diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9388 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 3120,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.564102564102564e-07,
+      "loss": 1.153,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 1.0848,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 1.0058,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 0.9626,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.282051282051282e-06,
+      "loss": 0.9113,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 0.87,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.794871794871795e-06,
+      "loss": 0.9152,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 0.8951,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 0.8945,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 0.8762,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8205128205128207e-06,
+      "loss": 0.8699,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.8808,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.9051,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 0.8512,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.8972,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 0.8724,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 0.8813,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 0.9005,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 0.8754,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.9101,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 0.9124,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 0.9446,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 0.918,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.8927,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 0.8798,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.9412,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 0.8864,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 0.9589,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 0.8901,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.9314,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 0.9623,
+      "step": 62
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 0.9221,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 0.9284,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 0.9762,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 0.9132,
+      "step": 70
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.9044,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 0.9738,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 0.9718,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 0.9838,
+      "step": 78
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 0.9725,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0512820512820514e-05,
+      "loss": 0.9578,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.9779,
+      "step": 84
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1025641025641028e-05,
+      "loss": 0.9377,
+      "step": 86
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 0.9781,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 0.9616,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 0.9874,
+      "step": 92
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2051282051282051e-05,
+      "loss": 0.9791,
+      "step": 94
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.994,
+      "step": 96
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2564102564102565e-05,
+      "loss": 0.9929,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 1.0049,
+      "step": 100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 0.9933,
+      "step": 102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.0206,
+      "step": 104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3589743589743592e-05,
+      "loss": 1.0226,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.9692,
+      "step": 108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4102564102564105e-05,
+      "loss": 1.0258,
+      "step": 110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 0.9944,
+      "step": 112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 0.9905,
+      "step": 114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 1.0415,
+      "step": 116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5128205128205129e-05,
+      "loss": 0.9845,
+      "step": 118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 1.0322,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5641025641025644e-05,
+      "loss": 1.0112,
+      "step": 122
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 1.0107,
+      "step": 124
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 0.9721,
+      "step": 126
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.9967,
+      "step": 128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.0599,
+      "step": 130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.9962,
+      "step": 132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.717948717948718e-05,
+      "loss": 1.0334,
+      "step": 134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 1.0493,
+      "step": 136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 1.0102,
+      "step": 138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 1.0721,
+      "step": 140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8205128205128208e-05,
+      "loss": 0.9963,
+      "step": 142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 1.0204,
+      "step": 144
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717948717948718e-05,
+      "loss": 1.0109,
+      "step": 146
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 1.0429,
+      "step": 148
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 1.0544,
+      "step": 150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 1.0515,
+      "step": 152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9743589743589745e-05,
+      "loss": 1.0876,
+      "step": 154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2e-05,
+      "loss": 1.0579,
+      "step": 156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9999977531546567e-05,
+      "loss": 1.0929,
+      "step": 158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999991012628722e-05,
+      "loss": 1.1048,
+      "step": 160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9999797784524866e-05,
+      "loss": 1.063,
+      "step": 162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.999964050676434e-05,
+      "loss": 1.0594,
+      "step": 164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9999438293712382e-05,
+      "loss": 1.0332,
+      "step": 166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999919114627769e-05,
+      "loss": 1.0667,
+      "step": 168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999889906557086e-05,
+      "loss": 1.0598,
+      "step": 170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999856205290442e-05,
+      "loss": 1.0371,
+      "step": 172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9998180109792793e-05,
+      "loss": 1.0385,
+      "step": 174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999775323795232e-05,
+      "loss": 1.0782,
+      "step": 176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9997281439301218e-05,
+      "loss": 1.0578,
+      "step": 178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999676471595962e-05,
+      "loss": 1.0543,
+      "step": 180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996203070249516e-05,
+      "loss": 1.0499,
+      "step": 182
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9995596504694764e-05,
+      "loss": 1.0398,
+      "step": 184
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9994945022021085e-05,
+      "loss": 1.041,
+      "step": 186
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.999424862515604e-05,
+      "loss": 1.0063,
+      "step": 188
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.999350731722902e-05,
+      "loss": 1.0446,
+      "step": 190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9992721101571238e-05,
+      "loss": 1.0901,
+      "step": 192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9991889981715696e-05,
+      "loss": 1.0695,
+      "step": 194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99910139613972e-05,
+      "loss": 0.993,
+      "step": 196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9990093044552304e-05,
+      "loss": 1.0332,
+      "step": 198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.998912723531933e-05,
+      "loss": 1.0452,
+      "step": 200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9988116538038326e-05,
+      "loss": 1.0354,
+      "step": 202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9987060957251047e-05,
+      "loss": 1.1051,
+      "step": 204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9985960497700953e-05,
+      "loss": 1.0247,
+      "step": 206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9984815164333163e-05,
+      "loss": 1.049,
+      "step": 208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.998362496229446e-05,
+      "loss": 1.084,
+      "step": 210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.998238989693323e-05,
+      "loss": 1.0296,
+      "step": 212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.998110997379949e-05,
+      "loss": 1.0731,
+      "step": 214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.997978519864481e-05,
+      "loss": 1.051,
+      "step": 216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.997841557742232e-05,
+      "loss": 1.0049,
+      "step": 218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9977001116286675e-05,
+      "loss": 1.0536,
+      "step": 220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9975541821594028e-05,
+      "loss": 1.0438,
+      "step": 222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9974037699901993e-05,
+      "loss": 1.0396,
+      "step": 224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9972488757969635e-05,
+      "loss": 1.0609,
+      "step": 226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9970895002757413e-05,
+      "loss": 0.995,
+      "step": 228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9969256441427173e-05,
+      "loss": 1.0334,
+      "step": 230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9967573081342103e-05,
+      "loss": 1.0591,
+      "step": 232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.99658449300667e-05,
+      "loss": 1.0601,
+      "step": 234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9964071995366744e-05,
+      "loss": 1.1103,
+      "step": 236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9962254285209256e-05,
+      "loss": 1.0248,
+      "step": 238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9960391807762462e-05,
+      "loss": 1.0353,
+      "step": 240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9958484571395756e-05,
+      "loss": 1.0258,
+      "step": 242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9956532584679676e-05,
+      "loss": 1.0278,
+      "step": 244
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.995453585638584e-05,
+      "loss": 1.0839,
+      "step": 246
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.995249439548693e-05,
+      "loss": 0.9785,
+      "step": 248
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9950408211156637e-05,
+      "loss": 1.0363,
+      "step": 250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.994827731276963e-05,
+      "loss": 1.0558,
+      "step": 252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9946101709901513e-05,
+      "loss": 1.0088,
+      "step": 254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.994388141232876e-05,
+      "loss": 1.0482,
+      "step": 256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9941616430028713e-05,
+      "loss": 1.0404,
+      "step": 258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9939306773179498e-05,
+      "loss": 1.0761,
+      "step": 260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9936952452159995e-05,
+      "loss": 1.0288,
+      "step": 262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9934553477549795e-05,
+      "loss": 1.0902,
+      "step": 264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9932109860129154e-05,
+      "loss": 1.0529,
+      "step": 266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.992962161087893e-05,
+      "loss": 1.0118,
+      "step": 268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.992708874098054e-05,
+      "loss": 1.0239,
+      "step": 270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9924511261815928e-05,
+      "loss": 1.0184,
+      "step": 272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.992188918496748e-05,
+      "loss": 1.0117,
+      "step": 274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9919222522217998e-05,
+      "loss": 1.0192,
+      "step": 276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.991651128555064e-05,
+      "loss": 1.0019,
+      "step": 278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9913755487148874e-05,
+      "loss": 1.0181,
+      "step": 280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910955139396395e-05,
+      "loss": 1.02,
+      "step": 282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908110254877107e-05,
+      "loss": 1.0314,
+      "step": 284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990522084637503e-05,
+      "loss": 1.0318,
+      "step": 286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990228692687429e-05,
+      "loss": 0.9852,
+      "step": 288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899308509558998e-05,
+      "loss": 1.0348,
+      "step": 290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9896285607813245e-05,
+      "loss": 1.0273,
+      "step": 292
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9893218235221016e-05,
+      "loss": 1.0033,
+      "step": 294
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989010640556614e-05,
+      "loss": 1.0633,
+      "step": 296
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988695013283221e-05,
+      "loss": 1.0679,
+      "step": 298
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.988374943120254e-05,
+      "loss": 0.9936,
+      "step": 300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9880504315060096e-05,
+      "loss": 1.0311,
+      "step": 302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9877214798987428e-05,
+      "loss": 1.0168,
+      "step": 304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9873880897766597e-05,
+      "loss": 1.0162,
+      "step": 306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9870502626379127e-05,
+      "loss": 1.0188,
+      "step": 308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9867080000005933e-05,
+      "loss": 0.968,
+      "step": 310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9863613034027224e-05,
+      "loss": 1.0709,
+      "step": 312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9860101744022485e-05,
+      "loss": 1.0015,
+      "step": 314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985654614577036e-05,
+      "loss": 1.0152,
+      "step": 316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985294625524861e-05,
+      "loss": 1.008,
+      "step": 318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9849302088634034e-05,
+      "loss": 1.0249,
+      "step": 320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9845613662302384e-05,
+      "loss": 1.0255,
+      "step": 322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9841880992828306e-05,
+      "loss": 1.0112,
+      "step": 324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9838104096985266e-05,
+      "loss": 1.1042,
+      "step": 326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9834282991745465e-05,
+      "loss": 0.9916,
+      "step": 328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9830417694279766e-05,
+      "loss": 1.0568,
+      "step": 330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9826508221957624e-05,
+      "loss": 0.9906,
+      "step": 332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9822554592346996e-05,
+      "loss": 1.0166,
+      "step": 334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.981855682321427e-05,
+      "loss": 1.0167,
+      "step": 336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.981451493252418e-05,
+      "loss": 1.0531,
+      "step": 338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.981042893843974e-05,
+      "loss": 0.989,
+      "step": 340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9806298859322143e-05,
+      "loss": 1.0199,
+      "step": 342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.980212471373068e-05,
+      "loss": 0.9728,
+      "step": 344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.979790652042268e-05,
+      "loss": 0.997,
+      "step": 346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.979364429835339e-05,
+      "loss": 0.9729,
+      "step": 348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9789338066675924e-05,
+      "loss": 1.0042,
+      "step": 350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.978498784474115e-05,
+      "loss": 1.0274,
+      "step": 352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.978059365209762e-05,
+      "loss": 1.0348,
+      "step": 354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9776155508491482e-05,
+      "loss": 0.9827,
+      "step": 356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.977167343386638e-05,
+      "loss": 1.0481,
+      "step": 358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9767147448363366e-05,
+      "loss": 1.0609,
+      "step": 360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9762577572320826e-05,
+      "loss": 0.9862,
+      "step": 362
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9757963826274357e-05,
+      "loss": 1.0423,
+      "step": 364
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.975330623095672e-05,
+      "loss": 0.9966,
+      "step": 366
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.97486048072977e-05,
+      "loss": 1.0406,
+      "step": 368
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9743859576424044e-05,
+      "loss": 0.9728,
+      "step": 370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9739070559659347e-05,
+      "loss": 1.0276,
+      "step": 372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9734237778523977e-05,
+      "loss": 1.0173,
+      "step": 374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.972936125473495e-05,
+      "loss": 1.0049,
+      "step": 376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9724441010205865e-05,
+      "loss": 1.0092,
+      "step": 378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9719477067046768e-05,
+      "loss": 1.0488,
+      "step": 380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.971446944756409e-05,
+      "loss": 1.0164,
+      "step": 382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 0.9694,
+      "step": 384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9704323269834925e-05,
+      "loss": 0.9877,
+      "step": 386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9699184757182225e-05,
+      "loss": 1.038,
+      "step": 388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9694002659393306e-05,
+      "loss": 1.0039,
+      "step": 390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9688776999754913e-05,
+      "loss": 0.9953,
+      "step": 392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9683507801749547e-05,
+      "loss": 1.0098,
+      "step": 394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9678195089055347e-05,
+      "loss": 0.9893,
+      "step": 396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9672838885546007e-05,
+      "loss": 0.9521,
+      "step": 398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.966743921529065e-05,
+      "loss": 0.9806,
+      "step": 400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9661996102553716e-05,
+      "loss": 1.0153,
+      "step": 402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.965650957179488e-05,
+      "loss": 1.0054,
+      "step": 404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9650979647668908e-05,
+      "loss": 1.0026,
+      "step": 406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9645406355025565e-05,
+      "loss": 1.005,
+      "step": 408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.963978971890951e-05,
+      "loss": 0.9792,
+      "step": 410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.963412976456017e-05,
+      "loss": 1.0335,
+      "step": 412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9628426517411625e-05,
+      "loss": 1.0231,
+      "step": 414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9622680003092503e-05,
+      "loss": 1.0409,
+      "step": 416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9616890247425867e-05,
+      "loss": 0.9979,
+      "step": 418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9611057276429085e-05,
+      "loss": 1.0009,
+      "step": 420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9605181116313725e-05,
+      "loss": 0.9753,
+      "step": 422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9599261793485432e-05,
+      "loss": 0.9813,
+      "step": 424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.959329933454381e-05,
+      "loss": 1.0204,
+      "step": 426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.958729376628231e-05,
+      "loss": 1.045,
+      "step": 428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9581245115688093e-05,
+      "loss": 0.9576,
+      "step": 430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.957515340994193e-05,
+      "loss": 1.0385,
+      "step": 432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9569018676418053e-05,
+      "loss": 0.9917,
+      "step": 434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.956284094268407e-05,
+      "loss": 1.0301,
+      "step": 436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9556620236500794e-05,
+      "loss": 1.0141,
+      "step": 438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.955035658582216e-05,
+      "loss": 1.0256,
+      "step": 440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9544050018795076e-05,
+      "loss": 1.0178,
+      "step": 442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9537700563759303e-05,
+      "loss": 0.9886,
+      "step": 444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9531308249247328e-05,
+      "loss": 1.0009,
+      "step": 446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9524873103984234e-05,
+      "loss": 1.0224,
+      "step": 448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9518395156887574e-05,
+      "loss": 0.983,
+      "step": 450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9511874437067243e-05,
+      "loss": 1.022,
+      "step": 452
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.950531097382533e-05,
+      "loss": 1.0,
+      "step": 454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.949870479665602e-05,
+      "loss": 1.0357,
+      "step": 456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9492055935245418e-05,
+      "loss": 0.9761,
+      "step": 458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9485364419471454e-05,
+      "loss": 1.0038,
+      "step": 460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9478630279403737e-05,
+      "loss": 1.0139,
+      "step": 462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9471853545303407e-05,
+      "loss": 0.9504,
+      "step": 464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9465034247623005e-05,
+      "loss": 1.0035,
+      "step": 466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9458172417006347e-05,
+      "loss": 1.0145,
+      "step": 468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9451268084288385e-05,
+      "loss": 0.9738,
+      "step": 470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9444321280495045e-05,
+      "loss": 0.9573,
+      "step": 472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.943733203684312e-05,
+      "loss": 0.9938,
+      "step": 474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9430300384740108e-05,
+      "loss": 0.9887,
+      "step": 476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.942322635578408e-05,
+      "loss": 0.9877,
+      "step": 478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9416109981763526e-05,
+      "loss": 0.9856,
+      "step": 480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.940895129465724e-05,
+      "loss": 1.0039,
+      "step": 482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9401750326634144e-05,
+      "loss": 1.0454,
+      "step": 484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.939450711005316e-05,
+      "loss": 1.0293,
+      "step": 486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9387221677463064e-05,
+      "loss": 1.0311,
+      "step": 488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9379894061602335e-05,
+      "loss": 0.9813,
+      "step": 490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9372524295399014e-05,
+      "loss": 0.9582,
+      "step": 492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.936511241197055e-05,
+      "loss": 1.016,
+      "step": 494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9357658444623655e-05,
+      "loss": 0.9816,
+      "step": 496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9350162426854152e-05,
+      "loss": 0.9778,
+      "step": 498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9342624392346826e-05,
+      "loss": 1.0288,
+      "step": 500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.933504437497527e-05,
+      "loss": 1.0361,
+      "step": 502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9327422408801744e-05,
+      "loss": 0.9652,
+      "step": 504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9319758528076998e-05,
+      "loss": 1.0004,
+      "step": 506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9312052767240153e-05,
+      "loss": 0.9903,
+      "step": 508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.93043051609185e-05,
+      "loss": 1.0056,
+      "step": 510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.92965157439274e-05,
+      "loss": 1.002,
+      "step": 512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9288684551270072e-05,
+      "loss": 0.9903,
+      "step": 514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9280811618137486e-05,
+      "loss": 1.0369,
+      "step": 516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9272896979908153e-05,
+      "loss": 0.9782,
+      "step": 518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9264940672148018e-05,
+      "loss": 0.9736,
+      "step": 520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9256942730610268e-05,
+      "loss": 1.0189,
+      "step": 522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9248903191235177e-05,
+      "loss": 0.9975,
+      "step": 524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9240822090149944e-05,
+      "loss": 1.0273,
+      "step": 526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9232699463668543e-05,
+      "loss": 1.0047,
+      "step": 528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9224535348291543e-05,
+      "loss": 0.9929,
+      "step": 530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9216329780705955e-05,
+      "loss": 1.0068,
+      "step": 532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9208082797785057e-05,
+      "loss": 1.0018,
+      "step": 534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9199794436588244e-05,
+      "loss": 0.9997,
+      "step": 536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9191464734360844e-05,
+      "loss": 0.9969,
+      "step": 538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9183093728533966e-05,
+      "loss": 0.9868,
+      "step": 540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.917468145672432e-05,
+      "loss": 0.9876,
+      "step": 542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.916622795673405e-05,
+      "loss": 0.9647,
+      "step": 544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9157733266550577e-05,
+      "loss": 0.9968,
+      "step": 546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9149197424346405e-05,
+      "loss": 0.9566,
+      "step": 548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9140620468478967e-05,
+      "loss": 1.0137,
+      "step": 550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.913200243749046e-05,
+      "loss": 1.0246,
+      "step": 552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9123343370107636e-05,
+      "loss": 0.9635,
+      "step": 554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9114643305241678e-05,
+      "loss": 1.0191,
+      "step": 556
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.910590228198798e-05,
+      "loss": 0.9818,
+      "step": 558
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9097120339625994e-05,
+      "loss": 0.981,
+      "step": 560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9088297517619056e-05,
+      "loss": 0.95,
+      "step": 562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9079433855614203e-05,
+      "loss": 0.9521,
+      "step": 564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9070529393441985e-05,
+      "loss": 0.9817,
+      "step": 566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9061584171116302e-05,
+      "loss": 0.9701,
+      "step": 568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9052598228834217e-05,
+      "loss": 1.0093,
+      "step": 570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9043571606975776e-05,
+      "loss": 0.9842,
+      "step": 572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9034504346103825e-05,
+      "loss": 0.9396,
+      "step": 574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9025396486963827e-05,
+      "loss": 0.9673,
+      "step": 576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9016248070483686e-05,
+      "loss": 0.9469,
+      "step": 578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.900705913777356e-05,
+      "loss": 0.9759,
+      "step": 580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8997829730125662e-05,
+      "loss": 0.9942,
+      "step": 582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.89885598890141e-05,
+      "loss": 0.948,
+      "step": 584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8979249656094672e-05,
+      "loss": 0.9954,
+      "step": 586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8969899073204687e-05,
+      "loss": 0.9547,
+      "step": 588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.896050818236277e-05,
+      "loss": 0.9652,
+      "step": 590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.895107702576868e-05,
+      "loss": 0.9418,
+      "step": 592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8941605645803115e-05,
+      "loss": 0.9573,
+      "step": 594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8932094085027534e-05,
+      "loss": 0.9951,
+      "step": 596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8922542386183942e-05,
+      "loss": 0.992,
+      "step": 598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.891295059219472e-05,
+      "loss": 0.9909,
+      "step": 600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8903318746162428e-05,
+      "loss": 0.9776,
+      "step": 602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.88936468913696e-05,
+      "loss": 0.9524,
+      "step": 604
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.888393507127856e-05,
+      "loss": 0.9784,
+      "step": 606
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8874183329531222e-05,
+      "loss": 0.9445,
+      "step": 608
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.88643917099489e-05,
+      "loss": 0.9838,
+      "step": 610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 0.9846,
+      "step": 612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8844689013460335e-05,
+      "loss": 0.9751,
+      "step": 614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.883477802509192e-05,
+      "loss": 0.9735,
+      "step": 616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8824827335963767e-05,
+      "loss": 0.9763,
+      "step": 618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.88148369907912e-05,
+      "loss": 0.9518,
+      "step": 620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8804807034467735e-05,
+      "loss": 0.9663,
+      "step": 622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.879473751206489e-05,
+      "loss": 1.02,
+      "step": 624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8784628468831997e-05,
+      "loss": 0.976,
+      "step": 626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.877447995019596e-05,
+      "loss": 0.9871,
+      "step": 628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.876429200176108e-05,
+      "loss": 0.9734,
+      "step": 630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875406466930886e-05,
+      "loss": 0.9715,
+      "step": 632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874379799879775e-05,
+      "loss": 0.951,
+      "step": 634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8733492036363007e-05,
+      "loss": 0.965,
+      "step": 636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.872314682831643e-05,
+      "loss": 1.0045,
+      "step": 638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8712762421146185e-05,
+      "loss": 1.0237,
+      "step": 640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8702338861516587e-05,
+      "loss": 0.972,
+      "step": 642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8691876196267892e-05,
+      "loss": 0.9857,
+      "step": 644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8681374472416075e-05,
+      "loss": 0.9652,
+      "step": 646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.867083373715264e-05,
+      "loss": 0.9234,
+      "step": 648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.9598,
+      "step": 650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8649635422033218e-05,
+      "loss": 0.9785,
+      "step": 652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.86389779374359e-05,
+      "loss": 0.9512,
+      "step": 654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.862828163194388e-05,
+      "loss": 0.9479,
+      "step": 656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.861754655362304e-05,
+      "loss": 0.9631,
+      "step": 658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8606772750713503e-05,
+      "loss": 0.9858,
+      "step": 660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8595960271629412e-05,
+      "loss": 0.9934,
+      "step": 662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8585109164958698e-05,
+      "loss": 0.974,
+      "step": 664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.857421947946288e-05,
+      "loss": 0.9948,
+      "step": 666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8563291264076834e-05,
+      "loss": 0.9524,
+      "step": 668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8552324567908585e-05,
+      "loss": 0.9399,
+      "step": 670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8541319440239066e-05,
+      "loss": 0.9611,
+      "step": 672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8530275930521925e-05,
+      "loss": 0.96,
+      "step": 674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.851919408838327e-05,
+      "loss": 0.9282,
+      "step": 676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8508073963621482e-05,
+      "loss": 0.9576,
+      "step": 678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8496915606206952e-05,
+      "loss": 0.9863,
+      "step": 680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8485719066281894e-05,
+      "loss": 0.9812,
+      "step": 682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.847448439416009e-05,
+      "loss": 0.9907,
+      "step": 684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8463211640326686e-05,
+      "loss": 1.0007,
+      "step": 686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.845190085543795e-05,
+      "loss": 0.9565,
+      "step": 688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8440552090321047e-05,
+      "loss": 0.9416,
+      "step": 690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.842916539597382e-05,
+      "loss": 0.9936,
+      "step": 692
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8417740823564552e-05,
+      "loss": 0.9561,
+      "step": 694
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8406278424431737e-05,
+      "loss": 0.9934,
+      "step": 696
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.839477825008385e-05,
+      "loss": 0.946,
+      "step": 698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8383240352199118e-05,
+      "loss": 0.9518,
+      "step": 700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8371664782625287e-05,
+      "loss": 0.9353,
+      "step": 702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8360051593379383e-05,
+      "loss": 0.9797,
+      "step": 704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8348400836647492e-05,
+      "loss": 0.949,
+      "step": 706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8336712564784506e-05,
+      "loss": 0.9681,
+      "step": 708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.83249868303139e-05,
+      "loss": 0.9341,
+      "step": 710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8313223685927507e-05,
+      "loss": 0.974,
+      "step": 712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8301423184485253e-05,
+      "loss": 0.9027,
+      "step": 714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8289585379014942e-05,
+      "loss": 0.9613,
+      "step": 716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.827771032271201e-05,
+      "loss": 0.966,
+      "step": 718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8265798068939295e-05,
+      "loss": 1.0244,
+      "step": 720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.825384867122677e-05,
+      "loss": 0.9949,
+      "step": 722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8241862183271338e-05,
+      "loss": 0.9307,
+      "step": 724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8229838658936566e-05,
+      "loss": 0.9577,
+      "step": 726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.821777815225245e-05,
+      "loss": 0.947,
+      "step": 728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8205680717415188e-05,
+      "loss": 0.9278,
+      "step": 730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.81935464087869e-05,
+      "loss": 0.932,
+      "step": 732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8181375280895417e-05,
+      "loss": 0.9878,
+      "step": 734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8169167388434024e-05,
+      "loss": 0.9873,
+      "step": 736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.815692278626122e-05,
+      "loss": 0.9393,
+      "step": 738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8144641529400445e-05,
+      "loss": 0.964,
+      "step": 740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8132323673039883e-05,
+      "loss": 0.9571,
+      "step": 742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8119969272532164e-05,
+      "loss": 0.9907,
+      "step": 744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8107578383394145e-05,
+      "loss": 0.952,
+      "step": 746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8095151061306647e-05,
+      "loss": 0.9455,
+      "step": 748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.808268736211421e-05,
+      "loss": 0.9461,
+      "step": 750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8070187341824848e-05,
+      "loss": 0.9424,
+      "step": 752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8057651056609784e-05,
+      "loss": 0.9303,
+      "step": 754
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8045078562803203e-05,
+      "loss": 1.0035,
+      "step": 756
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8032469916902004e-05,
+      "loss": 0.9598,
+      "step": 758
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8019825175565544e-05,
+      "loss": 0.9471,
+      "step": 760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.800714439561538e-05,
+      "loss": 0.9732,
+      "step": 762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7994427634035016e-05,
+      "loss": 0.9672,
+      "step": 764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7981674947969635e-05,
+      "loss": 0.948,
+      "step": 766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7968886394725876e-05,
+      "loss": 0.9364,
+      "step": 768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7956062031771533e-05,
+      "loss": 0.9265,
+      "step": 770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7943201916735337e-05,
+      "loss": 0.9267,
+      "step": 772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.793030610740665e-05,
+      "loss": 0.966,
+      "step": 774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.791737466173527e-05,
+      "loss": 0.9343,
+      "step": 776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7904407637831097e-05,
+      "loss": 0.9437,
+      "step": 778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.789140509396394e-05,
+      "loss": 0.9446,
+      "step": 780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7878367088563195e-05,
+      "loss": 0.7208,
+      "step": 782
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7865293680217636e-05,
+      "loss": 0.6581,
+      "step": 784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7852184927675113e-05,
+      "loss": 0.6642,
+      "step": 786
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7839040889842307e-05,
+      "loss": 0.6761,
+      "step": 788
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7825861625784457e-05,
+      "loss": 0.6743,
+      "step": 790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7812647194725093e-05,
+      "loss": 0.6978,
+      "step": 792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.779939765604579e-05,
+      "loss": 0.6894,
+      "step": 794
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7786113069285877e-05,
+      "loss": 0.6569,
+      "step": 796
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.777279349414217e-05,
+      "loss": 0.6834,
+      "step": 798
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7759438990468726e-05,
+      "loss": 0.6673,
+      "step": 800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7746049618276545e-05,
+      "loss": 0.706,
+      "step": 802
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7732625437733338e-05,
+      "loss": 0.6816,
+      "step": 804
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.771916650916321e-05,
+      "loss": 0.6727,
+      "step": 806
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7705672893046425e-05,
+      "loss": 0.6521,
+      "step": 808
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7692144650019125e-05,
+      "loss": 0.6575,
+      "step": 810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.767858184087304e-05,
+      "loss": 0.6536,
+      "step": 812
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7664984526555247e-05,
+      "loss": 0.6786,
+      "step": 814
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.765135276816787e-05,
+      "loss": 0.6776,
+      "step": 816
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7637686626967815e-05,
+      "loss": 0.6591,
+      "step": 818
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7623986164366487e-05,
+      "loss": 0.681,
+      "step": 820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7610251441929532e-05,
+      "loss": 0.6805,
+      "step": 822
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7596482521376546e-05,
+      "loss": 0.6553,
+      "step": 824
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7582679464580796e-05,
+      "loss": 0.6591,
+      "step": 826
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7568842333568952e-05,
+      "loss": 0.7032,
+      "step": 828
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.75549711905208e-05,
+      "loss": 0.7043,
+      "step": 830
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7541066097768965e-05,
+      "loss": 0.6576,
+      "step": 832
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7527127117798635e-05,
+      "loss": 0.7119,
+      "step": 834
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7513154313247273e-05,
+      "loss": 0.7079,
+      "step": 836
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7499147746904337e-05,
+      "loss": 0.6893,
+      "step": 838
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 0.676,
+      "step": 840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7471033580759903e-05,
+      "loss": 0.6884,
+      "step": 842
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7456926107294765e-05,
+      "loss": 0.6845,
+      "step": 844
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7442785124710227e-05,
+      "loss": 0.6677,
+      "step": 846
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.742861069655148e-05,
+      "loss": 0.6762,
+      "step": 848
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.741440288651403e-05,
+      "loss": 0.6927,
+      "step": 850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7400161758443377e-05,
+      "loss": 0.6563,
+      "step": 852
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7385887376334743e-05,
+      "loss": 0.6766,
+      "step": 854
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.737157980433279e-05,
+      "loss": 0.685,
+      "step": 856
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.735723910673132e-05,
+      "loss": 0.6739,
+      "step": 858
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7342865347972987e-05,
+      "loss": 0.694,
+      "step": 860
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7328458592649028e-05,
+      "loss": 0.6966,
+      "step": 862
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7314018905498932e-05,
+      "loss": 0.6762,
+      "step": 864
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7299546351410197e-05,
+      "loss": 0.6665,
+      "step": 866
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7285040995418003e-05,
+      "loss": 0.7266,
+      "step": 868
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7270502902704925e-05,
+      "loss": 0.6916,
+      "step": 870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7255932138600665e-05,
+      "loss": 0.676,
+      "step": 872
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7241328768581725e-05,
+      "loss": 0.687,
+      "step": 874
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7226692858271133e-05,
+      "loss": 0.681,
+      "step": 876
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7212024473438145e-05,
+      "loss": 0.6643,
+      "step": 878
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7197323679997943e-05,
+      "loss": 0.681,
+      "step": 880
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.718259054401135e-05,
+      "loss": 0.666,
+      "step": 882
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7167825131684516e-05,
+      "loss": 0.677,
+      "step": 884
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7153027509368637e-05,
+      "loss": 0.6968,
+      "step": 886
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7138197743559656e-05,
+      "loss": 0.6781,
+      "step": 888
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7123335900897946e-05,
+      "loss": 0.6608,
+      "step": 890
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7108442048168038e-05,
+      "loss": 0.7042,
+      "step": 892
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7093516252298296e-05,
+      "loss": 0.6992,
+      "step": 894
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.707855858036063e-05,
+      "loss": 0.6746,
+      "step": 896
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7063569099570196e-05,
+      "loss": 0.6718,
+      "step": 898
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7048547877285078e-05,
+      "loss": 0.6895,
+      "step": 900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7033494981006003e-05,
+      "loss": 0.6815,
+      "step": 902
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7018410478376033e-05,
+      "loss": 0.6403,
+      "step": 904
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7003294437180254e-05,
+      "loss": 0.6526,
+      "step": 906
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6988146925345487e-05,
+      "loss": 0.6716,
+      "step": 908
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6972968010939953e-05,
+      "loss": 0.6815,
+      "step": 910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.695775776217301e-05,
+      "loss": 0.6915,
+      "step": 912
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6942516247394806e-05,
+      "loss": 0.6547,
+      "step": 914
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6927243535095995e-05,
+      "loss": 0.6959,
+      "step": 916
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6911939693907422e-05,
+      "loss": 0.6991,
+      "step": 918
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6896604792599813e-05,
+      "loss": 0.6732,
+      "step": 920
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6881238900083472e-05,
+      "loss": 0.6604,
+      "step": 922
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.686584208540797e-05,
+      "loss": 0.7089,
+      "step": 924
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.685041441776183e-05,
+      "loss": 0.6744,
+      "step": 926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6834955966472214e-05,
+      "loss": 0.6912,
+      "step": 928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6819466801004622e-05,
+      "loss": 0.6964,
+      "step": 930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6803946990962577e-05,
+      "loss": 0.6497,
+      "step": 932
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.67883966060873e-05,
+      "loss": 0.6606,
+      "step": 934
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6772815716257414e-05,
+      "loss": 0.6708,
+      "step": 936
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6757204391488617e-05,
+      "loss": 0.695,
+      "step": 938
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6741562701933366e-05,
+      "loss": 0.6839,
+      "step": 940
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.672589071788059e-05,
+      "loss": 0.6916,
+      "step": 942
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.671018850975533e-05,
+      "loss": 0.7089,
+      "step": 944
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6694456148118453e-05,
+      "loss": 0.666,
+      "step": 946
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6678693703666327e-05,
+      "loss": 0.6705,
+      "step": 948
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.66629012472305e-05,
+      "loss": 0.693,
+      "step": 950
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.664707884977739e-05,
+      "loss": 0.6799,
+      "step": 952
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6631226582407954e-05,
+      "loss": 0.6794,
+      "step": 954
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.661534451635738e-05,
+      "loss": 0.6721,
+      "step": 956
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6599432722994754e-05,
+      "loss": 0.6717,
+      "step": 958
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6583491273822763e-05,
+      "loss": 0.6708,
+      "step": 960
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6567520240477344e-05,
+      "loss": 0.6726,
+      "step": 962
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.655151969472738e-05,
+      "loss": 0.6824,
+      "step": 964
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.653548970847438e-05,
+      "loss": 0.6867,
+      "step": 966
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6519430353752138e-05,
+      "loss": 0.6749,
+      "step": 968
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6503341702726428e-05,
+      "loss": 0.6589,
+      "step": 970
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6487223827694673e-05,
+      "loss": 0.6842,
+      "step": 972
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6471076801085618e-05,
+      "loss": 0.7212,
+      "step": 974
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6454900695459e-05,
+      "loss": 0.6861,
+      "step": 976
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.643869558350524e-05,
+      "loss": 0.6584,
+      "step": 978
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6422461538045104e-05,
+      "loss": 0.6463,
+      "step": 980
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6406198632029358e-05,
+      "loss": 0.6669,
+      "step": 982
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.638990693853848e-05,
+      "loss": 0.6782,
+      "step": 984
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.63735865307823e-05,
+      "loss": 0.6849,
+      "step": 986
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.6357237482099682e-05,
+      "loss": 0.6453,
+      "step": 988
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.6340859865958193e-05,
+      "loss": 0.6681,
+      "step": 990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.6324453755953772e-05,
+      "loss": 0.6767,
+      "step": 992
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.630801922581041e-05,
+      "loss": 0.6767,
+      "step": 994
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.6291556349379794e-05,
+      "loss": 0.6729,
+      "step": 996
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.6275065200641004e-05,
+      "loss": 0.6778,
+      "step": 998
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.6258545853700157e-05,
+      "loss": 0.661,
+      "step": 1000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.6241998382790095e-05,
+      "loss": 0.7027,
+      "step": 1002
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.622542286227003e-05,
+      "loss": 0.6713,
+      "step": 1004
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.620881936662522e-05,
+      "loss": 0.6996,
+      "step": 1006
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.6192187970466646e-05,
+      "loss": 0.6657,
+      "step": 1008
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.617552874853065e-05,
+      "loss": 0.6614,
+      "step": 1010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.615884177567863e-05,
+      "loss": 0.6811,
+      "step": 1012
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6142127126896682e-05,
+      "loss": 0.66,
+      "step": 1014
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6125384877295255e-05,
+      "loss": 0.6528,
+      "step": 1016
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6108615102108854e-05,
+      "loss": 0.7191,
+      "step": 1018
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6091817876695655e-05,
+      "loss": 0.6658,
+      "step": 1020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6074993276537198e-05,
+      "loss": 0.6688,
+      "step": 1022
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6058141377238026e-05,
+      "loss": 0.6993,
+      "step": 1024
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.6041262254525362e-05,
+      "loss": 0.7091,
+      "step": 1026
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.602435598424877e-05,
+      "loss": 0.6668,
+      "step": 1028
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.600742264237979e-05,
+      "loss": 0.6942,
+      "step": 1030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.599046230501163e-05,
+      "loss": 0.6601,
+      "step": 1032
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5973475048358795e-05,
+      "loss": 0.6937,
+      "step": 1034
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5956460948756765e-05,
+      "loss": 0.7019,
+      "step": 1036
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.593942008266164e-05,
+      "loss": 0.6782,
+      "step": 1038
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5922352526649803e-05,
+      "loss": 0.6759,
+      "step": 1040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5905258357417567e-05,
+      "loss": 0.6486,
+      "step": 1042
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5888137651780847e-05,
+      "loss": 0.6811,
+      "step": 1044
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5870990486674794e-05,
+      "loss": 0.6868,
+      "step": 1046
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.585381693915346e-05,
+      "loss": 0.6565,
+      "step": 1048
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.583661708638947e-05,
+      "loss": 0.6563,
+      "step": 1050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.581939100567363e-05,
+      "loss": 0.6765,
+      "step": 1052
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.580213877441462e-05,
+      "loss": 0.6917,
+      "step": 1054
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5784860470138633e-05,
+      "loss": 0.6991,
+      "step": 1056
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5767556170489027e-05,
+      "loss": 0.6678,
+      "step": 1058
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5750225953225968e-05,
+      "loss": 0.6775,
+      "step": 1060
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.573286989622609e-05,
+      "loss": 0.6888,
+      "step": 1062
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5715488077482152e-05,
+      "loss": 0.7152,
+      "step": 1064
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5698080575102662e-05,
+      "loss": 0.6731,
+      "step": 1066
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 0.6565,
+      "step": 1068
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5663188832447833e-05,
+      "loss": 0.7098,
+      "step": 1070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5645704748965193e-05,
+      "loss": 0.6824,
+      "step": 1072
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5628195295431696e-05,
+      "loss": 0.6652,
+      "step": 1074
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5610660550529413e-05,
+      "loss": 0.6767,
+      "step": 1076
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5593100593054063e-05,
+      "loss": 0.6672,
+      "step": 1078
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.557551550191467e-05,
+      "loss": 0.696,
+      "step": 1080
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5557905356133182e-05,
+      "loss": 0.6986,
+      "step": 1082
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.554027023484416e-05,
+      "loss": 0.7039,
+      "step": 1084
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5522610217294377e-05,
+      "loss": 0.6816,
+      "step": 1086
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.550492538284249e-05,
+      "loss": 0.6606,
+      "step": 1088
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5487215810958675e-05,
+      "loss": 0.6978,
+      "step": 1090
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5469481581224274e-05,
+      "loss": 0.6792,
+      "step": 1092
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.545172277333142e-05,
+      "loss": 0.6698,
+      "step": 1094
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5433939467082713e-05,
+      "loss": 0.6946,
+      "step": 1096
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5416131742390827e-05,
+      "loss": 0.6636,
+      "step": 1098
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5398299679278172e-05,
+      "loss": 0.7072,
+      "step": 1100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5380443357876516e-05,
+      "loss": 0.6604,
+      "step": 1102
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5362562858426655e-05,
+      "loss": 0.6487,
+      "step": 1104
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5344658261278013e-05,
+      "loss": 0.654,
+      "step": 1106
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5326729646888314e-05,
+      "loss": 0.7315,
+      "step": 1108
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.530877709582321e-05,
+      "loss": 0.6849,
+      "step": 1110
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.5290800688755906e-05,
+      "loss": 0.7132,
+      "step": 1112
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.527280050646682e-05,
+      "loss": 0.7095,
+      "step": 1114
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.5254776629843204e-05,
+      "loss": 0.6958,
+      "step": 1116
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.523672913987878e-05,
+      "loss": 0.6912,
+      "step": 1118
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.5218658117673389e-05,
+      "loss": 0.6872,
+      "step": 1120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.5200563644432614e-05,
+      "loss": 0.6528,
+      "step": 1122
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.518244580146742e-05,
+      "loss": 0.6489,
+      "step": 1124
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.5164304670193792e-05,
+      "loss": 0.6625,
+      "step": 1126
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5146140332132359e-05,
+      "loss": 0.6994,
+      "step": 1128
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5127952868908043e-05,
+      "loss": 0.6519,
+      "step": 1130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5109742362249673e-05,
+      "loss": 0.7028,
+      "step": 1132
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5091508893989633e-05,
+      "loss": 0.7121,
+      "step": 1134
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5073252546063493e-05,
+      "loss": 0.6831,
+      "step": 1136
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5054973400509631e-05,
+      "loss": 0.6785,
+      "step": 1138
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5036671539468879e-05,
+      "loss": 0.7107,
+      "step": 1140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5018347045184132e-05,
+      "loss": 0.6388,
+      "step": 1142
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.6863,
+      "step": 1144
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4981630486362435e-05,
+      "loss": 0.7048,
+      "step": 1146
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4963238586818346e-05,
+      "loss": 0.6739,
+      "step": 1148
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4944824384015238e-05,
+      "loss": 0.713,
+      "step": 1150
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4926387960700843e-05,
+      "loss": 0.6581,
+      "step": 1152
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4907929399722752e-05,
+      "loss": 0.6761,
+      "step": 1154
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.488944878402802e-05,
+      "loss": 0.6654,
+      "step": 1156
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4870946196662822e-05,
+      "loss": 0.6789,
+      "step": 1158
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4852421720772064e-05,
+      "loss": 0.719,
+      "step": 1160
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4833875439599004e-05,
+      "loss": 0.6891,
+      "step": 1162
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4815307436484898e-05,
+      "loss": 0.6749,
+      "step": 1164
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4796717794868606e-05,
+      "loss": 0.6907,
+      "step": 1166
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4778106598286235e-05,
+      "loss": 0.6898,
+      "step": 1168
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4759473930370738e-05,
+      "loss": 0.6835,
+      "step": 1170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4740819874851562e-05,
+      "loss": 0.6484,
+      "step": 1172
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4722144515554263e-05,
+      "loss": 0.6236,
+      "step": 1174
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4703447936400135e-05,
+      "loss": 0.6484,
+      "step": 1176
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4684730221405814e-05,
+      "loss": 0.694,
+      "step": 1178
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4665991454682924e-05,
+      "loss": 0.6673,
+      "step": 1180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4647231720437687e-05,
+      "loss": 0.6609,
+      "step": 1182
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4628451102970546e-05,
+      "loss": 0.7178,
+      "step": 1184
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4609649686675786e-05,
+      "loss": 0.6564,
+      "step": 1186
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4590827556041158e-05,
+      "loss": 0.6898,
+      "step": 1188
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4571984795647494e-05,
+      "loss": 0.6892,
+      "step": 1190
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4553121490168335e-05,
+      "loss": 0.6854,
+      "step": 1192
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4534237724369534e-05,
+      "loss": 0.6946,
+      "step": 1194
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4515333583108896e-05,
+      "loss": 0.6577,
+      "step": 1196
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4496409151335786e-05,
+      "loss": 0.667,
+      "step": 1198
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4477464514090745e-05,
+      "loss": 0.6932,
+      "step": 1200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4458499756505116e-05,
+      "loss": 0.6625,
+      "step": 1202
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.443951496380065e-05,
+      "loss": 0.6827,
+      "step": 1204
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4420510221289137e-05,
+      "loss": 0.6946,
+      "step": 1206
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4401485614372009e-05,
+      "loss": 0.67,
+      "step": 1208
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4382441228539962e-05,
+      "loss": 0.6913,
+      "step": 1210
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4363377149372584e-05,
+      "loss": 0.6376,
+      "step": 1212
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.434429346253794e-05,
+      "loss": 0.6709,
+      "step": 1214
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4325190253792222e-05,
+      "loss": 0.7009,
+      "step": 1216
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.430606760897934e-05,
+      "loss": 0.7033,
+      "step": 1218
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4286925614030542e-05,
+      "loss": 0.6607,
+      "step": 1220
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4267764354964038e-05,
+      "loss": 0.6627,
+      "step": 1222
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4248583917884595e-05,
+      "loss": 0.7147,
+      "step": 1224
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4229384388983167e-05,
+      "loss": 0.6639,
+      "step": 1226
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4210165854536495e-05,
+      "loss": 0.6839,
+      "step": 1228
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4190928400906731e-05,
+      "loss": 0.6975,
+      "step": 1230
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4171672114541042e-05,
+      "loss": 0.6952,
+      "step": 1232
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.415239708197122e-05,
+      "loss": 0.6946,
+      "step": 1234
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4133103389813302e-05,
+      "loss": 0.6661,
+      "step": 1236
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.411379112476717e-05,
+      "loss": 0.6631,
+      "step": 1238
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.409446037361617e-05,
+      "loss": 0.6417,
+      "step": 1240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4075111223226721e-05,
+      "loss": 0.677,
+      "step": 1242
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4055743760547918e-05,
+      "loss": 0.6788,
+      "step": 1244
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4036358072611147e-05,
+      "loss": 0.6612,
+      "step": 1246
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4016954246529697e-05,
+      "loss": 0.6838,
+      "step": 1248
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3997532369498357e-05,
+      "loss": 0.7189,
+      "step": 1250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3978092528793032e-05,
+      "loss": 0.6397,
+      "step": 1252
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3958634811770361e-05,
+      "loss": 0.6933,
+      "step": 1254
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.39391593058673e-05,
+      "loss": 0.6961,
+      "step": 1256
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3919666098600753e-05,
+      "loss": 0.6556,
+      "step": 1258
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3900155277567157e-05,
+      "loss": 0.6771,
+      "step": 1260
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3880626930442115e-05,
+      "loss": 0.7032,
+      "step": 1262
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3861081144979975e-05,
+      "loss": 0.6713,
+      "step": 1264
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3841518009013446e-05,
+      "loss": 0.7388,
+      "step": 1266
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.382193761045322e-05,
+      "loss": 0.6844,
+      "step": 1268
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.380234003728754e-05,
+      "loss": 0.6441,
+      "step": 1270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.378272537758185e-05,
+      "loss": 0.6997,
+      "step": 1272
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3763093719478357e-05,
+      "loss": 0.6808,
+      "step": 1274
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3743445151195658e-05,
+      "loss": 0.6849,
+      "step": 1276
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3723779761028349e-05,
+      "loss": 0.6796,
+      "step": 1278
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.37040976373466e-05,
+      "loss": 0.6406,
+      "step": 1280
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.36843988685958e-05,
+      "loss": 0.727,
+      "step": 1282
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3664683543296114e-05,
+      "loss": 0.6628,
+      "step": 1284
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3644951750042115e-05,
+      "loss": 0.6442,
+      "step": 1286
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3625203577502384e-05,
+      "loss": 0.6828,
+      "step": 1288
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3605439114419095e-05,
+      "loss": 0.696,
+      "step": 1290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3585658449607632e-05,
+      "loss": 0.6949,
+      "step": 1292
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3565861671956189e-05,
+      "loss": 0.6909,
+      "step": 1294
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 0.7021,
+      "step": 1296
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.352622013404774e-05,
+      "loss": 0.6814,
+      "step": 1298
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3506375551927546e-05,
+      "loss": 0.6632,
+      "step": 1300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3486515213240188e-05,
+      "loss": 0.6376,
+      "step": 1302
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3466639207231882e-05,
+      "loss": 0.698,
+      "step": 1304
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3446747623219256e-05,
+      "loss": 0.6847,
+      "step": 1306
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3426840550588933e-05,
+      "loss": 0.6574,
+      "step": 1308
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3406918078797142e-05,
+      "loss": 0.6717,
+      "step": 1310
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3386980297369308e-05,
+      "loss": 0.6421,
+      "step": 1312
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3367027295899652e-05,
+      "loss": 0.6802,
+      "step": 1314
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3347059164050796e-05,
+      "loss": 0.7005,
+      "step": 1316
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.332707599155334e-05,
+      "loss": 0.7182,
+      "step": 1318
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3307077868205487e-05,
+      "loss": 0.6878,
+      "step": 1320
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3287064883872614e-05,
+      "loss": 0.7044,
+      "step": 1322
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3267037128486883e-05,
+      "loss": 0.6779,
+      "step": 1324
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3246994692046837e-05,
+      "loss": 0.667,
+      "step": 1326
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3226937664616977e-05,
+      "loss": 0.6598,
+      "step": 1328
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3206866136327387e-05,
+      "loss": 0.6672,
+      "step": 1330
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3186780197373306e-05,
+      "loss": 0.6835,
+      "step": 1332
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3166679938014728e-05,
+      "loss": 0.6901,
+      "step": 1334
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3146565448576002e-05,
+      "loss": 0.654,
+      "step": 1336
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3126436819445423e-05,
+      "loss": 0.661,
+      "step": 1338
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3106294141074825e-05,
+      "loss": 0.6613,
+      "step": 1340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.308613750397917e-05,
+      "loss": 0.6717,
+      "step": 1342
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3065966998736155e-05,
+      "loss": 0.6694,
+      "step": 1344
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.304578271598579e-05,
+      "loss": 0.6875,
+      "step": 1346
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.302558474643e-05,
+      "loss": 0.6482,
+      "step": 1348
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.300537318083221e-05,
+      "loss": 0.6687,
+      "step": 1350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2985148110016947e-05,
+      "loss": 0.6992,
+      "step": 1352
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2964909624869422e-05,
+      "loss": 0.6722,
+      "step": 1354
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2944657816335124e-05,
+      "loss": 0.6592,
+      "step": 1356
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2924392775419421e-05,
+      "loss": 0.7048,
+      "step": 1358
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2904114593187136e-05,
+      "loss": 0.6543,
+      "step": 1360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2883823360762149e-05,
+      "loss": 0.6909,
+      "step": 1362
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2863519169326984e-05,
+      "loss": 0.658,
+      "step": 1364
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.284320211012239e-05,
+      "loss": 0.6825,
+      "step": 1366
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2822872274446958e-05,
+      "loss": 0.648,
+      "step": 1368
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.280252975365667e-05,
+      "loss": 0.642,
+      "step": 1370
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2782174639164528e-05,
+      "loss": 0.6841,
+      "step": 1372
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.276180702244012e-05,
+      "loss": 0.674,
+      "step": 1374
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2741426995009214e-05,
+      "loss": 0.6792,
+      "step": 1376
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2721034648453353e-05,
+      "loss": 0.6756,
+      "step": 1378
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2700630074409427e-05,
+      "loss": 0.6637,
+      "step": 1380
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.268021336456929e-05,
+      "loss": 0.6692,
+      "step": 1382
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2659784610679318e-05,
+      "loss": 0.6655,
+      "step": 1384
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2639343904540008e-05,
+      "loss": 0.6875,
+      "step": 1386
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2618891338005574e-05,
+      "loss": 0.645,
+      "step": 1388
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2598427002983523e-05,
+      "loss": 0.6756,
+      "step": 1390
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2577950991434249e-05,
+      "loss": 0.6815,
+      "step": 1392
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2557463395370609e-05,
+      "loss": 0.6962,
+      "step": 1394
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2536964306857526e-05,
+      "loss": 0.6513,
+      "step": 1396
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2516453818011567e-05,
+      "loss": 0.6645,
+      "step": 1398
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2495932021000516e-05,
+      "loss": 0.6686,
+      "step": 1400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.247539900804299e-05,
+      "loss": 0.6504,
+      "step": 1402
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2454854871407993e-05,
+      "loss": 0.6844,
+      "step": 1404
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2434299703414523e-05,
+      "loss": 0.6789,
+      "step": 1406
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2413733596431141e-05,
+      "loss": 0.6846,
+      "step": 1408
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2393156642875579e-05,
+      "loss": 0.6725,
+      "step": 1410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2372568935214298e-05,
+      "loss": 0.6694,
+      "step": 1412
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2351970565962083e-05,
+      "loss": 0.6641,
+      "step": 1414
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2331361627681645e-05,
+      "loss": 0.6932,
+      "step": 1416
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2310742212983167e-05,
+      "loss": 0.68,
+      "step": 1418
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2290112414523927e-05,
+      "loss": 0.6809,
+      "step": 1420
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2269472325007858e-05,
+      "loss": 0.6702,
+      "step": 1422
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2248822037185137e-05,
+      "loss": 0.6505,
+      "step": 1424
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.222816164385177e-05,
+      "loss": 0.6658,
+      "step": 1426
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2207491237849174e-05,
+      "loss": 0.6188,
+      "step": 1428
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.218681091206376e-05,
+      "loss": 0.6844,
+      "step": 1430
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2166120759426515e-05,
+      "loss": 0.6565,
+      "step": 1432
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2145420872912586e-05,
+      "loss": 0.6683,
+      "step": 1434
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2124711345540861e-05,
+      "loss": 0.6803,
+      "step": 1436
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2103992270373548e-05,
+      "loss": 0.66,
+      "step": 1438
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2083263740515764e-05,
+      "loss": 0.6937,
+      "step": 1440
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2062525849115107e-05,
+      "loss": 0.6634,
+      "step": 1442
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2041778689361254e-05,
+      "loss": 0.694,
+      "step": 1444
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2021022354485514e-05,
+      "loss": 0.6357,
+      "step": 1446
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2000256937760446e-05,
+      "loss": 0.6816,
+      "step": 1448
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1979482532499402e-05,
+      "loss": 0.6762,
+      "step": 1450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1958699232056135e-05,
+      "loss": 0.6622,
+      "step": 1452
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1937907129824371e-05,
+      "loss": 0.6793,
+      "step": 1454
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1917106319237386e-05,
+      "loss": 0.6721,
+      "step": 1456
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1896296893767588e-05,
+      "loss": 0.6635,
+      "step": 1458
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1875478946926094e-05,
+      "loss": 0.6895,
+      "step": 1460
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1854652572262324e-05,
+      "loss": 0.6156,
+      "step": 1462
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1833817863363563e-05,
+      "loss": 0.6605,
+      "step": 1464
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1812974913854545e-05,
+      "loss": 0.6532,
+      "step": 1466
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1792123817397041e-05,
+      "loss": 0.6317,
+      "step": 1468
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1771264667689428e-05,
+      "loss": 0.6231,
+      "step": 1470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1750397558466273e-05,
+      "loss": 0.6719,
+      "step": 1472
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1729522583497913e-05,
+      "loss": 0.6393,
+      "step": 1474
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1708639836590024e-05,
+      "loss": 0.6373,
+      "step": 1476
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1687749411583213e-05,
+      "loss": 0.687,
+      "step": 1478
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1666851402352587e-05,
+      "loss": 0.6593,
+      "step": 1480
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.164594590280734e-05,
+      "loss": 0.6433,
+      "step": 1482
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1625033006890316e-05,
+      "loss": 0.6282,
+      "step": 1484
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1604112808577603e-05,
+      "loss": 0.6786,
+      "step": 1486
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.15831854018781e-05,
+      "loss": 0.6407,
+      "step": 1488
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1562250880833103e-05,
+      "loss": 0.6557,
+      "step": 1490
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.154130933951587e-05,
+      "loss": 0.6797,
+      "step": 1492
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1520360872031208e-05,
+      "loss": 0.6817,
+      "step": 1494
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1499405572515059e-05,
+      "loss": 0.6411,
+      "step": 1496
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.147844353513405e-05,
+      "loss": 0.6404,
+      "step": 1498
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1457474854085095e-05,
+      "loss": 0.702,
+      "step": 1500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1436499623594963e-05,
+      "loss": 0.6632,
+      "step": 1502
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1415517937919846e-05,
+      "loss": 0.6872,
+      "step": 1504
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.139452989134496e-05,
+      "loss": 0.6829,
+      "step": 1506
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1373535578184083e-05,
+      "loss": 0.6821,
+      "step": 1508
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1352535092779174e-05,
+      "loss": 0.6751,
+      "step": 1510
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1331528529499909e-05,
+      "loss": 0.647,
+      "step": 1512
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1310515982743293e-05,
+      "loss": 0.6522,
+      "step": 1514
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1289497546933212e-05,
+      "loss": 0.6586,
+      "step": 1516
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1268473316520007e-05,
+      "loss": 0.6563,
+      "step": 1518
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.124744338598008e-05,
+      "loss": 0.6724,
+      "step": 1520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1226407849815422e-05,
+      "loss": 0.6247,
+      "step": 1522
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 0.6723,
+      "step": 1524
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1184320338745467e-05,
+      "loss": 0.6379,
+      "step": 1526
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1163268552968422e-05,
+      "loss": 0.6515,
+      "step": 1528
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1142211539822318e-05,
+      "loss": 0.6712,
+      "step": 1530
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.112114939393085e-05,
+      "loss": 0.6779,
+      "step": 1532
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1100082209940795e-05,
+      "loss": 0.6767,
+      "step": 1534
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1079010082521557e-05,
+      "loss": 0.6324,
+      "step": 1536
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1057933106364758e-05,
+      "loss": 0.6992,
+      "step": 1538
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1036851376183812e-05,
+      "loss": 0.6635,
+      "step": 1540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.101576498671349e-05,
+      "loss": 0.6659,
+      "step": 1542
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0994674032709514e-05,
+      "loss": 0.6141,
+      "step": 1544
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0973578608948094e-05,
+      "loss": 0.6541,
+      "step": 1546
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.095247881022555e-05,
+      "loss": 0.6571,
+      "step": 1548
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0931374731357843e-05,
+      "loss": 0.6636,
+      "step": 1550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.091026646718018e-05,
+      "loss": 0.6615,
+      "step": 1552
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0889154112546569e-05,
+      "loss": 0.6636,
+      "step": 1554
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0868037762329405e-05,
+      "loss": 0.6838,
+      "step": 1556
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0846917511419031e-05,
+      "loss": 0.6399,
+      "step": 1558
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0825793454723325e-05,
+      "loss": 0.6657,
+      "step": 1560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0804665687167262e-05,
+      "loss": 0.3384,
+      "step": 1562
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0783534303692493e-05,
+      "loss": 0.3217,
+      "step": 1564
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0762399399256917e-05,
+      "loss": 0.3226,
+      "step": 1566
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0741261068834266e-05,
+      "loss": 0.3234,
+      "step": 1568
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0720119407413647e-05,
+      "loss": 0.3116,
+      "step": 1570
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0698974509999159e-05,
+      "loss": 0.3009,
+      "step": 1572
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0677826471609423e-05,
+      "loss": 0.3139,
+      "step": 1574
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0656675387277183e-05,
+      "loss": 0.3013,
+      "step": 1576
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0635521352048873e-05,
+      "loss": 0.3091,
+      "step": 1578
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0614364460984178e-05,
+      "loss": 0.2875,
+      "step": 1580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0593204809155628e-05,
+      "loss": 0.276,
+      "step": 1582
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.057204249164815e-05,
+      "loss": 0.2999,
+      "step": 1584
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0550877603558656e-05,
+      "loss": 0.2988,
+      "step": 1586
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0529710239995606e-05,
+      "loss": 0.3077,
+      "step": 1588
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0508540496078582e-05,
+      "loss": 0.332,
+      "step": 1590
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0487368466937866e-05,
+      "loss": 0.2911,
+      "step": 1592
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0466194247714008e-05,
+      "loss": 0.2854,
+      "step": 1594
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0445017933557404e-05,
+      "loss": 0.2989,
+      "step": 1596
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0423839619627854e-05,
+      "loss": 0.2874,
+      "step": 1598
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0402659401094154e-05,
+      "loss": 0.3164,
+      "step": 1600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0381477373133652e-05,
+      "loss": 0.2996,
+      "step": 1602
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.036029363093183e-05,
+      "loss": 0.3044,
+      "step": 1604
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0339108269681874e-05,
+      "loss": 0.3236,
+      "step": 1606
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0317921384584245e-05,
+      "loss": 0.3203,
+      "step": 1608
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0296733070846252e-05,
+      "loss": 0.3005,
+      "step": 1610
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0275543423681622e-05,
+      "loss": 0.2955,
+      "step": 1612
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0254352538310075e-05,
+      "loss": 0.3222,
+      "step": 1614
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0233160509956893e-05,
+      "loss": 0.3012,
+      "step": 1616
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0211967433852501e-05,
+      "loss": 0.2842,
+      "step": 1618
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0190773405232024e-05,
+      "loss": 0.2941,
+      "step": 1620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0169578519334873e-05,
+      "loss": 0.3058,
+      "step": 1622
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.014838287140431e-05,
+      "loss": 0.31,
+      "step": 1624
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.012718655668702e-05,
+      "loss": 0.3019,
+      "step": 1626
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.010598967043268e-05,
+      "loss": 0.3079,
+      "step": 1628
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0084792307893551e-05,
+      "loss": 0.2947,
+      "step": 1630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0063594564324014e-05,
+      "loss": 0.2863,
+      "step": 1632
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0042396534980176e-05,
+      "loss": 0.3025,
+      "step": 1634
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0021198315119426e-05,
+      "loss": 0.2999,
+      "step": 1636
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1e-05,
+      "loss": 0.29,
+      "step": 1638
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.97880168488058e-06,
+      "loss": 0.2966,
+      "step": 1640
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.957603465019827e-06,
+      "loss": 0.3113,
+      "step": 1642
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.936405435675991e-06,
+      "loss": 0.2922,
+      "step": 1644
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.91520769210645e-06,
+      "loss": 0.2946,
+      "step": 1646
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.894010329567322e-06,
+      "loss": 0.2964,
+      "step": 1648
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.872813443312984e-06,
+      "loss": 0.2843,
+      "step": 1650
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.851617128595694e-06,
+      "loss": 0.3191,
+      "step": 1652
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.830421480665128e-06,
+      "loss": 0.2967,
+      "step": 1654
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.809226594767979e-06,
+      "loss": 0.3098,
+      "step": 1656
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.788032566147506e-06,
+      "loss": 0.2984,
+      "step": 1658
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.766839490043108e-06,
+      "loss": 0.2802,
+      "step": 1660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.745647461689932e-06,
+      "loss": 0.2973,
+      "step": 1662
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.724456576318383e-06,
+      "loss": 0.2894,
+      "step": 1664
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.703266929153751e-06,
+      "loss": 0.2967,
+      "step": 1666
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.682078615415755e-06,
+      "loss": 0.2898,
+      "step": 1668
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.660891730318129e-06,
+      "loss": 0.3209,
+      "step": 1670
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.63970636906817e-06,
+      "loss": 0.2937,
+      "step": 1672
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.618522626866351e-06,
+      "loss": 0.3105,
+      "step": 1674
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.597340598905851e-06,
+      "loss": 0.312,
+      "step": 1676
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.57616038037215e-06,
+      "loss": 0.2894,
+      "step": 1678
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.554982066442601e-06,
+      "loss": 0.3153,
+      "step": 1680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.533805752285992e-06,
+      "loss": 0.2843,
+      "step": 1682
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.512631533062138e-06,
+      "loss": 0.2865,
+      "step": 1684
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.49145950392142e-06,
+      "loss": 0.3006,
+      "step": 1686
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.470289760004398e-06,
+      "loss": 0.2958,
+      "step": 1688
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.449122396441344e-06,
+      "loss": 0.2926,
+      "step": 1690
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.427957508351852e-06,
+      "loss": 0.2962,
+      "step": 1692
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.406795190844376e-06,
+      "loss": 0.2961,
+      "step": 1694
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.385635539015824e-06,
+      "loss": 0.3022,
+      "step": 1696
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.364478647951132e-06,
+      "loss": 0.3152,
+      "step": 1698
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.343324612722819e-06,
+      "loss": 0.2987,
+      "step": 1700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.32217352839058e-06,
+      "loss": 0.3159,
+      "step": 1702
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.301025490000843e-06,
+      "loss": 0.3118,
+      "step": 1704
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.279880592586355e-06,
+      "loss": 0.294,
+      "step": 1706
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.25873893116574e-06,
+      "loss": 0.3079,
+      "step": 1708
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.237600600743086e-06,
+      "loss": 0.2939,
+      "step": 1710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.216465696307513e-06,
+      "loss": 0.2924,
+      "step": 1712
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.195334312832742e-06,
+      "loss": 0.279,
+      "step": 1714
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.174206545276678e-06,
+      "loss": 0.2767,
+      "step": 1716
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.153082488580969e-06,
+      "loss": 0.2925,
+      "step": 1718
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.131962237670599e-06,
+      "loss": 0.2965,
+      "step": 1720
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.11084588745343e-06,
+      "loss": 0.2946,
+      "step": 1722
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.089733532819825e-06,
+      "loss": 0.3002,
+      "step": 1724
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.068625268642162e-06,
+      "loss": 0.2934,
+      "step": 1726
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.047521189774456e-06,
+      "loss": 0.308,
+      "step": 1728
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.026421391051908e-06,
+      "loss": 0.2945,
+      "step": 1730
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.005325967290489e-06,
+      "loss": 0.2886,
+      "step": 1732
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.984235013286512e-06,
+      "loss": 0.3069,
+      "step": 1734
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.963148623816191e-06,
+      "loss": 0.3006,
+      "step": 1736
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.942066893635245e-06,
+      "loss": 0.2986,
+      "step": 1738
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.920989917478446e-06,
+      "loss": 0.2964,
+      "step": 1740
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.899917790059208e-06,
+      "loss": 0.2792,
+      "step": 1742
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.878850606069152e-06,
+      "loss": 0.3028,
+      "step": 1744
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.857788460177685e-06,
+      "loss": 0.3172,
+      "step": 1746
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.836731447031581e-06,
+      "loss": 0.2767,
+      "step": 1748
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.815679661254538e-06,
+      "loss": 0.2948,
+      "step": 1750
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 0.2942,
+      "step": 1752
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.77359215018458e-06,
+      "loss": 0.2977,
+      "step": 1754
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.752556614019924e-06,
+      "loss": 0.2903,
+      "step": 1756
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.731526683479991e-06,
+      "loss": 0.3057,
+      "step": 1758
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.710502453066791e-06,
+      "loss": 0.2959,
+      "step": 1760
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.689484017256712e-06,
+      "loss": 0.3104,
+      "step": 1762
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.668471470500094e-06,
+      "loss": 0.2941,
+      "step": 1764
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.647464907220833e-06,
+      "loss": 0.3048,
+      "step": 1766
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.626464421815919e-06,
+      "loss": 0.2999,
+      "step": 1768
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.605470108655046e-06,
+      "loss": 0.2885,
+      "step": 1770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.584482062080154e-06,
+      "loss": 0.289,
+      "step": 1772
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.56350037640504e-06,
+      "loss": 0.2921,
+      "step": 1774
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.542525145914907e-06,
+      "loss": 0.2813,
+      "step": 1776
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.521556464865955e-06,
+      "loss": 0.2843,
+      "step": 1778
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.500594427484946e-06,
+      "loss": 0.3057,
+      "step": 1780
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.479639127968793e-06,
+      "loss": 0.2842,
+      "step": 1782
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.458690660484134e-06,
+      "loss": 0.2703,
+      "step": 1784
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.4377491191669e-06,
+      "loss": 0.2838,
+      "step": 1786
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.416814598121901e-06,
+      "loss": 0.2977,
+      "step": 1788
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.395887191422397e-06,
+      "loss": 0.2726,
+      "step": 1790
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.374966993109689e-06,
+      "loss": 0.2827,
+      "step": 1792
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.35405409719266e-06,
+      "loss": 0.2787,
+      "step": 1794
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.333148597647414e-06,
+      "loss": 0.3053,
+      "step": 1796
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.312250588416792e-06,
+      "loss": 0.2894,
+      "step": 1798
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.291360163409978e-06,
+      "loss": 0.2838,
+      "step": 1800
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.27047741650209e-06,
+      "loss": 0.3018,
+      "step": 1802
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.249602441533727e-06,
+      "loss": 0.2988,
+      "step": 1804
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.228735332310575e-06,
+      "loss": 0.2973,
+      "step": 1806
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.207876182602959e-06,
+      "loss": 0.2712,
+      "step": 1808
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.187025086145458e-06,
+      "loss": 0.2916,
+      "step": 1810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 8.16618213663644e-06,
+      "loss": 0.2871,
+      "step": 1812
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.14534742773768e-06,
+      "loss": 0.2842,
+      "step": 1814
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.12452105307391e-06,
+      "loss": 0.2888,
+      "step": 1816
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.103703106232416e-06,
+      "loss": 0.3169,
+      "step": 1818
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 8.082893680762619e-06,
+      "loss": 0.2979,
+      "step": 1820
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.062092870175629e-06,
+      "loss": 0.2896,
+      "step": 1822
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.041300767943867e-06,
+      "loss": 0.2961,
+      "step": 1824
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 8.0205174675006e-06,
+      "loss": 0.2889,
+      "step": 1826
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.999743062239557e-06,
+      "loss": 0.301,
+      "step": 1828
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.978977645514488e-06,
+      "loss": 0.2911,
+      "step": 1830
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.958221310638749e-06,
+      "loss": 0.2906,
+      "step": 1832
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.937474150884898e-06,
+      "loss": 0.297,
+      "step": 1834
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.916736259484239e-06,
+      "loss": 0.2907,
+      "step": 1836
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.896007729626457e-06,
+      "loss": 0.2909,
+      "step": 1838
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.875288654459144e-06,
+      "loss": 0.292,
+      "step": 1840
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.854579127087418e-06,
+      "loss": 0.2956,
+      "step": 1842
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.833879240573487e-06,
+      "loss": 0.2963,
+      "step": 1844
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.813189087936243e-06,
+      "loss": 0.3009,
+      "step": 1846
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.792508762150833e-06,
+      "loss": 0.2882,
+      "step": 1848
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.771838356148231e-06,
+      "loss": 0.3053,
+      "step": 1850
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.751177962814867e-06,
+      "loss": 0.2846,
+      "step": 1852
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.730527674992143e-06,
+      "loss": 0.2963,
+      "step": 1854
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.709887585476075e-06,
+      "loss": 0.2911,
+      "step": 1856
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.689257787016834e-06,
+      "loss": 0.2831,
+      "step": 1858
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.668638372318359e-06,
+      "loss": 0.2957,
+      "step": 1860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.648029434037915e-06,
+      "loss": 0.2871,
+      "step": 1862
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.627431064785705e-06,
+      "loss": 0.26,
+      "step": 1864
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.606843357124426e-06,
+      "loss": 0.2909,
+      "step": 1866
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.5862664035688604e-06,
+      "loss": 0.2835,
+      "step": 1868
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.565700296585483e-06,
+      "loss": 0.303,
+      "step": 1870
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.545145128592009e-06,
+      "loss": 0.2841,
+      "step": 1872
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.524600991957013e-06,
+      "loss": 0.2881,
+      "step": 1874
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.504067978999484e-06,
+      "loss": 0.2958,
+      "step": 1876
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.483546181988437e-06,
+      "loss": 0.2905,
+      "step": 1878
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.463035693142473e-06,
+      "loss": 0.2774,
+      "step": 1880
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.442536604629396e-06,
+      "loss": 0.2847,
+      "step": 1882
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.422049008565757e-06,
+      "loss": 0.3078,
+      "step": 1884
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.40157299701648e-06,
+      "loss": 0.3036,
+      "step": 1886
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.38110866199443e-06,
+      "loss": 0.3172,
+      "step": 1888
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.360656095459995e-06,
+      "loss": 0.2815,
+      "step": 1890
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.340215389320686e-06,
+      "loss": 0.2841,
+      "step": 1892
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.31978663543071e-06,
+      "loss": 0.2834,
+      "step": 1894
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.299369925590575e-06,
+      "loss": 0.2891,
+      "step": 1896
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.278965351546648e-06,
+      "loss": 0.2928,
+      "step": 1898
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.258573004990789e-06,
+      "loss": 0.3074,
+      "step": 1900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.2381929775598835e-06,
+      "loss": 0.2841,
+      "step": 1902
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.217825360835475e-06,
+      "loss": 0.3,
+      "step": 1904
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.197470246343333e-06,
+      "loss": 0.2773,
+      "step": 1906
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.1771277255530456e-06,
+      "loss": 0.28,
+      "step": 1908
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.156797889877613e-06,
+      "loss": 0.2912,
+      "step": 1910
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.136480830673018e-06,
+      "loss": 0.2723,
+      "step": 1912
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 7.116176639237853e-06,
+      "loss": 0.285,
+      "step": 1914
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.095885406812866e-06,
+      "loss": 0.2734,
+      "step": 1916
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.075607224580581e-06,
+      "loss": 0.2895,
+      "step": 1918
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.05534218366488e-06,
+      "loss": 0.2972,
+      "step": 1920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.035090375130581e-06,
+      "loss": 0.285,
+      "step": 1922
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 7.014851889983058e-06,
+      "loss": 0.2692,
+      "step": 1924
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.994626819167789e-06,
+      "loss": 0.2743,
+      "step": 1926
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.974415253570003e-06,
+      "loss": 0.2884,
+      "step": 1928
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.9542172840142105e-06,
+      "loss": 0.2915,
+      "step": 1930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.934033001263847e-06,
+      "loss": 0.2994,
+      "step": 1932
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.913862496020831e-06,
+      "loss": 0.3014,
+      "step": 1934
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.893705858925179e-06,
+      "loss": 0.2946,
+      "step": 1936
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.873563180554583e-06,
+      "loss": 0.3003,
+      "step": 1938
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.853434551424001e-06,
+      "loss": 0.2781,
+      "step": 1940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.833320061985278e-06,
+      "loss": 0.2902,
+      "step": 1942
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.813219802626698e-06,
+      "loss": 0.2727,
+      "step": 1944
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.793133863672616e-06,
+      "loss": 0.2725,
+      "step": 1946
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.773062335383024e-06,
+      "loss": 0.2882,
+      "step": 1948
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.7530053079531664e-06,
+      "loss": 0.2661,
+      "step": 1950
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.73296287151312e-06,
+      "loss": 0.2777,
+      "step": 1952
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.712935116127389e-06,
+      "loss": 0.3,
+      "step": 1954
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.692922131794517e-06,
+      "loss": 0.2839,
+      "step": 1956
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.672924008446663e-06,
+      "loss": 0.2947,
+      "step": 1958
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.652940835949208e-06,
+      "loss": 0.2744,
+      "step": 1960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.632972704100349e-06,
+      "loss": 0.2881,
+      "step": 1962
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.6130197026306945e-06,
+      "loss": 0.2877,
+      "step": 1964
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.59308192120286e-06,
+      "loss": 0.2955,
+      "step": 1966
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.573159449411071e-06,
+      "loss": 0.2739,
+      "step": 1968
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.553252376780749e-06,
+      "loss": 0.2795,
+      "step": 1970
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.533360792768122e-06,
+      "loss": 0.2874,
+      "step": 1972
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.513484786759818e-06,
+      "loss": 0.2883,
+      "step": 1974
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.4936244480724575e-06,
+      "loss": 0.2895,
+      "step": 1976
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.473779865952263e-06,
+      "loss": 0.2929,
+      "step": 1978
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 0.2896,
+      "step": 1980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.434138328043815e-06,
+      "loss": 0.2796,
+      "step": 1982
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.4143415503923676e-06,
+      "loss": 0.2947,
+      "step": 1984
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.39456088558091e-06,
+      "loss": 0.2776,
+      "step": 1986
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.374796422497622e-06,
+      "loss": 0.2864,
+      "step": 1988
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.355048249957887e-06,
+      "loss": 0.2671,
+      "step": 1990
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.335316456703891e-06,
+      "loss": 0.292,
+      "step": 1992
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.315601131404201e-06,
+      "loss": 0.2857,
+      "step": 1994
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.295902362653401e-06,
+      "loss": 0.284,
+      "step": 1996
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.276220238971653e-06,
+      "loss": 0.2903,
+      "step": 1998
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.256554848804343e-06,
+      "loss": 0.2727,
+      "step": 2000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.236906280521646e-06,
+      "loss": 0.2874,
+      "step": 2002
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.2172746224181524e-06,
+      "loss": 0.2713,
+      "step": 2004
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.197659962712461e-06,
+      "loss": 0.2758,
+      "step": 2006
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.178062389546784e-06,
+      "loss": 0.2754,
+      "step": 2008
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.158481990986558e-06,
+      "loss": 0.2708,
+      "step": 2010
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.138918855020028e-06,
+      "loss": 0.2837,
+      "step": 2012
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.119373069557889e-06,
+      "loss": 0.2761,
+      "step": 2014
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.099844722432844e-06,
+      "loss": 0.267,
+      "step": 2016
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.080333901399252e-06,
+      "loss": 0.2831,
+      "step": 2018
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.060840694132701e-06,
+      "loss": 0.2714,
+      "step": 2020
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.041365188229641e-06,
+      "loss": 0.28,
+      "step": 2022
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.021907471206971e-06,
+      "loss": 0.2567,
+      "step": 2024
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.002467630501646e-06,
+      "loss": 0.2911,
+      "step": 2026
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.983045753470308e-06,
+      "loss": 0.2929,
+      "step": 2028
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.963641927388854e-06,
+      "loss": 0.2825,
+      "step": 2030
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.944256239452085e-06,
+      "loss": 0.2765,
+      "step": 2032
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.924888776773281e-06,
+      "loss": 0.2777,
+      "step": 2034
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.905539626383831e-06,
+      "loss": 0.2886,
+      "step": 2036
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.886208875232833e-06,
+      "loss": 0.2633,
+      "step": 2038
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.866896610186701e-06,
+      "loss": 0.2765,
+      "step": 2040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.847602918028785e-06,
+      "loss": 0.2805,
+      "step": 2042
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.82832788545896e-06,
+      "loss": 0.275,
+      "step": 2044
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.809071599093272e-06,
+      "loss": 0.2917,
+      "step": 2046
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.789834145463506e-06,
+      "loss": 0.2831,
+      "step": 2048
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.7706156110168384e-06,
+      "loss": 0.2996,
+      "step": 2050
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.7514160821154085e-06,
+      "loss": 0.2871,
+      "step": 2052
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.732235645035964e-06,
+      "loss": 0.2763,
+      "step": 2054
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.713074385969457e-06,
+      "loss": 0.2852,
+      "step": 2056
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.693932391020664e-06,
+      "loss": 0.2823,
+      "step": 2058
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.67480974620778e-06,
+      "loss": 0.2793,
+      "step": 2060
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.65570653746206e-06,
+      "loss": 0.2825,
+      "step": 2062
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.63662285062742e-06,
+      "loss": 0.2792,
+      "step": 2064
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.617558771460039e-06,
+      "loss": 0.2879,
+      "step": 2066
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.598514385627997e-06,
+      "loss": 0.2845,
+      "step": 2068
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.579489778710867e-06,
+      "loss": 0.2784,
+      "step": 2070
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.56048503619935e-06,
+      "loss": 0.279,
+      "step": 2072
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.541500243494889e-06,
+      "loss": 0.2707,
+      "step": 2074
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.522535485909258e-06,
+      "loss": 0.2837,
+      "step": 2076
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.5035908486642195e-06,
+      "loss": 0.264,
+      "step": 2078
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.484666416891109e-06,
+      "loss": 0.2671,
+      "step": 2080
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.465762275630471e-06,
+      "loss": 0.2794,
+      "step": 2082
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.446878509831668e-06,
+      "loss": 0.2742,
+      "step": 2084
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.428015204352508e-06,
+      "loss": 0.268,
+      "step": 2086
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.409172443958844e-06,
+      "loss": 0.2719,
+      "step": 2088
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.390350313324214e-06,
+      "loss": 0.2547,
+      "step": 2090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.371548897029457e-06,
+      "loss": 0.294,
+      "step": 2092
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.352768279562315e-06,
+      "loss": 0.2635,
+      "step": 2094
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.334008545317082e-06,
+      "loss": 0.2829,
+      "step": 2096
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.315269778594189e-06,
+      "loss": 0.2619,
+      "step": 2098
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.2965520635998676e-06,
+      "loss": 0.2679,
+      "step": 2100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.277855484445736e-06,
+      "loss": 0.2649,
+      "step": 2102
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.259180125148442e-06,
+      "loss": 0.2837,
+      "step": 2104
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.240526069629265e-06,
+      "loss": 0.2901,
+      "step": 2106
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.22189340171377e-06,
+      "loss": 0.2852,
+      "step": 2108
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.203282205131395e-06,
+      "loss": 0.2878,
+      "step": 2110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.184692563515104e-06,
+      "loss": 0.2811,
+      "step": 2112
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.166124560401002e-06,
+      "loss": 0.2644,
+      "step": 2114
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.147578279227943e-06,
+      "loss": 0.2773,
+      "step": 2116
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.129053803337181e-06,
+      "loss": 0.2733,
+      "step": 2118
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.110551215971981e-06,
+      "loss": 0.2809,
+      "step": 2120
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.0920706002772525e-06,
+      "loss": 0.2848,
+      "step": 2122
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.073612039299157e-06,
+      "loss": 0.2946,
+      "step": 2124
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.055175615984766e-06,
+      "loss": 0.2686,
+      "step": 2126
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.036761413181659e-06,
+      "loss": 0.2877,
+      "step": 2128
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.018369513637567e-06,
+      "loss": 0.2693,
+      "step": 2130
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.263,
+      "step": 2132
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.981652954815871e-06,
+      "loss": 0.2569,
+      "step": 2134
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.963328460531127e-06,
+      "loss": 0.2827,
+      "step": 2136
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.9450265994903666e-06,
+      "loss": 0.2666,
+      "step": 2138
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.926747453936509e-06,
+      "loss": 0.2729,
+      "step": 2140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.908491106010368e-06,
+      "loss": 0.2886,
+      "step": 2142
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.890257637750332e-06,
+      "loss": 0.2792,
+      "step": 2144
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.872047131091961e-06,
+      "loss": 0.2821,
+      "step": 2146
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.853859667867641e-06,
+      "loss": 0.2768,
+      "step": 2148
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.8356953298062115e-06,
+      "loss": 0.2915,
+      "step": 2150
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.817554198532582e-06,
+      "loss": 0.2761,
+      "step": 2152
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.799436355567391e-06,
+      "loss": 0.2864,
+      "step": 2154
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.781341882326615e-06,
+      "loss": 0.2785,
+      "step": 2156
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.763270860121222e-06,
+      "loss": 0.2914,
+      "step": 2158
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.745223370156797e-06,
+      "loss": 0.2718,
+      "step": 2160
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.727199493533181e-06,
+      "loss": 0.276,
+      "step": 2162
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.709199311244098e-06,
+      "loss": 0.2768,
+      "step": 2164
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.691222904176791e-06,
+      "loss": 0.2677,
+      "step": 2166
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.673270353111687e-06,
+      "loss": 0.2644,
+      "step": 2168
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.655341738721989e-06,
+      "loss": 0.2595,
+      "step": 2170
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.63743714157335e-06,
+      "loss": 0.2655,
+      "step": 2172
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.619556642123484e-06,
+      "loss": 0.2844,
+      "step": 2174
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.6017003207218294e-06,
+      "loss": 0.277,
+      "step": 2176
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.583868257609171e-06,
+      "loss": 0.2819,
+      "step": 2178
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.566060532917288e-06,
+      "loss": 0.2544,
+      "step": 2180
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.548277226668585e-06,
+      "loss": 0.2706,
+      "step": 2182
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.530518418775734e-06,
+      "loss": 0.2859,
+      "step": 2184
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.512784189041328e-06,
+      "loss": 0.2814,
+      "step": 2186
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.4950746171575135e-06,
+      "loss": 0.2922,
+      "step": 2188
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.477389782705628e-06,
+      "loss": 0.2624,
+      "step": 2190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.459729765155842e-06,
+      "loss": 0.2671,
+      "step": 2192
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.442094643866817e-06,
+      "loss": 0.2673,
+      "step": 2194
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.424484498085335e-06,
+      "loss": 0.2757,
+      "step": 2196
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.406899406945938e-06,
+      "loss": 0.2703,
+      "step": 2198
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.389339449470592e-06,
+      "loss": 0.2667,
+      "step": 2200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.371804704568309e-06,
+      "loss": 0.2826,
+      "step": 2202
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.354295251034811e-06,
+      "loss": 0.2695,
+      "step": 2204
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.336811167552167e-06,
+      "loss": 0.2766,
+      "step": 2206
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 0.2765,
+      "step": 2208
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.301919424897339e-06,
+      "loss": 0.2739,
+      "step": 2210
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.284511922517853e-06,
+      "loss": 0.2828,
+      "step": 2212
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.267130103773911e-06,
+      "loss": 0.2632,
+      "step": 2214
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.249774046774034e-06,
+      "loss": 0.2912,
+      "step": 2216
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.232443829510977e-06,
+      "loss": 0.2659,
+      "step": 2218
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.2151395298613675e-06,
+      "loss": 0.2584,
+      "step": 2220
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.19786122558538e-06,
+      "loss": 0.2818,
+      "step": 2222
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.180608994326371e-06,
+      "loss": 0.2446,
+      "step": 2224
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.163382913610533e-06,
+      "loss": 0.2798,
+      "step": 2226
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.1461830608465385e-06,
+      "loss": 0.264,
+      "step": 2228
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.129009513325211e-06,
+      "loss": 0.2681,
+      "step": 2230
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.111862348219158e-06,
+      "loss": 0.2717,
+      "step": 2232
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0947416425824334e-06,
+      "loss": 0.2645,
+      "step": 2234
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.077647473350201e-06,
+      "loss": 0.2685,
+      "step": 2236
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.060579917338362e-06,
+      "loss": 0.2598,
+      "step": 2238
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.04353905124324e-06,
+      "loss": 0.286,
+      "step": 2240
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.026524951641204e-06,
+      "loss": 0.2704,
+      "step": 2242
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.009537694988372e-06,
+      "loss": 0.2661,
+      "step": 2244
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.99257735762021e-06,
+      "loss": 0.2662,
+      "step": 2246
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.975644015751234e-06,
+      "loss": 0.2635,
+      "step": 2248
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.958737745474638e-06,
+      "loss": 0.2765,
+      "step": 2250
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.941858622761975e-06,
+      "loss": 0.2805,
+      "step": 2252
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.925006723462806e-06,
+      "loss": 0.2895,
+      "step": 2254
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.908182123304344e-06,
+      "loss": 0.2776,
+      "step": 2256
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.891384897891148e-06,
+      "loss": 0.2528,
+      "step": 2258
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.8746151227047455e-06,
+      "loss": 0.2572,
+      "step": 2260
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.857872873103322e-06,
+      "loss": 0.2679,
+      "step": 2262
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.84115822432137e-06,
+      "loss": 0.258,
+      "step": 2264
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.824471251469352e-06,
+      "loss": 0.2815,
+      "step": 2266
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.807812029533362e-06,
+      "loss": 0.2669,
+      "step": 2268
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.7911806333747847e-06,
+      "loss": 0.2831,
+      "step": 2270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.7745771377299758e-06,
+      "loss": 0.2597,
+      "step": 2272
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7580016172099067e-06,
+      "loss": 0.2482,
+      "step": 2274
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7414541462998446e-06,
+      "loss": 0.2715,
+      "step": 2276
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.724934799358998e-06,
+      "loss": 0.2757,
+      "step": 2278
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.708443650620206e-06,
+      "loss": 0.2657,
+      "step": 2280
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.691980774189591e-06,
+      "loss": 0.2694,
+      "step": 2282
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.6755462440462288e-06,
+      "loss": 0.2826,
+      "step": 2284
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.659140134041812e-06,
+      "loss": 0.2612,
+      "step": 2286
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.6427625179003223e-06,
+      "loss": 0.2574,
+      "step": 2288
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.626413469217702e-06,
+      "loss": 0.2723,
+      "step": 2290
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.6100930614615204e-06,
+      "loss": 0.274,
+      "step": 2292
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5938013679706453e-06,
+      "loss": 0.276,
+      "step": 2294
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5775384619549e-06,
+      "loss": 0.2547,
+      "step": 2296
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5613044164947617e-06,
+      "loss": 0.2656,
+      "step": 2298
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5450993045409997e-06,
+      "loss": 0.2686,
+      "step": 2300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5289231989143868e-06,
+      "loss": 0.2559,
+      "step": 2302
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5127761723053313e-06,
+      "loss": 0.2546,
+      "step": 2304
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.496658297273574e-06,
+      "loss": 0.2567,
+      "step": 2306
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.4805696462478634e-06,
+      "loss": 0.2884,
+      "step": 2308
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.46451029152562e-06,
+      "loss": 0.2527,
+      "step": 2310
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.448480305272619e-06,
+      "loss": 0.2686,
+      "step": 2312
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4324797595226567e-06,
+      "loss": 0.2757,
+      "step": 2314
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.41650872617724e-06,
+      "loss": 0.272,
+      "step": 2316
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.400567277005247e-06,
+      "loss": 0.2606,
+      "step": 2318
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.384655483642624e-06,
+      "loss": 0.2664,
+      "step": 2320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3687734175920505e-06,
+      "loss": 0.2516,
+      "step": 2322
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.352921150222612e-06,
+      "loss": 0.2712,
+      "step": 2324
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.337098752769503e-06,
+      "loss": 0.2666,
+      "step": 2326
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.321306296333673e-06,
+      "loss": 0.264,
+      "step": 2328
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.3055438518815486e-06,
+      "loss": 0.2579,
+      "step": 2330
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2898114902446708e-06,
+      "loss": 0.2589,
+      "step": 2332
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.274109282119413e-06,
+      "loss": 0.2539,
+      "step": 2334
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2584372980666344e-06,
+      "loss": 0.2508,
+      "step": 2336
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2427956085113875e-06,
+      "loss": 0.2569,
+      "step": 2338
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2271842837425917e-06,
+      "loss": 0.2525,
+      "step": 2340
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2116033939127024e-06,
+      "loss": 0.0771,
+      "step": 2342
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.1960530090374277e-06,
+      "loss": 0.07,
+      "step": 2344
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.180533198995379e-06,
+      "loss": 0.0702,
+      "step": 2346
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.165044033527789e-06,
+      "loss": 0.067,
+      "step": 2348
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.1495855822381716e-06,
+      "loss": 0.0683,
+      "step": 2350
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.134157914592032e-06,
+      "loss": 0.0676,
+      "step": 2352
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.1187610999165317e-06,
+      "loss": 0.0689,
+      "step": 2354
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.1033952074001882e-06,
+      "loss": 0.0745,
+      "step": 2356
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.088060306092582e-06,
+      "loss": 0.0665,
+      "step": 2358
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.0727564649040066e-06,
+      "loss": 0.0704,
+      "step": 2360
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.057483752605196e-06,
+      "loss": 0.0712,
+      "step": 2362
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.042242237826991e-06,
+      "loss": 0.0719,
+      "step": 2364
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.0270319890600465e-06,
+      "loss": 0.0716,
+      "step": 2366
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.011853074654515e-06,
+      "loss": 0.0706,
+      "step": 2368
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.996705562819747e-06,
+      "loss": 0.0726,
+      "step": 2370
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.981589521623973e-06,
+      "loss": 0.0667,
+      "step": 2372
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.9665050189940013e-06,
+      "loss": 0.072,
+      "step": 2374
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.951452122714926e-06,
+      "loss": 0.0643,
+      "step": 2376
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.936430900429805e-06,
+      "loss": 0.0729,
+      "step": 2378
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.9214414196393702e-06,
+      "loss": 0.0707,
+      "step": 2380
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.906483747701705e-06,
+      "loss": 0.076,
+      "step": 2382
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.8915579518319626e-06,
+      "loss": 0.0693,
+      "step": 2384
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.8766640991020533e-06,
+      "loss": 0.0642,
+      "step": 2386
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.861802256440348e-06,
+      "loss": 0.069,
+      "step": 2388
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.8469724906313677e-06,
+      "loss": 0.0657,
+      "step": 2390
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.8321748683154893e-06,
+      "loss": 0.071,
+      "step": 2392
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.8174094559886535e-06,
+      "loss": 0.0672,
+      "step": 2394
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.8026763200020557e-06,
+      "loss": 0.0701,
+      "step": 2396
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7879755265618558e-06,
+      "loss": 0.0659,
+      "step": 2398
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.773307141728867e-06,
+      "loss": 0.064,
+      "step": 2400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.7586712314182775e-06,
+      "loss": 0.0673,
+      "step": 2402
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.744067861399333e-06,
+      "loss": 0.0718,
+      "step": 2404
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.729497097295075e-06,
+      "loss": 0.0704,
+      "step": 2406
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.714959004582003e-06,
+      "loss": 0.0666,
+      "step": 2408
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.700453648589805e-06,
+      "loss": 0.0639,
+      "step": 2410
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6859810945010687e-06,
+      "loss": 0.0678,
+      "step": 2412
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.671541407350975e-06,
+      "loss": 0.0748,
+      "step": 2414
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.6571346520270147e-06,
+      "loss": 0.0742,
+      "step": 2416
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.642760893268684e-06,
+      "loss": 0.0655,
+      "step": 2418
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.628420195667214e-06,
+      "loss": 0.0666,
+      "step": 2420
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.614112623665259e-06,
+      "loss": 0.067,
+      "step": 2422
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.5998382415566258e-06,
+      "loss": 0.0642,
+      "step": 2424
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.5855971134859736e-06,
+      "loss": 0.0656,
+      "step": 2426
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.5713893034485216e-06,
+      "loss": 0.0677,
+      "step": 2428
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.55721487528978e-06,
+      "loss": 0.0719,
+      "step": 2430
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.5430738927052346e-06,
+      "loss": 0.0746,
+      "step": 2432
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.5289664192400996e-06,
+      "loss": 0.0657,
+      "step": 2434
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 0.0662,
+      "step": 2436
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.5008522530956647e-06,
+      "loss": 0.0666,
+      "step": 2438
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.4868456867527315e-06,
+      "loss": 0.062,
+      "step": 2440
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.4728728822013683e-06,
+      "loss": 0.0654,
+      "step": 2442
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.4589339022310386e-06,
+      "loss": 0.0723,
+      "step": 2444
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.445028809479203e-06,
+      "loss": 0.0644,
+      "step": 2446
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.431157666431052e-06,
+      "loss": 0.0642,
+      "step": 2448
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.4173205354192065e-06,
+      "loss": 0.0708,
+      "step": 2450
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.403517478623456e-06,
+      "loss": 0.0626,
+      "step": 2452
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.3897485580704684e-06,
+      "loss": 0.0689,
+      "step": 2454
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.3760138356335172e-06,
+      "loss": 0.0727,
+      "step": 2456
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.3623133730321923e-06,
+      "loss": 0.0725,
+      "step": 2458
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.348647231832131e-06,
+      "loss": 0.0628,
+      "step": 2460
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.335015473444754e-06,
+      "loss": 0.0675,
+      "step": 2462
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.3214181591269603e-06,
+      "loss": 0.0599,
+      "step": 2464
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.30785534998088e-06,
+      "loss": 0.0657,
+      "step": 2466
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2943271069535754e-06,
+      "loss": 0.0655,
+      "step": 2468
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2808334908367914e-06,
+      "loss": 0.0636,
+      "step": 2470
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.267374562266662e-06,
+      "loss": 0.0728,
+      "step": 2472
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.2539503817234553e-06,
+      "loss": 0.0705,
+      "step": 2474
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.240561009531281e-06,
+      "loss": 0.0623,
+      "step": 2476
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.227206505857834e-06,
+      "loss": 0.0663,
+      "step": 2478
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.2138869307141266e-06,
+      "loss": 0.0638,
+      "step": 2480
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.200602343954209e-06,
+      "loss": 0.0695,
+      "step": 2482
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1873528052749094e-06,
+      "loss": 0.0625,
+      "step": 2484
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.1741383742155476e-06,
+      "loss": 0.0649,
+      "step": 2486
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.1609591101576945e-06,
+      "loss": 0.0636,
+      "step": 2488
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.147815072324886e-06,
+      "loss": 0.0647,
+      "step": 2490
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.1347063197823648e-06,
+      "loss": 0.0686,
+      "step": 2492
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.1216329114368083e-06,
+      "loss": 0.0639,
+      "step": 2494
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.1085949060360654e-06,
+      "loss": 0.0707,
+      "step": 2496
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0955923621689034e-06,
+      "loss": 0.0652,
+      "step": 2498
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0826253382647334e-06,
+      "loss": 0.0685,
+      "step": 2500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0696938925933505e-06,
+      "loss": 0.0694,
+      "step": 2502
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.056798083264667e-06,
+      "loss": 0.0599,
+      "step": 2504
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0439379682284686e-06,
+      "loss": 0.068,
+      "step": 2506
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0311136052741274e-06,
+      "loss": 0.069,
+      "step": 2508
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0183250520303653e-06,
+      "loss": 0.0658,
+      "step": 2510
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.0055723659649907e-06,
+      "loss": 0.06,
+      "step": 2512
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9928556043846215e-06,
+      "loss": 0.069,
+      "step": 2514
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.9801748244344587e-06,
+      "loss": 0.0642,
+      "step": 2516
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.967530083097996e-06,
+      "loss": 0.0663,
+      "step": 2518
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.9549214371968008e-06,
+      "loss": 0.0648,
+      "step": 2520
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.9423489433902186e-06,
+      "loss": 0.0635,
+      "step": 2522
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.9298126581751542e-06,
+      "loss": 0.0646,
+      "step": 2524
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.917312637885791e-06,
+      "loss": 0.0631,
+      "step": 2526
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.9048489386933545e-06,
+      "loss": 0.0711,
+      "step": 2528
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8924216166058574e-06,
+      "loss": 0.0706,
+      "step": 2530
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8800307274678364e-06,
+      "loss": 0.0762,
+      "step": 2532
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8676763269601195e-06,
+      "loss": 0.0664,
+      "step": 2534
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8553584705995564e-06,
+      "loss": 0.0706,
+      "step": 2536
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8430772137387853e-06,
+      "loss": 0.0724,
+      "step": 2538
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8308326115659757e-06,
+      "loss": 0.0627,
+      "step": 2540
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8186247191045857e-06,
+      "loss": 0.0637,
+      "step": 2542
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8064535912131032e-06,
+      "loss": 0.0683,
+      "step": 2544
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7943192825848133e-06,
+      "loss": 0.0602,
+      "step": 2546
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7822218477475496e-06,
+      "loss": 0.0666,
+      "step": 2548
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7701613410634367e-06,
+      "loss": 0.0652,
+      "step": 2550
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7581378167286655e-06,
+      "loss": 0.0683,
+      "step": 2552
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7461513287732313e-06,
+      "loss": 0.0609,
+      "step": 2554
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7342019310607062e-06,
+      "loss": 0.0671,
+      "step": 2556
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7222896772879871e-06,
+      "loss": 0.0625,
+      "step": 2558
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7104146209850591e-06,
+      "loss": 0.0644,
+      "step": 2560
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6985768155147498e-06,
+      "loss": 0.0648,
+      "step": 2562
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6867763140724969e-06,
+      "loss": 0.0604,
+      "step": 2564
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6750131696861016e-06,
+      "loss": 0.066,
+      "step": 2566
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6632874352154982e-06,
+      "loss": 0.0657,
+      "step": 2568
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6515991633525118e-06,
+      "loss": 0.0622,
+      "step": 2570
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.6399484066206183e-06,
+      "loss": 0.0624,
+      "step": 2572
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.6283352173747148e-06,
+      "loss": 0.0648,
+      "step": 2574
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.6167596478008817e-06,
+      "loss": 0.0709,
+      "step": 2576
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6052217499161516e-06,
+      "loss": 0.064,
+      "step": 2578
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5937215755682667e-06,
+      "loss": 0.0666,
+      "step": 2580
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5822591764354512e-06,
+      "loss": 0.0625,
+      "step": 2582
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5708346040261812e-06,
+      "loss": 0.0667,
+      "step": 2584
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.559447909678954e-06,
+      "loss": 0.068,
+      "step": 2586
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.5480991445620541e-06,
+      "loss": 0.0638,
+      "step": 2588
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.5367883596733157e-06,
+      "loss": 0.0609,
+      "step": 2590
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.5255156058399124e-06,
+      "loss": 0.0697,
+      "step": 2592
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.5142809337181063e-06,
+      "loss": 0.0657,
+      "step": 2594
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.5030843937930485e-06,
+      "loss": 0.0605,
+      "step": 2596
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4919260363785215e-06,
+      "loss": 0.0645,
+      "step": 2598
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4808059116167306e-06,
+      "loss": 0.0658,
+      "step": 2600
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4697240694780769e-06,
+      "loss": 0.0613,
+      "step": 2602
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4586805597609333e-06,
+      "loss": 0.0725,
+      "step": 2604
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4476754320914187e-06,
+      "loss": 0.0596,
+      "step": 2606
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4367087359231668e-06,
+      "loss": 0.0593,
+      "step": 2608
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4257805205371233e-06,
+      "loss": 0.0668,
+      "step": 2610
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4148908350413048e-06,
+      "loss": 0.0638,
+      "step": 2612
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4040397283705898e-06,
+      "loss": 0.0664,
+      "step": 2614
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3932272492864984e-06,
+      "loss": 0.0646,
+      "step": 2616
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3824534463769633e-06,
+      "loss": 0.0686,
+      "step": 2618
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3717183680561253e-06,
+      "loss": 0.0648,
+      "step": 2620
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3610220625641002e-06,
+      "loss": 0.0663,
+      "step": 2622
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3503645779667852e-06,
+      "loss": 0.0649,
+      "step": 2624
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.0692,
+      "step": 2626
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3291662628473634e-06,
+      "loss": 0.0683,
+      "step": 2628
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.3186255275839266e-06,
+      "loss": 0.0672,
+      "step": 2630
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.308123803732111e-06,
+      "loss": 0.0703,
+      "step": 2632
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2976611384834148e-06,
+      "loss": 0.0676,
+      "step": 2634
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2872375788538171e-06,
+      "loss": 0.0636,
+      "step": 2636
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2768531716835754e-06,
+      "loss": 0.0593,
+      "step": 2638
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.266507963636997e-06,
+      "loss": 0.0661,
+      "step": 2640
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2562020012022512e-06,
+      "loss": 0.0641,
+      "step": 2642
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2459353306911438e-06,
+      "loss": 0.0636,
+      "step": 2644
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2357079982389197e-06,
+      "loss": 0.0609,
+      "step": 2646
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2255200498040432e-06,
+      "loss": 0.062,
+      "step": 2648
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2153715311680048e-06,
+      "loss": 0.0593,
+      "step": 2650
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2052624879351105e-06,
+      "loss": 0.0644,
+      "step": 2652
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1951929655322692e-06,
+      "loss": 0.0661,
+      "step": 2654
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1851630092088051e-06,
+      "loss": 0.0655,
+      "step": 2656
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.175172664036235e-06,
+      "loss": 0.0664,
+      "step": 2658
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1652219749080817e-06,
+      "loss": 0.0662,
+      "step": 2660
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.155310986539665e-06,
+      "loss": 0.0655,
+      "step": 2662
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 0.0603,
+      "step": 2664
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1356082900511023e-06,
+      "loss": 0.067,
+      "step": 2666
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.12581667046878e-06,
+      "loss": 0.0625,
+      "step": 2668
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.116064928721442e-06,
+      "loss": 0.0593,
+      "step": 2670
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1063531086304003e-06,
+      "loss": 0.0659,
+      "step": 2672
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0966812538375738e-06,
+      "loss": 0.0647,
+      "step": 2674
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0870494078052796e-06,
+      "loss": 0.0606,
+      "step": 2676
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0774576138160596e-06,
+      "loss": 0.0612,
+      "step": 2678
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.067905914972468e-06,
+      "loss": 0.0671,
+      "step": 2680
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0583943541968856e-06,
+      "loss": 0.0697,
+      "step": 2682
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0489229742313223e-06,
+      "loss": 0.0685,
+      "step": 2684
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0394918176372327e-06,
+      "loss": 0.0656,
+      "step": 2686
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0301009267953145e-06,
+      "loss": 0.0686,
+      "step": 2688
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0207503439053279e-06,
+      "loss": 0.0602,
+      "step": 2690
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0114401109859019e-06,
+      "loss": 0.0577,
+      "step": 2692
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.0021702698743408e-06,
+      "loss": 0.057,
+      "step": 2694
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.929408622264448e-07,
+      "loss": 0.0681,
+      "step": 2696
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.837519295163155e-07,
+      "loss": 0.0615,
+      "step": 2698
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.746035130361741e-07,
+      "loss": 0.0603,
+      "step": 2700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.65495653896179e-07,
+      "loss": 0.0601,
+      "step": 2702
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.564283930242258e-07,
+      "loss": 0.0594,
+      "step": 2704
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.474017711657835e-07,
+      "loss": 0.0585,
+      "step": 2706
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.384158288836987e-07,
+      "loss": 0.0617,
+      "step": 2708
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.294706065580172e-07,
+      "loss": 0.0601,
+      "step": 2710
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.205661443857994e-07,
+      "loss": 0.0729,
+      "step": 2712
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.117024823809462e-07,
+      "loss": 0.0626,
+      "step": 2714
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.028796603740097e-07,
+      "loss": 0.0644,
+      "step": 2716
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.940977180120247e-07,
+      "loss": 0.0556,
+      "step": 2718
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.853566947583259e-07,
+      "loss": 0.0648,
+      "step": 2720
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.766566298923651e-07,
+      "loss": 0.061,
+      "step": 2722
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.67997562509546e-07,
+      "loss": 0.0605,
+      "step": 2724
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.593795315210341e-07,
+      "loss": 0.0638,
+      "step": 2726
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.508025756535987e-07,
+      "loss": 0.0634,
+      "step": 2728
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.42266733449425e-07,
+      "loss": 0.0634,
+      "step": 2730
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.337720432659513e-07,
+      "loss": 0.0606,
+      "step": 2732
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.253185432756816e-07,
+      "loss": 0.0611,
+      "step": 2734
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.169062714660347e-07,
+      "loss": 0.0572,
+      "step": 2736
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.085352656391576e-07,
+      "loss": 0.0656,
+      "step": 2738
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 8.002055634117578e-07,
+      "loss": 0.0612,
+      "step": 2740
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.919172022149458e-07,
+      "loss": 0.0684,
+      "step": 2742
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.836702192940493e-07,
+      "loss": 0.0687,
+      "step": 2744
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.754646517084585e-07,
+      "loss": 0.0661,
+      "step": 2746
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.673005363314578e-07,
+      "loss": 0.0593,
+      "step": 2748
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.591779098500584e-07,
+      "loss": 0.0681,
+      "step": 2750
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.510968087648262e-07,
+      "loss": 0.0604,
+      "step": 2752
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.430572693897342e-07,
+      "loss": 0.0629,
+      "step": 2754
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.350593278519824e-07,
+      "loss": 0.0612,
+      "step": 2756
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.271030200918483e-07,
+      "loss": 0.0646,
+      "step": 2758
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.19188381862519e-07,
+      "loss": 0.0655,
+      "step": 2760
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.113154487299267e-07,
+      "loss": 0.0684,
+      "step": 2762
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.034842560726008e-07,
+      "loss": 0.0643,
+      "step": 2764
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.956948390814977e-07,
+      "loss": 0.0602,
+      "step": 2766
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.879472327598502e-07,
+      "loss": 0.0622,
+      "step": 2768
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.802414719230021e-07,
+      "loss": 0.0583,
+      "step": 2770
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.725775911982602e-07,
+      "loss": 0.0614,
+      "step": 2772
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.649556250247325e-07,
+      "loss": 0.0589,
+      "step": 2774
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.573756076531779e-07,
+      "loss": 0.063,
+      "step": 2776
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.498375731458529e-07,
+      "loss": 0.0681,
+      "step": 2778
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.423415553763479e-07,
+      "loss": 0.0656,
+      "step": 2780
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.348875880294536e-07,
+      "loss": 0.0608,
+      "step": 2782
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.274757046009871e-07,
+      "loss": 0.0644,
+      "step": 2784
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.201059383976671e-07,
+      "loss": 0.062,
+      "step": 2786
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.127783225369377e-07,
+      "loss": 0.0624,
+      "step": 2788
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 6.054928899468427e-07,
+      "loss": 0.0683,
+      "step": 2790
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.982496733658582e-07,
+      "loss": 0.0612,
+      "step": 2792
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.910487053427616e-07,
+      "loss": 0.0636,
+      "step": 2794
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.83890018236476e-07,
+      "loss": 0.0633,
+      "step": 2796
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.767736442159256e-07,
+      "loss": 0.0677,
+      "step": 2798
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.696996152598966e-07,
+      "loss": 0.0719,
+      "step": 2800
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.626679631568832e-07,
+      "loss": 0.0622,
+      "step": 2802
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.556787195049573e-07,
+      "loss": 0.0618,
+      "step": 2804
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.487319157116178e-07,
+      "loss": 0.0565,
+      "step": 2806
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.418275829936537e-07,
+      "loss": 0.0627,
+      "step": 2808
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.349657523769991e-07,
+      "loss": 0.0613,
+      "step": 2810
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.281464546965953e-07,
+      "loss": 0.0629,
+      "step": 2812
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.213697205962631e-07,
+      "loss": 0.0654,
+      "step": 2814
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.146355805285452e-07,
+      "loss": 0.0647,
+      "step": 2816
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.079440647545852e-07,
+      "loss": 0.0614,
+      "step": 2818
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.012952033439844e-07,
+      "loss": 0.0665,
+      "step": 2820
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.946890261746706e-07,
+      "loss": 0.0604,
+      "step": 2822
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.881255629327608e-07,
+      "loss": 0.059,
+      "step": 2824
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.816048431124265e-07,
+      "loss": 0.0611,
+      "step": 2826
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.7512689601576843e-07,
+      "loss": 0.0652,
+      "step": 2828
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.686917507526756e-07,
+      "loss": 0.0571,
+      "step": 2830
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.6229943624069963e-07,
+      "loss": 0.0646,
+      "step": 2832
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5594998120492505e-07,
+      "loss": 0.0661,
+      "step": 2834
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.4964341417784165e-07,
+      "loss": 0.0599,
+      "step": 2836
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.4337976349920763e-07,
+      "loss": 0.0572,
+      "step": 2838
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3715905731593233e-07,
+      "loss": 0.0626,
+      "step": 2840
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.309813235819471e-07,
+      "loss": 0.0606,
+      "step": 2842
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.248465900580734e-07,
+      "loss": 0.0625,
+      "step": 2844
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.1875488431190934e-07,
+      "loss": 0.0604,
+      "step": 2846
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.127062337176935e-07,
+      "loss": 0.0603,
+      "step": 2848
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.0670066545619224e-07,
+      "loss": 0.0656,
+      "step": 2850
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.0073820651457043e-07,
+      "loss": 0.0622,
+      "step": 2852
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.9481888368627764e-07,
+      "loss": 0.058,
+      "step": 2854
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.889427235709153e-07,
+      "loss": 0.0585,
+      "step": 2856
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.8310975257413343e-07,
+      "loss": 0.0724,
+      "step": 2858
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.773199969074959e-07,
+      "loss": 0.0646,
+      "step": 2860
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.715734825883766e-07,
+      "loss": 0.0654,
+      "step": 2862
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.658702354398325e-07,
+      "loss": 0.0578,
+      "step": 2864
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.6021028109049063e-07,
+      "loss": 0.0563,
+      "step": 2866
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.5459364497443696e-07,
+      "loss": 0.0569,
+      "step": 2868
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4902035233109644e-07,
+      "loss": 0.0643,
+      "step": 2870
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4349042820512325e-07,
+      "loss": 0.0567,
+      "step": 2872
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.38003897446284e-07,
+      "loss": 0.0527,
+      "step": 2874
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.325607847093537e-07,
+      "loss": 0.0654,
+      "step": 2876
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.2716111445399324e-07,
+      "loss": 0.0652,
+      "step": 2878
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.2180491094465414e-07,
+      "loss": 0.0599,
+      "step": 2880
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.164921982504565e-07,
+      "loss": 0.0641,
+      "step": 2882
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.112230002450889e-07,
+      "loss": 0.0623,
+      "step": 2884
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.059973406066963e-07,
+      "loss": 0.0664,
+      "step": 2886
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0081524281777687e-07,
+      "loss": 0.0643,
+      "step": 2888
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9567673016507646e-07,
+      "loss": 0.0628,
+      "step": 2890
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 0.067,
+      "step": 2892
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.855305524359131e-07,
+      "loss": 0.0606,
+      "step": 2894
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.805229329532344e-07,
+      "loss": 0.0618,
+      "step": 2896
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.7555898979413796e-07,
+      "loss": 0.0596,
+      "step": 2898
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.706387452650494e-07,
+      "loss": 0.0536,
+      "step": 2900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.657622214760247e-07,
+      "loss": 0.0576,
+      "step": 2902
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.609294403406537e-07,
+      "loss": 0.0638,
+      "step": 2904
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.56140423575959e-07,
+      "loss": 0.0598,
+      "step": 2906
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.513951927023017e-07,
+      "loss": 0.0587,
+      "step": 2908
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.4669376904328244e-07,
+      "loss": 0.0592,
+      "step": 2910
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.420361737256438e-07,
+      "loss": 0.0592,
+      "step": 2912
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.3742242767917921e-07,
+      "loss": 0.0622,
+      "step": 2914
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.3285255163663535e-07,
+      "loss": 0.0598,
+      "step": 2916
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2832656613362092e-07,
+      "loss": 0.0542,
+      "step": 2918
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2384449150851695e-07,
+      "loss": 0.0597,
+      "step": 2920
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.1940634790238003e-07,
+      "loss": 0.0628,
+      "step": 2922
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.1501215525885245e-07,
+      "loss": 0.0574,
+      "step": 2924
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.1066193332407891e-07,
+      "loss": 0.0649,
+      "step": 2926
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.063557016466111e-07,
+      "loss": 0.0595,
+      "step": 2928
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.0209347957732328e-07,
+      "loss": 0.0662,
+      "step": 2930
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.978752862693212e-07,
+      "loss": 0.0571,
+      "step": 2932
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9370114067785995e-07,
+      "loss": 0.0625,
+      "step": 2934
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8957106156026084e-07,
+      "loss": 0.064,
+      "step": 2936
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.854850674758213e-07,
+      "loss": 0.0624,
+      "step": 2938
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8144317678573497e-07,
+      "loss": 0.06,
+      "step": 2940
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7744540765300743e-07,
+      "loss": 0.0625,
+      "step": 2942
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7349177804237837e-07,
+      "loss": 0.0622,
+      "step": 2944
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6958230572023504e-07,
+      "loss": 0.0642,
+      "step": 2946
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6571700825453674e-07,
+      "loss": 0.061,
+      "step": 2948
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.618959030147349e-07,
+      "loss": 0.0613,
+      "step": 2950
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5811900717169537e-07,
+      "loss": 0.0646,
+      "step": 2952
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.5438633769761845e-07,
+      "loss": 0.0672,
+      "step": 2954
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.506979113659679e-07,
+      "loss": 0.0586,
+      "step": 2956
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4705374475138978e-07,
+      "loss": 0.0627,
+      "step": 2958
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4345385422964043e-07,
+      "loss": 0.0631,
+      "step": 2960
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.398982559775175e-07,
+      "loss": 0.0643,
+      "step": 2962
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3638696597277678e-07,
+      "loss": 0.0622,
+      "step": 2964
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3291999999407e-07,
+      "loss": 0.0606,
+      "step": 2966
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2949737362087156e-07,
+      "loss": 0.0609,
+      "step": 2968
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2611910223340408e-07,
+      "loss": 0.0586,
+      "step": 2970
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.227852010125752e-07,
+      "loss": 0.0609,
+      "step": 2972
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1949568493990428e-07,
+      "loss": 0.0609,
+      "step": 2974
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1625056879746133e-07,
+      "loss": 0.0614,
+      "step": 2976
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1304986716779265e-07,
+      "loss": 0.0641,
+      "step": 2978
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0989359443386305e-07,
+      "loss": 0.0646,
+      "step": 2980
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0678176477898372e-07,
+      "loss": 0.0588,
+      "step": 2982
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.0371439218675671e-07,
+      "loss": 0.0593,
+      "step": 2984
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.0069149044100502e-07,
+      "loss": 0.0596,
+      "step": 2986
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.771307312571254e-08,
+      "loss": 0.0671,
+      "step": 2988
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.47791536249676e-08,
+      "loss": 0.0577,
+      "step": 2990
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.188974512289617e-08,
+      "loss": 0.0621,
+      "step": 2992
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.90448606036054e-08,
+      "loss": 0.0616,
+      "step": 2994
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.624451285112689e-08,
+      "loss": 0.0633,
+      "step": 2996
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.348871444935902e-08,
+      "loss": 0.0655,
+      "step": 2998
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.077747778200474e-08,
+      "loss": 0.0615,
+      "step": 3000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.811081503252382e-08,
+      "loss": 0.059,
+      "step": 3002
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.54887381840752e-08,
+      "loss": 0.0604,
+      "step": 3004
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.291125901946027e-08,
+      "loss": 0.0582,
+      "step": 3006
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 7.037838912107298e-08,
+      "loss": 0.0619,
+      "step": 3008
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.789013987084647e-08,
+      "loss": 0.0685,
+      "step": 3010
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.544652245020433e-08,
+      "loss": 0.0585,
+      "step": 3012
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.304754784000721e-08,
+      "loss": 0.0571,
+      "step": 3014
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.069322682050516e-08,
+      "loss": 0.0568,
+      "step": 3016
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.838356997128869e-08,
+      "loss": 0.0594,
+      "step": 3018
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.611858767124001e-08,
+      "loss": 0.0659,
+      "step": 3020
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.389829009849079e-08,
+      "loss": 0.061,
+      "step": 3022
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.1722687230369995e-08,
+      "loss": 0.057,
+      "step": 3024
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.959178884336502e-08,
+      "loss": 0.0615,
+      "step": 3026
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.7505604513072845e-08,
+      "loss": 0.0608,
+      "step": 3028
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.5464143614162294e-08,
+      "loss": 0.0603,
+      "step": 3030
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.346741532032628e-08,
+      "loss": 0.0601,
+      "step": 3032
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.151542860424407e-08,
+      "loss": 0.0539,
+      "step": 3034
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.96081922375402e-08,
+      "loss": 0.06,
+      "step": 3036
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7745714790745623e-08,
+      "loss": 0.0564,
+      "step": 3038
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.592800463325663e-08,
+      "loss": 0.0592,
+      "step": 3040
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4155069933301535e-08,
+      "loss": 0.0601,
+      "step": 3042
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.242691865790071e-08,
+      "loss": 0.0637,
+      "step": 3044
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.074355857282996e-08,
+      "loss": 0.0578,
+      "step": 3046
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.9104997242590528e-08,
+      "loss": 0.0612,
+      "step": 3048
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7511242030368035e-08,
+      "loss": 0.0605,
+      "step": 3050
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.5962300098008042e-08,
+      "loss": 0.0636,
+      "step": 3052
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.4458178405974974e-08,
+      "loss": 0.06,
+      "step": 3054
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.2998883713326592e-08,
+      "loss": 0.0618,
+      "step": 3056
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.1584422577682894e-08,
+      "loss": 0.0574,
+      "step": 3058
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.0214801355192826e-08,
+      "loss": 0.0579,
+      "step": 3060
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.889002620051206e-08,
+      "loss": 0.0609,
+      "step": 3062
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.761010306676969e-08,
+      "loss": 0.0548,
+      "step": 3064
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.6375037705543827e-08,
+      "loss": 0.0594,
+      "step": 3066
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.518483566683826e-08,
+      "loss": 0.0594,
+      "step": 3068
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.4039502299050266e-08,
+      "loss": 0.066,
+      "step": 3070
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2939042748955078e-08,
+      "loss": 0.0579,
+      "step": 3072
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1883461961678112e-08,
+      "loss": 0.0615,
+      "step": 3074
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0872764680671666e-08,
+      "loss": 0.061,
+      "step": 3076
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.906955447697153e-09,
+      "loss": 0.0589,
+      "step": 3078
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.986038602802894e-09,
+      "loss": 0.0616,
+      "step": 3080
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.110018284304132e-09,
+      "loss": 0.061,
+      "step": 3082
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.278898428764169e-09,
+      "loss": 0.0591,
+      "step": 3084
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.492682770979919e-09,
+      "loss": 0.0631,
+      "step": 3086
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.751374843961932e-09,
+      "loss": 0.0617,
+      "step": 3088
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.054977978916631e-09,
+      "loss": 0.0629,
+      "step": 3090
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.403495305237426e-09,
+      "loss": 0.0686,
+      "step": 3092
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.7969297504858445e-09,
+      "loss": 0.0573,
+      "step": 3094
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.2352840403804264e-09,
+      "loss": 0.0607,
+      "step": 3096
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.7185606987811808e-09,
+      "loss": 0.0648,
+      "step": 3098
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.246762047685147e-09,
+      "loss": 0.0621,
+      "step": 3100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.8198902072097402e-09,
+      "loss": 0.0621,
+      "step": 3102
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.437947095582759e-09,
+      "loss": 0.0561,
+      "step": 3104
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.1009344291401657e-09,
+      "loss": 0.0584,
+      "step": 3106
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.088537223116533e-10,
+      "loss": 0.0597,
+      "step": 3108
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.617062876195345e-10,
+      "loss": 0.0579,
+      "step": 3110
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.594932356654202e-10,
+      "loss": 0.0577,
+      "step": 3112
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.02215475132439e-10,
+      "loss": 0.0564,
+      "step": 3114
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.987371278079693e-11,
+      "loss": 0.0599,
+      "step": 3116
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.246845343667481e-11,
+      "loss": 0.0611,
+      "step": 3118
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0,
+      "loss": 0.0613,
+      "step": 3120
+    },
+    {
+      "epoch": 4.0,
+      "step": 3120,
+      "total_flos": 9.431345112220172e+18,
+      "train_loss": 0.5043172255636026,
+      "train_runtime": 26129.153,
+      "train_samples_per_second": 7.642,
+      "train_steps_per_second": 0.119
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 3120,
+  "num_train_epochs": 4,
+  "save_steps": 780,
+  "total_flos": 9.431345112220172e+18,
+  "trial_name": null,
+  "trial_params": null
+}