4112
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3192616850999018e-05,
+      "loss": 0.3071,
+      "step": 4114
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.318683868687465e-05,
+      "loss": 0.2804,
+      "step": 4116
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3181059338238109e-05,
+      "loss": 0.2482,
+      "step": 4118
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3175278807237517e-05,
+      "loss": 0.2903,
+      "step": 4120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3169497096021429e-05,
+      "loss": 0.3009,
+      "step": 4122
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3163714206738847e-05,
+      "loss": 0.2636,
+      "step": 4124
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.31579301415392e-05,
+      "loss": 0.3288,
+      "step": 4126
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3152144902572363e-05,
+      "loss": 0.2758,
+      "step": 4128
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3146358491988647e-05,
+      "loss": 0.2766,
+      "step": 4130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3140570911938795e-05,
+      "loss": 0.2612,
+      "step": 4132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3134782164573983e-05,
+      "loss": 0.2706,
+      "step": 4134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3128992252045823e-05,
+      "loss": 0.2287,
+      "step": 4136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3123201176506363e-05,
+      "loss": 0.3362,
+      "step": 4138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3117408940108081e-05,
+      "loss": 0.2595,
+      "step": 4140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3111615545003884e-05,
+      "loss": 0.2841,
+      "step": 4142
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3105820993347112e-05,
+      "loss": 0.2926,
+      "step": 4144
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3100025287291538e-05,
+      "loss": 0.2965,
+      "step": 4146
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3094228428991356e-05,
+      "loss": 0.251,
+      "step": 4148
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.30884304206012e-05,
+      "loss": 0.2363,
+      "step": 4150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3082631264276113e-05,
+      "loss": 0.2842,
+      "step": 4152
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3076830962171591e-05,
+      "loss": 0.2848,
+      "step": 4154
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3071029516443532e-05,
+      "loss": 0.3466,
+      "step": 4156
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3065226929248272e-05,
+      "loss": 0.3308,
+      "step": 4158
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3059423202742564e-05,
+      "loss": 0.2861,
+      "step": 4160
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.305361833908359e-05,
+      "loss": 0.2775,
+      "step": 4162
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3047812340428954e-05,
+      "loss": 0.2803,
+      "step": 4164
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3042005208936677e-05,
+      "loss": 0.2638,
+      "step": 4166
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3036196946765207e-05,
+      "loss": 0.3372,
+      "step": 4168
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3030387556073407e-05,
+      "loss": 0.2991,
+      "step": 4170
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3024577039020571e-05,
+      "loss": 0.3154,
+      "step": 4172
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3018765397766393e-05,
+      "loss": 0.2756,
+      "step": 4174
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3012952634470999e-05,
+      "loss": 0.2606,
+      "step": 4176
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3007138751294926e-05,
+      "loss": 0.253,
+      "step": 4178
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3001323750399136e-05,
+      "loss": 0.238,
+      "step": 4180
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2995507633944988e-05,
+      "loss": 0.2746,
+      "step": 4182
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2989690404094275e-05,
+      "loss": 0.3015,
+      "step": 4184
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2983872063009198e-05,
+      "loss": 0.2744,
+      "step": 4186
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2978052612852362e-05,
+      "loss": 0.2517,
+      "step": 4188
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2972232055786798e-05,
+      "loss": 0.3049,
+      "step": 4190
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2966410393975933e-05,
+      "loss": 0.2916,
+      "step": 4192
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2960587629583623e-05,
+      "loss": 0.2982,
+      "step": 4194
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2954763764774119e-05,
+      "loss": 0.2783,
+      "step": 4196
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.294893880171209e-05,
+      "loss": 0.2737,
+      "step": 4198
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.29431127425626e-05,
+      "loss": 0.246,
+      "step": 4200
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2937285589491144e-05,
+      "loss": 0.2698,
+      "step": 4202
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2931457344663599e-05,
+      "loss": 0.2386,
+      "step": 4204
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2925628010246261e-05,
+      "loss": 0.2688,
+      "step": 4206
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2919797588405828e-05,
+      "loss": 0.2716,
+      "step": 4208
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2913966081309402e-05,
+      "loss": 0.3407,
+      "step": 4210
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.290813349112449e-05,
+      "loss": 0.268,
+      "step": 4212
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2902299820019e-05,
+      "loss": 0.2479,
+      "step": 4214
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2896465070161238e-05,
+      "loss": 0.3325,
+      "step": 4216
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2890629243719923e-05,
+      "loss": 0.2781,
+      "step": 4218
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2884792342864164e-05,
+      "loss": 0.322,
+      "step": 4220
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2878954369763464e-05,
+      "loss": 0.3298,
+      "step": 4222
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2873115326587739e-05,
+      "loss": 0.2814,
+      "step": 4224
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2867275215507296e-05,
+      "loss": 0.2831,
+      "step": 4226
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2861434038692838e-05,
+      "loss": 0.2546,
+      "step": 4228
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.285559179831546e-05,
+      "loss": 0.2942,
+      "step": 4230
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2849748496546663e-05,
+      "loss": 0.2877,
+      "step": 4232
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2843904135558338e-05,
+      "loss": 0.2911,
+      "step": 4234
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2838058717522762e-05,
+      "loss": 0.258,
+      "step": 4236
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2832212244612614e-05,
+      "loss": 0.2752,
+      "step": 4238
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.282636471900096e-05,
+      "loss": 0.3064,
+      "step": 4240
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.282051614286127e-05,
+      "loss": 0.2491,
+      "step": 4242
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2814666518367382e-05,
+      "loss": 0.2717,
+      "step": 4244
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2808815847693543e-05,
+      "loss": 0.2945,
+      "step": 4246
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2802964133014375e-05,
+      "loss": 0.2918,
+      "step": 4248
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.27971113765049e-05,
+      "loss": 0.2678,
+      "step": 4250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2791257580340522e-05,
+      "loss": 0.2808,
+      "step": 4252
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2785402746697031e-05,
+      "loss": 0.2973,
+      "step": 4254
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2779546877750596e-05,
+      "loss": 0.2438,
+      "step": 4256
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.277368997567779e-05,
+      "loss": 0.2581,
+      "step": 4258
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2767832042655548e-05,
+      "loss": 0.2621,
+      "step": 4260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2761973080861206e-05,
+      "loss": 0.3014,
+      "step": 4262
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2756113092472464e-05,
+      "loss": 0.2508,
+      "step": 4264
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.275025207966742e-05,
+      "loss": 0.2408,
+      "step": 4266
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.274439004462455e-05,
+      "loss": 0.2254,
+      "step": 4268
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.27385269895227e-05,
+      "loss": 0.2803,
+      "step": 4270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2732662916541105e-05,
+      "loss": 0.2745,
+      "step": 4272
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2726797827859374e-05,
+      "loss": 0.2454,
+      "step": 4274
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2720931725657497e-05,
+      "loss": 0.28,
+      "step": 4276
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2715064612115836e-05,
+      "loss": 0.2478,
+      "step": 4278
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2709196489415131e-05,
+      "loss": 0.2961,
+      "step": 4280
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.27033273597365e-05,
+      "loss": 0.2447,
+      "step": 4282
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2697457225261432e-05,
+      "loss": 0.2661,
+      "step": 4284
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2691586088171785e-05,
+      "loss": 0.2814,
+      "step": 4286
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2685713950649803e-05,
+      "loss": 0.2308,
+      "step": 4288
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2679840814878089e-05,
+      "loss": 0.2803,
+      "step": 4290
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.267396668303962e-05,
+      "loss": 0.2608,
+      "step": 4292
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.266809155731775e-05,
+      "loss": 0.2892,
+      "step": 4294
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2662215439896188e-05,
+      "loss": 0.2734,
+      "step": 4296
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2656338332959033e-05,
+      "loss": 0.2837,
+      "step": 4298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2650460238690731e-05,
+      "loss": 0.256,
+      "step": 4300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.264458115927611e-05,
+      "loss": 0.2499,
+      "step": 4302
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.263870109690035e-05,
+      "loss": 0.2945,
+      "step": 4304
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2632820053749011e-05,
+      "loss": 0.2628,
+      "step": 4306
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2626938032008006e-05,
+      "loss": 0.317,
+      "step": 4308
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2621055033863622e-05,
+      "loss": 0.3312,
+      "step": 4310
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2615171061502498e-05,
+      "loss": 0.2269,
+      "step": 4312
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2609286117111642e-05,
+      "loss": 0.2972,
+      "step": 4314
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2603400202878427e-05,
+      "loss": 0.307,
+      "step": 4316
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.259751332099057e-05,
+      "loss": 0.2769,
+      "step": 4318
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2591625473636169e-05,
+      "loss": 0.2522,
+      "step": 4320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2585736663003668e-05,
+      "loss": 0.2802,
+      "step": 4322
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.257984689128187e-05,
+      "loss": 0.2613,
+      "step": 4324
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2573956160659937e-05,
+      "loss": 0.2432,
+      "step": 4326
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.256806447332739e-05,
+      "loss": 0.2916,
+      "step": 4328
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2562171831474101e-05,
+      "loss": 0.2635,
+      "step": 4330
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.25562782372903e-05,
+      "loss": 0.2647,
+      "step": 4332
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2550383692966568e-05,
+      "loss": 0.2916,
+      "step": 4334
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2544488200693846e-05,
+      "loss": 0.3113,
+      "step": 4336
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2538591762663415e-05,
+      "loss": 0.2656,
+      "step": 4338
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2532694381066919e-05,
+      "loss": 0.3187,
+      "step": 4340
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.252679605809635e-05,
+      "loss": 0.2375,
+      "step": 4342
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2520896795944041e-05,
+      "loss": 0.2637,
+      "step": 4344
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2514996596802696e-05,
+      "loss": 0.2798,
+      "step": 4346
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2509095462865337e-05,
+      "loss": 0.3261,
+      "step": 4348
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2503193396325359e-05,
+      "loss": 0.2693,
+      "step": 4350
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2497290399376488e-05,
+      "loss": 0.2601,
+      "step": 4352
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2491386474212807e-05,
+      "loss": 0.2706,
+      "step": 4354
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2485481623028734e-05,
+      "loss": 0.2653,
+      "step": 4356
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2479575848019043e-05,
+      "loss": 0.2677,
+      "step": 4358
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2473669151378836e-05,
+      "loss": 0.2446,
+      "step": 4360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2467761535303568e-05,
+      "loss": 0.3198,
+      "step": 4362
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.246185300198904e-05,
+      "loss": 0.2729,
+      "step": 4364
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2455943553631378e-05,
+      "loss": 0.2422,
+      "step": 4366
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2450033192427063e-05,
+      "loss": 0.2687,
+      "step": 4368
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2444121920572908e-05,
+      "loss": 0.3218,
+      "step": 4370
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2438209740266069e-05,
+      "loss": 0.2633,
+      "step": 4372
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2432296653704037e-05,
+      "loss": 0.2588,
+      "step": 4374
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2426382663084634e-05,
+      "loss": 0.2574,
+      "step": 4376
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.242046777060603e-05,
+      "loss": 0.268,
+      "step": 4378
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2414551978466727e-05,
+      "loss": 0.2583,
+      "step": 4380
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.240863528886555e-05,
+      "loss": 0.2839,
+      "step": 4382
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.240271770400167e-05,
+      "loss": 0.2551,
+      "step": 4384
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.239679922607459e-05,
+      "loss": 0.2792,
+      "step": 4386
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2390879857284134e-05,
+      "loss": 0.3267,
+      "step": 4388
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2384959599830475e-05,
+      "loss": 0.2988,
+      "step": 4390
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2379038455914097e-05,
+      "loss": 0.2989,
+      "step": 4392
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2373116427735827e-05,
+      "loss": 0.2884,
+      "step": 4394
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2367193517496819e-05,
+      "loss": 0.2963,
+      "step": 4396
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2361269727398549e-05,
+      "loss": 0.2952,
+      "step": 4398
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2355345059642819e-05,
+      "loss": 0.2833,
+      "step": 4400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.234941951643177e-05,
+      "loss": 0.2946,
+      "step": 4402
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2343493099967856e-05,
+      "loss": 0.2507,
+      "step": 4404
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2337565812453862e-05,
+      "loss": 0.2619,
+      "step": 4406
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2331637656092888e-05,
+      "loss": 0.2543,
+      "step": 4408
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2325708633088369e-05,
+      "loss": 0.3366,
+      "step": 4410
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2319778745644056e-05,
+      "loss": 0.3157,
+      "step": 4412
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2313847995964015e-05,
+      "loss": 0.2593,
+      "step": 4414
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2307916386252648e-05,
+      "loss": 0.2612,
+      "step": 4416
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2301983918714662e-05,
+      "loss": 0.2939,
+      "step": 4418
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2296050595555092e-05,
+      "loss": 0.2456,
+      "step": 4420
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2290116418979284e-05,
+      "loss": 0.2586,
+      "step": 4422
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2284181391192907e-05,
+      "loss": 0.2847,
+      "step": 4424
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2278245514401945e-05,
+      "loss": 0.3056,
+      "step": 4426
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2272308790812697e-05,
+      "loss": 0.2976,
+      "step": 4428
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2266371222631772e-05,
+      "loss": 0.2942,
+      "step": 4430
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2260432812066104e-05,
+      "loss": 0.2356,
+      "step": 4432
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.225449356132293e-05,
+      "loss": 0.2989,
+      "step": 4434
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2248553472609803e-05,
+      "loss": 0.2499,
+      "step": 4436
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2242612548134586e-05,
+      "loss": 0.2717,
+      "step": 4438
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2236670790105456e-05,
+      "loss": 0.3012,
+      "step": 4440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2230728200730899e-05,
+      "loss": 0.2708,
+      "step": 4442
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2224784782219705e-05,
+      "loss": 0.2708,
+      "step": 4444
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.221884053678098e-05,
+      "loss": 0.2703,
+      "step": 4446
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2212895466624128e-05,
+      "loss": 0.3206,
+      "step": 4448
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.220694957395887e-05,
+      "loss": 0.2792,
+      "step": 4450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2201002860995226e-05,
+      "loss": 0.309,
+      "step": 4452
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2195055329943524e-05,
+      "loss": 0.2914,
+      "step": 4454
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2189106983014391e-05,
+      "loss": 0.267,
+      "step": 4456
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.218315782241876e-05,
+      "loss": 0.2631,
+      "step": 4458
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2177207850367872e-05,
+      "loss": 0.2833,
+      "step": 4460
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2171257069073264e-05,
+      "loss": 0.2376,
+      "step": 4462
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2165305480746772e-05,
+      "loss": 0.2987,
+      "step": 4464
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2159353087600534e-05,
+      "loss": 0.279,
+      "step": 4466
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2153399891846992e-05,
+      "loss": 0.2722,
+      "step": 4468
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2147445895698878e-05,
+      "loss": 0.3197,
+      "step": 4470
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2141491101369228e-05,
+      "loss": 0.2666,
+      "step": 4472
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2135535511071371e-05,
+      "loss": 0.3251,
+      "step": 4474
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2129579127018934e-05,
+      "loss": 0.2612,
+      "step": 4476
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2123621951425834e-05,
+      "loss": 0.2566,
+      "step": 4478
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2117663986506292e-05,
+      "loss": 0.3155,
+      "step": 4480
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2111705234474815e-05,
+      "loss": 0.2807,
+      "step": 4482
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2105745697546206e-05,
+      "loss": 0.3561,
+      "step": 4484
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2099785377935552e-05,
+      "loss": 0.3551,
+      "step": 4486
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.209382427785824e-05,
+      "loss": 0.2858,
+      "step": 4488
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2087862399529948e-05,
+      "loss": 0.2802,
+      "step": 4490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2081899745166631e-05,
+      "loss": 0.279,
+      "step": 4492
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2075936316984551e-05,
+      "loss": 0.2512,
+      "step": 4494
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2069972117200237e-05,
+      "loss": 0.2342,
+      "step": 4496
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2064007148030523e-05,
+      "loss": 0.2223,
+      "step": 4498
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2058041411692518e-05,
+      "loss": 0.2656,
+      "step": 4500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2052074910403619e-05,
+      "loss": 0.2624,
+      "step": 4502
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2046107646381507e-05,
+      "loss": 0.3105,
+      "step": 4504
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.204013962184415e-05,
+      "loss": 0.2979,
+      "step": 4506
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.203417083900979e-05,
+      "loss": 0.2676,
+      "step": 4508
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2028201300096964e-05,
+      "loss": 0.2613,
+      "step": 4510
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2022231007324477e-05,
+      "loss": 0.282,
+      "step": 4512
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.201625996291142e-05,
+      "loss": 0.2757,
+      "step": 4514
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2010288169077171e-05,
+      "loss": 0.2736,
+      "step": 4516
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2004315628041368e-05,
+      "loss": 0.2711,
+      "step": 4518
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.199834234202394e-05,
+      "loss": 0.2527,
+      "step": 4520
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1992368313245094e-05,
+      "loss": 0.2749,
+      "step": 4522
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1986393543925312e-05,
+      "loss": 0.2144,
+      "step": 4524
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.198041803628534e-05,
+      "loss": 0.2903,
+      "step": 4526
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1974441792546212e-05,
+      "loss": 0.3277,
+      "step": 4528
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1968464814929232e-05,
+      "loss": 0.3081,
+      "step": 4530
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1962487105655975e-05,
+      "loss": 0.3328,
+      "step": 4532
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1956508666948286e-05,
+      "loss": 0.2663,
+      "step": 4534
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1950529501028284e-05,
+      "loss": 0.2785,
+      "step": 4536
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1944549610118362e-05,
+      "loss": 0.2682,
+      "step": 4538
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.193856899644117e-05,
+      "loss": 0.2944,
+      "step": 4540
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1932587662219643e-05,
+      "loss": 0.2626,
+      "step": 4542
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.192660560967697e-05,
+      "loss": 0.3177,
+      "step": 4544
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1920622841036622e-05,
+      "loss": 0.2108,
+      "step": 4546
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1914639358522315e-05,
+      "loss": 0.254,
+      "step": 4548
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1908655164358053e-05,
+      "loss": 0.2763,
+      "step": 4550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1902670260768083e-05,
+      "loss": 0.2879,
+      "step": 4552
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1896684649976938e-05,
+      "loss": 0.3552,
+      "step": 4554
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1890698334209391e-05,
+      "loss": 0.2865,
+      "step": 4556
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1884711315690497e-05,
+      "loss": 0.2985,
+      "step": 4558
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1878723596645561e-05,
+      "loss": 0.2773,
+      "step": 4560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.187273517930015e-05,
+      "loss": 0.2391,
+      "step": 4562
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1866746065880093e-05,
+      "loss": 0.2863,
+      "step": 4564
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1860756258611473e-05,
+      "loss": 0.2903,
+      "step": 4566
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1854765759720637e-05,
+      "loss": 0.2446,
+      "step": 4568
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1848774571434186e-05,
+      "loss": 0.2778,
+      "step": 4570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1842782695978979e-05,
+      "loss": 0.3315,
+      "step": 4572
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1836790135582124e-05,
+      "loss": 0.2348,
+      "step": 4574
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1830796892470992e-05,
+      "loss": 0.2788,
+      "step": 4576
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1824802968873203e-05,
+      "loss": 0.3188,
+      "step": 4578
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1818808367016635e-05,
+      "loss": 0.2309,
+      "step": 4580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1812813089129406e-05,
+      "loss": 0.2342,
+      "step": 4582
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1806817137439901e-05,
+      "loss": 0.2716,
+      "step": 4584
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1800820514176748e-05,
+      "loss": 0.2684,
+      "step": 4586
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1794823221568819e-05,
+      "loss": 0.2725,
+      "step": 4588
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1788825261845243e-05,
+      "loss": 0.3006,
+      "step": 4590
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1782826637235395e-05,
+      "loss": 0.2447,
+      "step": 4592
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1776827349968895e-05,
+      "loss": 0.2889,
+      "step": 4594
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1770827402275613e-05,
+      "loss": 0.2388,
+      "step": 4596
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1764826796385662e-05,
+      "loss": 0.2827,
+      "step": 4598
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1758825534529394e-05,
+      "loss": 0.2634,
+      "step": 4600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1752823618937422e-05,
+      "loss": 0.3043,
+      "step": 4602
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.174682105184058e-05,
+      "loss": 0.2534,
+      "step": 4604
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1740817835469963e-05,
+      "loss": 0.2607,
+      "step": 4606
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1734813972056891e-05,
+      "loss": 0.3103,
+      "step": 4608
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1728809463832938e-05,
+      "loss": 0.2617,
+      "step": 4610
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1722804313029913e-05,
+      "loss": 0.3074,
+      "step": 4612
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1716798521879857e-05,
+      "loss": 0.3007,
+      "step": 4614
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1710792092615061e-05,
+      "loss": 0.3103,
+      "step": 4616
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.170478502746805e-05,
+      "loss": 0.2153,
+      "step": 4618
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1698777328671573e-05,
+      "loss": 0.2843,
+      "step": 4620
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1692768998458632e-05,
+      "loss": 0.2748,
+      "step": 4622
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1686760039062453e-05,
+      "loss": 0.2843,
+      "step": 4624
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1680750452716498e-05,
+      "loss": 0.2973,
+      "step": 4626
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.167474024165447e-05,
+      "loss": 0.2618,
+      "step": 4628
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1668729408110284e-05,
+      "loss": 0.2388,
+      "step": 4630
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.166271795431811e-05,
+      "loss": 0.2346,
+      "step": 4632
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1656705882512335e-05,
+      "loss": 0.3074,
+      "step": 4634
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1650693194927576e-05,
+      "loss": 0.2843,
+      "step": 4636
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1644679893798685e-05,
+      "loss": 0.2677,
+      "step": 4638
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1638665981360735e-05,
+      "loss": 0.2816,
+      "step": 4640
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1632651459849034e-05,
+      "loss": 0.2434,
+      "step": 4642
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1626636331499109e-05,
+      "loss": 0.215,
+      "step": 4644
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1620620598546716e-05,
+      "loss": 0.292,
+      "step": 4646
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1614604263227836e-05,
+      "loss": 0.2455,
+      "step": 4648
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1608587327778676e-05,
+      "loss": 0.2813,
+      "step": 4650
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1602569794435657e-05,
+      "loss": 0.5178,
+      "step": 4652
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1596551665435434e-05,
+      "loss": 0.2499,
+      "step": 4654
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1590532943014876e-05,
+      "loss": 0.2602,
+      "step": 4656
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1584513629411074e-05,
+      "loss": 0.272,
+      "step": 4658
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1578493726861344e-05,
+      "loss": 0.2929,
+      "step": 4660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1572473237603209e-05,
+      "loss": 0.2924,
+      "step": 4662
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.156645216387442e-05,
+      "loss": 0.2731,
+      "step": 4664
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1560430507912945e-05,
+      "loss": 0.2806,
+      "step": 4666
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1554408271956964e-05,
+      "loss": 0.2784,
+      "step": 4668
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1548385458244875e-05,
+      "loss": 0.2841,
+      "step": 4670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.154236206901529e-05,
+      "loss": 0.2104,
+      "step": 4672
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1536338106507033e-05,
+      "loss": 0.2625,
+      "step": 4674
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1530313572959148e-05,
+      "loss": 0.2787,
+      "step": 4676
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1524288470610883e-05,
+      "loss": 0.2465,
+      "step": 4678
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1518262801701701e-05,
+      "loss": 0.2883,
+      "step": 4680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1512236568471278e-05,
+      "loss": 0.2871,
+      "step": 4682
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1506209773159492e-05,
+      "loss": 0.2784,
+      "step": 4684
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.150018241800644e-05,
+      "loss": 0.2483,
+      "step": 4686
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1494154505252422e-05,
+      "loss": 0.2957,
+      "step": 4688
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1488126037137945e-05,
+      "loss": 0.2654,
+      "step": 4690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1482097015903721e-05,
+      "loss": 0.2659,
+      "step": 4692
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1476067443790675e-05,
+      "loss": 0.2449,
+      "step": 4694
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.147003732303992e-05,
+      "loss": 0.2914,
+      "step": 4696
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1464006655892797e-05,
+      "loss": 0.3081,
+      "step": 4698
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.145797544459083e-05,
+      "loss": 0.3069,
+      "step": 4700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1451943691375755e-05,
+      "loss": 0.2961,
+      "step": 4702
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1445911398489506e-05,
+      "loss": 0.2874,
+      "step": 4704
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1439878568174219e-05,
+      "loss": 0.2078,
+      "step": 4706
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1433845202672226e-05,
+      "loss": 0.3053,
+      "step": 4708
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1427811304226068e-05,
+      "loss": 0.2977,
+      "step": 4710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1421776875078471e-05,
+      "loss": 0.2882,
+      "step": 4712
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.141574191747237e-05,
+      "loss": 0.2621,
+      "step": 4714
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1409706433650889e-05,
+      "loss": 0.2529,
+      "step": 4716
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1403670425857344e-05,
+      "loss": 0.2443,
+      "step": 4718
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.139763389633526e-05,
+      "loss": 0.2775,
+      "step": 4720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1391596847328342e-05,
+      "loss": 0.2521,
+      "step": 4722
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1385559281080497e-05,
+      "loss": 0.2564,
+      "step": 4724
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1379521199835817e-05,
+      "loss": 0.2575,
+      "step": 4726
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1373482605838589e-05,
+      "loss": 0.2778,
+      "step": 4728
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1367443501333292e-05,
+      "loss": 0.2637,
+      "step": 4730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1361403888564596e-05,
+      "loss": 0.2878,
+      "step": 4732
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1355363769777354e-05,
+      "loss": 0.2449,
+      "step": 4734
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1349323147216611e-05,
+      "loss": 0.2691,
+      "step": 4736
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.13432820231276e-05,
+      "loss": 0.2789,
+      "step": 4738
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1337240399755736e-05,
+      "loss": 0.3356,
+      "step": 4740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1331198279346625e-05,
+      "loss": 0.2616,
+      "step": 4742
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.132515566414605e-05,
+      "loss": 0.2806,
+      "step": 4744
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1319112556399994e-05,
+      "loss": 0.2757,
+      "step": 4746
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1313068958354606e-05,
+      "loss": 0.3369,
+      "step": 4748
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1307024872256222e-05,
+      "loss": 0.265,
+      "step": 4750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.130098030035136e-05,
+      "loss": 0.2611,
+      "step": 4752
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1294935244886727e-05,
+      "loss": 0.2788,
+      "step": 4754
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1288889708109194e-05,
+      "loss": 0.2457,
+      "step": 4756
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1282843692265826e-05,
+      "loss": 0.2923,
+      "step": 4758
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1276797199603853e-05,
+      "loss": 0.3472,
+      "step": 4760
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1270750232370693e-05,
+      "loss": 0.2639,
+      "step": 4762
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1264702792813936e-05,
+      "loss": 0.2735,
+      "step": 4764
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1258654883181344e-05,
+      "loss": 0.218,
+      "step": 4766
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1252606505720856e-05,
+      "loss": 0.2219,
+      "step": 4768
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1246557662680591e-05,
+      "loss": 0.2447,
+      "step": 4770
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1240508356308836e-05,
+      "loss": 0.2559,
+      "step": 4772
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1234458588854045e-05,
+      "loss": 0.3216,
+      "step": 4774
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1228408362564853e-05,
+      "loss": 0.2358,
+      "step": 4776
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1222357679690059e-05,
+      "loss": 0.2435,
+      "step": 4778
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1216306542478639e-05,
+      "loss": 0.3059,
+      "step": 4780
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1210254953179724e-05,
+      "loss": 0.3054,
+      "step": 4782
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1204202914042627e-05,
+      "loss": 0.3302,
+      "step": 4784
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1198150427316828e-05,
+      "loss": 0.2153,
+      "step": 4786
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1192097495251957e-05,
+      "loss": 0.236,
+      "step": 4788
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.118604412009783e-05,
+      "loss": 0.1922,
+      "step": 4790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.117999030410441e-05,
+      "loss": 0.2382,
+      "step": 4792
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1173936049521846e-05,
+      "loss": 0.2785,
+      "step": 4794
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1167881358600422e-05,
+      "loss": 0.2802,
+      "step": 4796
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1161826233590611e-05,
+      "loss": 0.2934,
+      "step": 4798
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1155770676743023e-05,
+      "loss": 0.2704,
+      "step": 4800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1149714690308451e-05,
+      "loss": 0.2753,
+      "step": 4802
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1143658276537829e-05,
+      "loss": 0.2014,
+      "step": 4804
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1137601437682265e-05,
+      "loss": 0.2184,
+      "step": 4806
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1131544175993015e-05,
+      "loss": 0.2556,
+      "step": 4808
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1125486493721494e-05,
+      "loss": 0.251,
+      "step": 4810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1119428393119276e-05,
+      "loss": 0.2665,
+      "step": 4812
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1113369876438087e-05,
+      "loss": 0.2161,
+      "step": 4814
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.110731094592981e-05,
+      "loss": 0.2772,
+      "step": 4816
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1101251603846486e-05,
+      "loss": 0.2514,
+      "step": 4818
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1095191852440301e-05,
+      "loss": 0.2909,
+      "step": 4820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1089131693963592e-05,
+      "loss": 0.2802,
+      "step": 4822
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.108307113066886e-05,
+      "loss": 0.2572,
+      "step": 4824
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1077010164808745e-05,
+      "loss": 0.2548,
+      "step": 4826
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1070948798636042e-05,
+      "loss": 0.2938,
+      "step": 4828
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1064887034403686e-05,
+      "loss": 0.2776,
+      "step": 4830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1058824874364773e-05,
+      "loss": 0.262,
+      "step": 4832
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.105276232077254e-05,
+      "loss": 0.2786,
+      "step": 4834
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1046699375880368e-05,
+      "loss": 0.2569,
+      "step": 4836
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1040636041941786e-05,
+      "loss": 0.2604,
+      "step": 4838
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1034572321210463e-05,
+      "loss": 0.2636,
+      "step": 4840
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1028508215940226e-05,
+      "loss": 0.2712,
+      "step": 4842
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1022443728385025e-05,
+      "loss": 0.2593,
+      "step": 4844
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1016378860798969e-05,
+      "loss": 0.2176,
+      "step": 4846
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1010313615436292e-05,
+      "loss": 0.2276,
+      "step": 4848
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.100424799455139e-05,
+      "loss": 0.3456,
+      "step": 4850
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0998182000398776e-05,
+      "loss": 0.2667,
+      "step": 4852
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.099211563523312e-05,
+      "loss": 0.3515,
+      "step": 4854
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0986048901309213e-05,
+      "loss": 0.2482,
+      "step": 4856
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0979981800882001e-05,
+      "loss": 0.3036,
+      "step": 4858
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.097391433620655e-05,
+      "loss": 0.2848,
+      "step": 4860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0967846509538071e-05,
+      "loss": 0.3159,
+      "step": 4862
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.096177832313191e-05,
+      "loss": 0.2688,
+      "step": 4864
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0955709779243536e-05,
+      "loss": 0.2655,
+      "step": 4866
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.094964088012857e-05,
+      "loss": 0.2427,
+      "step": 4868
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0943571628042742e-05,
+      "loss": 0.3016,
+      "step": 4870
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0937502025241933e-05,
+      "loss": 0.323,
+      "step": 4872
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0931432073982142e-05,
+      "loss": 0.3223,
+      "step": 4874
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0925361776519505e-05,
+      "loss": 0.287,
+      "step": 4876
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.091929113511028e-05,
+      "loss": 0.276,
+      "step": 4878
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.091322015201086e-05,
+      "loss": 0.2928,
+      "step": 4880
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.090714882947776e-05,
+      "loss": 0.2521,
+      "step": 4882
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.090107716976762e-05,
+      "loss": 0.2721,
+      "step": 4884
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.089500517513721e-05,
+      "loss": 0.3578,
+      "step": 4886
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0888932847843418e-05,
+      "loss": 0.2941,
+      "step": 4888
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.088286019014327e-05,
+      "loss": 0.2422,
+      "step": 4890
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0876787204293895e-05,
+      "loss": 0.2594,
+      "step": 4892
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0870713892552559e-05,
+      "loss": 0.3286,
+      "step": 4894
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0864640257176637e-05,
+      "loss": 0.2499,
+      "step": 4896
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.085856630042364e-05,
+      "loss": 0.2804,
+      "step": 4898
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0852492024551185e-05,
+      "loss": 0.289,
+      "step": 4900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0846417431817014e-05,
+      "loss": 0.2451,
+      "step": 4902
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0840342524478981e-05,
+      "loss": 0.2754,
+      "step": 4904
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0834267304795066e-05,
+      "loss": 0.2584,
+      "step": 4906
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0828191775023356e-05,
+      "loss": 0.2575,
+      "step": 4908
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0822115937422058e-05,
+      "loss": 0.2941,
+      "step": 4910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0816039794249494e-05,
+      "loss": 0.2666,
+      "step": 4912
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0809963347764097e-05,
+      "loss": 0.2792,
+      "step": 4914
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0803886600224416e-05,
+      "loss": 0.2986,
+      "step": 4916
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0797809553889106e-05,
+      "loss": 0.2716,
+      "step": 4918
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.079173221101694e-05,
+      "loss": 0.3128,
+      "step": 4920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0785654573866797e-05,
+      "loss": 0.2403,
+      "step": 4922
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.077957664469767e-05,
+      "loss": 0.2484,
+      "step": 4924
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0773498425768652e-05,
+      "loss": 0.3091,
+      "step": 4926
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0767419919338949e-05,
+      "loss": 0.2876,
+      "step": 4928
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0761341127667879e-05,
+      "loss": 0.2398,
+      "step": 4930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0755262053014855e-05,
+      "loss": 0.2635,
+      "step": 4932
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0749182697639401e-05,
+      "loss": 0.2737,
+      "step": 4934
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0743103063801148e-05,
+      "loss": 0.2504,
+      "step": 4936
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.073702315375983e-05,
+      "loss": 0.2525,
+      "step": 4938
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0730942969775273e-05,
+      "loss": 0.2326,
+      "step": 4940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0724862514107424e-05,
+      "loss": 0.3504,
+      "step": 4942
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0718781789016306e-05,
+      "loss": 0.2504,
+      "step": 4944
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0712700796762072e-05,
+      "loss": 0.3109,
+      "step": 4946
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0706619539604948e-05,
+      "loss": 0.2499,
+      "step": 4948
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0700538019805273e-05,
+      "loss": 0.2854,
+      "step": 4950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0694456239623476e-05,
+      "loss": 0.255,
+      "step": 4952
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.068837420132009e-05,
+      "loss": 0.2827,
+      "step": 4954
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0682291907155742e-05,
+      "loss": 0.3006,
+      "step": 4956
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0676209359391149e-05,
+      "loss": 0.3063,
+      "step": 4958
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0670126560287127e-05,
+      "loss": 0.2621,
+      "step": 4960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0664043512104582e-05,
+      "loss": 0.2579,
+      "step": 4962
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.065796021710452e-05,
+      "loss": 0.2411,
+      "step": 4964
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0651876677548027e-05,
+      "loss": 0.2426,
+      "step": 4966
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.064579289569629e-05,
+      "loss": 0.2153,
+      "step": 4968
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0639708873810582e-05,
+      "loss": 0.2399,
+      "step": 4970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0633624614152267e-05,
+      "loss": 0.2837,
+      "step": 4972
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0627540118982793e-05,
+      "loss": 0.2801,
+      "step": 4974
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.06214553905637e-05,
+      "loss": 0.2254,
+      "step": 4976
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0615370431156612e-05,
+      "loss": 0.2679,
+      "step": 4978
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0609285243023247e-05,
+      "loss": 0.3156,
+      "step": 4980
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0603199828425393e-05,
+      "loss": 0.2717,
+      "step": 4982
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0597114189624931e-05,
+      "loss": 0.2932,
+      "step": 4984
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0591028328883827e-05,
+      "loss": 0.2751,
+      "step": 4986
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.058494224846413e-05,
+      "loss": 0.3174,
+      "step": 4988
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0578855950627962e-05,
+      "loss": 0.2683,
+      "step": 4990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0572769437637531e-05,
+      "loss": 0.2704,
+      "step": 4992
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0566682711755134e-05,
+      "loss": 0.2555,
+      "step": 4994
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.056059577524313e-05,
+      "loss": 0.2979,
+      "step": 4996
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0554508630363971e-05,
+      "loss": 0.3001,
+      "step": 4998
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0548421279380175e-05,
+      "loss": 0.2594,
+      "step": 5000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0542333724554347e-05,
+      "loss": 0.3283,
+      "step": 5002
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0536245968149159e-05,
+      "loss": 0.2479,
+      "step": 5004
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0530158012427367e-05,
+      "loss": 0.2787,
+      "step": 5006
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0524069859651788e-05,
+      "loss": 0.1967,
+      "step": 5008
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0517981512085325e-05,
+      "loss": 0.2684,
+      "step": 5010
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0511892971990949e-05,
+      "loss": 0.2719,
+      "step": 5012
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0505804241631701e-05,
+      "loss": 0.2525,
+      "step": 5014
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0499715323270692e-05,
+      "loss": 0.2552,
+      "step": 5016
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0493626219171105e-05,
+      "loss": 0.2725,
+      "step": 5018
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0487536931596196e-05,
+      "loss": 0.2971,
+      "step": 5020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0481447462809278e-05,
+      "loss": 0.2785,
+      "step": 5022
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0475357815073744e-05,
+      "loss": 0.2172,
+      "step": 5024
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0469267990653043e-05,
+      "loss": 0.2494,
+      "step": 5026
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.04631779918107e-05,
+      "loss": 0.2429,
+      "step": 5028
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0457087820810294e-05,
+      "loss": 0.2597,
+      "step": 5030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0450997479915477e-05,
+      "loss": 0.2738,
+      "step": 5032
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.044490697138996e-05,
+      "loss": 0.2155,
+      "step": 5034
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0438816297497512e-05,
+      "loss": 0.2743,
+      "step": 5036
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0432725460501973e-05,
+      "loss": 0.2979,
+      "step": 5038
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0426634462667233e-05,
+      "loss": 0.2902,
+      "step": 5040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0420543306257258e-05,
+      "loss": 0.3477,
+      "step": 5042
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0414451993536051e-05,
+      "loss": 0.3041,
+      "step": 5044
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0408360526767693e-05,
+      "loss": 0.2829,
+      "step": 5046
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0402268908216308e-05,
+      "loss": 0.2724,
+      "step": 5048
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0396177140146088e-05,
+      "loss": 0.261,
+      "step": 5050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0390085224821268e-05,
+      "loss": 0.2665,
+      "step": 5052
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0383993164506152e-05,
+      "loss": 0.2549,
+      "step": 5054
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0377900961465087e-05,
+      "loss": 0.2653,
+      "step": 5056
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0371808617962471e-05,
+      "loss": 0.2307,
+      "step": 5058
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.036571613626277e-05,
+      "loss": 0.2264,
+      "step": 5060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0359623518630482e-05,
+      "loss": 0.262,
+      "step": 5062
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0353530767330171e-05,
+      "loss": 0.2395,
+      "step": 5064
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.034743788462644e-05,
+      "loss": 0.2441,
+      "step": 5066
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0341344872783949e-05,
+      "loss": 0.3383,
+      "step": 5068
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0335251734067397e-05,
+      "loss": 0.2938,
+      "step": 5070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.032915847074154e-05,
+      "loss": 0.2907,
+      "step": 5072
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0323065085071173e-05,
+      "loss": 0.2592,
+      "step": 5074
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0316971579321142e-05,
+      "loss": 0.3315,
+      "step": 5076
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0310877955756331e-05,
+      "loss": 0.2713,
+      "step": 5078
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0304784216641673e-05,
+      "loss": 0.3252,
+      "step": 5080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0298690364242143e-05,
+      "loss": 0.2824,
+      "step": 5082
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0292596400822756e-05,
+      "loss": 0.2697,
+      "step": 5084
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0286502328648573e-05,
+      "loss": 0.2705,
+      "step": 5086
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0280408149984686e-05,
+      "loss": 0.2206,
+      "step": 5088
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0274313867096242e-05,
+      "loss": 0.2537,
+      "step": 5090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.026821948224841e-05,
+      "loss": 0.2354,
+      "step": 5092
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0262124997706407e-05,
+      "loss": 0.2888,
+      "step": 5094
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0256030415735483e-05,
+      "loss": 0.2484,
+      "step": 5096
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.024993573860093e-05,
+      "loss": 0.2856,
+      "step": 5098
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0243840968568065e-05,
+      "loss": 0.256,
+      "step": 5100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0237746107902251e-05,
+      "loss": 0.3381,
+      "step": 5102
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0231651158868876e-05,
+      "loss": 0.2467,
+      "step": 5104
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0225556123733367e-05,
+      "loss": 0.1866,
+      "step": 5106
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0219461004761178e-05,
+      "loss": 0.2478,
+      "step": 5108
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0213365804217794e-05,
+      "loss": 0.2194,
+      "step": 5110
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0207270524368737e-05,
+      "loss": 0.2744,
+      "step": 5112
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0201175167479554e-05,
+      "loss": 0.2514,
+      "step": 5114
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0195079735815818e-05,
+      "loss": 0.258,
+      "step": 5116
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0188984231643133e-05,
+      "loss": 0.2826,
+      "step": 5118
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.018288865722713e-05,
+      "loss": 0.2796,
+      "step": 5120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0176793014833466e-05,
+      "loss": 0.2476,
+      "step": 5122
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0170697306727826e-05,
+      "loss": 0.2894,
+      "step": 5124
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.016460153517591e-05,
+      "loss": 0.257,
+      "step": 5126
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0158505702443449e-05,
+      "loss": 0.2762,
+      "step": 5128
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0152409810796203e-05,
+      "loss": 0.2346,
+      "step": 5130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0146313862499935e-05,
+      "loss": 0.2681,
+      "step": 5132
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0140217859820451e-05,
+      "loss": 0.3059,
+      "step": 5134
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0134121805023558e-05,
+      "loss": 0.2777,
+      "step": 5136
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.01280257003751e-05,
+      "loss": 0.222,
+      "step": 5138
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.012192954814092e-05,
+      "loss": 0.2505,
+      "step": 5140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0115833350586904e-05,
+      "loss": 0.2534,
+      "step": 5142
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0109737109978923e-05,
+      "loss": 0.2376,
+      "step": 5144
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0103640828582895e-05,
+      "loss": 0.2638,
+      "step": 5146
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0097544508664732e-05,
+      "loss": 0.2465,
+      "step": 5148
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0091448152490373e-05,
+      "loss": 0.2344,
+      "step": 5150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0085351762325759e-05,
+      "loss": 0.2185,
+      "step": 5152
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0079255340436853e-05,
+      "loss": 0.2066,
+      "step": 5154
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.007315888908963e-05,
+      "loss": 0.2654,
+      "step": 5156
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0067062410550066e-05,
+      "loss": 0.2465,
+      "step": 5158
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0060965907084158e-05,
+      "loss": 0.2321,
+      "step": 5160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0054869380957904e-05,
+      "loss": 0.243,
+      "step": 5162
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0048772834437321e-05,
+      "loss": 0.2786,
+      "step": 5164
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.004267626978842e-05,
+      "loss": 0.262,
+      "step": 5166
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0036579689277229e-05,
+      "loss": 0.2151,
+      "step": 5168
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0030483095169776e-05,
+      "loss": 0.2559,
+      "step": 5170
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0024386489732104e-05,
+      "loss": 0.2849,
+      "step": 5172
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0018289875230242e-05,
+      "loss": 0.2702,
+      "step": 5174
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0012193253930242e-05,
+      "loss": 0.2782,
+      "step": 5176
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0006096628098147e-05,
+      "loss": 0.2734,
+      "step": 5178
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1e-05,
+      "loss": 0.3819,
+      "step": 5180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.993903371901856e-06,
+      "loss": 0.2531,
+      "step": 5182
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.987806746069762e-06,
+      "loss": 0.2364,
+      "step": 5184
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.98171012476976e-06,
+      "loss": 0.3075,
+      "step": 5186
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.975613510267901e-06,
+      "loss": 0.2383,
+      "step": 5188
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.969516904830224e-06,
+      "loss": 0.2865,
+      "step": 5190
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.963420310722773e-06,
+      "loss": 0.2724,
+      "step": 5192
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.957323730211583e-06,
+      "loss": 0.1925,
+      "step": 5194
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.951227165562682e-06,
+      "loss": 0.2707,
+      "step": 5196
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.945130619042098e-06,
+      "loss": 0.2458,
+      "step": 5198
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.939034092915846e-06,
+      "loss": 0.3262,
+      "step": 5200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.932937589449939e-06,
+      "loss": 0.2187,
+      "step": 5202
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.926841110910374e-06,
+      "loss": 0.2518,
+      "step": 5204
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.920744659563147e-06,
+      "loss": 0.3043,
+      "step": 5206
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.914648237674244e-06,
+      "loss": 0.2323,
+      "step": 5208
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.90855184750963e-06,
+      "loss": 0.2518,
+      "step": 5210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.90245549133527e-06,
+      "loss": 0.3419,
+      "step": 5212
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.896359171417109e-06,
+      "loss": 0.2628,
+      "step": 5214
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.89026289002108e-06,
+      "loss": 0.238,
+      "step": 5216
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.884166649413103e-06,
+      "loss": 0.2402,
+      "step": 5218
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.878070451859083e-06,
+      "loss": 0.2997,
+      "step": 5220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.871974299624905e-06,
+      "loss": 0.2834,
+      "step": 5222
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.865878194976444e-06,
+      "loss": 0.2725,
+      "step": 5224
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.859782140179552e-06,
+      "loss": 0.2392,
+      "step": 5226
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.853686137500068e-06,
+      "loss": 0.2434,
+      "step": 5228
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.847590189203802e-06,
+      "loss": 0.2357,
+      "step": 5230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.841494297556553e-06,
+      "loss": 0.3721,
+      "step": 5232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.835398464824094e-06,
+      "loss": 0.2825,
+      "step": 5234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.829302693272181e-06,
+      "loss": 0.2677,
+      "step": 5236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.823206985166534e-06,
+      "loss": 0.2707,
+      "step": 5238
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.81711134277287e-06,
+      "loss": 0.3084,
+      "step": 5240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.81101576835687e-06,
+      "loss": 0.2322,
+      "step": 5242
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.804920264184185e-06,
+      "loss": 0.2608,
+      "step": 5244
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.79882483252045e-06,
+      "loss": 0.214,
+      "step": 5246
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.792729475631265e-06,
+      "loss": 0.2754,
+      "step": 5248
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.78663419578221e-06,
+      "loss": 0.2838,
+      "step": 5250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.780538995238827e-06,
+      "loss": 0.232,
+      "step": 5252
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.774443876266634e-06,
+      "loss": 0.2563,
+      "step": 5254
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.768348841131126e-06,
+      "loss": 0.2668,
+      "step": 5256
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.76225389209775e-06,
+      "loss": 0.2755,
+      "step": 5258
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.756159031431937e-06,
+      "loss": 0.2632,
+      "step": 5260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.750064261399074e-06,
+      "loss": 0.2678,
+      "step": 5262
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.743969584264522e-06,
+      "loss": 0.2579,
+      "step": 5264
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.737875002293598e-06,
+      "loss": 0.3179,
+      "step": 5266
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.731780517751597e-06,
+      "loss": 0.28,
+      "step": 5268
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.72568613290376e-06,
+      "loss": 0.2083,
+      "step": 5270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.719591850015314e-06,
+      "loss": 0.2606,
+      "step": 5272
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.713497671351429e-06,
+      "loss": 0.3224,
+      "step": 5274
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.707403599177247e-06,
+      "loss": 0.2451,
+      "step": 5276
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.701309635757859e-06,
+      "loss": 0.2499,
+      "step": 5278
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.695215783358332e-06,
+      "loss": 0.2554,
+      "step": 5280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.689122044243674e-06,
+      "loss": 0.2395,
+      "step": 5282
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.683028420678864e-06,
+      "loss": 0.2569,
+      "step": 5284
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.676934914928828e-06,
+      "loss": 0.2448,
+      "step": 5286
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.670841529258462e-06,
+      "loss": 0.2521,
+      "step": 5288
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.664748265932604e-06,
+      "loss": 0.2818,
+      "step": 5290
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.658655127216054e-06,
+      "loss": 0.2941,
+      "step": 5292
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.652562115373564e-06,
+      "loss": 0.2843,
+      "step": 5294
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.646469232669832e-06,
+      "loss": 0.2461,
+      "step": 5296
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.640376481369523e-06,
+      "loss": 0.2508,
+      "step": 5298
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.634283863737235e-06,
+      "loss": 0.2755,
+      "step": 5300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.62819138203753e-06,
+      "loss": 0.3009,
+      "step": 5302
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.622099038534918e-06,
+      "loss": 0.2558,
+      "step": 5304
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.61600683549385e-06,
+      "loss": 0.2424,
+      "step": 5306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.609914775178733e-06,
+      "loss": 0.2951,
+      "step": 5308
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.603822859853915e-06,
+      "loss": 0.2793,
+      "step": 5310
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.597731091783695e-06,
+      "loss": 0.3261,
+      "step": 5312
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.59163947323231e-06,
+      "loss": 0.3159,
+      "step": 5314
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.585548006463954e-06,
+      "loss": 0.2126,
+      "step": 5316
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.579456693742744e-06,
+      "loss": 0.241,
+      "step": 5318
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.573365537332767e-06,
+      "loss": 0.2197,
+      "step": 5320
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.56727453949803e-06,
+      "loss": 0.2716,
+      "step": 5322
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.561183702502492e-06,
+      "loss": 0.282,
+      "step": 5324
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.555093028610044e-06,
+      "loss": 0.2867,
+      "step": 5326
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.549002520084526e-06,
+      "loss": 0.2737,
+      "step": 5328
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.542912179189709e-06,
+      "loss": 0.269,
+      "step": 5330
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.536822008189304e-06,
+      "loss": 0.2243,
+      "step": 5332
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.530732009346957e-06,
+      "loss": 0.2199,
+      "step": 5334
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.524642184926258e-06,
+      "loss": 0.2322,
+      "step": 5336
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.518552537190724e-06,
+      "loss": 0.2672,
+      "step": 5338
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.512463068403807e-06,
+      "loss": 0.2906,
+      "step": 5340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.506373780828898e-06,
+      "loss": 0.2801,
+      "step": 5342
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.500284676729312e-06,
+      "loss": 0.226,
+      "step": 5344
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.494195758368306e-06,
+      "loss": 0.2478,
+      "step": 5346
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.488107028009054e-06,
+      "loss": 0.2643,
+      "step": 5348
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.482018487914675e-06,
+      "loss": 0.2391,
+      "step": 5350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.475930140348215e-06,
+      "loss": 0.2747,
+      "step": 5352
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.469841987572636e-06,
+      "loss": 0.2738,
+      "step": 5354
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.463754031850844e-06,
+      "loss": 0.2784,
+      "step": 5356
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.457666275445654e-06,
+      "loss": 0.2673,
+      "step": 5358
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.451578720619828e-06,
+      "loss": 0.2462,
+      "step": 5360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.445491369636034e-06,
+      "loss": 0.2678,
+      "step": 5362
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.439404224756871e-06,
+      "loss": 0.239,
+      "step": 5364
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.433317288244866e-06,
+      "loss": 0.2686,
+      "step": 5366
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.427230562362469e-06,
+      "loss": 0.2598,
+      "step": 5368
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.42114404937204e-06,
+      "loss": 0.2845,
+      "step": 5370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.415057751535874e-06,
+      "loss": 0.2619,
+      "step": 5372
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.408971671116174e-06,
+      "loss": 0.2701,
+      "step": 5374
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.402885810375072e-06,
+      "loss": 0.2421,
+      "step": 5376
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.396800171574612e-06,
+      "loss": 0.2441,
+      "step": 5378
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.390714756976758e-06,
+      "loss": 0.2439,
+      "step": 5380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.384629568843388e-06,
+      "loss": 0.2543,
+      "step": 5382
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.378544609436301e-06,
+      "loss": 0.3429,
+      "step": 5384
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.37245988101721e-06,
+      "loss": 0.2362,
+      "step": 5386
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.366375385847737e-06,
+      "loss": 0.2184,
+      "step": 5388
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.360291126189421e-06,
+      "loss": 0.2735,
+      "step": 5390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.354207104303713e-06,
+      "loss": 0.2466,
+      "step": 5392
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.34812332245198e-06,
+      "loss": 0.2668,
+      "step": 5394
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.342039782895485e-06,
+      "loss": 0.2211,
+      "step": 5396
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.335956487895419e-06,
+      "loss": 0.293,
+      "step": 5398
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.329873439712877e-06,
+      "loss": 0.2785,
+      "step": 5400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.323790640608853e-06,
+      "loss": 0.2408,
+      "step": 5402
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.317708092844261e-06,
+      "loss": 0.2737,
+      "step": 5404
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.311625798679911e-06,
+      "loss": 0.2446,
+      "step": 5406
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.305543760376525e-06,
+      "loss": 0.3046,
+      "step": 5408
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.299461980194732e-06,
+      "loss": 0.2286,
+      "step": 5410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.293380460395057e-06,
+      "loss": 0.3168,
+      "step": 5412
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.28729920323793e-06,
+      "loss": 0.2313,
+      "step": 5414
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.281218210983694e-06,
+      "loss": 0.281,
+      "step": 5416
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.27513748589258e-06,
+      "loss": 0.252,
+      "step": 5418
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.269057030224729e-06,
+      "loss": 0.2603,
+      "step": 5420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.262976846240174e-06,
+      "loss": 0.3376,
+      "step": 5422
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.256896936198854e-06,
+      "loss": 0.2713,
+      "step": 5424
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.250817302360602e-06,
+      "loss": 0.2708,
+      "step": 5426
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.24473794698515e-06,
+      "loss": 0.2946,
+      "step": 5428
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.238658872332124e-06,
+      "loss": 0.2832,
+      "step": 5430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.23258008066105e-06,
+      "loss": 0.2694,
+      "step": 5432
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.226501574231351e-06,
+      "loss": 0.2252,
+      "step": 5434
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.220423355302332e-06,
+      "loss": 0.3397,
+      "step": 5436
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.214345426133205e-06,
+      "loss": 0.3,
+      "step": 5438
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.208267788983063e-06,
+      "loss": 0.2464,
+      "step": 5440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.202190446110899e-06,
+      "loss": 0.2437,
+      "step": 5442
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.196113399775589e-06,
+      "loss": 0.2963,
+      "step": 5444
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.190036652235903e-06,
+      "loss": 0.2922,
+      "step": 5446
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.18396020575051e-06,
+      "loss": 0.2181,
+      "step": 5448
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.177884062577946e-06,
+      "loss": 0.2237,
+      "step": 5450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.171808224976648e-06,
+      "loss": 0.2468,
+      "step": 5452
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.165732695204938e-06,
+      "loss": 0.1992,
+      "step": 5454
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.15965747552102e-06,
+      "loss": 0.2871,
+      "step": 5456
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.153582568182992e-06,
+      "loss": 0.2806,
+      "step": 5458
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.147507975448818e-06,
+      "loss": 0.215,
+      "step": 5460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.14143369957636e-06,
+      "loss": 0.2342,
+      "step": 5462
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.135359742823365e-06,
+      "loss": 0.2482,
+      "step": 5464
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.129286107447445e-06,
+      "loss": 0.2741,
+      "step": 5466
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.123212795706109e-06,
+      "loss": 0.2653,
+      "step": 5468
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.117139809856733e-06,
+      "loss": 0.3157,
+      "step": 5470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.111067152156584e-06,
+      "loss": 0.3292,
+      "step": 5472
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.104994824862795e-06,
+      "loss": 0.2953,
+      "step": 5474
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.098922830232386e-06,
+      "loss": 0.2669,
+      "step": 5476
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.092851170522244e-06,
+      "loss": 0.2586,
+      "step": 5478
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.08677984798914e-06,
+      "loss": 0.2962,
+      "step": 5480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.080708864889723e-06,
+      "loss": 0.2397,
+      "step": 5482
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.074638223480497e-06,
+      "loss": 0.2378,
+      "step": 5484
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.068567926017861e-06,
+      "loss": 0.2998,
+      "step": 5486
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.06249797475807e-06,
+      "loss": 0.2867,
+      "step": 5488
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.056428371957263e-06,
+      "loss": 0.2307,
+      "step": 5490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.050359119871437e-06,
+      "loss": 0.2359,
+      "step": 5492
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.044290220756464e-06,
+      "loss": 0.2883,
+      "step": 5494
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.038221676868092e-06,
+      "loss": 0.2749,
+      "step": 5496
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.03215349046193e-06,
+      "loss": 0.2484,
+      "step": 5498
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.026085663793453e-06,
+      "loss": 0.2298,
+      "step": 5500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.020018199118004e-06,
+      "loss": 0.2364,
+      "step": 5502
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.013951098690788e-06,
+      "loss": 0.2743,
+      "step": 5504
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.007884364766885e-06,
+      "loss": 0.2385,
+      "step": 5506
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.001817999601225e-06,
+      "loss": 0.2794,
+      "step": 5508
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.99575200544861e-06,
+      "loss": 0.2784,
+      "step": 5510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.989686384563708e-06,
+      "loss": 0.2409,
+      "step": 5512
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.983621139201034e-06,
+      "loss": 0.2338,
+      "step": 5514
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.977556271614977e-06,
+      "loss": 0.256,
+      "step": 5516
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.971491784059779e-06,
+      "loss": 0.2865,
+      "step": 5518
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.965427678789538e-06,
+      "loss": 0.243,
+      "step": 5520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.95936395805822e-06,
+      "loss": 0.3001,
+      "step": 5522
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.953300624119637e-06,
+      "loss": 0.2577,
+      "step": 5524
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.947237679227461e-06,
+      "loss": 0.2497,
+      "step": 5526
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.941175125635227e-06,
+      "loss": 0.2923,
+      "step": 5528
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.935112965596316e-06,
+      "loss": 0.2871,
+      "step": 5530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.929051201363962e-06,
+      "loss": 0.2709,
+      "step": 5532
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.922989835191258e-06,
+      "loss": 0.2649,
+      "step": 5534
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.916928869331141e-06,
+      "loss": 0.2342,
+      "step": 5536
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.910868306036411e-06,
+      "loss": 0.2375,
+      "step": 5538
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.904808147559704e-06,
+      "loss": 0.2209,
+      "step": 5540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.898748396153516e-06,
+      "loss": 0.2573,
+      "step": 5542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.89268905407019e-06,
+      "loss": 0.3008,
+      "step": 5544
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.886630123561916e-06,
+      "loss": 0.2711,
+      "step": 5546
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.880571606880726e-06,
+      "loss": 0.2491,
+      "step": 5548
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.874513506278511e-06,
+      "loss": 0.2456,
+      "step": 5550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.868455824006987e-06,
+      "loss": 0.2381,
+      "step": 5552
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.862398562317738e-06,
+      "loss": 0.2504,
+      "step": 5554
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.856341723462173e-06,
+      "loss": 0.2509,
+      "step": 5556
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.85028530969155e-06,
+      "loss": 0.307,
+      "step": 5558
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.844229323256977e-06,
+      "loss": 0.2433,
+      "step": 5560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.83817376640939e-06,
+      "loss": 0.2473,
+      "step": 5562
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.83211864139958e-06,
+      "loss": 0.2643,
+      "step": 5564
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.826063950478156e-06,
+      "loss": 0.2458,
+      "step": 5566
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.820009695895591e-06,
+      "loss": 0.2682,
+      "step": 5568
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.813955879902173e-06,
+      "loss": 0.2329,
+      "step": 5570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.807902504748048e-06,
+      "loss": 0.2908,
+      "step": 5572
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.801849572683177e-06,
+      "loss": 0.2819,
+      "step": 5574
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.795797085957373e-06,
+      "loss": 0.2298,
+      "step": 5576
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.789745046820278e-06,
+      "loss": 0.2391,
+      "step": 5578
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.783693457521365e-06,
+      "loss": 0.3112,
+      "step": 5580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.777642320309944e-06,
+      "loss": 0.2768,
+      "step": 5582
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.771591637435149e-06,
+      "loss": 0.2245,
+      "step": 5584
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.765541411145959e-06,
+      "loss": 0.2625,
+      "step": 5586
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.759491643691169e-06,
+      "loss": 0.3425,
+      "step": 5588
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.75344233731941e-06,
+      "loss": 0.2444,
+      "step": 5590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.747393494279144e-06,
+      "loss": 0.2296,
+      "step": 5592
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.741345116818661e-06,
+      "loss": 0.2707,
+      "step": 5594
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.735297207186067e-06,
+      "loss": 0.2315,
+      "step": 5596
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.729249767629309e-06,
+      "loss": 0.2589,
+      "step": 5598
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.723202800396149e-06,
+      "loss": 0.2648,
+      "step": 5600
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.717156307734178e-06,
+      "loss": 0.2557,
+      "step": 5602
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.711110291890809e-06,
+      "loss": 0.2612,
+      "step": 5604
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.705064755113275e-06,
+      "loss": 0.2809,
+      "step": 5606
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.699019699648641e-06,
+      "loss": 0.2568,
+      "step": 5608
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.692975127743781e-06,
+      "loss": 0.2594,
+      "step": 5610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.6869310416454e-06,
+      "loss": 0.2331,
+      "step": 5612
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.680887443600007e-06,
+      "loss": 0.3275,
+      "step": 5614
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.674844335853951e-06,
+      "loss": 0.2966,
+      "step": 5616
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.66880172065338e-06,
+      "loss": 0.2235,
+      "step": 5618
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.662759600244271e-06,
+      "loss": 0.2193,
+      "step": 5620
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.656717976872405e-06,
+      "loss": 0.2046,
+      "step": 5622
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.65067685278339e-06,
+      "loss": 0.2755,
+      "step": 5624
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.644636230222649e-06,
+      "loss": 0.2581,
+      "step": 5626
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.638596111435405e-06,
+      "loss": 0.293,
+      "step": 5628
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.63255649866671e-06,
+      "loss": 0.2684,
+      "step": 5630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.626517394161413e-06,
+      "loss": 0.2722,
+      "step": 5632
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.620478800164186e-06,
+      "loss": 0.2935,
+      "step": 5634
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.614440718919508e-06,
+      "loss": 0.2844,
+      "step": 5636
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.608403152671658e-06,
+      "loss": 0.2722,
+      "step": 5638
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.602366103664742e-06,
+      "loss": 0.2397,
+      "step": 5640
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.596329574142657e-06,
+      "loss": 0.3261,
+      "step": 5642
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.590293566349114e-06,
+      "loss": 0.2939,
+      "step": 5644
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.584258082527633e-06,
+      "loss": 0.2582,
+      "step": 5646
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.57822312492153e-06,
+      "loss": 0.2413,
+      "step": 5648
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.572188695773937e-06,
+      "loss": 0.2474,
+      "step": 5650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.566154797327777e-06,
+      "loss": 0.2501,
+      "step": 5652
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.560121431825783e-06,
+      "loss": 0.2439,
+      "step": 5654
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.554088601510497e-06,
+      "loss": 0.2179,
+      "step": 5656
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.548056308624247e-06,
+      "loss": 0.2511,
+      "step": 5658
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.542024555409173e-06,
+      "loss": 0.2693,
+      "step": 5660
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.535993344107204e-06,
+      "loss": 0.3089,
+      "step": 5662
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.529962676960082e-06,
+      "loss": 0.2206,
+      "step": 5664
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.52393255620933e-06,
+      "loss": 0.2619,
+      "step": 5666
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.517902984096282e-06,
+      "loss": 0.3381,
+      "step": 5668
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.511873962862057e-06,
+      "loss": 0.2499,
+      "step": 5670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.50584549474758e-06,
+      "loss": 0.2535,
+      "step": 5672
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.49981758199356e-06,
+      "loss": 0.2625,
+      "step": 5674
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.49379022684051e-06,
+      "loss": 0.2665,
+      "step": 5676
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.487763431528727e-06,
+      "loss": 0.2735,
+      "step": 5678
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.481737198298304e-06,
+      "loss": 0.2071,
+      "step": 5680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.475711529389122e-06,
+      "loss": 0.2224,
+      "step": 5682
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.469686427040857e-06,
+      "loss": 0.2313,
+      "step": 5684
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.463661893492969e-06,
+      "loss": 0.3122,
+      "step": 5686
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.457637930984712e-06,
+      "loss": 0.275,
+      "step": 5688
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.451614541755128e-06,
+      "loss": 0.2076,
+      "step": 5690
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.445591728043037e-06,
+      "loss": 0.304,
+      "step": 5692
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.439569492087058e-06,
+      "loss": 0.2496,
+      "step": 5694
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.433547836125582e-06,
+      "loss": 0.2636,
+      "step": 5696
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.427526762396796e-06,
+      "loss": 0.2666,
+      "step": 5698
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.421506273138663e-06,
+      "loss": 0.2604,
+      "step": 5700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.415486370588926e-06,
+      "loss": 0.2606,
+      "step": 5702
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.409467056985127e-06,
+      "loss": 0.2525,
+      "step": 5704
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.403448334564568e-06,
+      "loss": 0.2227,
+      "step": 5706
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.397430205564346e-06,
+      "loss": 0.2833,
+      "step": 5708
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.391412672221329e-06,
+      "loss": 0.2397,
+      "step": 5710
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.385395736772169e-06,
+      "loss": 0.2693,
+      "step": 5712
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.379379401453289e-06,
+      "loss": 0.23,
+      "step": 5714
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.373363668500896e-06,
+      "loss": 0.2737,
+      "step": 5716
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.367348540150967e-06,
+      "loss": 0.2648,
+      "step": 5718
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.361334018639267e-06,
+      "loss": 0.2706,
+      "step": 5720
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.355320106201317e-06,
+      "loss": 0.2326,
+      "step": 5722
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.349306805072427e-06,
+      "loss": 0.2634,
+      "step": 5724
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.343294117487668e-06,
+      "loss": 0.2168,
+      "step": 5726
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.337282045681894e-06,
+      "loss": 0.2331,
+      "step": 5728
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.331270591889717e-06,
+      "loss": 0.2312,
+      "step": 5730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.325259758345536e-06,
+      "loss": 0.2602,
+      "step": 5732
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.3192495472835e-06,
+      "loss": 0.2885,
+      "step": 5734
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.313239960937547e-06,
+      "loss": 0.279,
+      "step": 5736
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.30723100154137e-06,
+      "loss": 0.2675,
+      "step": 5738
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.301222671328429e-06,
+      "loss": 0.2912,
+      "step": 5740
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.295214972531956e-06,
+      "loss": 0.2785,
+      "step": 5742
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.28920790738494e-06,
+      "loss": 0.298,
+      "step": 5744
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.283201478120146e-06,
+      "loss": 0.2712,
+      "step": 5746
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.277195686970092e-06,
+      "loss": 0.2597,
+      "step": 5748
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.271190536167064e-06,
+      "loss": 0.243,
+      "step": 5750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.265186027943112e-06,
+      "loss": 0.2787,
+      "step": 5752
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.25918216453004e-06,
+      "loss": 0.217,
+      "step": 5754
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.253178948159422e-06,
+      "loss": 0.2441,
+      "step": 5756
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.24717638106258e-06,
+      "loss": 0.2662,
+      "step": 5758
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.241174465470607e-06,
+      "loss": 0.2406,
+      "step": 5760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.235173203614344e-06,
+      "loss": 0.2133,
+      "step": 5762
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.229172597724392e-06,
+      "loss": 0.2586,
+      "step": 5764
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.223172650031105e-06,
+      "loss": 0.3324,
+      "step": 5766
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.217173362764607e-06,
+      "loss": 0.2697,
+      "step": 5768
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.211174738154759e-06,
+      "loss": 0.2626,
+      "step": 5770
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.205176778431185e-06,
+      "loss": 0.2758,
+      "step": 5772
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.199179485823255e-06,
+      "loss": 0.2892,
+      "step": 5774
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.193182862560102e-06,
+      "loss": 0.2623,
+      "step": 5776
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.187186910870595e-06,
+      "loss": 0.2852,
+      "step": 5778
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.18119163298337e-06,
+      "loss": 0.2736,
+      "step": 5780
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.175197031126798e-06,
+      "loss": 0.2772,
+      "step": 5782
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.16920310752901e-06,
+      "loss": 0.2671,
+      "step": 5784
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.16320986441788e-06,
+      "loss": 0.2735,
+      "step": 5786
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.157217304021025e-06,
+      "loss": 0.2526,
+      "step": 5788
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.151225428565817e-06,
+      "loss": 0.2413,
+      "step": 5790
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.145234240279367e-06,
+      "loss": 0.2594,
+      "step": 5792
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.139243741388532e-06,
+      "loss": 0.237,
+      "step": 5794
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.133253934119912e-06,
+      "loss": 0.2281,
+      "step": 5796
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.127264820699852e-06,
+      "loss": 0.2515,
+      "step": 5798
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.12127640335444e-06,
+      "loss": 0.283,
+      "step": 5800
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.115288684309504e-06,
+      "loss": 0.2368,
+      "step": 5802
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.10930166579061e-06,
+      "loss": 0.283,
+      "step": 5804
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.103315350023067e-06,
+      "loss": 0.2547,
+      "step": 5806
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.09732973923192e-06,
+      "loss": 0.2441,
+      "step": 5808
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.091344835641952e-06,
+      "loss": 0.2358,
+      "step": 5810
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.085360641477688e-06,
+      "loss": 0.2423,
+      "step": 5812
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.07937715896338e-06,
+      "loss": 0.2355,
+      "step": 5814
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.07339439032303e-06,
+      "loss": 0.2219,
+      "step": 5816
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.067412337780359e-06,
+      "loss": 0.2946,
+      "step": 5818
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.061431003558833e-06,
+      "loss": 0.2738,
+      "step": 5820
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.055450389881643e-06,
+      "loss": 0.2372,
+      "step": 5822
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.04947049897172e-06,
+      "loss": 0.2515,
+      "step": 5824
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.043491333051719e-06,
+      "loss": 0.3073,
+      "step": 5826
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.03751289434403e-06,
+      "loss": 0.2241,
+      "step": 5828
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.03153518507077e-06,
+      "loss": 0.242,
+      "step": 5830
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.025558207453788e-06,
+      "loss": 0.2278,
+      "step": 5832
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.019581963714661e-06,
+      "loss": 0.2784,
+      "step": 5834
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.013606456074693e-06,
+      "loss": 0.2614,
+      "step": 5836
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.007631686754907e-06,
+      "loss": 0.2535,
+      "step": 5838
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.001657657976061e-06,
+      "loss": 0.3033,
+      "step": 5840
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.995684371958639e-06,
+      "loss": 0.2906,
+      "step": 5842
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.989711830922835e-06,
+      "loss": 0.281,
+      "step": 5844
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.983740037088578e-06,
+      "loss": 0.2743,
+      "step": 5846
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.977768992675525e-06,
+      "loss": 0.3055,
+      "step": 5848
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.971798699903038e-06,
+      "loss": 0.2735,
+      "step": 5850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.965829160990213e-06,
+      "loss": 0.2649,
+      "step": 5852
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.959860378155854e-06,
+      "loss": 0.2646,
+      "step": 5854
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.953892353618497e-06,
+      "loss": 0.2484,
+      "step": 5856
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.947925089596384e-06,
+      "loss": 0.23,
+      "step": 5858
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.941958588307485e-06,
+      "loss": 0.2226,
+      "step": 5860
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.935992851969477e-06,
+      "loss": 0.2671,
+      "step": 5862
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.930027882799763e-06,
+      "loss": 0.2985,
+      "step": 5864
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.92406368301545e-06,
+      "loss": 0.2682,
+      "step": 5866
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.91810025483337e-06,
+      "loss": 0.2593,
+      "step": 5868
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.912137600470055e-06,
+      "loss": 0.2518,
+      "step": 5870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.906175722141763e-06,
+      "loss": 0.2599,
+      "step": 5872
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.900214622064453e-06,
+      "loss": 0.2825,
+      "step": 5874
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.894254302453801e-06,
+      "loss": 0.2649,
+      "step": 5876
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.888294765525187e-06,
+      "loss": 0.2371,
+      "step": 5878
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.882336013493708e-06,
+      "loss": 0.2544,
+      "step": 5880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.876378048574168e-06,
+      "loss": 0.2051,
+      "step": 5882
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.870420872981071e-06,
+      "loss": 0.2177,
+      "step": 5884
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.864464488928634e-06,
+      "loss": 0.2388,
+      "step": 5886
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.858508898630775e-06,
+      "loss": 0.2504,
+      "step": 5888
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.852554104301127e-06,
+      "loss": 0.2802,
+      "step": 5890
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.846600108153013e-06,
+      "loss": 0.294,
+      "step": 5892
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.840646912399467e-06,
+      "loss": 0.2537,
+      "step": 5894
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.834694519253231e-06,
+      "loss": 0.2495,
+      "step": 5896
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.828742930926738e-06,
+      "loss": 0.2279,
+      "step": 5898
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.82279214963213e-06,
+      "loss": 0.2739,
+      "step": 5900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.816842177581243e-06,
+      "loss": 0.2431,
+      "step": 5902
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.810893016985612e-06,
+      "loss": 0.2612,
+      "step": 5904
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.80494467005648e-06,
+      "loss": 0.3042,
+      "step": 5906
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.798997139004776e-06,
+      "loss": 0.3064,
+      "step": 5908
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.793050426041128e-06,
+      "loss": 0.2941,
+      "step": 5910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.787104533375874e-06,
+      "loss": 0.1952,
+      "step": 5912
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.781159463219023e-06,
+      "loss": 0.275,
+      "step": 5914
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.775215217780298e-06,
+      "loss": 0.2734,
+      "step": 5916
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.769271799269104e-06,
+      "loss": 0.1762,
+      "step": 5918
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.763329209894548e-06,
+      "loss": 0.3051,
+      "step": 5920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.757387451865417e-06,
+      "loss": 0.2271,
+      "step": 5922
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.751446527390203e-06,
+      "loss": 0.2427,
+      "step": 5924
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.745506438677073e-06,
+      "loss": 0.3809,
+      "step": 5926
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.739567187933898e-06,
+      "loss": 0.3004,
+      "step": 5928
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.73362877736823e-06,
+      "loss": 0.2142,
+      "step": 5930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.727691209187306e-06,
+      "loss": 0.2855,
+      "step": 5932
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.721754485598058e-06,
+      "loss": 0.2654,
+      "step": 5934
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.715818608807097e-06,
+      "loss": 0.2674,
+      "step": 5936
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.709883581020721e-06,
+      "loss": 0.2585,
+      "step": 5938
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.703949404444913e-06,
+      "loss": 0.2039,
+      "step": 5940
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.69801608128534e-06,
+      "loss": 0.2789,
+      "step": 5942
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.692083613747353e-06,
+      "loss": 0.246,
+      "step": 5944
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.686152004035986e-06,
+      "loss": 0.231,
+      "step": 5946
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.680221254355948e-06,
+      "loss": 0.2432,
+      "step": 5948
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.674291366911634e-06,
+      "loss": 0.2414,
+      "step": 5950
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.668362343907113e-06,
+      "loss": 0.2493,
+      "step": 5952
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.662434187546143e-06,
+      "loss": 0.27,
+      "step": 5954
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.656506900032146e-06,
+      "loss": 0.2756,
+      "step": 5956
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.65058048356823e-06,
+      "loss": 0.2751,
+      "step": 5958
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.644654940357181e-06,
+      "loss": 0.1842,
+      "step": 5960
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.638730272601454e-06,
+      "loss": 0.2159,
+      "step": 5962
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.632806482503185e-06,
+      "loss": 0.2575,
+      "step": 5964
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.626883572264174e-06,
+      "loss": 0.2568,
+      "step": 5966
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.620961544085908e-06,
+      "loss": 0.2794,
+      "step": 5968
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.61504040016953e-06,
+      "loss": 0.2746,
+      "step": 5970
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.60912014271587e-06,
+      "loss": 0.2579,
+      "step": 5972
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.6032007739254145e-06,
+      "loss": 0.2355,
+      "step": 5974
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.597282295998332e-06,
+      "loss": 0.2838,
+      "step": 5976
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.591364711134453e-06,
+      "loss": 0.2582,
+      "step": 5978
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.585448021533276e-06,
+      "loss": 0.2908,
+      "step": 5980
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.579532229393971e-06,
+      "loss": 0.2741,
+      "step": 5982
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.573617336915367e-06,
+      "loss": 0.194,
+      "step": 5984
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.567703346295969e-06,
+      "loss": 0.2488,
+      "step": 5986
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.561790259733934e-06,
+      "loss": 0.2769,
+      "step": 5988
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.555878079427092e-06,
+      "loss": 0.2074,
+      "step": 5990
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.549966807572939e-06,
+      "loss": 0.2676,
+      "step": 5992
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.544056446368626e-06,
+      "loss": 0.2396,
+      "step": 5994
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.538146998010964e-06,
+      "loss": 0.2294,
+      "step": 5996
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.532238464696434e-06,
+      "loss": 0.24,
+      "step": 5998
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.526330848621167e-06,
+      "loss": 0.2426,
+      "step": 6000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.5204241519809625e-06,
+      "loss": 0.2379,
+      "step": 6002
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.5145183769712685e-06,
+      "loss": 0.2622,
+      "step": 6004
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.508613525787194e-06,
+      "loss": 0.3402,
+      "step": 6006
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.502709600623513e-06,
+      "loss": 0.2415,
+      "step": 6008
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.496806603674643e-06,
+      "loss": 0.2347,
+      "step": 6010
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.490904537134667e-06,
+      "loss": 0.303,
+      "step": 6012
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.485003403197309e-06,
+      "loss": 0.2411,
+      "step": 6014
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.4791032040559595e-06,
+      "loss": 0.2405,
+      "step": 6016
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.473203941903654e-06,
+      "loss": 0.2126,
+      "step": 6018
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.467305618933085e-06,
+      "loss": 0.2681,
+      "step": 6020
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.461408237336587e-06,
+      "loss": 0.2415,
+      "step": 6022
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.455511799306157e-06,
+      "loss": 0.2549,
+      "step": 6024
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.449616307033434e-06,
+      "loss": 0.3153,
+      "step": 6026
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.443721762709701e-06,
+      "loss": 0.2412,
+      "step": 6028
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.437828168525902e-06,
+      "loss": 0.2449,
+      "step": 6030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.431935526672613e-06,
+      "loss": 0.2378,
+      "step": 6032
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.426043839340067e-06,
+      "loss": 0.2825,
+      "step": 6034
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.420153108718135e-06,
+      "loss": 0.2589,
+      "step": 6036
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.414263336996335e-06,
+      "loss": 0.2316,
+      "step": 6038
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.4083745263638305e-06,
+      "loss": 0.2667,
+      "step": 6040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.40248667900943e-06,
+      "loss": 0.2538,
+      "step": 6042
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.396599797121578e-06,
+      "loss": 0.2768,
+      "step": 6044
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.390713882888359e-06,
+      "loss": 0.2935,
+      "step": 6046
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.384828938497504e-06,
+      "loss": 0.2586,
+      "step": 6048
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.378944966136382e-06,
+      "loss": 0.2463,
+      "step": 6050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.373061967991995e-06,
+      "loss": 0.2103,
+      "step": 6052
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.36717994625099e-06,
+      "loss": 0.276,
+      "step": 6054
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.361298903099652e-06,
+      "loss": 0.2622,
+      "step": 6056
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.355418840723893e-06,
+      "loss": 0.3027,
+      "step": 6058
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.34953976130927e-06,
+      "loss": 0.2708,
+      "step": 6060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.343661667040969e-06,
+      "loss": 0.2334,
+      "step": 6062
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.337784560103813e-06,
+      "loss": 0.215,
+      "step": 6064
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.331908442682256e-06,
+      "loss": 0.2989,
+      "step": 6066
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.3260333169603846e-06,
+      "loss": 0.2308,
+      "step": 6068
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.320159185121914e-06,
+      "loss": 0.2904,
+      "step": 6070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.314286049350199e-06,
+      "loss": 0.2375,
+      "step": 6072
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.308413911828216e-06,
+      "loss": 0.2502,
+      "step": 6074
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.30254277473857e-06,
+      "loss": 0.2595,
+      "step": 6076
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.2966726402635024e-06,
+      "loss": 0.2349,
+      "step": 6078
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.29080351058487e-06,
+      "loss": 0.2785,
+      "step": 6080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.284935387884169e-06,
+      "loss": 0.2517,
+      "step": 6082
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.2790682743425065e-06,
+      "loss": 0.2369,
+      "step": 6084
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.273202172140627e-06,
+      "loss": 0.2331,
+      "step": 6086
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.267337083458896e-06,
+      "loss": 0.2373,
+      "step": 6088
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.261473010477302e-06,
+      "loss": 0.2433,
+      "step": 6090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.2556099553754535e-06,
+      "loss": 0.2947,
+      "step": 6092
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.249747920332582e-06,
+      "loss": 0.2319,
+      "step": 6094
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.24388690752754e-06,
+      "loss": 0.2496,
+      "step": 6096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.238026919138801e-06,
+      "loss": 0.2303,
+      "step": 6098
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.232167957344453e-06,
+      "loss": 0.3032,
+      "step": 6100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.226310024322211e-06,
+      "loss": 0.2066,
+      "step": 6102
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.2204531222494035e-06,
+      "loss": 0.2932,
+      "step": 6104
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.214597253302971e-06,
+      "loss": 0.2366,
+      "step": 6106
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.20874241965948e-06,
+      "loss": 0.2925,
+      "step": 6108
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.202888623495101e-06,
+      "loss": 0.2411,
+      "step": 6110
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.197035866985629e-06,
+      "loss": 0.2269,
+      "step": 6112
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.191184152306463e-06,
+      "loss": 0.2601,
+      "step": 6114
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.185333481632623e-06,
+      "loss": 0.2553,
+      "step": 6116
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.179483857138733e-06,
+      "loss": 0.2767,
+      "step": 6118
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.173635280999038e-06,
+      "loss": 0.2407,
+      "step": 6120
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.16778775538739e-06,
+      "loss": 0.2566,
+      "step": 6122
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.1619412824772425e-06,
+      "loss": 0.2725,
+      "step": 6124
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.156095864441665e-06,
+      "loss": 0.3011,
+      "step": 6126
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.15025150345334e-06,
+      "loss": 0.2745,
+      "step": 6128
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.1444082016845425e-06,
+      "loss": 0.2461,
+      "step": 6130
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.138565961307169e-06,
+      "loss": 0.2937,
+      "step": 6132
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.132724784492706e-06,
+      "loss": 0.2717,
+      "step": 6134
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.126884673412262e-06,
+      "loss": 0.3007,
+      "step": 6136
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.121045630236538e-06,
+      "loss": 0.2469,
+      "step": 6138
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.115207657135841e-06,
+      "loss": 0.2729,
+      "step": 6140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.109370756280081e-06,
+      "loss": 0.2678,
+      "step": 6142
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.103534929838763e-06,
+      "loss": 0.2754,
+      "step": 6144
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.097700179981006e-06,
+      "loss": 0.2331,
+      "step": 6146
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.091866508875516e-06,
+      "loss": 0.3193,
+      "step": 6148
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.0860339186906e-06,
+      "loss": 0.2245,
+      "step": 6150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.080202411594176e-06,
+      "loss": 0.2321,
+      "step": 6152
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.074371989753744e-06,
+      "loss": 0.2925,
+      "step": 6154
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.0685426553364055e-06,
+      "loss": 0.2166,
+      "step": 6156
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.06271441050886e-06,
+      "loss": 0.2842,
+      "step": 6158
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.056887257437401e-06,
+      "loss": 0.2822,
+      "step": 6160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.051061198287916e-06,
+      "loss": 0.2496,
+      "step": 6162
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.045236235225886e-06,
+      "loss": 0.2515,
+      "step": 6164
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.03941237041638e-06,
+      "loss": 0.2669,
+      "step": 6166
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.033589606024068e-06,
+      "loss": 0.2624,
+      "step": 6168
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.027767944213206e-06,
+      "loss": 0.2363,
+      "step": 6170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.021947387147641e-06,
+      "loss": 0.2396,
+      "step": 6172
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.016127936990807e-06,
+      "loss": 0.2229,
+      "step": 6174
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.010309595905728e-06,
+      "loss": 0.2423,
+      "step": 6176
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.004492366055016e-06,
+      "loss": 0.2442,
+      "step": 6178
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.998676249600871e-06,
+      "loss": 0.2617,
+      "step": 6180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.992861248705074e-06,
+      "loss": 0.206,
+      "step": 6182
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.9870473655290025e-06,
+      "loss": 0.2317,
+      "step": 6184
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.981234602233611e-06,
+      "loss": 0.2467,
+      "step": 6186
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.975422960979432e-06,
+      "loss": 0.2349,
+      "step": 6188
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.969612443926594e-06,
+      "loss": 0.2106,
+      "step": 6190
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.963803053234797e-06,
+      "loss": 0.2493,
+      "step": 6192
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.95799479106333e-06,
+      "loss": 0.2778,
+      "step": 6194
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.952187659571052e-06,
+      "loss": 0.2381,
+      "step": 6196
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.946381660916412e-06,
+      "loss": 0.2954,
+      "step": 6198
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.940576797257439e-06,
+      "loss": 0.2517,
+      "step": 6200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.93477307075173e-06,
+      "loss": 0.2274,
+      "step": 6202
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.92897048355647e-06,
+      "loss": 0.2844,
+      "step": 6204
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.923169037828411e-06,
+      "loss": 0.2168,
+      "step": 6206
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.9173687357238886e-06,
+      "loss": 0.2163,
+      "step": 6208
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.911569579398806e-06,
+      "loss": 0.2298,
+      "step": 6210
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.9057715710086485e-06,
+      "loss": 0.2398,
+      "step": 6212
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.899974712708464e-06,
+      "loss": 0.2538,
+      "step": 6214
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.89417900665289e-06,
+      "loss": 0.2341,
+      "step": 6216
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.888384454996117e-06,
+      "loss": 0.247,
+      "step": 6218
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.882591059891923e-06,
+      "loss": 0.2197,
+      "step": 6220
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.8767988234936385e-06,
+      "loss": 0.2708,
+      "step": 6222
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.87100774795418e-06,
+      "loss": 0.2583,
+      "step": 6224
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.8652178354260215e-06,
+      "loss": 0.2546,
+      "step": 6226
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.859429088061209e-06,
+      "loss": 0.2574,
+      "step": 6228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.853641508011353e-06,
+      "loss": 0.2803,
+      "step": 6230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.847855097427636e-06,
+      "loss": 0.2852,
+      "step": 6232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.842069858460803e-06,
+      "loss": 0.2482,
+      "step": 6234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.836285793261156e-06,
+      "loss": 0.265,
+      "step": 6236
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.830502903978574e-06,
+      "loss": 0.2255,
+      "step": 6238
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.824721192762487e-06,
+      "loss": 0.2401,
+      "step": 6240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.818940661761894e-06,
+      "loss": 0.2727,
+      "step": 6242
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.813161313125354e-06,
+      "loss": 0.2628,
+      "step": 6244
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.8073831490009826e-06,
+      "loss": 0.2858,
+      "step": 6246
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.801606171536466e-06,
+      "loss": 0.2577,
+      "step": 6248
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.795830382879034e-06,
+      "loss": 0.2459,
+      "step": 6250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.790055785175489e-06,
+      "loss": 0.2438,
+      "step": 6252
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.7871689336021725e-06,
+      "loss": 0.2698,
+      "step": 6254
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.781396126353726e-06,
+      "loss": 0.2435,
+      "step": 6256
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.775624515424214e-06,
+      "loss": 0.2761,
+      "step": 6258
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.7698541029588785e-06,
+      "loss": 0.2666,
+      "step": 6260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.764084891102516e-06,
+      "loss": 0.2493,
+      "step": 6262
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.758316881999485e-06,
+      "loss": 0.2659,
+      "step": 6264
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.752550077793682e-06,
+      "loss": 0.2373,
+      "step": 6266
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.7467844806285696e-06,
+      "loss": 0.2533,
+      "step": 6268
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.74102009264715e-06,
+      "loss": 0.2762,
+      "step": 6270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.735256915991986e-06,
+      "loss": 0.246,
+      "step": 6272
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.72949495280518e-06,
+      "loss": 0.3053,
+      "step": 6274
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.723734205228396e-06,
+      "loss": 0.2206,
+      "step": 6276
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.7179746754028315e-06,
+      "loss": 0.2345,
+      "step": 6278
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.712216365469246e-06,
+      "loss": 0.3328,
+      "step": 6280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.706459277567937e-06,
+      "loss": 0.2655,
+      "step": 6282
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.700703413838747e-06,
+      "loss": 0.2469,
+      "step": 6284
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.6949487764210685e-06,
+      "loss": 0.2583,
+      "step": 6286
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.689195367453836e-06,
+      "loss": 0.2119,
+      "step": 6288
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.683443189075522e-06,
+      "loss": 0.266,
+      "step": 6290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.677692243424154e-06,
+      "loss": 0.2594,
+      "step": 6292
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.671942532637286e-06,
+      "loss": 0.2407,
+      "step": 6294
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.666194058852027e-06,
+      "loss": 0.296,
+      "step": 6296
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.66044682420502e-06,
+      "loss": 0.2651,
+      "step": 6298
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.6547008308324435e-06,
+      "loss": 0.2611,
+      "step": 6300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.648956080870024e-06,
+      "loss": 0.2771,
+      "step": 6302
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.643212576453016e-06,
+      "loss": 0.2077,
+      "step": 6304
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.63747031971622e-06,
+      "loss": 0.2521,
+      "step": 6306
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.631729312793962e-06,
+      "loss": 0.2243,
+      "step": 6308
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.625989557820114e-06,
+      "loss": 0.2329,
+      "step": 6310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.62025105692808e-06,
+      "loss": 0.2528,
+      "step": 6312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.614513812250793e-06,
+      "loss": 0.2269,
+      "step": 6314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.608777825920727e-06,
+      "loss": 0.2672,
+      "step": 6316
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.603043100069879e-06,
+      "loss": 0.2483,
+      "step": 6318
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.597309636829785e-06,
+      "loss": 0.2527,
+      "step": 6320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.591577438331507e-06,
+      "loss": 0.253,
+      "step": 6322
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.58584650670564e-06,
+      "loss": 0.2664,
+      "step": 6324
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.582981516635543e-06,
+      "loss": 0.2917,
+      "step": 6326
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.57725248931212e-06,
+      "loss": 0.2627,
+      "step": 6328
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.571524734185535e-06,
+      "loss": 0.2196,
+      "step": 6330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.565798253384728e-06,
+      "loss": 0.2505,
+      "step": 6332
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.560073049038162e-06,
+      "loss": 0.2842,
+      "step": 6334
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.554349123273839e-06,
+      "loss": 0.2762,
+      "step": 6336
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.54862647821927e-06,
+      "loss": 0.2459,
+      "step": 6338
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.5429051160015035e-06,
+      "loss": 0.2581,
+      "step": 6340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.5371850387471025e-06,
+      "loss": 0.2524,
+      "step": 6342
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.531466248582154e-06,
+      "loss": 0.2765,
+      "step": 6344
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.5257487476322745e-06,
+      "loss": 0.2452,
+      "step": 6346
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.520032538022589e-06,
+      "loss": 0.2272,
+      "step": 6348
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.514317621877755e-06,
+      "loss": 0.2301,
+      "step": 6350
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.508604001321935e-06,
+      "loss": 0.2373,
+      "step": 6352
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.50289167847882e-06,
+      "loss": 0.228,
+      "step": 6354
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.4971806554716195e-06,
+      "loss": 0.2254,
+      "step": 6356
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.491470934423054e-06,
+      "loss": 0.2758,
+      "step": 6358
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.48576251745536e-06,
+      "loss": 0.2878,
+      "step": 6360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.4800554066903e-06,
+      "loss": 0.2593,
+      "step": 6362
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.474349604249135e-06,
+      "loss": 0.2429,
+      "step": 6364
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.468645112252654e-06,
+      "loss": 0.2382,
+      "step": 6366
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.462941932821146e-06,
+      "loss": 0.2374,
+      "step": 6368
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.457240068074426e-06,
+      "loss": 0.2281,
+      "step": 6370
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.451539520131806e-06,
+      "loss": 0.2788,
+      "step": 6372
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.445840291112123e-06,
+      "loss": 0.2479,
+      "step": 6374
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.440142383133707e-06,
+      "loss": 0.228,
+      "step": 6376
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.434445798314416e-06,
+      "loss": 0.2311,
+      "step": 6378
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.428750538771604e-06,
+      "loss": 0.2734,
+      "step": 6380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.423056606622131e-06,
+      "loss": 0.2414,
+      "step": 6382
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.4173640039823724e-06,
+      "loss": 0.2613,
+      "step": 6384
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.4116727329682016e-06,
+      "loss": 0.2984,
+      "step": 6386
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.405982795695003e-06,
+      "loss": 0.3059,
+      "step": 6388
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.4002941942776585e-06,
+      "loss": 0.2375,
+      "step": 6390
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.394606930830558e-06,
+      "loss": 0.2307,
+      "step": 6392
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.388921007467599e-06,
+      "loss": 0.2939,
+      "step": 6394
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.38323642630217e-06,
+      "loss": 0.229,
+      "step": 6396
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.3775531894471665e-06,
+      "loss": 0.2694,
+      "step": 6398
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.3718712990149875e-06,
+      "loss": 0.2496,
+      "step": 6400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.366190757117523e-06,
+      "loss": 0.2445,
+      "step": 6402
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.360511565866172e-06,
+      "loss": 0.2976,
+      "step": 6404
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.35483372737182e-06,
+      "loss": 0.2366,
+      "step": 6406
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.349157243744856e-06,
+      "loss": 0.2859,
+      "step": 6408
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.343482117095174e-06,
+      "loss": 0.234,
+      "step": 6410
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.337808349532147e-06,
+      "loss": 0.295,
+      "step": 6412
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.332135943164654e-06,
+      "loss": 0.2881,
+      "step": 6414
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.3264649001010625e-06,
+      "loss": 0.2782,
+      "step": 6416
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.320795222449239e-06,
+      "loss": 0.2303,
+      "step": 6418
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.315126912316534e-06,
+      "loss": 0.2714,
+      "step": 6420
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.309459971809798e-06,
+      "loss": 0.286,
+      "step": 6422
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.303794403035366e-06,
+      "loss": 0.2689,
+      "step": 6424
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.298130208099072e-06,
+      "loss": 0.2204,
+      "step": 6426
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.29246738910623e-06,
+      "loss": 0.2284,
+      "step": 6428
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.286805948161647e-06,
+      "loss": 0.2256,
+      "step": 6430
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.281145887369617e-06,
+      "loss": 0.2368,
+      "step": 6432
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.275487208833921e-06,
+      "loss": 0.2271,
+      "step": 6434
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.269829914657828e-06,
+      "loss": 0.2061,
+      "step": 6436
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.264174006944086e-06,
+      "loss": 0.2847,
+      "step": 6438
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.258519487794934e-06,
+      "loss": 0.2511,
+      "step": 6440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.252866359312097e-06,
+      "loss": 0.3066,
+      "step": 6442
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.24721462359678e-06,
+      "loss": 0.2475,
+      "step": 6444
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.2415642827496635e-06,
+      "loss": 0.2606,
+      "step": 6446
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.235915338870923e-06,
+      "loss": 0.2684,
+      "step": 6448
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.2302677940602005e-06,
+      "loss": 0.2097,
+      "step": 6450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.22462165041663e-06,
+      "loss": 0.2773,
+      "step": 6452
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.2189769100388165e-06,
+      "loss": 0.2849,
+      "step": 6454
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.213333575024846e-06,
+      "loss": 0.2225,
+      "step": 6456
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.207691647472289e-06,
+      "loss": 0.2459,
+      "step": 6458
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.202051129478179e-06,
+      "loss": 0.2736,
+      "step": 6460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.196412023139041e-06,
+      "loss": 0.2216,
+      "step": 6462
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.19077433055086e-06,
+      "loss": 0.2946,
+      "step": 6464
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.18513805380911e-06,
+      "loss": 0.2771,
+      "step": 6466
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.179503195008725e-06,
+      "loss": 0.264,
+      "step": 6468
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.1738697562441245e-06,
+      "loss": 0.2431,
+      "step": 6470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.168237739609186e-06,
+      "loss": 0.2396,
+      "step": 6472
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.162607147197277e-06,
+      "loss": 0.1997,
+      "step": 6474
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.156977981101222e-06,
+      "loss": 0.2602,
+      "step": 6476
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.151350243413317e-06,
+      "loss": 0.2301,
+      "step": 6478
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.145723936225332e-06,
+      "loss": 0.2647,
+      "step": 6480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.140099061628498e-06,
+      "loss": 0.2592,
+      "step": 6482
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.134475621713523e-06,
+      "loss": 0.267,
+      "step": 6484
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.128853618570569e-06,
+      "loss": 0.2387,
+      "step": 6486
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.123233054289277e-06,
+      "loss": 0.2359,
+      "step": 6488
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.117613930958748e-06,
+      "loss": 0.2594,
+      "step": 6490
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.1119962506675465e-06,
+      "loss": 0.2396,
+      "step": 6492
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.106380015503699e-06,
+      "loss": 0.2957,
+      "step": 6494
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.100765227554701e-06,
+      "loss": 0.2563,
+      "step": 6496
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.095151888907502e-06,
+      "loss": 0.2274,
+      "step": 6498
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.089540001648519e-06,
+      "loss": 0.2463,
+      "step": 6500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.083929567863627e-06,
+      "loss": 0.3476,
+      "step": 6502
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.078320589638159e-06,
+      "loss": 0.271,
+      "step": 6504
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.072713069056916e-06,
+      "loss": 0.2451,
+      "step": 6506
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.067107008204144e-06,
+      "loss": 0.2913,
+      "step": 6508
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.061502409163556e-06,
+      "loss": 0.2311,
+      "step": 6510
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.0558992740183175e-06,
+      "loss": 0.2649,
+      "step": 6512
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.050297604851052e-06,
+      "loss": 0.3188,
+      "step": 6514
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.044697403743833e-06,
+      "loss": 0.2216,
+      "step": 6516
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.039098672778199e-06,
+      "loss": 0.24,
+      "step": 6518
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.0335014140351275e-06,
+      "loss": 0.2038,
+      "step": 6520
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.027905629595065e-06,
+      "loss": 0.2974,
+      "step": 6522
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.022311321537898e-06,
+      "loss": 0.255,
+      "step": 6524
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.016718491942967e-06,
+      "loss": 0.2504,
+      "step": 6526
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.0111271428890665e-06,
+      "loss": 0.2675,
+      "step": 6528
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.005537276454434e-06,
+      "loss": 0.2594,
+      "step": 6530
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.9999488947167665e-06,
+      "loss": 0.2528,
+      "step": 6532
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.994361999753196e-06,
+      "loss": 0.2533,
+      "step": 6534
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.988776593640311e-06,
+      "loss": 0.2979,
+      "step": 6536
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.983192678454145e-06,
+      "loss": 0.2079,
+      "step": 6538
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.97761025627018e-06,
+      "loss": 0.2794,
+      "step": 6540
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.972029329163334e-06,
+      "loss": 0.2645,
+      "step": 6542
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.966449899207979e-06,
+      "loss": 0.243,
+      "step": 6544
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.960871968477923e-06,
+      "loss": 0.2279,
+      "step": 6546
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.955295539046424e-06,
+      "loss": 0.2685,
+      "step": 6548
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.949720612986174e-06,
+      "loss": 0.2509,
+      "step": 6550
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.94414719236931e-06,
+      "loss": 0.2769,
+      "step": 6552
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.938575279267418e-06,
+      "loss": 0.2346,
+      "step": 6554
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.933004875751507e-06,
+      "loss": 0.25,
+      "step": 6556
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.927435983892039e-06,
+      "loss": 0.2519,
+      "step": 6558
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.921868605758905e-06,
+      "loss": 0.2703,
+      "step": 6560
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.916302743421439e-06,
+      "loss": 0.2799,
+      "step": 6562
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.9107383989484064e-06,
+      "loss": 0.2812,
+      "step": 6564
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.905175574408018e-06,
+      "loss": 0.289,
+      "step": 6566
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.899614271867903e-06,
+      "loss": 0.2409,
+      "step": 6568
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.8940544933951434e-06,
+      "loss": 0.2754,
+      "step": 6570
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.888496241056247e-06,
+      "loss": 0.2511,
+      "step": 6572
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.88293951691715e-06,
+      "loss": 0.2757,
+      "step": 6574
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.877384323043226e-06,
+      "loss": 0.2361,
+      "step": 6576
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.871830661499276e-06,
+      "loss": 0.2595,
+      "step": 6578
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.866278534349537e-06,
+      "loss": 0.2552,
+      "step": 6580
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.86072794365767e-06,
+      "loss": 0.2727,
+      "step": 6582
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.855178891486766e-06,
+      "loss": 0.2501,
+      "step": 6584
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.8496313798993485e-06,
+      "loss": 0.239,
+      "step": 6586
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.844085410957367e-06,
+      "loss": 0.2379,
+      "step": 6588
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.83854098672219e-06,
+      "loss": 0.3126,
+      "step": 6590
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.832998109254622e-06,
+      "loss": 0.215,
+      "step": 6592
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.827456780614891e-06,
+      "loss": 0.2408,
+      "step": 6594
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.821917002862639e-06,
+      "loss": 0.2324,
+      "step": 6596
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.816378778056943e-06,
+      "loss": 0.2153,
+      "step": 6598
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.810842108256298e-06,
+      "loss": 0.2709,
+      "step": 6600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.805306995518625e-06,
+      "loss": 0.2302,
+      "step": 6602
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.7997734419012605e-06,
+      "loss": 0.2959,
+      "step": 6604
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.794241449460966e-06,
+      "loss": 0.2467,
+      "step": 6606
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.7887110202539234e-06,
+      "loss": 0.251,
+      "step": 6608
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.783182156335722e-06,
+      "loss": 0.2262,
+      "step": 6610
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.777654859761388e-06,
+      "loss": 0.2427,
+      "step": 6612
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.77212913258535e-06,
+      "loss": 0.2081,
+      "step": 6614
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.766604976861461e-06,
+      "loss": 0.2769,
+      "step": 6616
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.761082394642987e-06,
+      "loss": 0.2272,
+      "step": 6618
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.755561387982611e-06,
+      "loss": 0.2501,
+      "step": 6620
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.750041958932431e-06,
+      "loss": 0.25,
+      "step": 6622
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.744524109543952e-06,
+      "loss": 0.2715,
+      "step": 6624
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.739007841868098e-06,
+      "loss": 0.2265,
+      "step": 6626
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.733493157955205e-06,
+      "loss": 0.2253,
+      "step": 6628
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.727980059855019e-06,
+      "loss": 0.2491,
+      "step": 6630
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.722468549616695e-06,
+      "loss": 0.2415,
+      "step": 6632
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.716958629288802e-06,
+      "loss": 0.2365,
+      "step": 6634
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7114503009193145e-06,
+      "loss": 0.2204,
+      "step": 6636
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.70594356655562e-06,
+      "loss": 0.2447,
+      "step": 6638
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.700438428244503e-06,
+      "loss": 0.2617,
+      "step": 6640
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.694934888032164e-06,
+      "loss": 0.2231,
+      "step": 6642
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.689432947964208e-06,
+      "loss": 0.2487,
+      "step": 6644
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.68393261008565e-06,
+      "loss": 0.2527,
+      "step": 6646
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.6784338764408885e-06,
+      "loss": 0.2144,
+      "step": 6648
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.672936749073755e-06,
+      "loss": 0.2967,
+      "step": 6650
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.667441230027471e-06,
+      "loss": 0.2428,
+      "step": 6652
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.661947321344648e-06,
+      "loss": 0.2538,
+      "step": 6654
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.656455025067318e-06,
+      "loss": 0.2329,
+      "step": 6656
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.650964343236903e-06,
+      "loss": 0.3084,
+      "step": 6658
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.645475277894233e-06,
+      "loss": 0.2314,
+      "step": 6660
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.639987831079525e-06,
+      "loss": 0.2517,
+      "step": 6662
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.6345020048324005e-06,
+      "loss": 0.247,
+      "step": 6664
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.629017801191886e-06,
+      "loss": 0.2902,
+      "step": 6666
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.6235352221964e-06,
+      "loss": 0.236,
+      "step": 6668
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.618054269883748e-06,
+      "loss": 0.2606,
+      "step": 6670
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.612574946291142e-06,
+      "loss": 0.2272,
+      "step": 6672
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.607097253455185e-06,
+      "loss": 0.2612,
+      "step": 6674
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.6016211934118765e-06,
+      "loss": 0.2728,
+      "step": 6676
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.5961467681966005e-06,
+      "loss": 0.2547,
+      "step": 6678
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.590673979844139e-06,
+      "loss": 0.2474,
+      "step": 6680
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.585202830388679e-06,
+      "loss": 0.2782,
+      "step": 6682
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.5797333218637705e-06,
+      "loss": 0.2473,
+      "step": 6684
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.574265456302375e-06,
+      "loss": 0.198,
+      "step": 6686
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.568799235736836e-06,
+      "loss": 0.2112,
+      "step": 6688
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.56333466219889e-06,
+      "loss": 0.2398,
+      "step": 6690
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.5578717377196486e-06,
+      "loss": 0.2793,
+      "step": 6692
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.552410464329627e-06,
+      "loss": 0.2594,
+      "step": 6694
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.546950844058716e-06,
+      "loss": 0.3136,
+      "step": 6696
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.541492878936196e-06,
+      "loss": 0.2692,
+      "step": 6698
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.536036570990731e-06,
+      "loss": 0.2418,
+      "step": 6700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.5305819222503706e-06,
+      "loss": 0.2948,
+      "step": 6702
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.52512893474255e-06,
+      "loss": 0.2562,
+      "step": 6704
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.519677610494075e-06,
+      "loss": 0.2785,
+      "step": 6706
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.5142279515311455e-06,
+      "loss": 0.2775,
+      "step": 6708
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.50877995987934e-06,
+      "loss": 0.2999,
+      "step": 6710
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.503333637563615e-06,
+      "loss": 0.2938,
+      "step": 6712
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.497888986608308e-06,
+      "loss": 0.2446,
+      "step": 6714
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.492446009037131e-06,
+      "loss": 0.2742,
+      "step": 6716
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.4870047068731835e-06,
+      "loss": 0.2539,
+      "step": 6718
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.481565082138937e-06,
+      "loss": 0.2274,
+      "step": 6720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.476127136856231e-06,
+      "loss": 0.202,
+      "step": 6722
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.470690873046292e-06,
+      "loss": 0.2481,
+      "step": 6724
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.465256292729722e-06,
+      "loss": 0.2206,
+      "step": 6726
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.45982339792649e-06,
+      "loss": 0.2699,
+      "step": 6728
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.454392190655942e-06,
+      "loss": 0.3208,
+      "step": 6730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.448962672936797e-06,
+      "loss": 0.2421,
+      "step": 6732
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.443534846787153e-06,
+      "loss": 0.25,
+      "step": 6734
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.43810871422446e-06,
+      "loss": 0.2705,
+      "step": 6736
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.432684277265558e-06,
+      "loss": 0.2351,
+      "step": 6738
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.427261537926646e-06,
+      "loss": 0.199,
+      "step": 6740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.421840498223301e-06,
+      "loss": 0.2363,
+      "step": 6742
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.416421160170451e-06,
+      "loss": 0.2595,
+      "step": 6744
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.411003525782415e-06,
+      "loss": 0.2256,
+      "step": 6746
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.405587597072865e-06,
+      "loss": 0.2731,
+      "step": 6748
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.400173376054836e-06,
+      "loss": 0.2907,
+      "step": 6750
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.3947608647407345e-06,
+      "loss": 0.2456,
+      "step": 6752
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.389350065142331e-06,
+      "loss": 0.2291,
+      "step": 6754
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.383940979270765e-06,
+      "loss": 0.2661,
+      "step": 6756
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.378533609136523e-06,
+      "loss": 0.265,
+      "step": 6758
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.373127956749463e-06,
+      "loss": 0.2687,
+      "step": 6760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.367724024118814e-06,
+      "loss": 0.3484,
+      "step": 6762
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.362321813253159e-06,
+      "loss": 0.2981,
+      "step": 6764
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.356921326160429e-06,
+      "loss": 0.2555,
+      "step": 6766
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.35152256484793e-06,
+      "loss": 0.2221,
+      "step": 6768
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.34612553132232e-06,
+      "loss": 0.3089,
+      "step": 6770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.340730227589619e-06,
+      "loss": 0.2841,
+      "step": 6772
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.3353366556551944e-06,
+      "loss": 0.2418,
+      "step": 6774
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.329944817523775e-06,
+      "loss": 0.2739,
+      "step": 6776
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.324554715199459e-06,
+      "loss": 0.2654,
+      "step": 6778
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.3191663506856765e-06,
+      "loss": 0.2791,
+      "step": 6780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.3137797259852246e-06,
+      "loss": 0.2821,
+      "step": 6782
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.3083948431002505e-06,
+      "loss": 0.1933,
+      "step": 6784
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.303011704032259e-06,
+      "loss": 0.3176,
+      "step": 6786
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.297630310782096e-06,
+      "loss": 0.2185,
+      "step": 6788
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.29225066534997e-06,
+      "loss": 0.2629,
+      "step": 6790
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.2868727697354295e-06,
+      "loss": 0.2144,
+      "step": 6792
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.2814966259373835e-06,
+      "loss": 0.2517,
+      "step": 6794
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.276122235954081e-06,
+      "loss": 0.2449,
+      "step": 6796
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.2707496017831235e-06,
+      "loss": 0.2553,
+      "step": 6798
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.2653787254214635e-06,
+      "loss": 0.2306,
+      "step": 6800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.260009608865387e-06,
+      "loss": 0.2614,
+      "step": 6802
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.254642254110538e-06,
+      "loss": 0.2459,
+      "step": 6804
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.249276663151903e-06,
+      "loss": 0.2402,
+      "step": 6806
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.24391283798381e-06,
+      "loss": 0.234,
+      "step": 6808
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.238550780599934e-06,
+      "loss": 0.235,
+      "step": 6810
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.233190492993292e-06,
+      "loss": 0.2659,
+      "step": 6812
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.227831977156242e-06,
+      "loss": 0.2673,
+      "step": 6814
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.222475235080487e-06,
+      "loss": 0.2721,
+      "step": 6816
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.217120268757062e-06,
+      "loss": 0.2491,
+      "step": 6818
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.211767080176351e-06,
+      "loss": 0.2637,
+      "step": 6820
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.206415671328076e-06,
+      "loss": 0.2096,
+      "step": 6822
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.201066044201291e-06,
+      "loss": 0.265,
+      "step": 6824
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.195718200784397e-06,
+      "loss": 0.3046,
+      "step": 6826
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.190372143065123e-06,
+      "loss": 0.2577,
+      "step": 6828
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.185027873030548e-06,
+      "loss": 0.2329,
+      "step": 6830
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.179685392667066e-06,
+      "loss": 0.2337,
+      "step": 6832
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.17434470396042e-06,
+      "loss": 0.2321,
+      "step": 6834
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.169005808895686e-06,
+      "loss": 0.2274,
+      "step": 6836
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.163668709457274e-06,
+      "loss": 0.2717,
+      "step": 6838
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.158333407628915e-06,
+      "loss": 0.2595,
+      "step": 6840
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.152999905393688e-06,
+      "loss": 0.2487,
+      "step": 6842
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.147668204734e-06,
+      "loss": 0.2408,
+      "step": 6844
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.142338307631573e-06,
+      "loss": 0.2333,
+      "step": 6846
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.137010216067477e-06,
+      "loss": 0.2331,
+      "step": 6848
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.131683932022103e-06,
+      "loss": 0.2547,
+      "step": 6850
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.12635945747517e-06,
+      "loss": 0.2709,
+      "step": 6852
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.12103679440573e-06,
+      "loss": 0.217,
+      "step": 6854
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.1157159447921454e-06,
+      "loss": 0.2176,
+      "step": 6856
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.110396910612127e-06,
+      "loss": 0.2093,
+      "step": 6858
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.105079693842702e-06,
+      "loss": 0.2177,
+      "step": 6860
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.09976429646021e-06,
+      "loss": 0.243,
+      "step": 6862
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.09445072044033e-06,
+      "loss": 0.2835,
+      "step": 6864
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.0891389677580595e-06,
+      "loss": 0.2847,
+      "step": 6866
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.08382904038772e-06,
+      "loss": 0.2222,
+      "step": 6868
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.078520940302944e-06,
+      "loss": 0.2324,
+      "step": 6870
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.073214669476691e-06,
+      "loss": 0.243,
+      "step": 6872
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.067910229881259e-06,
+      "loss": 0.1348,
+      "step": 6874
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.062607623488231e-06,
+      "loss": 0.1056,
+      "step": 6876
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.057306852268535e-06,
+      "loss": 0.1577,
+      "step": 6878
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.052007918192406e-06,
+      "loss": 0.0847,
+      "step": 6880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.046710823229402e-06,
+      "loss": 0.0831,
+      "step": 6882
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.041415569348387e-06,
+      "loss": 0.0941,
+      "step": 6884
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.036122158517552e-06,
+      "loss": 0.0958,
+      "step": 6886
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.0308305927043945e-06,
+      "loss": 0.0937,
+      "step": 6888
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.025540873875734e-06,
+      "loss": 0.1263,
+      "step": 6890
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.020253003997698e-06,
+      "loss": 0.0839,
+      "step": 6892
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.01496698503573e-06,
+      "loss": 0.0918,
+      "step": 6894
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.009682818954581e-06,
+      "loss": 0.0851,
+      "step": 6896
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.004400507718323e-06,
+      "loss": 0.0921,
+      "step": 6898
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.999120053290321e-06,
+      "loss": 0.0845,
+      "step": 6900
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.993841457633268e-06,
+      "loss": 0.0714,
+      "step": 6902
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.988564722709157e-06,
+      "loss": 0.0756,
+      "step": 6904
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9832898504792915e-06,
+      "loss": 0.0851,
+      "step": 6906
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.978016842904282e-06,
+      "loss": 0.0991,
+      "step": 6908
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9727457019440465e-06,
+      "loss": 0.1137,
+      "step": 6910
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.967476429557811e-06,
+      "loss": 0.1178,
+      "step": 6912
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9622090277041e-06,
+      "loss": 0.1066,
+      "step": 6914
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.95694349834075e-06,
+      "loss": 0.0818,
+      "step": 6916
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9516798434249e-06,
+      "loss": 0.0834,
+      "step": 6918
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9464180649129915e-06,
+      "loss": 0.1097,
+      "step": 6920
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9411581647607665e-06,
+      "loss": 0.0644,
+      "step": 6922
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.935900144923273e-06,
+      "loss": 0.0928,
+      "step": 6924
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.930644007354861e-06,
+      "loss": 0.0993,
+      "step": 6926
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.92538975400917e-06,
+      "loss": 0.0951,
+      "step": 6928
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.920137386839151e-06,
+      "loss": 0.0831,
+      "step": 6930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.91488690779705e-06,
+      "loss": 0.1027,
+      "step": 6932
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.909638318834415e-06,
+      "loss": 0.098,
+      "step": 6934
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.904391621902076e-06,
+      "loss": 0.1186,
+      "step": 6936
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.899146818950183e-06,
+      "loss": 0.1091,
+      "step": 6938
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.893903911928167e-06,
+      "loss": 0.11,
+      "step": 6940
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8886629027847606e-06,
+      "loss": 0.0766,
+      "step": 6942
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8834237934679805e-06,
+      "loss": 0.0779,
+      "step": 6944
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.87818658592515e-06,
+      "loss": 0.1004,
+      "step": 6946
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.872951282102878e-06,
+      "loss": 0.1642,
+      "step": 6948
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.867717883947074e-06,
+      "loss": 0.0931,
+      "step": 6950
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.862486393402922e-06,
+      "loss": 0.084,
+      "step": 6952
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.857256812414918e-06,
+      "loss": 0.083,
+      "step": 6954
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.85202914292684e-06,
+      "loss": 0.081,
+      "step": 6956
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.846803386881745e-06,
+      "loss": 0.1056,
+      "step": 6958
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.841579546221994e-06,
+      "loss": 0.0955,
+      "step": 6960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.8363576228892285e-06,
+      "loss": 0.0805,
+      "step": 6962
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.831137618824382e-06,
+      "loss": 0.0758,
+      "step": 6964
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.825919535967665e-06,
+      "loss": 0.0774,
+      "step": 6966
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.8207033762585795e-06,
+      "loss": 0.0846,
+      "step": 6968
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.815489141635926e-06,
+      "loss": 0.0918,
+      "step": 6970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.810276834037765e-06,
+      "loss": 0.0768,
+      "step": 6972
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.805066455401455e-06,
+      "loss": 0.0805,
+      "step": 6974
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.799858007663638e-06,
+      "loss": 0.1012,
+      "step": 6976
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.794651492760237e-06,
+      "loss": 0.0895,
+      "step": 6978
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.78944691262645e-06,
+      "loss": 0.0837,
+      "step": 6980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.784244269196763e-06,
+      "loss": 0.0982,
+      "step": 6982
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.779043564404936e-06,
+      "loss": 0.0869,
+      "step": 6984
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.773844800184025e-06,
+      "loss": 0.0839,
+      "step": 6986
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.7686479784663405e-06,
+      "loss": 0.0891,
+      "step": 6988
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.763453101183489e-06,
+      "loss": 0.0871,
+      "step": 6990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.7582601702663436e-06,
+      "loss": 0.1256,
+      "step": 6992
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.753069187645064e-06,
+      "loss": 0.0808,
+      "step": 6994
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.747880155249075e-06,
+      "loss": 0.0835,
+      "step": 6996
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.742693075007082e-06,
+      "loss": 0.0871,
+      "step": 6998
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.737507948847067e-06,
+      "loss": 0.123,
+      "step": 7000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.7323247786962826e-06,
+      "loss": 0.1159,
+      "step": 7002
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.7271435664812535e-06,
+      "loss": 0.0862,
+      "step": 7004
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.721964314127781e-06,
+      "loss": 0.0962,
+      "step": 7006
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.716787023560936e-06,
+      "loss": 0.0739,
+      "step": 7008
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.71419911454888e-06,
+      "loss": 0.1747,
+      "step": 7010
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.70902477026988e-06,
+      "loss": 0.0984,
+      "step": 7012
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.703852392586988e-06,
+      "loss": 0.0973,
+      "step": 7014
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6986819834227185e-06,
+      "loss": 0.11,
+      "step": 7016
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.693513544698855e-06,
+      "loss": 0.0881,
+      "step": 7018
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.688347078336448e-06,
+      "loss": 0.0826,
+      "step": 7020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.683182586255815e-06,
+      "loss": 0.1163,
+      "step": 7022
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.6780200703765435e-06,
+      "loss": 0.0751,
+      "step": 7024
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.672859532617475e-06,
+      "loss": 0.0823,
+      "step": 7026
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.667700974896728e-06,
+      "loss": 0.088,
+      "step": 7028
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.662544399131681e-06,
+      "loss": 0.0846,
+      "step": 7030
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.657389807238973e-06,
+      "loss": 0.0752,
+      "step": 7032
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6522372011345095e-06,
+      "loss": 0.0845,
+      "step": 7034
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.647086582733457e-06,
+      "loss": 0.0884,
+      "step": 7036
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.641937953950244e-06,
+      "loss": 0.1042,
+      "step": 7038
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.636791316698554e-06,
+      "loss": 0.1038,
+      "step": 7040
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.631646672891333e-06,
+      "loss": 0.0858,
+      "step": 7042
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.626504024440793e-06,
+      "loss": 0.0856,
+      "step": 7044
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.6213633732584e-06,
+      "loss": 0.1068,
+      "step": 7046
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.616224721254863e-06,
+      "loss": 0.1019,
+      "step": 7048
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.611088070340176e-06,
+      "loss": 0.0695,
+      "step": 7050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.605953422423569e-06,
+      "loss": 0.1211,
+      "step": 7052
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.600820779413536e-06,
+      "loss": 0.0734,
+      "step": 7054
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.595690143217818e-06,
+      "loss": 0.092,
+      "step": 7056
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.590561515743415e-06,
+      "loss": 0.1033,
+      "step": 7058
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.585434898896584e-06,
+      "loss": 0.0818,
+      "step": 7060
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.580310294582832e-06,
+      "loss": 0.0835,
+      "step": 7062
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.575187704706906e-06,
+      "loss": 0.1017,
+      "step": 7064
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.5700671311728286e-06,
+      "loss": 0.0847,
+      "step": 7066
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.564948575883857e-06,
+      "loss": 0.077,
+      "step": 7068
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.559832040742493e-06,
+      "loss": 0.1003,
+      "step": 7070
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.5547175276505005e-06,
+      "loss": 0.081,
+      "step": 7072
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.5496050385088875e-06,
+      "loss": 0.1029,
+      "step": 7074
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.54449457521791e-06,
+      "loss": 0.101,
+      "step": 7076
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.5393861396770655e-06,
+      "loss": 0.0715,
+      "step": 7078
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.534279733785099e-06,
+      "loss": 0.1009,
+      "step": 7080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.5291753594400165e-06,
+      "loss": 0.1222,
+      "step": 7082
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.524073018539047e-06,
+      "loss": 0.0802,
+      "step": 7084
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.518972712978675e-06,
+      "loss": 0.0698,
+      "step": 7086
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.513874444654628e-06,
+      "loss": 0.0728,
+      "step": 7088
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.508778215461878e-06,
+      "loss": 0.0752,
+      "step": 7090
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.503684027294629e-06,
+      "loss": 0.1107,
+      "step": 7092
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.498591882046339e-06,
+      "loss": 0.0985,
+      "step": 7094
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.493501781609695e-06,
+      "loss": 0.0943,
+      "step": 7096
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.488413727876642e-06,
+      "loss": 0.1051,
+      "step": 7098
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.483327722738343e-06,
+      "loss": 0.0844,
+      "step": 7100
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.478243768085211e-06,
+      "loss": 0.0848,
+      "step": 7102
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.473161865806895e-06,
+      "loss": 0.0843,
+      "step": 7104
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.468082017792286e-06,
+      "loss": 0.0897,
+      "step": 7106
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.463004225929497e-06,
+      "loss": 0.0961,
+      "step": 7108
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.4579284921058915e-06,
+      "loss": 0.0702,
+      "step": 7110
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.452854818208063e-06,
+      "loss": 0.111,
+      "step": 7112
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.447783206121838e-06,
+      "loss": 0.1257,
+      "step": 7114
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.442713657732279e-06,
+      "loss": 0.1174,
+      "step": 7116
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.437646174923679e-06,
+      "loss": 0.0687,
+      "step": 7118
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.432580759579569e-06,
+      "loss": 0.0939,
+      "step": 7120
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.427517413582699e-06,
+      "loss": 0.0874,
+      "step": 7122
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.422456138815063e-06,
+      "loss": 0.09,
+      "step": 7124
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.417396937157878e-06,
+      "loss": 0.1,
+      "step": 7126
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.412339810491594e-06,
+      "loss": 0.1198,
+      "step": 7128
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.407284760695888e-06,
+      "loss": 0.0748,
+      "step": 7130
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.402231789649665e-06,
+      "loss": 0.0967,
+      "step": 7132
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.397180899231061e-06,
+      "loss": 0.0955,
+      "step": 7134
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.392132091317427e-06,
+      "loss": 0.091,
+      "step": 7136
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.387085367785354e-06,
+      "loss": 0.0773,
+      "step": 7138
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.382040730510652e-06,
+      "loss": 0.0964,
+      "step": 7140
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.376998181368359e-06,
+      "loss": 0.0945,
+      "step": 7142
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.3719577222327235e-06,
+      "loss": 0.1093,
+      "step": 7144
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.366919354977238e-06,
+      "loss": 0.0902,
+      "step": 7146
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.361883081474605e-06,
+      "loss": 0.1114,
+      "step": 7148
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.356848903596752e-06,
+      "loss": 0.0822,
+      "step": 7150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.35181682321482e-06,
+      "loss": 0.0627,
+      "step": 7152
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.346786842199182e-06,
+      "loss": 0.0768,
+      "step": 7154
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.341758962419423e-06,
+      "loss": 0.0963,
+      "step": 7156
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.336733185744355e-06,
+      "loss": 0.1251,
+      "step": 7158
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.331709514041992e-06,
+      "loss": 0.0819,
+      "step": 7160
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.326687949179585e-06,
+      "loss": 0.094,
+      "step": 7162
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.321668493023596e-06,
+      "loss": 0.0951,
+      "step": 7164
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.316651147439692e-06,
+      "loss": 0.0938,
+      "step": 7166
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.311635914292768e-06,
+      "loss": 0.084,
+      "step": 7168
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.306622795446929e-06,
+      "loss": 0.0725,
+      "step": 7170
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.3016117927654984e-06,
+      "loss": 0.0741,
+      "step": 7172
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.296602908111005e-06,
+      "loss": 0.078,
+      "step": 7174
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.291596143345193e-06,
+      "loss": 0.0914,
+      "step": 7176
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.2865915003290315e-06,
+      "loss": 0.0761,
+      "step": 7178
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.28158898092268e-06,
+      "loss": 0.0865,
+      "step": 7180
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.276588586985521e-06,
+      "loss": 0.0842,
+      "step": 7182
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.2715903203761465e-06,
+      "loss": 0.0898,
+      "step": 7184
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.266594182952358e-06,
+      "loss": 0.1176,
+      "step": 7186
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.261600176571159e-06,
+      "loss": 0.0698,
+      "step": 7188
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.2566083030887676e-06,
+      "loss": 0.0881,
+      "step": 7190
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.251618564360603e-06,
+      "loss": 0.0812,
+      "step": 7192
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.246630962241307e-06,
+      "loss": 0.0741,
+      "step": 7194
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.241645498584704e-06,
+      "loss": 0.0709,
+      "step": 7196
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.23666217524384e-06,
+      "loss": 0.0889,
+      "step": 7198
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.231680994070957e-06,
+      "loss": 0.1373,
+      "step": 7200
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.226701956917511e-06,
+      "loss": 0.0731,
+      "step": 7202
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.221725065634144e-06,
+      "loss": 0.0732,
+      "step": 7204
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.216750322070717e-06,
+      "loss": 0.1018,
+      "step": 7206
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.211777728076286e-06,
+      "loss": 0.0965,
+      "step": 7208
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.206807285499107e-06,
+      "loss": 0.1021,
+      "step": 7210
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.201838996186639e-06,
+      "loss": 0.0914,
+      "step": 7212
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.196872861985538e-06,
+      "loss": 0.1041,
+      "step": 7214
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1919088847416655e-06,
+      "loss": 0.1021,
+      "step": 7216
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1869470663000664e-06,
+      "loss": 0.0873,
+      "step": 7218
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.181987408505e-06,
+      "loss": 0.1084,
+      "step": 7220
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.177029913199914e-06,
+      "loss": 0.0911,
+      "step": 7222
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.172074582227452e-06,
+      "loss": 0.1053,
+      "step": 7224
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.167121417429457e-06,
+      "loss": 0.0892,
+      "step": 7226
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.162170420646963e-06,
+      "loss": 0.0951,
+      "step": 7228
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.157221593720206e-06,
+      "loss": 0.0988,
+      "step": 7230
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.152274938488602e-06,
+      "loss": 0.0925,
+      "step": 7232
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1473304567907675e-06,
+      "loss": 0.0873,
+      "step": 7234
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1423881504645145e-06,
+      "loss": 0.0952,
+      "step": 7236
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.1374480213468405e-06,
+      "loss": 0.1006,
+      "step": 7238
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.132510071273938e-06,
+      "loss": 0.0981,
+      "step": 7240
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.127574302081185e-06,
+      "loss": 0.1,
+      "step": 7242
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.122640715603154e-06,
+      "loss": 0.0898,
+      "step": 7244
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.117709313673607e-06,
+      "loss": 0.0901,
+      "step": 7246
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.112780098125481e-06,
+      "loss": 0.0962,
+      "step": 7248
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.107853070790917e-06,
+      "loss": 0.076,
+      "step": 7250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.102928233501235e-06,
+      "loss": 0.0729,
+      "step": 7252
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.098005588086943e-06,
+      "loss": 0.0746,
+      "step": 7254
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.093085136377724e-06,
+      "loss": 0.0779,
+      "step": 7256
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.0881668802024665e-06,
+      "loss": 0.0891,
+      "step": 7258
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.08325082138923e-06,
+      "loss": 0.1064,
+      "step": 7260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.078336961765251e-06,
+      "loss": 0.0844,
+      "step": 7262
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.073425303156959e-06,
+      "loss": 0.0809,
+      "step": 7264
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.068515847389964e-06,
+      "loss": 0.0743,
+      "step": 7266
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.063608596289059e-06,
+      "loss": 0.0851,
+      "step": 7268
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.058703551678206e-06,
+      "loss": 0.1032,
+      "step": 7270
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.053800715380555e-06,
+      "loss": 0.0951,
+      "step": 7272
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.048900089218442e-06,
+      "loss": 0.0948,
+      "step": 7274
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.044001675013377e-06,
+      "loss": 0.1051,
+      "step": 7276
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.039105474586036e-06,
+      "loss": 0.1069,
+      "step": 7278
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.034211489756287e-06,
+      "loss": 0.0702,
+      "step": 7280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.029319722343168e-06,
+      "loss": 0.0943,
+      "step": 7282
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.024430174164897e-06,
+      "loss": 0.0811,
+      "step": 7284
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.01954284703886e-06,
+      "loss": 0.0793,
+      "step": 7286
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.014657742781619e-06,
+      "loss": 0.0888,
+      "step": 7288
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.009774863208924e-06,
+      "loss": 0.1177,
+      "step": 7290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.0048942101356756e-06,
+      "loss": 0.1123,
+      "step": 7292
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.000015785375961e-06,
+      "loss": 0.1745,
+      "step": 7294
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9951395907430375e-06,
+      "loss": 0.0676,
+      "step": 7296
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9902656280493346e-06,
+      "loss": 0.088,
+      "step": 7298
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9853938991064415e-06,
+      "loss": 0.1018,
+      "step": 7300
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.980524405725129e-06,
+      "loss": 0.0761,
+      "step": 7302
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.975657149715334e-06,
+      "loss": 0.1132,
+      "step": 7304
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.97079213288616e-06,
+      "loss": 0.0897,
+      "step": 7306
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9659293570458805e-06,
+      "loss": 0.0794,
+      "step": 7308
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.961068824001933e-06,
+      "loss": 0.1033,
+      "step": 7310
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.956210535560928e-06,
+      "loss": 0.0779,
+      "step": 7312
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.95135449352863e-06,
+      "loss": 0.0828,
+      "step": 7314
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.946500699709977e-06,
+      "loss": 0.1089,
+      "step": 7316
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.94164915590907e-06,
+      "loss": 0.0987,
+      "step": 7318
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.936799863929175e-06,
+      "loss": 0.1017,
+      "step": 7320
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.931952825572718e-06,
+      "loss": 0.078,
+      "step": 7322
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.92710804264129e-06,
+      "loss": 0.0925,
+      "step": 7324
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9222655169356415e-06,
+      "loss": 0.0678,
+      "step": 7326
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.917425250255688e-06,
+      "loss": 0.0828,
+      "step": 7328
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.912587244400496e-06,
+      "loss": 0.1024,
+      "step": 7330
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.907751501168302e-06,
+      "loss": 0.1078,
+      "step": 7332
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.9029180223564966e-06,
+      "loss": 0.0955,
+      "step": 7334
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.89808680976163e-06,
+      "loss": 0.0859,
+      "step": 7336
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.89325786517941e-06,
+      "loss": 0.098,
+      "step": 7338
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.8884311904047e-06,
+      "loss": 0.106,
+      "step": 7340
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.883606787231525e-06,
+      "loss": 0.1019,
+      "step": 7342
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.878784657453055e-06,
+      "loss": 0.0888,
+      "step": 7344
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.873964802861625e-06,
+      "loss": 0.0922,
+      "step": 7346
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.86914722524872e-06,
+      "loss": 0.0728,
+      "step": 7348
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.864331926404983e-06,
+      "loss": 0.0946,
+      "step": 7350
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.859518908120197e-06,
+      "loss": 0.0851,
+      "step": 7352
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.854708172183317e-06,
+      "loss": 0.0899,
+      "step": 7354
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.849899720382439e-06,
+      "loss": 0.0919,
+      "step": 7356
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.845093554504804e-06,
+      "loss": 0.1072,
+      "step": 7358
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.840289676336814e-06,
+      "loss": 0.0816,
+      "step": 7360
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.835488087664016e-06,
+      "loss": 0.0777,
+      "step": 7362
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.830688790271109e-06,
+      "loss": 0.0896,
+      "step": 7364
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.825891785941933e-06,
+      "loss": 0.0953,
+      "step": 7366
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.821097076459479e-06,
+      "loss": 0.0936,
+      "step": 7368
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.816304663605895e-06,
+      "loss": 0.0792,
+      "step": 7370
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8115145491624673e-06,
+      "loss": 0.0767,
+      "step": 7372
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8067267349096195e-06,
+      "loss": 0.088,
+      "step": 7374
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8019412226269316e-06,
+      "loss": 0.0928,
+      "step": 7376
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.797158014093125e-06,
+      "loss": 0.0846,
+      "step": 7378
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.792377111086071e-06,
+      "loss": 0.0742,
+      "step": 7380
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7875985153827667e-06,
+      "loss": 0.0871,
+      "step": 7382
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.782822228759364e-06,
+      "loss": 0.0787,
+      "step": 7384
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7780482529911655e-06,
+      "loss": 0.0811,
+      "step": 7386
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7732765898525938e-06,
+      "loss": 0.0809,
+      "step": 7388
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7685072411172276e-06,
+      "loss": 0.0912,
+      "step": 7390
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7637402085577778e-06,
+      "loss": 0.1052,
+      "step": 7392
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.758975493946101e-06,
+      "loss": 0.101,
+      "step": 7394
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7542130990531824e-06,
+      "loss": 0.0712,
+      "step": 7396
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7494530256491534e-06,
+      "loss": 0.1161,
+      "step": 7398
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7446952755032797e-06,
+      "loss": 0.0923,
+      "step": 7400
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7399398503839646e-06,
+      "loss": 0.1169,
+      "step": 7402
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.7351867520587448e-06,
+      "loss": 0.0859,
+      "step": 7404
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.730435982294295e-06,
+      "loss": 0.1345,
+      "step": 7406
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7256875428564245e-06,
+      "loss": 0.0732,
+      "step": 7408
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7209414355100704e-06,
+      "loss": 0.0806,
+      "step": 7410
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7161976620193097e-06,
+      "loss": 0.1039,
+      "step": 7412
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7114562241473494e-06,
+      "loss": 0.0791,
+      "step": 7414
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7067171236565293e-06,
+      "loss": 0.0903,
+      "step": 7416
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7019803623083204e-06,
+      "loss": 0.085,
+      "step": 7418
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.697245941863323e-06,
+      "loss": 0.0637,
+      "step": 7420
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6925138640812687e-06,
+      "loss": 0.0948,
+      "step": 7422
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6877841307210193e-06,
+      "loss": 0.064,
+      "step": 7424
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6830567435405597e-06,
+      "loss": 0.0825,
+      "step": 7426
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.678331704297008e-06,
+      "loss": 0.0937,
+      "step": 7428
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6736090147466076e-06,
+      "loss": 0.0994,
+      "step": 7430
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.668888676644732e-06,
+      "loss": 0.0763,
+      "step": 7432
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6641706917458753e-06,
+      "loss": 0.0716,
+      "step": 7434
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6594550618036596e-06,
+      "loss": 0.0875,
+      "step": 7436
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.654741788570837e-06,
+      "loss": 0.0888,
+      "step": 7438
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6500308737992707e-06,
+      "loss": 0.0881,
+      "step": 7440
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6453223192399566e-06,
+      "loss": 0.092,
+      "step": 7442
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.640616126643014e-06,
+      "loss": 0.1265,
+      "step": 7444
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6359122977576835e-06,
+      "loss": 0.087,
+      "step": 7446
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.631210834332316e-06,
+      "loss": 0.0672,
+      "step": 7448
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.626511738114403e-06,
+      "loss": 0.1038,
+      "step": 7450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.621815010850546e-06,
+      "loss": 0.1261,
+      "step": 7452
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.617120654286458e-06,
+      "loss": 0.0838,
+      "step": 7454
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.612428670166983e-06,
+      "loss": 0.0756,
+      "step": 7456
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6077390602360783e-06,
+      "loss": 0.0992,
+      "step": 7458
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6030518262368196e-06,
+      "loss": 0.1042,
+      "step": 7460
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5983669699114033e-06,
+      "loss": 0.0893,
+      "step": 7462
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.593684493001126e-06,
+      "loss": 0.0978,
+      "step": 7464
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5890043972464227e-06,
+      "loss": 0.096,
+      "step": 7466
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.584326684386833e-06,
+      "loss": 0.0904,
+      "step": 7468
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5796513561610035e-06,
+      "loss": 0.0982,
+      "step": 7470
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5749784143067046e-06,
+      "loss": 0.0843,
+      "step": 7472
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5703078605608155e-06,
+      "loss": 0.0924,
+      "step": 7474
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.565639696659332e-06,
+      "loss": 0.0735,
+      "step": 7476
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.560973924337352e-06,
+      "loss": 0.0835,
+      "step": 7478
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5563105453290915e-06,
+      "loss": 0.0874,
+      "step": 7480
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5516495613678847e-06,
+      "loss": 0.0805,
+      "step": 7482
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5469909741861573e-06,
+      "loss": 0.0871,
+      "step": 7484
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5423347855154588e-06,
+      "loss": 0.0954,
+      "step": 7486
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.53768099708644e-06,
+      "loss": 0.0774,
+      "step": 7488
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5330296106288664e-06,
+      "loss": 0.09,
+      "step": 7490
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.528380627871599e-06,
+      "loss": 0.0926,
+      "step": 7492
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5237340505426177e-06,
+      "loss": 0.0923,
+      "step": 7494
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5190898803689998e-06,
+      "loss": 0.0818,
+      "step": 7496
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.514448119076933e-06,
+      "loss": 0.102,
+      "step": 7498
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.5098087683917094e-06,
+      "loss": 0.0905,
+      "step": 7500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.50517183003772e-06,
+      "loss": 0.0976,
+      "step": 7502
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.500537305738466e-06,
+      "loss": 0.0862,
+      "step": 7504
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.4959051972165503e-06,
+      "loss": 0.0766,
+      "step": 7506
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.491275506193669e-06,
+      "loss": 0.0869,
+      "step": 7508
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.486648234390629e-06,
+      "loss": 0.1081,
+      "step": 7510
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4820233835273353e-06,
+      "loss": 0.1013,
+      "step": 7512
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.477400955322795e-06,
+      "loss": 0.0792,
+      "step": 7514
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4727809514951107e-06,
+      "loss": 0.0925,
+      "step": 7516
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.468163373761486e-06,
+      "loss": 0.0926,
+      "step": 7518
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.463548223838227e-06,
+      "loss": 0.0901,
+      "step": 7520
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.458935503440726e-06,
+      "loss": 0.1064,
+      "step": 7522
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4543252142834828e-06,
+      "loss": 0.076,
+      "step": 7524
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.44971735808009e-06,
+      "loss": 0.0786,
+      "step": 7526
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.445111936543236e-06,
+      "loss": 0.0871,
+      "step": 7528
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4405089513847057e-06,
+      "loss": 0.1,
+      "step": 7530
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4359084043153745e-06,
+      "loss": 0.078,
+      "step": 7532
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4313102970452193e-06,
+      "loss": 0.0943,
+      "step": 7534
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4267146312832978e-06,
+      "loss": 0.1134,
+      "step": 7536
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4221214087377707e-06,
+      "loss": 0.1024,
+      "step": 7538
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4175306311158874e-06,
+      "loss": 0.0982,
+      "step": 7540
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.4129423001239913e-06,
+      "loss": 0.0985,
+      "step": 7542
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4083564174675034e-06,
+      "loss": 0.1023,
+      "step": 7544
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.403772984850956e-06,
+      "loss": 0.0943,
+      "step": 7546
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3991920039779558e-06,
+      "loss": 0.097,
+      "step": 7548
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.394613476551204e-06,
+      "loss": 0.0887,
+      "step": 7550
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3900374042724828e-06,
+      "loss": 0.0939,
+      "step": 7552
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.385463788842669e-06,
+      "loss": 0.097,
+      "step": 7554
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.380892631961725e-06,
+      "loss": 0.0831,
+      "step": 7556
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3763239353287016e-06,
+      "loss": 0.0924,
+      "step": 7558
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3717577006417214e-06,
+      "loss": 0.1077,
+      "step": 7560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3671939295980138e-06,
+      "loss": 0.0962,
+      "step": 7562
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.362632623893881e-06,
+      "loss": 0.1024,
+      "step": 7564
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3580737852247015e-06,
+      "loss": 0.0828,
+      "step": 7566
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3535174152849494e-06,
+      "loss": 0.1246,
+      "step": 7568
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3489635157681745e-06,
+      "loss": 0.098,
+      "step": 7570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.344412088367014e-06,
+      "loss": 0.0855,
+      "step": 7572
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3398631347731758e-06,
+      "loss": 0.103,
+      "step": 7574
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.3353166566774565e-06,
+      "loss": 0.0677,
+      "step": 7576
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3307726557697386e-06,
+      "loss": 0.0849,
+      "step": 7578
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3262311337389674e-06,
+      "loss": 0.0862,
+      "step": 7580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3216920922731787e-06,
+      "loss": 0.0861,
+      "step": 7582
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.317155533059483e-06,
+      "loss": 0.1041,
+      "step": 7584
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.312621457784073e-06,
+      "loss": 0.0821,
+      "step": 7586
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3080898681322047e-06,
+      "loss": 0.0938,
+      "step": 7588
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.303560765788225e-06,
+      "loss": 0.0963,
+      "step": 7590
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.2990341524355463e-06,
+      "loss": 0.1011,
+      "step": 7592
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.294510029756669e-06,
+      "loss": 0.0981,
+      "step": 7594
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.2899883994331496e-06,
+      "loss": 0.1114,
+      "step": 7596
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.2854692631456297e-06,
+      "loss": 0.0856,
+      "step": 7598
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.280952622573822e-06,
+      "loss": 0.0721,
+      "step": 7600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.276438479396515e-06,
+      "loss": 0.1038,
+      "step": 7602
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.271926835291558e-06,
+      "loss": 0.0852,
+      "step": 7604
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.2674176919358826e-06,
+      "loss": 0.1123,
+      "step": 7606
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.262911051005485e-06,
+      "loss": 0.0687,
+      "step": 7608
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.258406914175435e-06,
+      "loss": 0.0852,
+      "step": 7610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.253905283119868e-06,
+      "loss": 0.115,
+      "step": 7612
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.249406159511991e-06,
+      "loss": 0.1559,
+      "step": 7614
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2449095450240797e-06,
+      "loss": 0.1163,
+      "step": 7616
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.24041544132747e-06,
+      "loss": 0.1007,
+      "step": 7618
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2359238500925728e-06,
+      "loss": 0.0721,
+      "step": 7620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.231434772988862e-06,
+      "loss": 0.0706,
+      "step": 7622
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.226948211684876e-06,
+      "loss": 0.1254,
+      "step": 7624
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2224641678482215e-06,
+      "loss": 0.068,
+      "step": 7626
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.217982643145566e-06,
+      "loss": 0.0751,
+      "step": 7628
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2135036392426457e-06,
+      "loss": 0.0809,
+      "step": 7630
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2090271578042487e-06,
+      "loss": 0.0909,
+      "step": 7632
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2045532004942372e-06,
+      "loss": 0.1034,
+      "step": 7634
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2000817689755304e-06,
+      "loss": 0.0795,
+      "step": 7636
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.1956128649101115e-06,
+      "loss": 0.0874,
+      "step": 7638
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.191146489959014e-06,
+      "loss": 0.0986,
+      "step": 7640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.1866826457823484e-06,
+      "loss": 0.0794,
+      "step": 7642
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.1822213340392717e-06,
+      "loss": 0.0897,
+      "step": 7644
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1777625563880067e-06,
+      "loss": 0.0831,
+      "step": 7646
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1733063144858255e-06,
+      "loss": 0.0632,
+      "step": 7648
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.168852609989066e-06,
+      "loss": 0.0804,
+      "step": 7650
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.164401444553119e-06,
+      "loss": 0.0721,
+      "step": 7652
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.159952819832438e-06,
+      "loss": 0.1719,
+      "step": 7654
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1555067374805157e-06,
+      "loss": 0.075,
+      "step": 7656
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.151063199149921e-06,
+      "loss": 0.0848,
+      "step": 7658
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.146622206492268e-06,
+      "loss": 0.1108,
+      "step": 7660
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.142183761158216e-06,
+      "loss": 0.0927,
+      "step": 7662
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.13774786479749e-06,
+      "loss": 0.1113,
+      "step": 7664
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.133314519058861e-06,
+      "loss": 0.0765,
+      "step": 7666
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1288837255901596e-06,
+      "loss": 0.0827,
+      "step": 7668
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1244554860382537e-06,
+      "loss": 0.0869,
+      "step": 7670
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1200298020490704e-06,
+      "loss": 0.0899,
+      "step": 7672
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1156066752675972e-06,
+      "loss": 0.1107,
+      "step": 7674
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.11118610733785e-06,
+      "loss": 0.0934,
+      "step": 7676
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1067680999029092e-06,
+      "loss": 0.0898,
+      "step": 7678
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1023526546048964e-06,
+      "loss": 0.0966,
+      "step": 7680
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0979397730849847e-06,
+      "loss": 0.0903,
+      "step": 7682
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0935294569833953e-06,
+      "loss": 0.1151,
+      "step": 7684
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0891217079393865e-06,
+      "loss": 0.0963,
+      "step": 7686
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0847165275912694e-06,
+      "loss": 0.1,
+      "step": 7688
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0803139175764097e-06,
+      "loss": 0.0862,
+      "step": 7690
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.075913879531198e-06,
+      "loss": 0.0854,
+      "step": 7692
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0715164150910837e-06,
+      "loss": 0.2041,
+      "step": 7694
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0671215258905542e-06,
+      "loss": 0.108,
+      "step": 7696
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0627292135631426e-06,
+      "loss": 0.1018,
+      "step": 7698
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.058339479741418e-06,
+      "loss": 0.086,
+      "step": 7700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.053952326056996e-06,
+      "loss": 0.0866,
+      "step": 7702
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0495677541405335e-06,
+      "loss": 0.0776,
+      "step": 7704
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0451857656217264e-06,
+      "loss": 0.0853,
+      "step": 7706
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0408063621293103e-06,
+      "loss": 0.1037,
+      "step": 7708
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.036429545291062e-06,
+      "loss": 0.0749,
+      "step": 7710
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.032055316733797e-06,
+      "loss": 0.0924,
+      "step": 7712
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.027683678083361e-06,
+      "loss": 0.0796,
+      "step": 7714
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0233146309646454e-06,
+      "loss": 0.0879,
+      "step": 7716
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0189481770015774e-06,
+      "loss": 0.0911,
+      "step": 7718
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0145843178171175e-06,
+      "loss": 0.0957,
+      "step": 7720
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0102230550332646e-06,
+      "loss": 0.1039,
+      "step": 7722
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0058643902710494e-06,
+      "loss": 0.0978,
+      "step": 7724
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0015083251505394e-06,
+      "loss": 0.1037,
+      "step": 7726
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.9971548612908397e-06,
+      "loss": 0.0852,
+      "step": 7728
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.9928040003100755e-06,
+      "loss": 0.0989,
+      "step": 7730
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.988455743825418e-06,
+      "loss": 0.1008,
+      "step": 7732
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.984110093453063e-06,
+      "loss": 0.0895,
+      "step": 7734
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.9797670508082434e-06,
+      "loss": 0.0775,
+      "step": 7736
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.975426617505216e-06,
+      "loss": 0.0928,
+      "step": 7738
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.971088795157272e-06,
+      "loss": 0.0924,
+      "step": 7740
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.966753585376736e-06,
+      "loss": 0.0937,
+      "step": 7742
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.9624209897749488e-06,
+      "loss": 0.0859,
+      "step": 7744
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.9580910099622917e-06,
+      "loss": 0.0869,
+      "step": 7746
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.95376364754817e-06,
+      "loss": 0.112,
+      "step": 7748
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9494389041410176e-06,
+      "loss": 0.0792,
+      "step": 7750
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.945116781348284e-06,
+      "loss": 0.0911,
+      "step": 7752
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.940797280776463e-06,
+      "loss": 0.0913,
+      "step": 7754
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9364804040310656e-06,
+      "loss": 0.0859,
+      "step": 7756
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9321661527166188e-06,
+      "loss": 0.0772,
+      "step": 7758
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9278545284366855e-06,
+      "loss": 0.1016,
+      "step": 7760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.925699701935507e-06,
+      "loss": 0.2616,
+      "step": 7762
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9213920212118796e-06,
+      "loss": 0.1129,
+      "step": 7764
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.917086971527393e-06,
+      "loss": 0.0725,
+      "step": 7766
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9127845544821786e-06,
+      "loss": 0.081,
+      "step": 7768
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9084847716753962e-06,
+      "loss": 0.0849,
+      "step": 7770
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.904187624705237e-06,
+      "loss": 0.073,
+      "step": 7772
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.899893115168894e-06,
+      "loss": 0.078,
+      "step": 7774
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8956012446625883e-06,
+      "loss": 0.0961,
+      "step": 7776
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8913120147815623e-06,
+      "loss": 0.0948,
+      "step": 7778
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.887025427120077e-06,
+      "loss": 0.112,
+      "step": 7780
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.882741483271403e-06,
+      "loss": 0.105,
+      "step": 7782
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.878460184827836e-06,
+      "loss": 0.0714,
+      "step": 7784
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.874181533380689e-06,
+      "loss": 0.0795,
+      "step": 7786
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.869905530520286e-06,
+      "loss": 0.0803,
+      "step": 7788
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.865632177835972e-06,
+      "loss": 0.1127,
+      "step": 7790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8613614769161024e-06,
+      "loss": 0.1074,
+      "step": 7792
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8570934293480525e-06,
+      "loss": 0.0776,
+      "step": 7794
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8528280367182006e-06,
+      "loss": 0.0815,
+      "step": 7796
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8485653006119496e-06,
+      "loss": 0.0813,
+      "step": 7798
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8443052226137102e-06,
+      "loss": 0.0799,
+      "step": 7800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8400478043069047e-06,
+      "loss": 0.08,
+      "step": 7802
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8357930472739693e-06,
+      "loss": 0.0804,
+      "step": 7804
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.831540953096349e-06,
+      "loss": 0.0782,
+      "step": 7806
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8272915233544984e-06,
+      "loss": 0.0792,
+      "step": 7808
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8230447596278875e-06,
+      "loss": 0.0835,
+      "step": 7810
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8188006634949827e-06,
+      "loss": 0.0893,
+      "step": 7812
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.814559236533272e-06,
+      "loss": 0.0902,
+      "step": 7814
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.810320480319244e-06,
+      "loss": 0.143,
+      "step": 7816
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.806084396428399e-06,
+      "loss": 0.1043,
+      "step": 7818
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8018509864352404e-06,
+      "loss": 0.0726,
+      "step": 7820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.79762025191328e-06,
+      "loss": 0.0923,
+      "step": 7822
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7933921944350375e-06,
+      "loss": 0.0904,
+      "step": 7824
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7891668155720263e-06,
+      "loss": 0.0757,
+      "step": 7826
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.784944116894778e-06,
+      "loss": 0.0868,
+      "step": 7828
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.78072409997282e-06,
+      "loss": 0.0924,
+      "step": 7830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7765067663746916e-06,
+      "loss": 0.0949,
+      "step": 7832
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7722921176679153e-06,
+      "loss": 0.1005,
+      "step": 7834
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7680801554190394e-06,
+      "loss": 0.0816,
+      "step": 7836
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7638708811936043e-06,
+      "loss": 0.0956,
+      "step": 7838
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7596642965561425e-06,
+      "loss": 0.0988,
+      "step": 7840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.755460403070198e-06,
+      "loss": 0.1052,
+      "step": 7842
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7512592022983096e-06,
+      "loss": 0.1007,
+      "step": 7844
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7470606958020184e-06,
+      "loss": 0.0794,
+      "step": 7846
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.742864885141863e-06,
+      "loss": 0.0896,
+      "step": 7848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7386717718773715e-06,
+      "loss": 0.091,
+      "step": 7850
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.734481357567086e-06,
+      "loss": 0.0993,
+      "step": 7852
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.730293643768537e-06,
+      "loss": 0.0811,
+      "step": 7854
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7261086320382423e-06,
+      "loss": 0.0791,
+      "step": 7856
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7219263239317295e-06,
+      "loss": 0.0756,
+      "step": 7858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7177467210035146e-06,
+      "loss": 0.0744,
+      "step": 7860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.713569824807112e-06,
+      "loss": 0.0752,
+      "step": 7862
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7093956368950204e-06,
+      "loss": 0.0919,
+      "step": 7864
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.70522415881874e-06,
+      "loss": 0.1233,
+      "step": 7866
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.701055392128771e-06,
+      "loss": 0.0711,
+      "step": 7868
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6968893383745877e-06,
+      "loss": 0.0968,
+      "step": 7870
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6927259991046683e-06,
+      "loss": 0.0824,
+      "step": 7872
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.68856537586648e-06,
+      "loss": 0.0911,
+      "step": 7874
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6844074702064828e-06,
+      "loss": 0.0904,
+      "step": 7876
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.680252283670116e-06,
+      "loss": 0.0786,
+      "step": 7878
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6760998178018215e-06,
+      "loss": 0.0785,
+      "step": 7880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6719500741450188e-06,
+      "loss": 0.098,
+      "step": 7882
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.667803054242132e-06,
+      "loss": 0.1083,
+      "step": 7884
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.663658759634551e-06,
+      "loss": 0.0793,
+      "step": 7886
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6595171918626674e-06,
+      "loss": 0.1019,
+      "step": 7888
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.655378352465856e-06,
+      "loss": 0.0982,
+      "step": 7890
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6512422429824793e-06,
+      "loss": 0.0793,
+      "step": 7892
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6471088649498776e-06,
+      "loss": 0.073,
+      "step": 7894
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.642978219904383e-06,
+      "loss": 0.0827,
+      "step": 7896
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.63885030938131e-06,
+      "loss": 0.0772,
+      "step": 7898
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6347251349149572e-06,
+      "loss": 0.0963,
+      "step": 7900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6306026980386058e-06,
+      "loss": 0.0825,
+      "step": 7902
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6264830002845187e-06,
+      "loss": 0.0898,
+      "step": 7904
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.622366043183945e-06,
+      "loss": 0.0928,
+      "step": 7906
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6182518282671044e-06,
+      "loss": 0.0946,
+      "step": 7908
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6141403570632084e-06,
+      "loss": 0.0899,
+      "step": 7910
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6100316311004446e-06,
+      "loss": 0.0875,
+      "step": 7912
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6059256519059795e-06,
+      "loss": 0.0716,
+      "step": 7914
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6018224210059607e-06,
+      "loss": 0.0722,
+      "step": 7916
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5977219399255116e-06,
+      "loss": 0.0854,
+      "step": 7918
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.593624210188739e-06,
+      "loss": 0.0772,
+      "step": 7920
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5895292333187173e-06,
+      "loss": 0.0863,
+      "step": 7922
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.585437010837506e-06,
+      "loss": 0.0855,
+      "step": 7924
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.581347544266137e-06,
+      "loss": 0.0988,
+      "step": 7926
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5772608351246254e-06,
+      "loss": 0.0772,
+      "step": 7928
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5731768849319427e-06,
+      "loss": 0.0987,
+      "step": 7930
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.569095695206059e-06,
+      "loss": 0.0988,
+      "step": 7932
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5650172674639005e-06,
+      "loss": 0.0883,
+      "step": 7934
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.56094160322138e-06,
+      "loss": 0.0931,
+      "step": 7936
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.556868703993367e-06,
+      "loss": 0.1053,
+      "step": 7938
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5527985712937162e-06,
+      "loss": 0.0937,
+      "step": 7940
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.548731206635252e-06,
+      "loss": 0.0997,
+      "step": 7942
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.544666611529769e-06,
+      "loss": 0.0996,
+      "step": 7944
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5406047874880235e-06,
+      "loss": 0.0765,
+      "step": 7946
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.536545736019759e-06,
+      "loss": 0.1246,
+      "step": 7948
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.532489458633679e-06,
+      "loss": 0.1046,
+      "step": 7950
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5284359568374507e-06,
+      "loss": 0.0868,
+      "step": 7952
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.524385232137717e-06,
+      "loss": 0.1014,
+      "step": 7954
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5203372860400886e-06,
+      "loss": 0.1157,
+      "step": 7956
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.516292120049144e-06,
+      "loss": 0.0728,
+      "step": 7958
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5122497356684195e-06,
+      "loss": 0.086,
+      "step": 7960
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.508210134400424e-06,
+      "loss": 0.0955,
+      "step": 7962
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.504173317746642e-06,
+      "loss": 0.0868,
+      "step": 7964
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5001392872075026e-06,
+      "loss": 0.078,
+      "step": 7966
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.4961080442824127e-06,
+      "loss": 0.0885,
+      "step": 7968
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.4920795904697413e-06,
+      "loss": 0.0765,
+      "step": 7970
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.4880539272668224e-06,
+      "loss": 0.121,
+      "step": 7972
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.484031056169942e-06,
+      "loss": 0.093,
+      "step": 7974
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.480010978674362e-06,
+      "loss": 0.0899,
+      "step": 7976
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.4759936962742935e-06,
+      "loss": 0.082,
+      "step": 7978
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.471979210462928e-06,
+      "loss": 0.1203,
+      "step": 7980
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.467967522732393e-06,
+      "loss": 0.107,
+      "step": 7982
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.463958634573792e-06,
+      "loss": 0.0771,
+      "step": 7984
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.4599525474771833e-06,
+      "loss": 0.1161,
+      "step": 7986
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.4559492629315883e-06,
+      "loss": 0.0807,
+      "step": 7988
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4519487824249767e-06,
+      "loss": 0.0713,
+      "step": 7990
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.447951107444284e-06,
+      "loss": 0.0733,
+      "step": 7992
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4439562394754023e-06,
+      "loss": 0.0832,
+      "step": 7994
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4399641800031794e-06,
+      "loss": 0.0754,
+      "step": 7996
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.435974930511419e-06,
+      "loss": 0.1006,
+      "step": 7998
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4319884924828786e-06,
+      "loss": 0.0931,
+      "step": 8000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.428004867399277e-06,
+      "loss": 0.0861,
+      "step": 8002
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4240240567412775e-06,
+      "loss": 0.0904,
+      "step": 8004
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4200460619885046e-06,
+      "loss": 0.0997,
+      "step": 8006
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4160708846195333e-06,
+      "loss": 0.072,
+      "step": 8008
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4120985261118957e-06,
+      "loss": 0.0798,
+      "step": 8010
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.40812898794207e-06,
+      "loss": 0.0908,
+      "step": 8012
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4041622715854906e-06,
+      "loss": 0.0884,
+      "step": 8014
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.400198378516543e-06,
+      "loss": 0.1011,
+      "step": 8016
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.396237310208557e-06,
+      "loss": 0.0768,
+      "step": 8018
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.392279068133819e-06,
+      "loss": 0.0862,
+      "step": 8020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3883236537635644e-06,
+      "loss": 0.068,
+      "step": 8022
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3843710685679767e-06,
+      "loss": 0.0898,
+      "step": 8024
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.380421314016185e-06,
+      "loss": 0.0913,
+      "step": 8026
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3764743915762712e-06,
+      "loss": 0.0771,
+      "step": 8028
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3725303027152615e-06,
+      "loss": 0.0851,
+      "step": 8030
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3685890488991316e-06,
+      "loss": 0.1161,
+      "step": 8032
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3646506315927953e-06,
+      "loss": 0.088,
+      "step": 8034
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.360715052260123e-06,
+      "loss": 0.0692,
+      "step": 8036
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3567823123639233e-06,
+      "loss": 0.088,
+      "step": 8038
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3528524133659547e-06,
+      "loss": 0.0727,
+      "step": 8040
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.348925356726909e-06,
+      "loss": 0.1054,
+      "step": 8042
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3450011439064366e-06,
+      "loss": 0.0959,
+      "step": 8044
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3410797763631245e-06,
+      "loss": 0.1204,
+      "step": 8046
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3371612555544944e-06,
+      "loss": 0.0902,
+      "step": 8048
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3332455829370183e-06,
+      "loss": 0.1022,
+      "step": 8050
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.32933275996611e-06,
+      "loss": 0.1247,
+      "step": 8052
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3254227880961233e-06,
+      "loss": 0.0963,
+      "step": 8054
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3215156687803463e-06,
+      "loss": 0.0725,
+      "step": 8056
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3176114034710096e-06,
+      "loss": 0.0851,
+      "step": 8058
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.313709993619295e-06,
+      "loss": 0.0875,
+      "step": 8060
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3098114406753037e-06,
+      "loss": 0.0715,
+      "step": 8062
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3059157460880867e-06,
+      "loss": 0.09,
+      "step": 8064
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.30202291130563e-06,
+      "loss": 0.081,
+      "step": 8066
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2981329377748573e-06,
+      "loss": 0.0953,
+      "step": 8068
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.294245826941629e-06,
+      "loss": 0.0768,
+      "step": 8070
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.290361580250737e-06,
+      "loss": 0.0795,
+      "step": 8072
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.286480199145912e-06,
+      "loss": 0.0804,
+      "step": 8074
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2826016850698273e-06,
+      "loss": 0.0702,
+      "step": 8076
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.278726039464074e-06,
+      "loss": 0.1157,
+      "step": 8078
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.274853263769191e-06,
+      "loss": 0.0813,
+      "step": 8080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2709833594246445e-06,
+      "loss": 0.0652,
+      "step": 8082
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2671163278688367e-06,
+      "loss": 0.0985,
+      "step": 8084
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2632521705390953e-06,
+      "loss": 0.0879,
+      "step": 8086
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2593908888716853e-06,
+      "loss": 0.0734,
+      "step": 8088
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2555324843018033e-06,
+      "loss": 0.0966,
+      "step": 8090
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2516769582635755e-06,
+      "loss": 0.1031,
+      "step": 8092
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.247824312190057e-06,
+      "loss": 0.1395,
+      "step": 8094
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2439745475132314e-06,
+      "loss": 0.2377,
+      "step": 8096
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.240127665664018e-06,
+      "loss": 0.0943,
+      "step": 8098
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2362836680722533e-06,
+      "loss": 0.0726,
+      "step": 8100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2324425561667107e-06,
+      "loss": 0.1064,
+      "step": 8102
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.228604331375088e-06,
+      "loss": 0.1031,
+      "step": 8104
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2247689951240125e-06,
+      "loss": 0.1112,
+      "step": 8106
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.220936548839032e-06,
+      "loss": 0.0987,
+      "step": 8108
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.217106993944628e-06,
+      "loss": 0.0842,
+      "step": 8110
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.213280331864198e-06,
+      "loss": 0.0895,
+      "step": 8112
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.209456564020077e-06,
+      "loss": 0.0882,
+      "step": 8114
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.205635691833509e-06,
+      "loss": 0.0731,
+      "step": 8116
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2018177167246703e-06,
+      "loss": 0.1082,
+      "step": 8118
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.198002640112662e-06,
+      "loss": 0.0975,
+      "step": 8120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1941904634155043e-06,
+      "loss": 0.0902,
+      "step": 8122
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.190381188050141e-06,
+      "loss": 0.0888,
+      "step": 8124
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1865748154324363e-06,
+      "loss": 0.0815,
+      "step": 8126
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1827713469771793e-06,
+      "loss": 0.0802,
+      "step": 8128
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1789707840980713e-06,
+      "loss": 0.1061,
+      "step": 8130
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.17517312820774e-06,
+      "loss": 0.086,
+      "step": 8132
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1713783807177346e-06,
+      "loss": 0.1052,
+      "step": 8134
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.167586543038519e-06,
+      "loss": 0.0828,
+      "step": 8136
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1637976165794715e-06,
+      "loss": 0.0928,
+      "step": 8138
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1600116027489017e-06,
+      "loss": 0.0858,
+      "step": 8140
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1562285029540264e-06,
+      "loss": 0.0751,
+      "step": 8142
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.152448318600977e-06,
+      "loss": 0.0604,
+      "step": 8144
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1486710510948084e-06,
+      "loss": 0.0789,
+      "step": 8146
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.144896701839488e-06,
+      "loss": 0.0928,
+      "step": 8148
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1411252722379027e-06,
+      "loss": 0.0999,
+      "step": 8150
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1373567636918447e-06,
+      "loss": 0.0894,
+      "step": 8152
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1335911776020256e-06,
+      "loss": 0.0942,
+      "step": 8154
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1298285153680787e-06,
+      "loss": 0.1191,
+      "step": 8156
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1260687783885415e-06,
+      "loss": 0.0845,
+      "step": 8158
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1223119680608616e-06,
+      "loss": 0.0974,
+      "step": 8160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1185580857814047e-06,
+      "loss": 0.1029,
+      "step": 8162
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.114807132945449e-06,
+      "loss": 0.0885,
+      "step": 8164
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1110591109471824e-06,
+      "loss": 0.0811,
+      "step": 8166
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1073140211796973e-06,
+      "loss": 0.0836,
+      "step": 8168
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1035718650350023e-06,
+      "loss": 0.0803,
+      "step": 8170
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0998326439040227e-06,
+      "loss": 0.0844,
+      "step": 8172
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.096096359176577e-06,
+      "loss": 0.0796,
+      "step": 8174
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.092363012241403e-06,
+      "loss": 0.0777,
+      "step": 8176
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0886326044861426e-06,
+      "loss": 0.0797,
+      "step": 8178
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.084905137297352e-06,
+      "loss": 0.0862,
+      "step": 8180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0811806120604803e-06,
+      "loss": 0.069,
+      "step": 8182
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0774590301598964e-06,
+      "loss": 0.0793,
+      "step": 8184
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0737403929788704e-06,
+      "loss": 0.1007,
+      "step": 8186
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0700247018995767e-06,
+      "loss": 0.1296,
+      "step": 8188
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0663119583030976e-06,
+      "loss": 0.0978,
+      "step": 8190
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.062602163569417e-06,
+      "loss": 0.0889,
+      "step": 8192
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0588953190774277e-06,
+      "loss": 0.0753,
+      "step": 8194
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0551914262049166e-06,
+      "loss": 0.0833,
+      "step": 8196
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.051490486328581e-06,
+      "loss": 0.0779,
+      "step": 8198
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0477925008240195e-06,
+      "loss": 0.0825,
+      "step": 8200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0440974710657313e-06,
+      "loss": 0.0799,
+      "step": 8202
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0404053984271175e-06,
+      "loss": 0.09,
+      "step": 8204
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.036716284280481e-06,
+      "loss": 0.1076,
+      "step": 8206
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0330301299970236e-06,
+      "loss": 0.0858,
+      "step": 8208
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.029346936946849e-06,
+      "loss": 0.0893,
+      "step": 8210
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.025666706498954e-06,
+      "loss": 0.0774,
+      "step": 8212
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0219894400212424e-06,
+      "loss": 0.0754,
+      "step": 8214
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0183151388805135e-06,
+      "loss": 0.099,
+      "step": 8216
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0146438044424623e-06,
+      "loss": 0.0914,
+      "step": 8218
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.010975438071682e-06,
+      "loss": 0.111,
+      "step": 8220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0073100411316646e-06,
+      "loss": 0.089,
+      "step": 8222
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0036476149847994e-06,
+      "loss": 0.0956,
+      "step": 8224
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9999881609923645e-06,
+      "loss": 0.0714,
+      "step": 8226
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.996331680514538e-06,
+      "loss": 0.0921,
+      "step": 8228
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9926781749103963e-06,
+      "loss": 0.0844,
+      "step": 8230
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9890276455379055e-06,
+      "loss": 0.0955,
+      "step": 8232
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.98538009375392e-06,
+      "loss": 0.1081,
+      "step": 8234
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.981735520914203e-06,
+      "loss": 0.1008,
+      "step": 8236
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9780939283734002e-06,
+      "loss": 0.0734,
+      "step": 8238
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9744553174850444e-06,
+      "loss": 0.0736,
+      "step": 8240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.970819689601571e-06,
+      "loss": 0.0723,
+      "step": 8242
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9671870460743015e-06,
+      "loss": 0.0927,
+      "step": 8244
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9635573882534498e-06,
+      "loss": 0.0777,
+      "step": 8246
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9599307174881154e-06,
+      "loss": 0.0838,
+      "step": 8248
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.95630703512629e-06,
+      "loss": 0.092,
+      "step": 8250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9526863425148624e-06,
+      "loss": 0.0944,
+      "step": 8252
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.949068640999603e-06,
+      "loss": 0.0921,
+      "step": 8254
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9454539319251642e-06,
+      "loss": 0.0965,
+      "step": 8256
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.941842216635096e-06,
+      "loss": 0.0697,
+      "step": 8258
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.938233496471834e-06,
+      "loss": 0.0827,
+      "step": 8260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.934627772776699e-06,
+      "loss": 0.0996,
+      "step": 8262
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.931025046889895e-06,
+      "loss": 0.0852,
+      "step": 8264
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9274253201505143e-06,
+      "loss": 0.0896,
+      "step": 8266
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9238285938965416e-06,
+      "loss": 0.0635,
+      "step": 8268
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9202348694648343e-06,
+      "loss": 0.072,
+      "step": 8270
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.916644148191139e-06,
+      "loss": 0.0875,
+      "step": 8272
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9130564314100876e-06,
+      "loss": 0.1061,
+      "step": 8274
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.909471720455198e-06,
+      "loss": 0.0801,
+      "step": 8276
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9058900166588602e-06,
+      "loss": 0.084,
+      "step": 8278
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9023113213523569e-06,
+      "loss": 0.0764,
+      "step": 8280
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8987356358658493e-06,
+      "loss": 0.0599,
+      "step": 8282
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8951629615283784e-06,
+      "loss": 0.0771,
+      "step": 8284
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8915932996678677e-06,
+      "loss": 0.0947,
+      "step": 8286
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.88802665161112e-06,
+      "loss": 0.0865,
+      "step": 8288
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8844630186838197e-06,
+      "loss": 0.7857,
+      "step": 8290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8809024022105303e-06,
+      "loss": 0.0687,
+      "step": 8292
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.877344803514689e-06,
+      "loss": 0.0852,
+      "step": 8294
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8737902239186167e-06,
+      "loss": 0.0878,
+      "step": 8296
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.870238664743511e-06,
+      "loss": 0.0982,
+      "step": 8298
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8666901273094473e-06,
+      "loss": 0.1073,
+      "step": 8300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.863144612935377e-06,
+      "loss": 0.0826,
+      "step": 8302
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.859602122939127e-06,
+      "loss": 0.082,
+      "step": 8304
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8560626586374041e-06,
+      "loss": 0.0995,
+      "step": 8306
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.852526221345784e-06,
+      "loss": 0.0762,
+      "step": 8308
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8489928123787205e-06,
+      "loss": 0.0865,
+      "step": 8310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8454624330495442e-06,
+      "loss": 0.0863,
+      "step": 8312
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8419350846704565e-06,
+      "loss": 0.0655,
+      "step": 8314
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8384107685525333e-06,
+      "loss": 0.1284,
+      "step": 8316
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8348894860057242e-06,
+      "loss": 0.0911,
+      "step": 8318
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8313712383388515e-06,
+      "loss": 0.094,
+      "step": 8320
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8278560268596036e-06,
+      "loss": 0.1732,
+      "step": 8322
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8243438528745483e-06,
+      "loss": 0.075,
+      "step": 8324
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8208347176891217e-06,
+      "loss": 0.0731,
+      "step": 8326
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8173286226076314e-06,
+      "loss": 0.1003,
+      "step": 8328
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8138255689332462e-06,
+      "loss": 0.0782,
+      "step": 8330
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.81032555796802e-06,
+      "loss": 0.067,
+      "step": 8332
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8068285910128646e-06,
+      "loss": 0.0924,
+      "step": 8334
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8033346693675658e-06,
+      "loss": 0.075,
+      "step": 8336
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7998437943307701e-06,
+      "loss": 0.1023,
+      "step": 8338
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7963559671999996e-06,
+      "loss": 0.0917,
+      "step": 8340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7928711892716399e-06,
+      "loss": 0.0886,
+      "step": 8342
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7893894618409468e-06,
+      "loss": 0.075,
+      "step": 8344
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.785910786202032e-06,
+      "loss": 0.0785,
+      "step": 8346
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7824351636478864e-06,
+      "loss": 0.0902,
+      "step": 8348
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7789625954703603e-06,
+      "loss": 0.1022,
+      "step": 8350
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7754930829601647e-06,
+      "loss": 0.0748,
+      "step": 8352
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7720266274068798e-06,
+      "loss": 0.0702,
+      "step": 8354
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7685632300989486e-06,
+      "loss": 0.0869,
+      "step": 8356
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.765102892323679e-06,
+      "loss": 0.1204,
+      "step": 8358
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7616456153672356e-06,
+      "loss": 0.1211,
+      "step": 8360
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7581914005146482e-06,
+      "loss": 0.0807,
+      "step": 8362
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7547402490498177e-06,
+      "loss": 0.1064,
+      "step": 8364
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7512921622554925e-06,
+      "loss": 0.0823,
+      "step": 8366
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7478471414132869e-06,
+      "loss": 0.1043,
+      "step": 8368
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7444051878036793e-06,
+      "loss": 0.1059,
+      "step": 8370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7409663027060076e-06,
+      "loss": 0.0693,
+      "step": 8372
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.73753048739846e-06,
+      "loss": 0.1189,
+      "step": 8374
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.734097743158094e-06,
+      "loss": 0.0911,
+      "step": 8376
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7306680712608204e-06,
+      "loss": 0.0787,
+      "step": 8378
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7272414729814158e-06,
+      "loss": 0.0939,
+      "step": 8380
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7238179495935015e-06,
+      "loss": 0.0761,
+      "step": 8382
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7203975023695653e-06,
+      "loss": 0.0735,
+      "step": 8384
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7169801325809498e-06,
+      "loss": 0.0831,
+      "step": 8386
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7135658414978528e-06,
+      "loss": 0.0987,
+      "step": 8388
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7101546303893267e-06,
+      "loss": 0.0932,
+      "step": 8390
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7067465005232798e-06,
+      "loss": 0.0881,
+      "step": 8392
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7033414531664771e-06,
+      "loss": 0.0718,
+      "step": 8394
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6999394895845366e-06,
+      "loss": 0.0827,
+      "step": 8396
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6965406110419292e-06,
+      "loss": 0.0944,
+      "step": 8398
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6931448188019817e-06,
+      "loss": 0.0764,
+      "step": 8400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6897521141268714e-06,
+      "loss": 0.0762,
+      "step": 8402
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.686362498277626e-06,
+      "loss": 0.0777,
+      "step": 8404
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6829759725141292e-06,
+      "loss": 0.0954,
+      "step": 8406
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.679592538095116e-06,
+      "loss": 0.086,
+      "step": 8408
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6762121962781685e-06,
+      "loss": 0.0877,
+      "step": 8410
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6728349483197236e-06,
+      "loss": 0.076,
+      "step": 8412
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6694607954750663e-06,
+      "loss": 0.0818,
+      "step": 8414
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6660897389983332e-06,
+      "loss": 0.0841,
+      "step": 8416
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6627217801425044e-06,
+      "loss": 0.0722,
+      "step": 8418
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6593569201594139e-06,
+      "loss": 0.084,
+      "step": 8420
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6559951602997416e-06,
+      "loss": 0.0793,
+      "step": 8422
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6526365018130197e-06,
+      "loss": 0.0723,
+      "step": 8424
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6492809459476167e-06,
+      "loss": 0.091,
+      "step": 8426
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6459284939507614e-06,
+      "loss": 0.0755,
+      "step": 8428
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6425791470685203e-06,
+      "loss": 0.0712,
+      "step": 8430
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6392329065458111e-06,
+      "loss": 0.077,
+      "step": 8432
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.635889773626388e-06,
+      "loss": 0.0932,
+      "step": 8434
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6325497495528597e-06,
+      "loss": 0.0965,
+      "step": 8436
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6292128355666747e-06,
+      "loss": 0.0763,
+      "step": 8438
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6258790329081298e-06,
+      "loss": 0.1,
+      "step": 8440
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6225483428163536e-06,
+      "loss": 0.0785,
+      "step": 8442
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6192207665293336e-06,
+      "loss": 0.0726,
+      "step": 8444
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6158963052838927e-06,
+      "loss": 0.1237,
+      "step": 8446
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6125749603156926e-06,
+      "loss": 0.0898,
+      "step": 8448
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6092567328592401e-06,
+      "loss": 0.0773,
+      "step": 8450
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6059416241478843e-06,
+      "loss": 0.073,
+      "step": 8452
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.602629635413816e-06,
+      "loss": 0.0762,
+      "step": 8454
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5993207678880596e-06,
+      "loss": 0.093,
+      "step": 8456
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5960150228004846e-06,
+      "loss": 0.085,
+      "step": 8458
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5927124013798045e-06,
+      "loss": 0.0865,
+      "step": 8460
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5894129048535623e-06,
+      "loss": 0.1278,
+      "step": 8462
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5861165344481433e-06,
+      "loss": 0.0714,
+      "step": 8464
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.582823291388773e-06,
+      "loss": 0.0897,
+      "step": 8466
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5795331768995159e-06,
+      "loss": 0.1024,
+      "step": 8468
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5762461922032645e-06,
+      "loss": 0.1045,
+      "step": 8470
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5729623385217562e-06,
+      "loss": 0.0868,
+      "step": 8472
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5696816170755624e-06,
+      "loss": 0.0779,
+      "step": 8474
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5664040290840954e-06,
+      "loss": 0.0957,
+      "step": 8476
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5631295757655917e-06,
+      "loss": 0.0774,
+      "step": 8478
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5598582583371302e-06,
+      "loss": 0.138,
+      "step": 8480
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5565900780146247e-06,
+      "loss": 0.0789,
+      "step": 8482
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5533250360128228e-06,
+      "loss": 0.0776,
+      "step": 8484
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5500631335452976e-06,
+      "loss": 0.102,
+      "step": 8486
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5468043718244663e-06,
+      "loss": 0.1027,
+      "step": 8488
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5435487520615723e-06,
+      "loss": 0.0691,
+      "step": 8490
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.540296275466695e-06,
+      "loss": 0.1004,
+      "step": 8492
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5370469432487412e-06,
+      "loss": 0.0824,
+      "step": 8494
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5338007566154523e-06,
+      "loss": 0.1028,
+      "step": 8496
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.530557716773402e-06,
+      "loss": 0.0727,
+      "step": 8498
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5273178249279863e-06,
+      "loss": 0.1209,
+      "step": 8500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5240810822834396e-06,
+      "loss": 0.0817,
+      "step": 8502
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5208474900428228e-06,
+      "loss": 0.0927,
+      "step": 8504
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5176170494080255e-06,
+      "loss": 0.077,
+      "step": 8506
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5143897615797664e-06,
+      "loss": 0.0884,
+      "step": 8508
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5111656277575926e-06,
+      "loss": 0.0792,
+      "step": 8510
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5079446491398808e-06,
+      "loss": 0.0988,
+      "step": 8512
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.504726826923827e-06,
+      "loss": 0.0756,
+      "step": 8514
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.501512162305462e-06,
+      "loss": 0.0738,
+      "step": 8516
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4983006564796421e-06,
+      "loss": 0.0844,
+      "step": 8518
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4950923106400462e-06,
+      "loss": 0.0849,
+      "step": 8520
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4918871259791812e-06,
+      "loss": 0.0962,
+      "step": 8522
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4886851036883776e-06,
+      "loss": 0.0776,
+      "step": 8524
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4854862449577923e-06,
+      "loss": 0.059,
+      "step": 8526
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4822905509764075e-06,
+      "loss": 0.0745,
+      "step": 8528
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4790980229320206e-06,
+      "loss": 0.0844,
+      "step": 8530
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4759086620112617e-06,
+      "loss": 0.0876,
+      "step": 8532
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4727224693995812e-06,
+      "loss": 0.079,
+      "step": 8534
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4695394462812529e-06,
+      "loss": 0.0631,
+      "step": 8536
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4663595938393638e-06,
+      "loss": 0.0734,
+      "step": 8538
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.463182913255835e-06,
+      "loss": 0.1365,
+      "step": 8540
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4600094057114055e-06,
+      "loss": 0.0981,
+      "step": 8542
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4568390723856251e-06,
+      "loss": 0.0899,
+      "step": 8544
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4536719144568768e-06,
+      "loss": 0.0675,
+      "step": 8546
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4505079331023542e-06,
+      "loss": 0.1267,
+      "step": 8548
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4473471294980779e-06,
+      "loss": 0.0774,
+      "step": 8550
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4441895048188781e-06,
+      "loss": 0.0771,
+      "step": 8552
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4410350602384082e-06,
+      "loss": 0.1079,
+      "step": 8554
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.437883796929147e-06,
+      "loss": 0.086,
+      "step": 8556
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4347357160623765e-06,
+      "loss": 0.0869,
+      "step": 8558
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4315908188082061e-06,
+      "loss": 0.0923,
+      "step": 8560
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4284491063355589e-06,
+      "loss": 0.0911,
+      "step": 8562
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4253105798121725e-06,
+      "loss": 0.1227,
+      "step": 8564
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4221752404046063e-06,
+      "loss": 0.0792,
+      "step": 8566
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4190430892782247e-06,
+      "loss": 0.0908,
+      "step": 8568
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4159141275972133e-06,
+      "loss": 0.0941,
+      "step": 8570
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4127883565245793e-06,
+      "loss": 0.0733,
+      "step": 8572
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.40966577722213e-06,
+      "loss": 0.0761,
+      "step": 8574
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4065463908504962e-06,
+      "loss": 0.09,
+      "step": 8576
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4034301985691168e-06,
+      "loss": 0.3568,
+      "step": 8578
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4003172015362487e-06,
+      "loss": 0.0777,
+      "step": 8580
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.397207400908953e-06,
+      "loss": 0.0846,
+      "step": 8582
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.394100797843112e-06,
+      "loss": 0.0766,
+      "step": 8584
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3909973934934129e-06,
+      "loss": 0.0973,
+      "step": 8586
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3878971890133564e-06,
+      "loss": 0.0682,
+      "step": 8588
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.384800185555255e-06,
+      "loss": 0.0795,
+      "step": 8590
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3817063842702282e-06,
+      "loss": 0.0765,
+      "step": 8592
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3786157863082127e-06,
+      "loss": 0.1044,
+      "step": 8594
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3755283928179408e-06,
+      "loss": 0.0776,
+      "step": 8596
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3724442049469666e-06,
+      "loss": 0.0932,
+      "step": 8598
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3693632238416476e-06,
+      "loss": 0.0716,
+      "step": 8600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3662854506471502e-06,
+      "loss": 0.0853,
+      "step": 8602
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3632108865074477e-06,
+      "loss": 0.1,
+      "step": 8604
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.360139532565321e-06,
+      "loss": 0.0958,
+      "step": 8606
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3570713899623577e-06,
+      "loss": 0.0799,
+      "step": 8608
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3540064598389558e-06,
+      "loss": 0.0773,
+      "step": 8610
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.350944743334308e-06,
+      "loss": 0.0647,
+      "step": 8612
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3478862415864246e-06,
+      "loss": 0.0827,
+      "step": 8614
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.344830955732116e-06,
+      "loss": 0.0941,
+      "step": 8616
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.341778886906997e-06,
+      "loss": 0.0604,
+      "step": 8618
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.338730036245488e-06,
+      "loss": 0.1029,
+      "step": 8620
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3356844048808126e-06,
+      "loss": 0.0802,
+      "step": 8622
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3326419939449986e-06,
+      "loss": 0.1232,
+      "step": 8624
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3296028045688736e-06,
+      "loss": 0.1105,
+      "step": 8626
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3265668378820707e-06,
+      "loss": 0.0842,
+      "step": 8628
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.323534095013026e-06,
+      "loss": 0.0862,
+      "step": 8630
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3205045770889769e-06,
+      "loss": 0.1031,
+      "step": 8632
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3174782852359558e-06,
+      "loss": 0.0703,
+      "step": 8634
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3144552205788065e-06,
+      "loss": 0.0734,
+      "step": 8636
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3114353842411698e-06,
+      "loss": 0.1073,
+      "step": 8638
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3084187773454805e-06,
+      "loss": 0.1539,
+      "step": 8640
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.305405401012978e-06,
+      "loss": 0.0905,
+      "step": 8642
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3023952563637032e-06,
+      "loss": 0.0799,
+      "step": 8644
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2993883445164924e-06,
+      "loss": 0.0827,
+      "step": 8646
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2963846665889779e-06,
+      "loss": 0.0715,
+      "step": 8648
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2933842236975935e-06,
+      "loss": 0.0814,
+      "step": 8650
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2903870169575749e-06,
+      "loss": 0.0784,
+      "step": 8652
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2873930474829477e-06,
+      "loss": 0.0877,
+      "step": 8654
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.284402316386536e-06,
+      "loss": 0.0972,
+      "step": 8656
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2814148247799597e-06,
+      "loss": 0.076,
+      "step": 8658
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.278430573773638e-06,
+      "loss": 0.0713,
+      "step": 8660
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.275449564476785e-06,
+      "loss": 0.0783,
+      "step": 8662
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2724717979974043e-06,
+      "loss": 0.0685,
+      "step": 8664
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2694972754422984e-06,
+      "loss": 0.0835,
+      "step": 8666
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2665259979170697e-06,
+      "loss": 0.099,
+      "step": 8668
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2635579665261032e-06,
+      "loss": 0.0715,
+      "step": 8670
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2605931823725847e-06,
+      "loss": 0.0911,
+      "step": 8672
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2576316465584904e-06,
+      "loss": 0.0755,
+      "step": 8674
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2546733601845928e-06,
+      "loss": 0.103,
+      "step": 8676
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2517183243504493e-06,
+      "loss": 0.0805,
+      "step": 8678
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2487665401544168e-06,
+      "loss": 0.084,
+      "step": 8680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2458180086936377e-06,
+      "loss": 0.0812,
+      "step": 8682
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.242872731064051e-06,
+      "loss": 0.0991,
+      "step": 8684
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2399307083603818e-06,
+      "loss": 0.0801,
+      "step": 8686
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2369919416761467e-06,
+      "loss": 0.0794,
+      "step": 8688
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2340564321036551e-06,
+      "loss": 0.0955,
+      "step": 8690
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2311241807339979e-06,
+      "loss": 0.0727,
+      "step": 8692
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.228195188657063e-06,
+      "loss": 0.0999,
+      "step": 8694
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2252694569615242e-06,
+      "loss": 0.0817,
+      "step": 8696
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2223469867348414e-06,
+      "loss": 0.0729,
+      "step": 8698
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2194277790632657e-06,
+      "loss": 0.0829,
+      "step": 8700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2165118350318327e-06,
+      "loss": 0.0826,
+      "step": 8702
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2135991557243677e-06,
+      "loss": 0.1128,
+      "step": 8704
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.210689742223481e-06,
+      "loss": 0.0789,
+      "step": 8706
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2077835956105655e-06,
+      "loss": 0.0911,
+      "step": 8708
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.204880716965806e-06,
+      "loss": 0.0968,
+      "step": 8710
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2019811073681697e-06,
+      "loss": 0.0705,
+      "step": 8712
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1990847678954076e-06,
+      "loss": 0.0949,
+      "step": 8714
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1961916996240587e-06,
+      "loss": 0.0964,
+      "step": 8716
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1933019036294424e-06,
+      "loss": 0.0872,
+      "step": 8718
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1904153809856666e-06,
+      "loss": 0.0872,
+      "step": 8720
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1875321327656142e-06,
+      "loss": 0.0823,
+      "step": 8722
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1846521600409589e-06,
+      "loss": 0.079,
+      "step": 8724
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1817754638821544e-06,
+      "loss": 0.0813,
+      "step": 8726
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1789020453584376e-06,
+      "loss": 0.0733,
+      "step": 8728
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.17603190553782e-06,
+      "loss": 0.0694,
+      "step": 8730
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1731650454871068e-06,
+      "loss": 0.0794,
+      "step": 8732
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1703014662718782e-06,
+      "loss": 0.0888,
+      "step": 8734
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1674411689564902e-06,
+      "loss": 0.0738,
+      "step": 8736
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1645841546040848e-06,
+      "loss": 0.0944,
+      "step": 8738
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1617304242765815e-06,
+      "loss": 0.0929,
+      "step": 8740
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1588799790346804e-06,
+      "loss": 0.0898,
+      "step": 8742
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1560328199378634e-06,
+      "loss": 0.1209,
+      "step": 8744
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1531889480443803e-06,
+      "loss": 0.0933,
+      "step": 8746
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1503483644112713e-06,
+      "loss": 0.0689,
+      "step": 8748
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1475110700943514e-06,
+      "loss": 0.0644,
+      "step": 8750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1446770661482065e-06,
+      "loss": 0.0633,
+      "step": 8752
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1418463536262059e-06,
+      "loss": 0.0705,
+      "step": 8754
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1390189335804924e-06,
+      "loss": 0.1082,
+      "step": 8756
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.136194807061991e-06,
+      "loss": 0.0766,
+      "step": 8758
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1333739751203909e-06,
+      "loss": 0.0778,
+      "step": 8760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1305564388041656e-06,
+      "loss": 0.0799,
+      "step": 8762
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1277421991605675e-06,
+      "loss": 0.0989,
+      "step": 8764
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.124931257235612e-06,
+      "loss": 0.0687,
+      "step": 8766
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1221236140740954e-06,
+      "loss": 0.0789,
+      "step": 8768
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1193192707195888e-06,
+      "loss": 0.0917,
+      "step": 8770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.116518228214436e-06,
+      "loss": 0.0868,
+      "step": 8772
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1137204875997488e-06,
+      "loss": 0.0903,
+      "step": 8774
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1109260499154196e-06,
+      "loss": 0.0964,
+      "step": 8776
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1081349162001076e-06,
+      "loss": 0.0786,
+      "step": 8778
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1053470874912476e-06,
+      "loss": 0.1389,
+      "step": 8780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1025625648250437e-06,
+      "loss": 0.0829,
+      "step": 8782
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0997813492364707e-06,
+      "loss": 0.0869,
+      "step": 8784
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0970034417592767e-06,
+      "loss": 0.0813,
+      "step": 8786
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0942288434259796e-06,
+      "loss": 0.083,
+      "step": 8788
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.091457555267863e-06,
+      "loss": 0.0705,
+      "step": 8790
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0886895783149852e-06,
+      "loss": 0.4134,
+      "step": 8792
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0859249135961724e-06,
+      "loss": 0.0669,
+      "step": 8794
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.083163562139019e-06,
+      "loss": 0.0819,
+      "step": 8796
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0804055249698898e-06,
+      "loss": 0.091,
+      "step": 8798
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0776508031139154e-06,
+      "loss": 0.0881,
+      "step": 8800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.074899397594996e-06,
+      "loss": 0.0874,
+      "step": 8802
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.072151309435796e-06,
+      "loss": 0.0838,
+      "step": 8804
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0694065396577503e-06,
+      "loss": 0.0955,
+      "step": 8806
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.066665089281057e-06,
+      "loss": 0.0867,
+      "step": 8808
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0639269593246859e-06,
+      "loss": 0.0919,
+      "step": 8810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.061192150806366e-06,
+      "loss": 0.0815,
+      "step": 8812
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0584606647425977e-06,
+      "loss": 0.1046,
+      "step": 8814
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0557325021486431e-06,
+      "loss": 0.0784,
+      "step": 8816
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.053007664038528e-06,
+      "loss": 0.0646,
+      "step": 8818
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.050286151425044e-06,
+      "loss": 0.0807,
+      "step": 8820
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0475679653197478e-06,
+      "loss": 0.0745,
+      "step": 8822
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0448531067329614e-06,
+      "loss": 0.0959,
+      "step": 8824
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0421415766737596e-06,
+      "loss": 0.0689,
+      "step": 8826
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0394333761499952e-06,
+      "loss": 0.0827,
+      "step": 8828
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0367285061682741e-06,
+      "loss": 0.0718,
+      "step": 8830
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0340269677339664e-06,
+      "loss": 0.0956,
+      "step": 8832
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0313287618512014e-06,
+      "loss": 0.0693,
+      "step": 8834
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.028633889522872e-06,
+      "loss": 0.0796,
+      "step": 8836
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.025942351750634e-06,
+      "loss": 0.0755,
+      "step": 8838
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0232541495349024e-06,
+      "loss": 0.0847,
+      "step": 8840
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.020569283874846e-06,
+      "loss": 0.0692,
+      "step": 8842
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0178877557684065e-06,
+      "loss": 0.0955,
+      "step": 8844
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0152095662122762e-06,
+      "loss": 0.0794,
+      "step": 8846
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0125347162019051e-06,
+      "loss": 0.086,
+      "step": 8848
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0098632067315061e-06,
+      "loss": 0.084,
+      "step": 8850
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0071950387940498e-06,
+      "loss": 0.12,
+      "step": 8852
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0045302133812661e-06,
+      "loss": 0.086,
+      "step": 8854
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0018687314836372e-06,
+      "loss": 0.0976,
+      "step": 8856
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.992105940904062e-07,
+      "loss": 0.1034,
+      "step": 8858
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.96555802189577e-07,
+      "loss": 0.0997,
+      "step": 8860
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.93904356767903e-07,
+      "loss": 0.09,
+      "step": 8862
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.912562588108954e-07,
+      "loss": 0.1098,
+      "step": 8864
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.886115093028247e-07,
+      "loss": 0.0712,
+      "step": 8866
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.859701092267171e-07,
+      "loss": 0.0859,
+      "step": 8868
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.833320595643447e-07,
+      "loss": 0.1024,
+      "step": 8870
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.806973612962456e-07,
+      "loss": 0.0755,
+      "step": 8872
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.780660154017042e-07,
+      "loss": 0.0845,
+      "step": 8874
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.754380228587689e-07,
+      "loss": 0.0782,
+      "step": 8876
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.728133846442277e-07,
+      "loss": 0.0824,
+      "step": 8878
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.701921017336335e-07,
+      "loss": 0.0703,
+      "step": 8880
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.675741751012857e-07,
+      "loss": 0.0674,
+      "step": 8882
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.649596057202414e-07,
+      "loss": 0.0666,
+      "step": 8884
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.623483945623014e-07,
+      "loss": 0.1045,
+      "step": 8886
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.597405425980277e-07,
+      "loss": 0.1128,
+      "step": 8888
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.571360507967275e-07,
+      "loss": 0.0818,
+      "step": 8890
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.54534920126462e-07,
+      "loss": 0.0882,
+      "step": 8892
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.519371515540421e-07,
+      "loss": 0.1066,
+      "step": 8894
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.4934274604503e-07,
+      "loss": 0.0672,
+      "step": 8896
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.467517045637375e-07,
+      "loss": 0.0751,
+      "step": 8898
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.441640280732234e-07,
+      "loss": 0.0745,
+      "step": 8900
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.415797175352992e-07,
+      "loss": 0.0972,
+      "step": 8902
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.389987739105244e-07,
+      "loss": 0.0715,
+      "step": 8904
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.364211981582061e-07,
+      "loss": 0.0934,
+      "step": 8906
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.338469912364012e-07,
+      "loss": 0.0913,
+      "step": 8908
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.312761541019133e-07,
+      "loss": 0.0758,
+      "step": 8910
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.287086877102947e-07,
+      "loss": 0.0802,
+      "step": 8912
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.261445930158408e-07,
+      "loss": 0.0779,
+      "step": 8914
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.235838709715983e-07,
+      "loss": 0.0924,
+      "step": 8916
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.210265225293602e-07,
+      "loss": 0.1044,
+      "step": 8918
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.18472548639664e-07,
+      "loss": 0.0795,
+      "step": 8920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.159219502517891e-07,
+      "loss": 0.0985,
+      "step": 8922
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.133747283137695e-07,
+      "loss": 0.0805,
+      "step": 8924
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.108308837723767e-07,
+      "loss": 0.1168,
+      "step": 8926
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.082904175731311e-07,
+      "loss": 0.0989,
+      "step": 8928
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.057533306602928e-07,
+      "loss": 0.0744,
+      "step": 8930
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.032196239768687e-07,
+      "loss": 0.0793,
+      "step": 8932
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.006892984646098e-07,
+      "loss": 0.0745,
+      "step": 8934
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.981623550640129e-07,
+      "loss": 0.069,
+      "step": 8936
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.956387947143063e-07,
+      "loss": 0.0796,
+      "step": 8938
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.931186183534779e-07,
+      "loss": 0.1104,
+      "step": 8940
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.906018269182459e-07,
+      "loss": 0.0733,
+      "step": 8942
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.880884213440711e-07,
+      "loss": 0.1119,
+      "step": 8944
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.855784025651603e-07,
+      "loss": 0.0908,
+      "step": 8946
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.830717715144587e-07,
+      "loss": 0.0798,
+      "step": 8948
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.805685291236554e-07,
+      "loss": 0.1027,
+      "step": 8950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.780686763231727e-07,
+      "loss": 0.0929,
+      "step": 8952
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.755722140421796e-07,
+      "loss": 0.0909,
+      "step": 8954
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.730791432085861e-07,
+      "loss": 0.0898,
+      "step": 8956
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.705894647490353e-07,
+      "loss": 0.0896,
+      "step": 8958
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.681031795889138e-07,
+      "loss": 0.0831,
+      "step": 8960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.656202886523457e-07,
+      "loss": 0.0734,
+      "step": 8962
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.63140792862196e-07,
+      "loss": 0.0942,
+      "step": 8964
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.60664693140062e-07,
+      "loss": 0.0985,
+      "step": 8966
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.58191990406283e-07,
+      "loss": 0.0702,
+      "step": 8968
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.557226855799339e-07,
+      "loss": 0.0928,
+      "step": 8970
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.532567795788326e-07,
+      "loss": 0.0773,
+      "step": 8972
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.507942733195241e-07,
+      "loss": 0.1022,
+      "step": 8974
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.483351677172946e-07,
+      "loss": 0.0828,
+      "step": 8976
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.458794636861667e-07,
+      "loss": 0.0785,
+      "step": 8978
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.434271621389001e-07,
+      "loss": 0.0677,
+      "step": 8980
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.409782639869846e-07,
+      "loss": 0.0601,
+      "step": 8982
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.385327701406487e-07,
+      "loss": 0.0881,
+      "step": 8984
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.360906815088565e-07,
+      "loss": 0.0799,
+      "step": 8986
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.336519989993042e-07,
+      "loss": 0.0823,
+      "step": 8988
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.31216723518422e-07,
+      "loss": 0.0953,
+      "step": 8990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.287848559713762e-07,
+      "loss": 0.0829,
+      "step": 8992
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.263563972620647e-07,
+      "loss": 0.0817,
+      "step": 8994
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.239313482931155e-07,
+      "loss": 0.0827,
+      "step": 8996
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.215097099658942e-07,
+      "loss": 0.0735,
+      "step": 8998
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.190914831804952e-07,
+      "loss": 0.0944,
+      "step": 9000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.166766688357486e-07,
+      "loss": 0.1249,
+      "step": 9002
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.142652678292118e-07,
+      "loss": 0.0876,
+      "step": 9004
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.118572810571768e-07,
+      "loss": 0.0811,
+      "step": 9006
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.094527094146653e-07,
+      "loss": 0.077,
+      "step": 9008
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.070515537954282e-07,
+      "loss": 0.0727,
+      "step": 9010
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.04653815091948e-07,
+      "loss": 0.0829,
+      "step": 9012
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.022594941954387e-07,
+      "loss": 0.0832,
+      "step": 9014
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.998685919958415e-07,
+      "loss": 0.0667,
+      "step": 9016
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.974811093818291e-07,
+      "loss": 0.0882,
+      "step": 9018
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.950970472408004e-07,
+      "loss": 0.0917,
+      "step": 9020
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.927164064588855e-07,
+      "loss": 0.0865,
+      "step": 9022
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.903391879209432e-07,
+      "loss": 0.086,
+      "step": 9024
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.879653925105557e-07,
+      "loss": 0.0794,
+      "step": 9026
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.855950211100382e-07,
+      "loss": 0.068,
+      "step": 9028
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.832280746004318e-07,
+      "loss": 0.0765,
+      "step": 9030
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.808645538615034e-07,
+      "loss": 0.1033,
+      "step": 9032
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.785044597717429e-07,
+      "loss": 0.1272,
+      "step": 9034
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.761477932083772e-07,
+      "loss": 0.0842,
+      "step": 9036
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.737945550473513e-07,
+      "loss": 0.0739,
+      "step": 9038
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.714447461633346e-07,
+      "loss": 0.0706,
+      "step": 9040
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.69098367429727e-07,
+      "loss": 0.0925,
+      "step": 9042
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.667554197186499e-07,
+      "loss": 0.1118,
+      "step": 9044
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.644159039009535e-07,
+      "loss": 0.0854,
+      "step": 9046
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.620798208462054e-07,
+      "loss": 0.0728,
+      "step": 9048
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.597471714227011e-07,
+      "loss": 0.074,
+      "step": 9050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.574179564974671e-07,
+      "loss": 0.1055,
+      "step": 9052
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.550921769362385e-07,
+      "loss": 0.0737,
+      "step": 9054
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.527698336034861e-07,
+      "loss": 0.0842,
+      "step": 9056
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.504509273623973e-07,
+      "loss": 0.1104,
+      "step": 9058
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.481354590748835e-07,
+      "loss": 0.076,
+      "step": 9060
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.458234296015809e-07,
+      "loss": 0.1014,
+      "step": 9062
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.43514839801841e-07,
+      "loss": 0.0672,
+      "step": 9064
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.412096905337396e-07,
+      "loss": 0.0629,
+      "step": 9066
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.389079826540813e-07,
+      "loss": 0.0761,
+      "step": 9068
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.366097170183806e-07,
+      "loss": 0.0962,
+      "step": 9070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.343148944808764e-07,
+      "loss": 0.0746,
+      "step": 9072
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.320235158945299e-07,
+      "loss": 0.0964,
+      "step": 9074
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.29735582111023e-07,
+      "loss": 0.0709,
+      "step": 9076
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.274510939807511e-07,
+      "loss": 0.0808,
+      "step": 9078
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.251700523528337e-07,
+      "loss": 0.1035,
+      "step": 9080
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.228924580751095e-07,
+      "loss": 0.0947,
+      "step": 9082
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.206183119941346e-07,
+      "loss": 0.0662,
+      "step": 9084
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.183476149551826e-07,
+      "loss": 0.0864,
+      "step": 9086
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.160803678022476e-07,
+      "loss": 0.0674,
+      "step": 9088
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.138165713780398e-07,
+      "loss": 0.0624,
+      "step": 9090
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.115562265239861e-07,
+      "loss": 0.0858,
+      "step": 9092
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.092993340802301e-07,
+      "loss": 0.0717,
+      "step": 9094
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.070458948856351e-07,
+      "loss": 0.0739,
+      "step": 9096
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.04795909777779e-07,
+      "loss": 0.0908,
+      "step": 9098
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.02549379592955e-07,
+      "loss": 0.0893,
+      "step": 9100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.003063051661752e-07,
+      "loss": 0.0903,
+      "step": 9102
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.980666873311636e-07,
+      "loss": 0.0821,
+      "step": 9104
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.958305269203624e-07,
+      "loss": 0.0706,
+      "step": 9106
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.935978247649256e-07,
+      "loss": 0.0737,
+      "step": 9108
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.913685816947235e-07,
+      "loss": 0.086,
+      "step": 9110
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.891427985383415e-07,
+      "loss": 0.0727,
+      "step": 9112
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.869204761230775e-07,
+      "loss": 0.0557,
+      "step": 9114
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.847016152749453e-07,
+      "loss": 0.1012,
+      "step": 9116
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.824862168186696e-07,
+      "loss": 0.0779,
+      "step": 9118
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.802742815776908e-07,
+      "loss": 0.0962,
+      "step": 9120
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.780658103741577e-07,
+      "loss": 0.0695,
+      "step": 9122
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.758608040289349e-07,
+      "loss": 0.0584,
+      "step": 9124
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.736592633616001e-07,
+      "loss": 0.1338,
+      "step": 9126
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.71461189190441e-07,
+      "loss": 0.0838,
+      "step": 9128
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.692665823324529e-07,
+      "loss": 0.0833,
+      "step": 9130
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.670754436033522e-07,
+      "loss": 0.0668,
+      "step": 9132
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.648877738175597e-07,
+      "loss": 0.0781,
+      "step": 9134
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.627035737882037e-07,
+      "loss": 0.0905,
+      "step": 9136
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.605228443271305e-07,
+      "loss": 0.0647,
+      "step": 9138
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.583455862448917e-07,
+      "loss": 0.0913,
+      "step": 9140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.561718003507511e-07,
+      "loss": 0.1209,
+      "step": 9142
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.540014874526779e-07,
+      "loss": 0.0741,
+      "step": 9144
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.518346483573524e-07,
+      "loss": 0.0907,
+      "step": 9146
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.49671283870168e-07,
+      "loss": 0.0984,
+      "step": 9148
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.475113947952216e-07,
+      "loss": 0.087,
+      "step": 9150
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.453549819353188e-07,
+      "loss": 0.0788,
+      "step": 9152
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.432020460919752e-07,
+      "loss": 0.0817,
+      "step": 9154
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.410525880654118e-07,
+      "loss": 0.0879,
+      "step": 9156
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.389066086545604e-07,
+      "loss": 0.0756,
+      "step": 9158
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.367641086570531e-07,
+      "loss": 0.0907,
+      "step": 9160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.346250888692351e-07,
+      "loss": 0.107,
+      "step": 9162
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.324895500861594e-07,
+      "loss": 0.0696,
+      "step": 9164
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.303574931015766e-07,
+      "loss": 0.073,
+      "step": 9166
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.282289187079516e-07,
+      "loss": 0.085,
+      "step": 9168
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.261038276964504e-07,
+      "loss": 0.1062,
+      "step": 9170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.23982220856948e-07,
+      "loss": 0.0828,
+      "step": 9172
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.218640989780178e-07,
+      "loss": 0.0788,
+      "step": 9174
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.197494628469425e-07,
+      "loss": 0.0863,
+      "step": 9176
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.17638313249711e-07,
+      "loss": 0.0769,
+      "step": 9178
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.155306509710135e-07,
+      "loss": 0.0749,
+      "step": 9180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.134264767942422e-07,
+      "loss": 0.0854,
+      "step": 9182
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.113257915014981e-07,
+      "loss": 0.0868,
+      "step": 9184
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.092285958735811e-07,
+      "loss": 0.0874,
+      "step": 9186
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.071348906899932e-07,
+      "loss": 0.0815,
+      "step": 9188
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.050446767289442e-07,
+      "loss": 0.1003,
+      "step": 9190
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.029579547673403e-07,
+      "loss": 0.0755,
+      "step": 9192
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.008747255807956e-07,
+      "loss": 0.0787,
+      "step": 9194
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.987949899436208e-07,
+      "loss": 0.0746,
+      "step": 9196
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.967187486288317e-07,
+      "loss": 0.0913,
+      "step": 9198
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.946460024081424e-07,
+      "loss": 0.0706,
+      "step": 9200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.92576752051972e-07,
+      "loss": 0.0758,
+      "step": 9202
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.905109983294344e-07,
+      "loss": 0.0647,
+      "step": 9204
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.884487420083496e-07,
+      "loss": 0.0715,
+      "step": 9206
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.863899838552334e-07,
+      "loss": 0.0933,
+      "step": 9208
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.843347246353026e-07,
+      "loss": 0.076,
+      "step": 9210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.822829651124772e-07,
+      "loss": 0.0769,
+      "step": 9212
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.802347060493696e-07,
+      "loss": 0.0809,
+      "step": 9214
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.781899482072984e-07,
+      "loss": 0.0755,
+      "step": 9216
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.761486923462745e-07,
+      "loss": 0.0852,
+      "step": 9218
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.741109392250088e-07,
+      "loss": 0.1052,
+      "step": 9220
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.720766896009134e-07,
+      "loss": 0.0694,
+      "step": 9222
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.700459442300966e-07,
+      "loss": 0.0731,
+      "step": 9224
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.680187038673601e-07,
+      "loss": 0.0722,
+      "step": 9226
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.659949692662103e-07,
+      "loss": 0.0968,
+      "step": 9228
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.639747411788465e-07,
+      "loss": 0.1023,
+      "step": 9230
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.619580203561614e-07,
+      "loss": 0.0863,
+      "step": 9232
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.599448075477498e-07,
+      "loss": 0.0807,
+      "step": 9234
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.579351035018998e-07,
+      "loss": 0.0657,
+      "step": 9236
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.559289089655972e-07,
+      "loss": 0.0898,
+      "step": 9238
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.539262246845212e-07,
+      "loss": 0.0575,
+      "step": 9240
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.519270514030439e-07,
+      "loss": 0.081,
+      "step": 9242
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.4993138986424e-07,
+      "loss": 0.1036,
+      "step": 9244
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.479392408098749e-07,
+      "loss": 0.0942,
+      "step": 9246
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.459506049804042e-07,
+      "loss": 0.0821,
+      "step": 9248
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.439654831149832e-07,
+      "loss": 0.0995,
+      "step": 9250
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.419838759514595e-07,
+      "loss": 0.0777,
+      "step": 9252
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.400057842263762e-07,
+      "loss": 0.0836,
+      "step": 9254
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.380312086749639e-07,
+      "loss": 0.0897,
+      "step": 9256
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.360601500311502e-07,
+      "loss": 0.0791,
+      "step": 9258
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.340926090275612e-07,
+      "loss": 0.0885,
+      "step": 9260
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.32128586395505e-07,
+      "loss": 0.0764,
+      "step": 9262
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.301680828649868e-07,
+      "loss": 0.0826,
+      "step": 9264
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.28211099164706e-07,
+      "loss": 0.1059,
+      "step": 9266
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.262576360220506e-07,
+      "loss": 0.081,
+      "step": 9268
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.243076941630998e-07,
+      "loss": 0.0815,
+      "step": 9270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.223612743126261e-07,
+      "loss": 0.0768,
+      "step": 9272
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.204183771940907e-07,
+      "loss": 0.0769,
+      "step": 9274
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.184790035296483e-07,
+      "loss": 0.0928,
+      "step": 9276
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.165431540401423e-07,
+      "loss": 0.0993,
+      "step": 9278
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.146108294451046e-07,
+      "loss": 0.0974,
+      "step": 9280
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.1268203046276e-07,
+      "loss": 0.0732,
+      "step": 9282
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.107567578100214e-07,
+      "loss": 0.0702,
+      "step": 9284
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.08835012202491e-07,
+      "loss": 0.07,
+      "step": 9286
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.069167943544584e-07,
+      "loss": 0.088,
+      "step": 9288
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.050021049789055e-07,
+      "loss": 0.0886,
+      "step": 9290
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.030909447874998e-07,
+      "loss": 0.0739,
+      "step": 9292
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.011833144905998e-07,
+      "loss": 0.0912,
+      "step": 9294
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.992792147972481e-07,
+      "loss": 0.0951,
+      "step": 9296
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.973786464151797e-07,
+      "loss": 0.0713,
+      "step": 9298
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.95481610050812e-07,
+      "loss": 0.0871,
+      "step": 9300
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.935881064092518e-07,
+      "loss": 0.0919,
+      "step": 9302
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.916981361942963e-07,
+      "loss": 0.0835,
+      "step": 9304
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.898117001084235e-07,
+      "loss": 0.0844,
+      "step": 9306
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.879287988528014e-07,
+      "loss": 0.0915,
+      "step": 9308
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.860494331272825e-07,
+      "loss": 0.075,
+      "step": 9310
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.84173603630409e-07,
+      "loss": 0.0758,
+      "step": 9312
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.823013110594021e-07,
+      "loss": 0.0801,
+      "step": 9314
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.804325561101731e-07,
+      "loss": 0.0848,
+      "step": 9316
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.785673394773171e-07,
+      "loss": 0.0674,
+      "step": 9318
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.767056618541155e-07,
+      "loss": 0.1009,
+      "step": 9320
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.748475239325301e-07,
+      "loss": 0.0687,
+      "step": 9322
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.729929264032129e-07,
+      "loss": 0.0843,
+      "step": 9324
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.711418699554959e-07,
+      "loss": 0.0804,
+      "step": 9326
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.692943552773965e-07,
+      "loss": 0.0795,
+      "step": 9328
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.674503830556132e-07,
+      "loss": 0.085,
+      "step": 9330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.656099539755321e-07,
+      "loss": 0.0713,
+      "step": 9332
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.637730687212183e-07,
+      "loss": 0.0904,
+      "step": 9334
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6193972797542454e-07,
+      "loss": 0.0755,
+      "step": 9336
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.601099324195768e-07,
+      "loss": 0.0818,
+      "step": 9338
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.582836827337955e-07,
+      "loss": 0.0924,
+      "step": 9340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5646097959687665e-07,
+      "loss": 0.0753,
+      "step": 9342
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5464182368629595e-07,
+      "loss": 0.0896,
+      "step": 9344
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.528262156782137e-07,
+      "loss": 0.1111,
+      "step": 9346
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5101415624747126e-07,
+      "loss": 0.085,
+      "step": 9348
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.492056460675931e-07,
+      "loss": 0.074,
+      "step": 9350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4740068581077824e-07,
+      "loss": 0.0923,
+      "step": 9352
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.455992761479111e-07,
+      "loss": 0.0842,
+      "step": 9354
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.438014177485583e-07,
+      "loss": 0.0748,
+      "step": 9356
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.420071112809621e-07,
+      "loss": 0.0794,
+      "step": 9358
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.402163574120444e-07,
+      "loss": 0.0792,
+      "step": 9360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.384291568074095e-07,
+      "loss": 0.0868,
+      "step": 9362
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3664551013134273e-07,
+      "loss": 0.0796,
+      "step": 9364
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.348654180468004e-07,
+      "loss": 0.0837,
+      "step": 9366
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3308888121542434e-07,
+      "loss": 0.0806,
+      "step": 9368
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.313159002975342e-07,
+      "loss": 0.0853,
+      "step": 9370
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.295464759521284e-07,
+      "loss": 0.0919,
+      "step": 9372
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.277806088368808e-07,
+      "loss": 0.0873,
+      "step": 9374
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.26018299608143e-07,
+      "loss": 0.0801,
+      "step": 9376
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2425954892094776e-07,
+      "loss": 0.0803,
+      "step": 9378
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.225043574290033e-07,
+      "loss": 0.0816,
+      "step": 9380
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2075272578469216e-07,
+      "loss": 0.0756,
+      "step": 9382
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.190046546390769e-07,
+      "loss": 0.1245,
+      "step": 9384
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1726014464189667e-07,
+      "loss": 0.0821,
+      "step": 9386
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.155191964415661e-07,
+      "loss": 0.0919,
+      "step": 9388
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.137818106851754e-07,
+      "loss": 0.0999,
+      "step": 9390
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.120479880184924e-07,
+      "loss": 0.0703,
+      "step": 9392
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1031772908596056e-07,
+      "loss": 0.0968,
+      "step": 9394
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.085910345306943e-07,
+      "loss": 0.092,
+      "step": 9396
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.06867904994489e-07,
+      "loss": 0.0827,
+      "step": 9398
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0514834111781256e-07,
+      "loss": 0.0883,
+      "step": 9400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.034323435398069e-07,
+      "loss": 0.0812,
+      "step": 9402
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0171991289828847e-07,
+      "loss": 0.0837,
+      "step": 9404
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.000110498297505e-07,
+      "loss": 0.0898,
+      "step": 9406
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9830575496935697e-07,
+      "loss": 0.087,
+      "step": 9408
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.966040289509465e-07,
+      "loss": 0.1,
+      "step": 9410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9490587240703204e-07,
+      "loss": 0.0993,
+      "step": 9412
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9321128596879866e-07,
+      "loss": 0.0726,
+      "step": 9414
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.915202702661081e-07,
+      "loss": 0.0815,
+      "step": 9416
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8983282592748527e-07,
+      "loss": 0.0925,
+      "step": 9418
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.881489535801408e-07,
+      "loss": 0.0713,
+      "step": 9420
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.864686538499496e-07,
+      "loss": 0.0806,
+      "step": 9422
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8479192736145976e-07,
+      "loss": 0.0795,
+      "step": 9424
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.831187747378906e-07,
+      "loss": 0.1148,
+      "step": 9426
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.814491966011347e-07,
+      "loss": 0.0952,
+      "step": 9428
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.797831935717555e-07,
+      "loss": 0.0901,
+      "step": 9430
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7812076626899006e-07,
+      "loss": 0.0867,
+      "step": 9432
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.764619153107385e-07,
+      "loss": 0.0813,
+      "step": 9434
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.74806641313582e-07,
+      "loss": 0.0822,
+      "step": 9436
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.731549448927674e-07,
+      "loss": 0.0689,
+      "step": 9438
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.715068266622079e-07,
+      "loss": 0.0768,
+      "step": 9440
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.698622872344937e-07,
+      "loss": 0.0753,
+      "step": 9442
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.682213272208801e-07,
+      "loss": 0.0988,
+      "step": 9444
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6658394723129597e-07,
+      "loss": 0.0886,
+      "step": 9446
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6495014787433335e-07,
+      "loss": 0.0885,
+      "step": 9448
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6331992975725763e-07,
+      "loss": 0.0781,
+      "step": 9450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.616932934860073e-07,
+      "loss": 0.0659,
+      "step": 9452
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6007023966518004e-07,
+      "loss": 0.1,
+      "step": 9454
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5845076889804764e-07,
+      "loss": 0.0862,
+      "step": 9456
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.568348817865497e-07,
+      "loss": 0.0937,
+      "step": 9458
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5522257893129463e-07,
+      "loss": 0.0932,
+      "step": 9460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5361386093155426e-07,
+      "loss": 0.0836,
+      "step": 9462
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5200872838527355e-07,
+      "loss": 0.0805,
+      "step": 9464
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.504071818890598e-07,
+      "loss": 0.0805,
+      "step": 9466
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4880922203819446e-07,
+      "loss": 0.0788,
+      "step": 9468
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.47214849426617e-07,
+      "loss": 0.0941,
+      "step": 9470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4562406464693885e-07,
+      "loss": 0.0728,
+      "step": 9472
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4403686829043714e-07,
+      "loss": 0.0914,
+      "step": 9474
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4245326094705655e-07,
+      "loss": 0.0799,
+      "step": 9476
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.408732432054029e-07,
+      "loss": 0.0715,
+      "step": 9478
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.392968156527532e-07,
+      "loss": 0.0814,
+      "step": 9480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3772397887504746e-07,
+      "loss": 0.0845,
+      "step": 9482
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3615473345689155e-07,
+      "loss": 0.0819,
+      "step": 9484
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.345890799815554e-07,
+      "loss": 0.072,
+      "step": 9486
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.33027019030977e-07,
+      "loss": 0.0784,
+      "step": 9488
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3146855118575515e-07,
+      "loss": 0.0754,
+      "step": 9490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.299136770251554e-07,
+      "loss": 0.1031,
+      "step": 9492
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2836239712710547e-07,
+      "loss": 0.0857,
+      "step": 9494
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.268147120682008e-07,
+      "loss": 0.0906,
+      "step": 9496
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.25270622423699e-07,
+      "loss": 0.0836,
+      "step": 9498
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.237301287675176e-07,
+      "loss": 0.0611,
+      "step": 9500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.221932316722443e-07,
+      "loss": 0.0741,
+      "step": 9502
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.206599317091263e-07,
+      "loss": 0.0825,
+      "step": 9504
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.191302294480714e-07,
+      "loss": 0.0756,
+      "step": 9506
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.176041254576534e-07,
+      "loss": 0.0963,
+      "step": 9508
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.160816203051098e-07,
+      "loss": 0.1293,
+      "step": 9510
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.145627145563368e-07,
+      "loss": 0.1081,
+      "step": 9512
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1304740877589614e-07,
+      "loss": 0.0737,
+      "step": 9514
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1153570352700947e-07,
+      "loss": 0.1287,
+      "step": 9516
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.100275993715596e-07,
+      "loss": 0.0967,
+      "step": 9518
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0852309687009476e-07,
+      "loss": 0.0788,
+      "step": 9520
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0702219658181764e-07,
+      "loss": 0.0808,
+      "step": 9522
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.055248990645987e-07,
+      "loss": 0.0837,
+      "step": 9524
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0403120487496606e-07,
+      "loss": 0.0786,
+      "step": 9526
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.025411145681101e-07,
+      "loss": 0.0809,
+      "step": 9528
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.010546286978777e-07,
+      "loss": 0.0897,
+      "step": 9530
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9957174781678144e-07,
+      "loss": 0.0962,
+      "step": 9532
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.980924724759926e-07,
+      "loss": 0.1068,
+      "step": 9534
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9661680322533804e-07,
+      "loss": 0.0864,
+      "step": 9536
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.95144740613309e-07,
+      "loss": 0.0679,
+      "step": 9538
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.936762851870556e-07,
+      "loss": 0.0776,
+      "step": 9540
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9221143749238456e-07,
+      "loss": 0.0977,
+      "step": 9542
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9075019807376483e-07,
+      "loss": 0.094,
+      "step": 9544
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8929256747431967e-07,
+      "loss": 0.0953,
+      "step": 9546
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.878385462358368e-07,
+      "loss": 0.0852,
+      "step": 9548
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8638813489876074e-07,
+      "loss": 0.0863,
+      "step": 9550
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8494133400219003e-07,
+      "loss": 0.1003,
+      "step": 9552
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8349814408388687e-07,
+      "loss": 0.0927,
+      "step": 9554
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8205856568026657e-07,
+      "loss": 0.0956,
+      "step": 9556
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8062259932640556e-07,
+      "loss": 0.0707,
+      "step": 9558
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.791902455560369e-07,
+      "loss": 0.0866,
+      "step": 9560
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7776150490154695e-07,
+      "loss": 0.0723,
+      "step": 9562
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.763363778939887e-07,
+      "loss": 0.0876,
+      "step": 9564
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7491486506306063e-07,
+      "loss": 0.0753,
+      "step": 9566
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7349696693712455e-07,
+      "loss": 0.1173,
+      "step": 9568
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.720826840431978e-07,
+      "loss": 0.1477,
+      "step": 9570
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.706720169069543e-07,
+      "loss": 0.0705,
+      "step": 9572
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.692649660527213e-07,
+      "loss": 0.0845,
+      "step": 9574
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.67861532003485e-07,
+      "loss": 0.0697,
+      "step": 9576
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.664617152808857e-07,
+      "loss": 0.0774,
+      "step": 9578
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.650655164052196e-07,
+      "loss": 0.0885,
+      "step": 9580
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6367293589543816e-07,
+      "loss": 0.0861,
+      "step": 9582
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.622839742691474e-07,
+      "loss": 0.0736,
+      "step": 9584
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.60898632042611e-07,
+      "loss": 0.0756,
+      "step": 9586
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.595169097307415e-07,
+      "loss": 0.063,
+      "step": 9588
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5813880784711254e-07,
+      "loss": 0.1117,
+      "step": 9590
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.567643269039477e-07,
+      "loss": 0.0862,
+      "step": 9592
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.553934674121261e-07,
+      "loss": 0.0675,
+      "step": 9594
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5402622988118244e-07,
+      "loss": 0.0911,
+      "step": 9596
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.526626148193012e-07,
+      "loss": 0.1105,
+      "step": 9598
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.513026227333248e-07,
+      "loss": 0.0825,
+      "step": 9600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4994625412874673e-07,
+      "loss": 0.073,
+      "step": 9602
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.485935095097125e-07,
+      "loss": 0.0909,
+      "step": 9604
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4724438937902217e-07,
+      "loss": 0.0823,
+      "step": 9606
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.458988942381302e-07,
+      "loss": 0.064,
+      "step": 9608
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4455702458714203e-07,
+      "loss": 0.0778,
+      "step": 9610
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4321878092481323e-07,
+      "loss": 0.0998,
+      "step": 9612
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4188416374855696e-07,
+      "loss": 0.0766,
+      "step": 9614
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.405531735544342e-07,
+      "loss": 0.1379,
+      "step": 9616
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3922581083715813e-07,
+      "loss": 0.0756,
+      "step": 9618
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3790207609009407e-07,
+      "loss": 0.0767,
+      "step": 9620
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3658196980526182e-07,
+      "loss": 0.0875,
+      "step": 9622
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3526549247332997e-07,
+      "loss": 0.1072,
+      "step": 9624
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.339526445836149e-07,
+      "loss": 0.0754,
+      "step": 9626
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3264342662409068e-07,
+      "loss": 0.0879,
+      "step": 9628
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.313378390813792e-07,
+      "loss": 0.0698,
+      "step": 9630
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3003588244075003e-07,
+      "loss": 0.0852,
+      "step": 9632
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2873755718612723e-07,
+      "loss": 0.1021,
+      "step": 9634
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2744286380008473e-07,
+      "loss": 0.1633,
+      "step": 9636
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2615180276384541e-07,
+      "loss": 0.0727,
+      "step": 9638
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2486437455727984e-07,
+      "loss": 0.0682,
+      "step": 9640
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2358057965891077e-07,
+      "loss": 0.0897,
+      "step": 9642
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2230041854591433e-07,
+      "loss": 0.0738,
+      "step": 9644
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2102389169410875e-07,
+      "loss": 0.0921,
+      "step": 9646
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1975099957796564e-07,
+      "loss": 0.0647,
+      "step": 9648
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.184817426706043e-07,
+      "loss": 0.1019,
+      "step": 9650
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.172161214437929e-07,
+      "loss": 0.1191,
+      "step": 9652
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.159541363679507e-07,
+      "loss": 0.0722,
+      "step": 9654
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.146957879121403e-07,
+      "loss": 0.0756,
+      "step": 9656
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1344107654407753e-07,
+      "loss": 0.0786,
+      "step": 9658
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1219000273012602e-07,
+      "loss": 0.063,
+      "step": 9660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.109425669352927e-07,
+      "loss": 0.0744,
+      "step": 9662
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0969876962323666e-07,
+      "loss": 0.0678,
+      "step": 9664
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0845861125626364e-07,
+      "loss": 0.0675,
+      "step": 9666
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.072220922953283e-07,
+      "loss": 0.0928,
+      "step": 9668
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.059892132000274e-07,
+      "loss": 0.0842,
+      "step": 9670
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0475997442861106e-07,
+      "loss": 0.0833,
+      "step": 9672
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0353437643797158e-07,
+      "loss": 0.0844,
+      "step": 9674
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0231241968365123e-07,
+      "loss": 0.0726,
+      "step": 9676
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.010941046198378e-07,
+      "loss": 0.0652,
+      "step": 9678
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9987943169936351e-07,
+      "loss": 0.0993,
+      "step": 9680
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.986684013737117e-07,
+      "loss": 0.0837,
+      "step": 9682
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9746101409300667e-07,
+      "loss": 0.0954,
+      "step": 9684
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.962572703060206e-07,
+      "loss": 0.0841,
+      "step": 9686
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9505717046017114e-07,
+      "loss": 0.0854,
+      "step": 9688
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.938607150015237e-07,
+      "loss": 0.077,
+      "step": 9690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9266790437478588e-07,
+      "loss": 0.0817,
+      "step": 9692
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9147873902331304e-07,
+      "loss": 0.1017,
+      "step": 9694
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9029321938910382e-07,
+      "loss": 0.0742,
+      "step": 9696
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8911134591280467e-07,
+      "loss": 0.0889,
+      "step": 9698
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8793311903370082e-07,
+      "loss": 0.0817,
+      "step": 9700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8675853918972864e-07,
+      "loss": 0.0713,
+      "step": 9702
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8558760681746558e-07,
+      "loss": 0.0861,
+      "step": 9704
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8442032235213347e-07,
+      "loss": 0.0805,
+      "step": 9706
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8325668622759972e-07,
+      "loss": 0.0838,
+      "step": 9708
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8209669887637505e-07,
+      "loss": 0.0733,
+      "step": 9710
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8094036072961453e-07,
+      "loss": 0.1017,
+      "step": 9712
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7978767221711323e-07,
+      "loss": 0.0793,
+      "step": 9714
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7863863376731516e-07,
+      "loss": 0.0659,
+      "step": 9716
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.774932458073031e-07,
+      "loss": 0.109,
+      "step": 9718
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7635150876280648e-07,
+      "loss": 0.0837,
+      "step": 9720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7521342305819478e-07,
+      "loss": 0.0868,
+      "step": 9722
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.740789891164829e-07,
+      "loss": 0.0828,
+      "step": 9724
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7294820735932694e-07,
+      "loss": 0.0944,
+      "step": 9726
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.718210782070251e-07,
+      "loss": 0.0703,
+      "step": 9728
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7069760207851892e-07,
+      "loss": 0.0859,
+      "step": 9730
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.695777793913922e-07,
+      "loss": 0.0697,
+      "step": 9732
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6846161056186973e-07,
+      "loss": 0.0874,
+      "step": 9734
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6734909600482073e-07,
+      "loss": 0.0626,
+      "step": 9736
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.662402361337523e-07,
+      "loss": 0.0706,
+      "step": 9738
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6513503136081577e-07,
+      "loss": 0.1002,
+      "step": 9740
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.640334820968048e-07,
+      "loss": 0.0772,
+      "step": 9742
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.629355887511508e-07,
+      "loss": 0.0722,
+      "step": 9744
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6184135173192948e-07,
+      "loss": 0.0653,
+      "step": 9746
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6075077144585562e-07,
+      "loss": 0.1019,
+      "step": 9748
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5966384829828818e-07,
+      "loss": 0.075,
+      "step": 9750
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5858058269322074e-07,
+      "loss": 0.0968,
+      "step": 9752
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.575009750332923e-07,
+      "loss": 0.0894,
+      "step": 9754
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.56425025719783e-07,
+      "loss": 0.0855,
+      "step": 9756
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5535273515260852e-07,
+      "loss": 0.1111,
+      "step": 9758
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5428410373032775e-07,
+      "loss": 0.0672,
+      "step": 9760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.532191318501397e-07,
+      "loss": 0.0698,
+      "step": 9762
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5215781990788325e-07,
+      "loss": 0.0785,
+      "step": 9764
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5110016829803288e-07,
+      "loss": 0.0774,
+      "step": 9766
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5004617741370963e-07,
+      "loss": 0.0539,
+      "step": 9768
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4899584764666796e-07,
+      "loss": 0.0938,
+      "step": 9770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4794917938730336e-07,
+      "loss": 0.0744,
+      "step": 9772
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.469061730246524e-07,
+      "loss": 0.0664,
+      "step": 9774
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4586682894638716e-07,
+      "loss": 0.0799,
+      "step": 9776
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4483114753882088e-07,
+      "loss": 0.0751,
+      "step": 9778
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4379912918690565e-07,
+      "loss": 0.0896,
+      "step": 9780
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4277077427423015e-07,
+      "loss": 0.0777,
+      "step": 9782
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4174608318302086e-07,
+      "loss": 0.0794,
+      "step": 9784
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4072505629414646e-07,
+      "loss": 0.0678,
+      "step": 9786
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3970769398711004e-07,
+      "loss": 0.0889,
+      "step": 9788
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3869399664005356e-07,
+      "loss": 0.068,
+      "step": 9790
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3768396462975786e-07,
+      "loss": 0.0971,
+      "step": 9792
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.366775983316404e-07,
+      "loss": 0.0806,
+      "step": 9794
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3567489811975532e-07,
+      "loss": 0.1006,
+      "step": 9796
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.346758643667956e-07,
+      "loss": 0.1694,
+      "step": 9798
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3368049744408973e-07,
+      "loss": 0.084,
+      "step": 9800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3268879772160625e-07,
+      "loss": 0.0962,
+      "step": 9802
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3170076556794808e-07,
+      "loss": 0.0861,
+      "step": 9804
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3071640135035479e-07,
+      "loss": 0.0875,
+      "step": 9806
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2973570543470592e-07,
+      "loss": 0.0993,
+      "step": 9808
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2875867818551323e-07,
+      "loss": 0.0657,
+      "step": 9810
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2778531996592625e-07,
+      "loss": 0.0681,
+      "step": 9812
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2681563113773222e-07,
+      "loss": 0.0815,
+      "step": 9814
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2584961206135393e-07,
+      "loss": 0.0842,
+      "step": 9816
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2488726309584863e-07,
+      "loss": 0.0825,
+      "step": 9818
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2392858459891243e-07,
+      "loss": 0.0888,
+      "step": 9820
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.229735769268736e-07,
+      "loss": 0.1117,
+      "step": 9822
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2202224043470046e-07,
+      "loss": 0.0825,
+      "step": 9824
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.210745754759901e-07,
+      "loss": 0.0916,
+      "step": 9826
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2013058240298193e-07,
+      "loss": 0.0863,
+      "step": 9828
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1919026156654634e-07,
+      "loss": 0.0897,
+      "step": 9830
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1825361331619046e-07,
+      "loss": 0.0768,
+      "step": 9832
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1732063800005578e-07,
+      "loss": 0.0806,
+      "step": 9834
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1639133596491825e-07,
+      "loss": 0.0822,
+      "step": 9836
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1546570755619158e-07,
+      "loss": 0.0955,
+      "step": 9838
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1454375311791722e-07,
+      "loss": 0.0722,
+      "step": 9840
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1362547299277881e-07,
+      "loss": 0.0948,
+      "step": 9842
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1271086752208893e-07,
+      "loss": 0.0749,
+      "step": 9844
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1179993704579783e-07,
+      "loss": 0.1014,
+      "step": 9846
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.108926819024858e-07,
+      "loss": 0.0687,
+      "step": 9848
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0998910242937088e-07,
+      "loss": 0.0784,
+      "step": 9850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.090891989623033e-07,
+      "loss": 0.0674,
+      "step": 9852
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.081929718357666e-07,
+      "loss": 0.0837,
+      "step": 9854
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0730042138287878e-07,
+      "loss": 0.0898,
+      "step": 9856
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0641154793539111e-07,
+      "loss": 0.0916,
+      "step": 9858
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.055263518236882e-07,
+      "loss": 0.0745,
+      "step": 9860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0464483337678577e-07,
+      "loss": 0.0766,
+      "step": 9862
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0376699292233505e-07,
+      "loss": 0.0743,
+      "step": 9864
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.028928307866206e-07,
+      "loss": 0.0908,
+      "step": 9866
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0202234729455806e-07,
+      "loss": 0.076,
+      "step": 9868
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0115554276969531e-07,
+      "loss": 0.0947,
+      "step": 9870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.002924175342157e-07,
+      "loss": 0.0889,
+      "step": 9872
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.943297190893154e-08,
+      "loss": 0.1039,
+      "step": 9874
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.857720621329059e-08,
+      "loss": 0.0871,
+      "step": 9876
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.772512076536956e-08,
+      "loss": 0.0815,
+      "step": 9878
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.687671588188064e-08,
+      "loss": 0.0765,
+      "step": 9880
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.603199187816603e-08,
+      "loss": 0.0982,
+      "step": 9882
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.519094906819904e-08,
+      "loss": 0.0838,
+      "step": 9884
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.435358776458736e-08,
+      "loss": 0.0871,
+      "step": 9886
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.351990827856872e-08,
+      "loss": 0.0712,
+      "step": 9888
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.26899109200119e-08,
+      "loss": 0.0946,
+      "step": 9890
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.186359599741679e-08,
+      "loss": 0.0943,
+      "step": 9892
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.10409638179166e-08,
+      "loss": 0.088,
+      "step": 9894
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.022201468727342e-08,
+      "loss": 0.0701,
+      "step": 9896
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.940674890988266e-08,
+      "loss": 0.0954,
+      "step": 9898
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.859516678876967e-08,
+      "loss": 0.0814,
+      "step": 9900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.778726862558873e-08,
+      "loss": 0.0779,
+      "step": 9902
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.698305472062852e-08,
+      "loss": 0.1011,
+      "step": 9904
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.618252537280436e-08,
+      "loss": 0.1005,
+      "step": 9906
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.538568087966492e-08,
+      "loss": 0.0726,
+      "step": 9908
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.459252153738772e-08,
+      "loss": 0.0669,
+      "step": 9910
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.380304764078251e-08,
+      "loss": 0.0741,
+      "step": 9912
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.301725948328677e-08,
+      "loss": 0.0806,
+      "step": 9914
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.223515735697019e-08,
+      "loss": 0.0746,
+      "step": 9916
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.145674155253025e-08,
+      "loss": 0.109,
+      "step": 9918
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.068201235929663e-08,
+      "loss": 0.0952,
+      "step": 9920
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.991097006522563e-08,
+      "loss": 0.1034,
+      "step": 9922
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.914361495690581e-08,
+      "loss": 0.0833,
+      "step": 9924
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.837994731955567e-08,
+      "loss": 0.1027,
+      "step": 9926
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.761996743702149e-08,
+      "loss": 0.058,
+      "step": 9928
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.686367559177621e-08,
+      "loss": 0.0812,
+      "step": 9930
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.61110720649294e-08,
+      "loss": 0.116,
+      "step": 9932
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.536215713621397e-08,
+      "loss": 0.0776,
+      "step": 9934
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.461693108399171e-08,
+      "loss": 0.075,
+      "step": 9936
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.387539418525547e-08,
+      "loss": 0.0888,
+      "step": 9938
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.3137546715627e-08,
+      "loss": 0.0863,
+      "step": 9940
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.240338894935583e-08,
+      "loss": 0.0737,
+      "step": 9942
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.167292115931812e-08,
+      "loss": 0.0857,
+      "step": 9944
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.094614361702334e-08,
+      "loss": 0.085,
+      "step": 9946
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.022305659260542e-08,
+      "loss": 0.0778,
+      "step": 9948
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.950366035482714e-08,
+      "loss": 0.1043,
+      "step": 9950
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.878795517108017e-08,
+      "loss": 0.0586,
+      "step": 9952
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.807594130738505e-08,
+      "loss": 0.0883,
+      "step": 9954
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.736761902838895e-08,
+      "loss": 0.0766,
+      "step": 9956
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.666298859736686e-08,
+      "loss": 0.0883,
+      "step": 9958
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.59620502762226e-08,
+      "loss": 0.1229,
+      "step": 9960
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.526480432548554e-08,
+      "loss": 0.0653,
+      "step": 9962
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.457125100431616e-08,
+      "loss": 0.1024,
+      "step": 9964
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.388139057049936e-08,
+      "loss": 0.1009,
+      "step": 9966
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.31952232804478e-08,
+      "loss": 0.0585,
+      "step": 9968
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.251274938920415e-08,
+      "loss": 0.0909,
+      "step": 9970
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.18339691504366e-08,
+      "loss": 0.088,
+      "step": 9972
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.115888281643667e-08,
+      "loss": 0.0869,
+      "step": 9974
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.048749063812919e-08,
+      "loss": 0.0715,
+      "step": 9976
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.981979286506234e-08,
+      "loss": 0.0777,
+      "step": 9978
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.915578974541203e-08,
+      "loss": 0.0955,
+      "step": 9980
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.849548152598084e-08,
+      "loss": 0.0791,
+      "step": 9982
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.783886845219799e-08,
+      "loss": 0.0941,
+      "step": 9984
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.718595076811939e-08,
+      "loss": 0.1002,
+      "step": 9986
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.653672871642646e-08,
+      "loss": 0.0706,
+      "step": 9988
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.589120253842728e-08,
+      "loss": 0.0752,
+      "step": 9990
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.524937247405882e-08,
+      "loss": 0.0983,
+      "step": 9992
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.4611238761879125e-08,
+      "loss": 0.0882,
+      "step": 9994
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.397680163907848e-08,
+      "loss": 0.0796,
+      "step": 9996
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.3346061341468246e-08,
+      "loss": 0.1218,
+      "step": 9998
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2719018103487563e-08,
+      "loss": 0.0815,
+      "step": 10000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2095672158202216e-08,
+      "loss": 0.0766,
+      "step": 10002
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.147602373730132e-08,
+      "loss": 0.0763,
+      "step": 10004
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.086007307110286e-08,
+      "loss": 0.0958,
+      "step": 10006
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.024782038854814e-08,
+      "loss": 0.0765,
+      "step": 10008
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9639265917205134e-08,
+      "loss": 0.082,
+      "step": 10010
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9034409883265135e-08,
+      "loss": 0.0876,
+      "step": 10012
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.843325251154829e-08,
+      "loss": 0.1109,
+      "step": 10014
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.783579402549698e-08,
+      "loss": 0.0758,
+      "step": 10016
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.7242034647180246e-08,
+      "loss": 0.0944,
+      "step": 10018
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.665197459729154e-08,
+      "loss": 0.0784,
+      "step": 10020
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.606561409514876e-08,
+      "loss": 0.1179,
+      "step": 10022
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.548295335869757e-08,
+      "loss": 0.0803,
+      "step": 10024
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.490399260450362e-08,
+      "loss": 0.0784,
+      "step": 10026
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.432873204776256e-08,
+      "loss": 0.0729,
+      "step": 10028
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.375717190229223e-08,
+      "loss": 0.0694,
+      "step": 10030
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.318931238053381e-08,
+      "loss": 0.0741,
+      "step": 10032
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2625153693554024e-08,
+      "loss": 0.103,
+      "step": 10034
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.206469605104624e-08,
+      "loss": 0.0677,
+      "step": 10036
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.150793966132494e-08,
+      "loss": 0.0809,
+      "step": 10038
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0954884731330134e-08,
+      "loss": 0.0844,
+      "step": 10040
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.040553146662518e-08,
+      "loss": 0.0908,
+      "step": 10042
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.985988007140118e-08,
+      "loss": 0.108,
+      "step": 10044
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9317930748469235e-08,
+      "loss": 0.0807,
+      "step": 10046
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8779683699265994e-08,
+      "loss": 0.0712,
+      "step": 10048
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8245139123851414e-08,
+      "loss": 0.0888,
+      "step": 10050
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.77142972209088e-08,
+      "loss": 0.0708,
+      "step": 10052
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.7187158187749205e-08,
+      "loss": 0.0874,
+      "step": 10054
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.666372222030146e-08,
+      "loss": 0.0934,
+      "step": 10056
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.6143989513121034e-08,
+      "loss": 0.0973,
+      "step": 10058
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.562796025938897e-08,
+      "loss": 0.0993,
+      "step": 10060
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.511563465090517e-08,
+      "loss": 0.0847,
+      "step": 10062
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.4607012878096204e-08,
+      "loss": 0.1079,
+      "step": 10064
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.410209513001084e-08,
+      "loss": 0.1101,
+      "step": 10066
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3600881594322285e-08,
+      "loss": 0.0928,
+      "step": 10068
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.310337245732487e-08,
+      "loss": 0.0925,
+      "step": 10070
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2609567903937325e-08,
+      "loss": 0.0805,
+      "step": 10072
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.211946811770172e-08,
+      "loss": 0.0833,
+      "step": 10074
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.163307328078236e-08,
+      "loss": 0.0944,
+      "step": 10076
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.115038357396683e-08,
+      "loss": 0.0979,
+      "step": 10078
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.067139917666606e-08,
+      "loss": 0.0987,
+      "step": 10080
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.01961202669121e-08,
+      "loss": 0.1047,
+      "step": 10082
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9724547021360296e-08,
+      "loss": 0.0701,
+      "step": 10084
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.925667961529155e-08,
+      "loss": 0.0861,
+      "step": 10086
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.8792518222604537e-08,
+      "loss": 0.0889,
+      "step": 10088
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.8332063015823474e-08,
+      "loss": 0.075,
+      "step": 10090
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.7875314166095902e-08,
+      "loss": 0.1039,
+      "step": 10092
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.742227184318824e-08,
+      "loss": 0.0874,
+      "step": 10094
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.697293621549135e-08,
+      "loss": 0.0787,
+      "step": 10096
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.6527307450020523e-08,
+      "loss": 0.0796,
+      "step": 10098
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.6085385712407706e-08,
+      "loss": 0.0902,
+      "step": 10100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.564717116691262e-08,
+      "loss": 0.0914,
+      "step": 10102
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.5212663976413866e-08,
+      "loss": 0.083,
+      "step": 10104
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.478186430241336e-08,
+      "loss": 0.0907,
+      "step": 10106
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4354772305034136e-08,
+      "loss": 0.0996,
+      "step": 10108
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.393138814302143e-08,
+      "loss": 0.0962,
+      "step": 10110
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.351171197374158e-08,
+      "loss": 0.0641,
+      "step": 10112
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3095743953185367e-08,
+      "loss": 0.0643,
+      "step": 10114
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.268348423596245e-08,
+      "loss": 0.0829,
+      "step": 10116
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2274932975304696e-08,
+      "loss": 0.0662,
+      "step": 10118
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1870090323066197e-08,
+      "loss": 0.0711,
+      "step": 10120
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.146895642972324e-08,
+      "loss": 0.1141,
+      "step": 10122
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.1071531444371018e-08,
+      "loss": 0.0914,
+      "step": 10124
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.067781551473136e-08,
+      "loss": 0.0672,
+      "step": 10126
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.028780878714054e-08,
+      "loss": 0.0864,
+      "step": 10128
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.990151140656038e-08,
+      "loss": 0.0927,
+      "step": 10130
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9518923516574917e-08,
+      "loss": 0.1032,
+      "step": 10132
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9140045259385952e-08,
+      "loss": 0.0821,
+      "step": 10134
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.876487677581973e-08,
+      "loss": 0.0734,
+      "step": 10136
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8393418205321368e-08,
+      "loss": 0.0672,
+      "step": 10138
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8025669685958203e-08,
+      "loss": 0.087,
+      "step": 10140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7661631354417562e-08,
+      "loss": 0.0863,
+      "step": 10142
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.730130334600899e-08,
+      "loss": 0.0904,
+      "step": 10144
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6944685794662018e-08,
+      "loss": 0.0788,
+      "step": 10146
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6591778832928395e-08,
+      "loss": 0.1024,
+      "step": 10148
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.624258259197875e-08,
+      "loss": 0.0698,
+      "step": 10150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6069376032411544e-08,
+      "loss": 0.0922,
+      "step": 10152
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5725746115566876e-08,
+      "loss": 0.0738,
+      "step": 10154
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.538582724140092e-08,
+      "loss": 0.0641,
+      "step": 10156
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5049619536259275e-08,
+      "loss": 0.0792,
+      "step": 10158
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4717123125103095e-08,
+      "loss": 0.0764,
+      "step": 10160
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4388338131521295e-08,
+      "loss": 0.0711,
+      "step": 10162
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4063264677716126e-08,
+      "loss": 0.0728,
+      "step": 10164
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.374190288451649e-08,
+      "loss": 0.071,
+      "step": 10166
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3424252871367948e-08,
+      "loss": 0.0687,
+      "step": 10168
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3110314756337173e-08,
+      "loss": 0.0767,
+      "step": 10170
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2800088656110821e-08,
+      "loss": 0.0716,
+      "step": 10172
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2493574685997767e-08,
+      "loss": 0.0891,
+      "step": 10174
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2190772959925768e-08,
+      "loss": 0.0733,
+      "step": 10176
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1891683590442572e-08,
+      "loss": 0.0766,
+      "step": 10178
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.159630668871592e-08,
+      "loss": 0.0943,
+      "step": 10180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1304642364533547e-08,
+      "loss": 0.0624,
+      "step": 10182
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1016690726305402e-08,
+      "loss": 0.1153,
+      "step": 10184
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0732451881058093e-08,
+      "loss": 0.0813,
+      "step": 10186
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0451925934442664e-08,
+      "loss": 0.0846,
+      "step": 10188
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.01751129907246e-08,
+      "loss": 0.0642,
+      "step": 10190
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.90201315279271e-09,
+      "loss": 0.0786,
+      "step": 10192
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.632626522156907e-09,
+      "loss": 0.0749,
+      "step": 10194
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.366953198943762e-09,
+      "loss": 0.0832,
+      "step": 10196
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.104993281900953e-09,
+      "loss": 0.0789,
+      "step": 10198
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.84674686839726e-09,
+      "loss": 0.1042,
+      "step": 10200
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.592214054418124e-09,
+      "loss": 0.0829,
+      "step": 10202
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.341394934572312e-09,
+      "loss": 0.0832,
+      "step": 10204
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.094289602085249e-09,
+      "loss": 0.0632,
+      "step": 10206
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.850898148803465e-09,
+      "loss": 0.0837,
+      "step": 10208
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.611220665193486e-09,
+      "loss": 0.0773,
+      "step": 10210
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.375257240340717e-09,
+      "loss": 0.0863,
+      "step": 10212
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.143007961949444e-09,
+      "loss": 0.1179,
+      "step": 10214
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.914472916343951e-09,
+      "loss": 0.0746,
+      "step": 10216
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.68965218846851e-09,
+      "loss": 0.0737,
+      "step": 10218
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.4685458618862776e-09,
+      "loss": 0.0844,
+      "step": 10220
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.251154018781513e-09,
+      "loss": 0.0937,
+      "step": 10222
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.037476739954029e-09,
+      "loss": 0.0859,
+      "step": 10224
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.827514104826959e-09,
+      "loss": 0.0749,
+      "step": 10226
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6212661914401e-09,
+      "loss": 0.0846,
+      "step": 10228
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.418733076453242e-09,
+      "loss": 0.1132,
+      "step": 10230
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.219914835146167e-09,
+      "loss": 0.0731,
+      "step": 10232
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.02481154141865e-09,
+      "loss": 0.0924,
+      "step": 10234
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.83342326778602e-09,
+      "loss": 0.11,
+      "step": 10236
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.645750085386924e-09,
+      "loss": 0.1036,
+      "step": 10238
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.553306675252334e-09,
+      "loss": 0.1796,
+      "step": 10240
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.3712062600631545e-09,
+      "loss": 0.0724,
+      "step": 10242
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.192821107907863e-09,
+      "loss": 0.0846,
+      "step": 10244
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.018151285090089e-09,
+      "loss": 0.0805,
+      "step": 10246
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.8471968565334525e-09,
+      "loss": 0.1011,
+      "step": 10248
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6799578857782402e-09,
+      "loss": 0.0941,
+      "step": 10250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5164344349858383e-09,
+      "loss": 0.111,
+      "step": 10252
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.356626564936516e-09,
+      "loss": 0.0926,
+      "step": 10254
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.200534335029426e-09,
+      "loss": 0.0772,
+      "step": 10256
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.048157803281493e-09,
+      "loss": 0.0719,
+      "step": 10258
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.899497026329634e-09,
+      "loss": 0.0806,
+      "step": 10260
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7545520594285392e-09,
+      "loss": 0.1004,
+      "step": 10262
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.613322956454001e-09,
+      "loss": 0.0659,
+      "step": 10264
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4758097698995844e-09,
+      "loss": 0.102,
+      "step": 10266
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3420125508755163e-09,
+      "loss": 0.0732,
+      "step": 10268
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.211931349114238e-09,
+      "loss": 0.1078,
+      "step": 10270
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.085566212964851e-09,
+      "loss": 0.0934,
+      "step": 10272
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9629171893953415e-09,
+      "loss": 0.0909,
+      "step": 10274
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8439843239947963e-09,
+      "loss": 0.0928,
+      "step": 10276
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7287676609667458e-09,
+      "loss": 0.0704,
+      "step": 10278
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.617267243138043e-09,
+      "loss": 0.071,
+      "step": 10280
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5094831119522036e-09,
+      "loss": 0.0961,
+      "step": 10282
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4054153074694044e-09,
+      "loss": 0.0598,
+      "step": 10284
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3050638683720363e-09,
+      "loss": 0.0817,
+      "step": 10286
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2084288319580417e-09,
+      "loss": 0.0688,
+      "step": 10288
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1155102341486867e-09,
+      "loss": 0.069,
+      "step": 10290
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0263081094774584e-09,
+      "loss": 0.0843,
+      "step": 10292
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.408224911011676e-10,
+      "loss": 0.1066,
+      "step": 10294
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.590534107943971e-10,
+      "loss": 0.0963,
+      "step": 10296
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.810008989483919e-10,
+      "loss": 0.1078,
+      "step": 10298
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.066649845755002e-10,
+      "loss": 0.0753,
+      "step": 10300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.360456953058425e-10,
+      "loss": 0.0832,
+      "step": 10302
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.691430573862011e-10,
+      "loss": 0.0856,
+      "step": 10304
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.059570956855719e-10,
+      "loss": 0.0882,
+      "step": 10306
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.4648783368850255e-10,
+      "loss": 0.072,
+      "step": 10308
+    },
+    {
+      "epoch": 3.0,
+      "step": 10308,
+      "total_flos": 218874711244800.0,
+      "train_loss": 0.0,
+      "train_runtime": 2.854,
+      "train_samples_per_second": 57786.49,
+      "train_steps_per_second": 3611.787
+    }
+  ],
+  "max_steps": 10308,
+  "num_train_epochs": 3,
+  "total_flos": 218874711244800.0,
+  "trial_name": null,
+  "trial_params": null
+}