diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6771 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.913265306122449,
+  "eval_steps": 800,
+  "global_step": 4500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0,
+      "loss": 2.7231,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0,
+      "loss": 2.8688,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 3.0604,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 3.3619,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0,
+      "loss": 3.1263,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0,
+      "loss": 3.1305,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5188198451414044e-06,
+      "loss": 1.2543,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0735116692035353e-06,
+      "loss": 1.0489,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.420532010460384e-06,
+      "loss": 0.8238,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6736916157045096e-06,
+      "loss": 0.6357,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.873103126046782e-06,
+      "loss": 0.9375,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0376396902828088e-06,
+      "loss": 1.0042,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.177703030066258e-06,
+      "loss": 0.7668,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.299639602133279e-06,
+      "loss": 0.9168,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.407608173344298e-06,
+      "loss": 0.8997,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5044824703694353e-06,
+      "loss": 1.0025,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5923315143449394e-06,
+      "loss": 0.6651,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6726945828900284e-06,
+      "loss": 0.699,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.746747919084026e-06,
+      "loss": 0.886,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8154106182475455e-06,
+      "loss": 0.8937,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8794145195304064e-06,
+      "loss": 0.6582,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9393518556017876e-06,
+      "loss": 0.7974,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.995708648306083e-06,
+      "loss": 0.9472,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.0488886794862905e-06,
+      "loss": 0.8867,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.099231058976525e-06,
+      "loss": 1.061,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1470233384070705e-06,
+      "loss": 0.7292,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.192511460845915e-06,
+      "loss": 0.9262,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.235907420412398e-06,
+      "loss": 0.7092,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.27739523661862e-06,
+      "loss": 0.7174,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.317135669700268e-06,
+      "loss": 0.8711,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.355269982485126e-06,
+      "loss": 0.9286,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.391922971188186e-06,
+      "loss": 0.9947,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.427205429264097e-06,
+      "loss": 0.8179,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.461216167003915e-06,
+      "loss": 0.6159,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.494043679663919e-06,
+      "loss": 0.8981,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.525767535063022e-06,
+      "loss": 0.9206,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.556459535424214e-06,
+      "loss": 0.7768,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.5861846961499975e-06,
+      "loss": 0.8515,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6150020750907925e-06,
+      "loss": 0.8131,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.642965478900328e-06,
+      "loss": 1.1253,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.670124067711698e-06,
+      "loss": 0.9236,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.689992082159791e-06,
+      "loss": 0.8775,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.715848379822425e-06,
+      "loss": 0.8848,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.74101509336297e-06,
+      "loss": 0.829,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.76552805154028e-06,
+      "loss": 0.9367,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.789420361336724e-06,
+      "loss": 0.6566,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.812722676847563e-06,
+      "loss": 0.8104,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.835463435763974e-06,
+      "loss": 0.74,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.857669068026358e-06,
+      "loss": 0.6171,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.879364180487766e-06,
+      "loss": 1.0545,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.900571720823068e-06,
+      "loss": 0.9191,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.921313123421507e-06,
+      "loss": 0.6995,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.941608439588058e-06,
+      "loss": 0.8479,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9614764540361516e-06,
+      "loss": 0.8976,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.980934789368156e-06,
+      "loss": 0.851,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5e-06,
+      "loss": 0.7368,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.997812135355893e-06,
+      "loss": 1.0125,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.994894982497083e-06,
+      "loss": 0.9607,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.991977829638274e-06,
+      "loss": 0.6321,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.989060676779464e-06,
+      "loss": 0.7108,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.986143523920654e-06,
+      "loss": 0.7881,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.983226371061844e-06,
+      "loss": 0.7147,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.980309218203034e-06,
+      "loss": 0.9225,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.977392065344224e-06,
+      "loss": 0.8849,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.974474912485414e-06,
+      "loss": 0.942,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.971557759626604e-06,
+      "loss": 1.1629,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.968640606767795e-06,
+      "loss": 0.8053,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.965723453908986e-06,
+      "loss": 0.6558,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.962806301050176e-06,
+      "loss": 0.9606,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.959889148191366e-06,
+      "loss": 0.9533,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.956971995332556e-06,
+      "loss": 0.8758,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.954054842473746e-06,
+      "loss": 0.5825,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.951137689614936e-06,
+      "loss": 0.8221,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.948220536756126e-06,
+      "loss": 0.8954,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.945303383897317e-06,
+      "loss": 0.9549,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.942386231038507e-06,
+      "loss": 0.8821,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.939469078179697e-06,
+      "loss": 0.9861,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.936551925320887e-06,
+      "loss": 0.9681,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.933634772462078e-06,
+      "loss": 0.647,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.930717619603268e-06,
+      "loss": 0.8151,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9278004667444575e-06,
+      "loss": 1.0373,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9248833138856475e-06,
+      "loss": 0.8858,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.921966161026838e-06,
+      "loss": 0.8392,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.919049008168029e-06,
+      "loss": 1.0194,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.916131855309218e-06,
+      "loss": 0.7719,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.913214702450409e-06,
+      "loss": 0.6672,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.910297549591599e-06,
+      "loss": 0.8939,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.90738039673279e-06,
+      "loss": 0.6781,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.90446324387398e-06,
+      "loss": 0.7391,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9015460910151695e-06,
+      "loss": 0.9476,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8986289381563595e-06,
+      "loss": 0.6729,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.89571178529755e-06,
+      "loss": 0.8046,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.89279463243874e-06,
+      "loss": 0.8521,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.88987747957993e-06,
+      "loss": 0.8688,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.886960326721121e-06,
+      "loss": 0.875,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.884043173862311e-06,
+      "loss": 0.6733,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.881126021003501e-06,
+      "loss": 0.7755,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.878208868144691e-06,
+      "loss": 0.7047,
+      "step": 392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8752917152858815e-06,
+      "loss": 0.6979,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8723745624270714e-06,
+      "loss": 0.8033,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.869457409568261e-06,
+      "loss": 0.8039,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.866540256709452e-06,
+      "loss": 0.9366,
+      "step": 408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.863623103850642e-06,
+      "loss": 0.609,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.860705950991833e-06,
+      "loss": 0.7258,
+      "step": 416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.857788798133022e-06,
+      "loss": 1.018,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.854871645274213e-06,
+      "loss": 0.7338,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.851954492415403e-06,
+      "loss": 0.6595,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8490373395565935e-06,
+      "loss": 0.7456,
+      "step": 432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8461201866977834e-06,
+      "loss": 0.637,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.843203033838973e-06,
+      "loss": 0.8026,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.840285880980164e-06,
+      "loss": 0.8439,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.837368728121354e-06,
+      "loss": 0.646,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.834451575262544e-06,
+      "loss": 0.8644,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.831534422403734e-06,
+      "loss": 0.8595,
+      "step": 456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.828617269544925e-06,
+      "loss": 1.023,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.825700116686115e-06,
+      "loss": 0.7745,
+      "step": 464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.822782963827305e-06,
+      "loss": 0.9293,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8198658109684954e-06,
+      "loss": 0.7542,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.816948658109685e-06,
+      "loss": 1.0059,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.814031505250875e-06,
+      "loss": 0.8302,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.811114352392065e-06,
+      "loss": 0.809,
+      "step": 484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.808197199533256e-06,
+      "loss": 0.7628,
+      "step": 488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.805280046674446e-06,
+      "loss": 0.6753,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.802362893815637e-06,
+      "loss": 0.8719,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.799445740956827e-06,
+      "loss": 0.7059,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.796528588098017e-06,
+      "loss": 0.6902,
+      "step": 504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.793611435239207e-06,
+      "loss": 1.104,
+      "step": 508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.790694282380397e-06,
+      "loss": 0.837,
+      "step": 512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.787777129521587e-06,
+      "loss": 0.794,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.784859976662777e-06,
+      "loss": 0.7997,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.781942823803968e-06,
+      "loss": 0.6738,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.779025670945158e-06,
+      "loss": 0.963,
+      "step": 528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.776108518086348e-06,
+      "loss": 0.6323,
+      "step": 532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.773191365227539e-06,
+      "loss": 0.7243,
+      "step": 536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.770274212368729e-06,
+      "loss": 0.8873,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7673570595099186e-06,
+      "loss": 0.7435,
+      "step": 544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7644399066511085e-06,
+      "loss": 0.8006,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.761522753792299e-06,
+      "loss": 0.9557,
+      "step": 552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.758605600933489e-06,
+      "loss": 0.7131,
+      "step": 556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.755688448074679e-06,
+      "loss": 0.8007,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.752771295215869e-06,
+      "loss": 0.8634,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.74985414235706e-06,
+      "loss": 0.7848,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.746936989498251e-06,
+      "loss": 0.6159,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.744019836639441e-06,
+      "loss": 0.7327,
+      "step": 576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7411026837806305e-06,
+      "loss": 0.733,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7381855309218205e-06,
+      "loss": 0.8081,
+      "step": 584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.735268378063011e-06,
+      "loss": 0.9809,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.732351225204201e-06,
+      "loss": 0.7101,
+      "step": 592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.729434072345391e-06,
+      "loss": 0.6195,
+      "step": 596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.726516919486582e-06,
+      "loss": 0.6483,
+      "step": 600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.723599766627772e-06,
+      "loss": 0.819,
+      "step": 604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.720682613768962e-06,
+      "loss": 0.7032,
+      "step": 608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.717765460910152e-06,
+      "loss": 0.7933,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7148483080513425e-06,
+      "loss": 0.9605,
+      "step": 616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7119311551925325e-06,
+      "loss": 0.7783,
+      "step": 620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.709014002333722e-06,
+      "loss": 0.7616,
+      "step": 624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.706096849474912e-06,
+      "loss": 0.6611,
+      "step": 628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.703179696616103e-06,
+      "loss": 0.656,
+      "step": 632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.700262543757294e-06,
+      "loss": 0.731,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.697345390898483e-06,
+      "loss": 0.6204,
+      "step": 640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.694428238039674e-06,
+      "loss": 0.695,
+      "step": 644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.691511085180864e-06,
+      "loss": 0.9691,
+      "step": 648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6885939323220545e-06,
+      "loss": 0.5099,
+      "step": 652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.6856767794632445e-06,
+      "loss": 0.6427,
+      "step": 656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.682759626604434e-06,
+      "loss": 0.9209,
+      "step": 660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.679842473745624e-06,
+      "loss": 0.891,
+      "step": 664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.676925320886815e-06,
+      "loss": 0.7068,
+      "step": 668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.674008168028005e-06,
+      "loss": 0.4388,
+      "step": 672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.671091015169195e-06,
+      "loss": 0.7769,
+      "step": 676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.668173862310386e-06,
+      "loss": 1.03,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.665256709451576e-06,
+      "loss": 0.9345,
+      "step": 684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.662339556592766e-06,
+      "loss": 0.8261,
+      "step": 688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.659422403733956e-06,
+      "loss": 0.5926,
+      "step": 692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.656505250875146e-06,
+      "loss": 0.618,
+      "step": 696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.653588098016336e-06,
+      "loss": 0.8133,
+      "step": 700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.650670945157526e-06,
+      "loss": 0.7822,
+      "step": 704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.647753792298717e-06,
+      "loss": 0.6367,
+      "step": 708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.644836639439907e-06,
+      "loss": 0.7457,
+      "step": 712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.641919486581098e-06,
+      "loss": 0.5508,
+      "step": 716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.639002333722287e-06,
+      "loss": 0.8247,
+      "step": 720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.636085180863478e-06,
+      "loss": 0.8024,
+      "step": 724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.633168028004668e-06,
+      "loss": 0.5489,
+      "step": 728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.630250875145858e-06,
+      "loss": 0.5789,
+      "step": 732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.627333722287048e-06,
+      "loss": 0.7541,
+      "step": 736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.624416569428238e-06,
+      "loss": 0.6148,
+      "step": 740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.621499416569429e-06,
+      "loss": 0.98,
+      "step": 744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.618582263710619e-06,
+      "loss": 0.635,
+      "step": 748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.615665110851809e-06,
+      "loss": 0.9664,
+      "step": 752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.612747957992999e-06,
+      "loss": 0.8736,
+      "step": 756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.60983080513419e-06,
+      "loss": 0.6281,
+      "step": 760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.60691365227538e-06,
+      "loss": 0.9843,
+      "step": 764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6039964994165695e-06,
+      "loss": 0.7795,
+      "step": 768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.60107934655776e-06,
+      "loss": 0.8437,
+      "step": 772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.59816219369895e-06,
+      "loss": 0.6626,
+      "step": 776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.59524504084014e-06,
+      "loss": 0.8164,
+      "step": 780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.59232788798133e-06,
+      "loss": 0.7359,
+      "step": 784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.589410735122521e-06,
+      "loss": 0.7709,
+      "step": 788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.586493582263711e-06,
+      "loss": 0.8953,
+      "step": 792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.583576429404902e-06,
+      "loss": 0.57,
+      "step": 796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.5806592765460916e-06,
+      "loss": 0.9151,
+      "step": 800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.5777421236872815e-06,
+      "loss": 0.7838,
+      "step": 804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.574824970828472e-06,
+      "loss": 0.8183,
+      "step": 808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.571907817969662e-06,
+      "loss": 0.9169,
+      "step": 812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.568990665110852e-06,
+      "loss": 0.6786,
+      "step": 816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.566073512252042e-06,
+      "loss": 0.7783,
+      "step": 820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.563156359393233e-06,
+      "loss": 0.811,
+      "step": 824
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.560239206534423e-06,
+      "loss": 0.8965,
+      "step": 828
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.557322053675613e-06,
+      "loss": 0.7526,
+      "step": 832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5544049008168036e-06,
+      "loss": 0.6549,
+      "step": 836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5514877479579935e-06,
+      "loss": 0.77,
+      "step": 840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5485705950991834e-06,
+      "loss": 0.5399,
+      "step": 844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.545653442240373e-06,
+      "loss": 0.7214,
+      "step": 848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.542736289381564e-06,
+      "loss": 0.7454,
+      "step": 852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.539819136522754e-06,
+      "loss": 0.6774,
+      "step": 856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.536901983663944e-06,
+      "loss": 0.855,
+      "step": 860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.533984830805134e-06,
+      "loss": 0.8038,
+      "step": 864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.531067677946325e-06,
+      "loss": 0.6897,
+      "step": 868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5281505250875156e-06,
+      "loss": 0.774,
+      "step": 872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.525233372228705e-06,
+      "loss": 0.9166,
+      "step": 876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5223162193698954e-06,
+      "loss": 0.5643,
+      "step": 880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.519399066511085e-06,
+      "loss": 0.5427,
+      "step": 884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.516481913652276e-06,
+      "loss": 0.6834,
+      "step": 888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.513564760793466e-06,
+      "loss": 0.8979,
+      "step": 892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.510647607934656e-06,
+      "loss": 0.8804,
+      "step": 896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.507730455075847e-06,
+      "loss": 0.6715,
+      "step": 900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.504813302217037e-06,
+      "loss": 0.8331,
+      "step": 904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.501896149358227e-06,
+      "loss": 0.9623,
+      "step": 908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.498978996499417e-06,
+      "loss": 0.7293,
+      "step": 912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.496061843640607e-06,
+      "loss": 0.6704,
+      "step": 916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.493144690781797e-06,
+      "loss": 0.6452,
+      "step": 920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.490227537922987e-06,
+      "loss": 0.7556,
+      "step": 924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.487310385064177e-06,
+      "loss": 0.6235,
+      "step": 928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.484393232205368e-06,
+      "loss": 0.789,
+      "step": 932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.481476079346558e-06,
+      "loss": 0.7959,
+      "step": 936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.478558926487748e-06,
+      "loss": 0.8286,
+      "step": 940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.475641773628939e-06,
+      "loss": 0.8604,
+      "step": 944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.472724620770129e-06,
+      "loss": 0.6135,
+      "step": 948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.469807467911319e-06,
+      "loss": 0.4793,
+      "step": 952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.4668903150525085e-06,
+      "loss": 0.7225,
+      "step": 956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.463973162193699e-06,
+      "loss": 0.5255,
+      "step": 960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.461056009334889e-06,
+      "loss": 0.4695,
+      "step": 964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.45813885647608e-06,
+      "loss": 0.8741,
+      "step": 968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.45522170361727e-06,
+      "loss": 0.6311,
+      "step": 972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.45230455075846e-06,
+      "loss": 0.6423,
+      "step": 976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.449387397899651e-06,
+      "loss": 0.7879,
+      "step": 980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.446470245040841e-06,
+      "loss": 0.9635,
+      "step": 984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4435530921820306e-06,
+      "loss": 0.6497,
+      "step": 988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.4406359393232205e-06,
+      "loss": 0.673,
+      "step": 992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.437718786464411e-06,
+      "loss": 0.7035,
+      "step": 996
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.434801633605601e-06,
+      "loss": 0.7664,
+      "step": 1000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.431884480746791e-06,
+      "loss": 0.6817,
+      "step": 1004
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.428967327887982e-06,
+      "loss": 0.7443,
+      "step": 1008
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.426050175029172e-06,
+      "loss": 0.7205,
+      "step": 1012
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.423133022170362e-06,
+      "loss": 0.6672,
+      "step": 1016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.420215869311552e-06,
+      "loss": 0.7427,
+      "step": 1020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4172987164527425e-06,
+      "loss": 0.659,
+      "step": 1024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.4143815635939325e-06,
+      "loss": 0.832,
+      "step": 1028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.411464410735123e-06,
+      "loss": 0.7823,
+      "step": 1032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.408547257876313e-06,
+      "loss": 0.712,
+      "step": 1036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.405630105017503e-06,
+      "loss": 0.8165,
+      "step": 1040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.402712952158694e-06,
+      "loss": 0.6221,
+      "step": 1044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.399795799299884e-06,
+      "loss": 0.684,
+      "step": 1048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.396878646441074e-06,
+      "loss": 0.6845,
+      "step": 1052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.393961493582264e-06,
+      "loss": 0.8462,
+      "step": 1056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.3910443407234545e-06,
+      "loss": 0.6242,
+      "step": 1060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.3881271878646445e-06,
+      "loss": 0.5025,
+      "step": 1064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.385210035005834e-06,
+      "loss": 0.6651,
+      "step": 1068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.382292882147025e-06,
+      "loss": 0.7653,
+      "step": 1072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.379375729288215e-06,
+      "loss": 0.9375,
+      "step": 1076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.376458576429405e-06,
+      "loss": 0.6936,
+      "step": 1080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.373541423570595e-06,
+      "loss": 0.6709,
+      "step": 1084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.370624270711786e-06,
+      "loss": 0.7321,
+      "step": 1088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.367707117852976e-06,
+      "loss": 1.0432,
+      "step": 1092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.364789964994166e-06,
+      "loss": 0.6687,
+      "step": 1096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.3618728121353565e-06,
+      "loss": 0.9193,
+      "step": 1100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.358955659276546e-06,
+      "loss": 0.5975,
+      "step": 1104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.356038506417737e-06,
+      "loss": 0.7527,
+      "step": 1108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.353121353558927e-06,
+      "loss": 0.9754,
+      "step": 1112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.350204200700117e-06,
+      "loss": 0.8299,
+      "step": 1116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.347287047841307e-06,
+      "loss": 0.8524,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.344369894982498e-06,
+      "loss": 0.8139,
+      "step": 1124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.341452742123688e-06,
+      "loss": 0.6144,
+      "step": 1128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.338535589264878e-06,
+      "loss": 0.8328,
+      "step": 1132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.3356184364060684e-06,
+      "loss": 0.5855,
+      "step": 1136
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.332701283547258e-06,
+      "loss": 0.8227,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.329784130688448e-06,
+      "loss": 0.6442,
+      "step": 1144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.326866977829638e-06,
+      "loss": 0.5148,
+      "step": 1148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.323949824970829e-06,
+      "loss": 0.8127,
+      "step": 1152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.321032672112019e-06,
+      "loss": 0.8909,
+      "step": 1156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.318115519253209e-06,
+      "loss": 0.6379,
+      "step": 1160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.315198366394399e-06,
+      "loss": 0.5912,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.31228121353559e-06,
+      "loss": 0.6885,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.3093640606767804e-06,
+      "loss": 0.8584,
+      "step": 1172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.3064469078179695e-06,
+      "loss": 0.6441,
+      "step": 1176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.30352975495916e-06,
+      "loss": 0.6933,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.30061260210035e-06,
+      "loss": 0.6647,
+      "step": 1184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.297695449241541e-06,
+      "loss": 0.7737,
+      "step": 1188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.294778296382731e-06,
+      "loss": 0.8209,
+      "step": 1192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.291861143523921e-06,
+      "loss": 0.9695,
+      "step": 1196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.288943990665112e-06,
+      "loss": 0.855,
+      "step": 1200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.286026837806302e-06,
+      "loss": 0.7051,
+      "step": 1204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.283109684947492e-06,
+      "loss": 0.4676,
+      "step": 1208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.2801925320886815e-06,
+      "loss": 0.6563,
+      "step": 1212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.277275379229872e-06,
+      "loss": 0.6791,
+      "step": 1216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.274358226371062e-06,
+      "loss": 0.6531,
+      "step": 1220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.271441073512252e-06,
+      "loss": 0.7638,
+      "step": 1224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.268523920653442e-06,
+      "loss": 0.8035,
+      "step": 1228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.265606767794633e-06,
+      "loss": 0.6947,
+      "step": 1232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.262689614935823e-06,
+      "loss": 0.4111,
+      "step": 1236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.259772462077013e-06,
+      "loss": 0.787,
+      "step": 1240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.2568553092182036e-06,
+      "loss": 0.7891,
+      "step": 1244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.2539381563593935e-06,
+      "loss": 0.6001,
+      "step": 1248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.251021003500584e-06,
+      "loss": 0.5617,
+      "step": 1252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.248103850641773e-06,
+      "loss": 0.9523,
+      "step": 1256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.245186697782964e-06,
+      "loss": 0.8455,
+      "step": 1260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.242269544924154e-06,
+      "loss": 0.7674,
+      "step": 1264
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.239352392065345e-06,
+      "loss": 0.5396,
+      "step": 1268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.236435239206535e-06,
+      "loss": 0.5736,
+      "step": 1272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.233518086347725e-06,
+      "loss": 0.5877,
+      "step": 1276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.2306009334889156e-06,
+      "loss": 0.6466,
+      "step": 1280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2276837806301055e-06,
+      "loss": 0.4551,
+      "step": 1284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2247666277712954e-06,
+      "loss": 0.5959,
+      "step": 1288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.221849474912485e-06,
+      "loss": 0.6903,
+      "step": 1292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.218932322053676e-06,
+      "loss": 0.7372,
+      "step": 1296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.216015169194866e-06,
+      "loss": 0.7741,
+      "step": 1300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.213098016336056e-06,
+      "loss": 0.6861,
+      "step": 1304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.210180863477247e-06,
+      "loss": 0.7325,
+      "step": 1308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.207263710618437e-06,
+      "loss": 0.6446,
+      "step": 1312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.204346557759627e-06,
+      "loss": 0.61,
+      "step": 1316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.201429404900817e-06,
+      "loss": 0.6286,
+      "step": 1320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1985122520420074e-06,
+      "loss": 0.7191,
+      "step": 1324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.195595099183197e-06,
+      "loss": 0.8621,
+      "step": 1328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.192677946324388e-06,
+      "loss": 0.7634,
+      "step": 1332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.189760793465578e-06,
+      "loss": 0.7067,
+      "step": 1336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.186843640606768e-06,
+      "loss": 0.6049,
+      "step": 1340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.183926487747959e-06,
+      "loss": 0.3731,
+      "step": 1344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.181009334889149e-06,
+      "loss": 0.5955,
+      "step": 1348
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.178092182030339e-06,
+      "loss": 0.5576,
+      "step": 1352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.175175029171529e-06,
+      "loss": 0.6731,
+      "step": 1356
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.172257876312719e-06,
+      "loss": 0.8847,
+      "step": 1360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.169340723453909e-06,
+      "loss": 0.6158,
+      "step": 1364
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.166423570595099e-06,
+      "loss": 0.7721,
+      "step": 1368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.16350641773629e-06,
+      "loss": 0.6517,
+      "step": 1372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.16058926487748e-06,
+      "loss": 0.6014,
+      "step": 1376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.15767211201867e-06,
+      "loss": 0.8049,
+      "step": 1380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.15475495915986e-06,
+      "loss": 0.7245,
+      "step": 1384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.151837806301051e-06,
+      "loss": 0.8411,
+      "step": 1388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.148920653442241e-06,
+      "loss": 0.6463,
+      "step": 1392
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1460035005834306e-06,
+      "loss": 0.5053,
+      "step": 1396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.143086347724621e-06,
+      "loss": 0.4919,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.140169194865811e-06,
+      "loss": 0.7352,
+      "step": 1404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.137252042007002e-06,
+      "loss": 0.5577,
+      "step": 1408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.134334889148192e-06,
+      "loss": 0.7224,
+      "step": 1412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.131417736289382e-06,
+      "loss": 0.9195,
+      "step": 1416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.128500583430572e-06,
+      "loss": 0.7652,
+      "step": 1420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.125583430571763e-06,
+      "loss": 0.756,
+      "step": 1424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.122666277712953e-06,
+      "loss": 0.7378,
+      "step": 1428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.1197491248541425e-06,
+      "loss": 0.7264,
+      "step": 1432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.116831971995333e-06,
+      "loss": 0.7489,
+      "step": 1436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.113914819136523e-06,
+      "loss": 0.661,
+      "step": 1440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.110997666277713e-06,
+      "loss": 0.6761,
+      "step": 1444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.108080513418903e-06,
+      "loss": 0.62,
+      "step": 1448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.105163360560094e-06,
+      "loss": 0.5477,
+      "step": 1452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.102246207701284e-06,
+      "loss": 0.7107,
+      "step": 1456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.099329054842474e-06,
+      "loss": 0.9508,
+      "step": 1460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.096411901983664e-06,
+      "loss": 0.7668,
+      "step": 1464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.0934947491248545e-06,
+      "loss": 0.7971,
+      "step": 1468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.0905775962660445e-06,
+      "loss": 0.7532,
+      "step": 1472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.087660443407234e-06,
+      "loss": 0.6721,
+      "step": 1476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.084743290548425e-06,
+      "loss": 0.702,
+      "step": 1480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.081826137689615e-06,
+      "loss": 0.5931,
+      "step": 1484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.078908984830806e-06,
+      "loss": 0.8941,
+      "step": 1488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.075991831971996e-06,
+      "loss": 0.4728,
+      "step": 1492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.073074679113186e-06,
+      "loss": 0.5366,
+      "step": 1496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.070157526254377e-06,
+      "loss": 0.6567,
+      "step": 1500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.0672403733955665e-06,
+      "loss": 0.6594,
+      "step": 1504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.0643232205367565e-06,
+      "loss": 0.5096,
+      "step": 1508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.061406067677946e-06,
+      "loss": 0.6852,
+      "step": 1512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.058488914819137e-06,
+      "loss": 0.7667,
+      "step": 1516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.055571761960327e-06,
+      "loss": 0.662,
+      "step": 1520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.052654609101517e-06,
+      "loss": 0.6275,
+      "step": 1524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.049737456242707e-06,
+      "loss": 0.715,
+      "step": 1528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.046820303383898e-06,
+      "loss": 0.4791,
+      "step": 1532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.043903150525088e-06,
+      "loss": 0.6893,
+      "step": 1536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.040985997666278e-06,
+      "loss": 0.4941,
+      "step": 1540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.0380688448074685e-06,
+      "loss": 0.5576,
+      "step": 1544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.035151691948658e-06,
+      "loss": 0.7089,
+      "step": 1548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.032234539089848e-06,
+      "loss": 0.4944,
+      "step": 1552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.029317386231038e-06,
+      "loss": 0.4799,
+      "step": 1556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.026400233372229e-06,
+      "loss": 0.6483,
+      "step": 1560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.023483080513419e-06,
+      "loss": 0.7882,
+      "step": 1564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.02056592765461e-06,
+      "loss": 0.6979,
+      "step": 1568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.0176487747958e-06,
+      "loss": 0.6079,
+      "step": 1572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.01473162193699e-06,
+      "loss": 0.7046,
+      "step": 1576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.0118144690781804e-06,
+      "loss": 0.5994,
+      "step": 1580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.00889731621937e-06,
+      "loss": 0.628,
+      "step": 1584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.00598016336056e-06,
+      "loss": 0.4581,
+      "step": 1588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.00306301050175e-06,
+      "loss": 0.6283,
+      "step": 1592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.000145857642941e-06,
+      "loss": 0.6736,
+      "step": 1596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.997228704784131e-06,
+      "loss": 0.6118,
+      "step": 1600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.994311551925321e-06,
+      "loss": 0.8692,
+      "step": 1604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.991394399066512e-06,
+      "loss": 0.7431,
+      "step": 1608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.988477246207702e-06,
+      "loss": 0.6712,
+      "step": 1612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.985560093348892e-06,
+      "loss": 0.5141,
+      "step": 1616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9826429404900815e-06,
+      "loss": 0.4037,
+      "step": 1620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.979725787631272e-06,
+      "loss": 0.5442,
+      "step": 1624
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.976808634772462e-06,
+      "loss": 0.7256,
+      "step": 1628
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.973891481913652e-06,
+      "loss": 0.7292,
+      "step": 1632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.970974329054843e-06,
+      "loss": 0.7739,
+      "step": 1636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.968057176196033e-06,
+      "loss": 0.6673,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.965140023337224e-06,
+      "loss": 0.8099,
+      "step": 1644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.962222870478414e-06,
+      "loss": 0.5271,
+      "step": 1648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9593057176196036e-06,
+      "loss": 0.7032,
+      "step": 1652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9563885647607935e-06,
+      "loss": 0.6766,
+      "step": 1656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.953471411901984e-06,
+      "loss": 0.6286,
+      "step": 1660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.950554259043174e-06,
+      "loss": 0.6236,
+      "step": 1664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.947637106184364e-06,
+      "loss": 0.7438,
+      "step": 1668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.944719953325555e-06,
+      "loss": 0.6499,
+      "step": 1672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.941802800466745e-06,
+      "loss": 0.7669,
+      "step": 1676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.938885647607935e-06,
+      "loss": 0.4944,
+      "step": 1680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.935968494749125e-06,
+      "loss": 0.4765,
+      "step": 1684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9330513418903156e-06,
+      "loss": 0.7785,
+      "step": 1688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9301341890315055e-06,
+      "loss": 0.6521,
+      "step": 1692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9272170361726954e-06,
+      "loss": 0.6744,
+      "step": 1696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924299883313886e-06,
+      "loss": 0.6819,
+      "step": 1700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921382730455076e-06,
+      "loss": 0.6566,
+      "step": 1704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.918465577596267e-06,
+      "loss": 0.7201,
+      "step": 1708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.915548424737456e-06,
+      "loss": 0.373,
+      "step": 1712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.912631271878647e-06,
+      "loss": 0.8481,
+      "step": 1716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.909714119019837e-06,
+      "loss": 0.6479,
+      "step": 1720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9067969661610276e-06,
+      "loss": 0.5682,
+      "step": 1724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9038798133022175e-06,
+      "loss": 0.651,
+      "step": 1728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9009626604434074e-06,
+      "loss": 0.637,
+      "step": 1732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.898045507584598e-06,
+      "loss": 0.5119,
+      "step": 1736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.895128354725788e-06,
+      "loss": 0.5434,
+      "step": 1740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.892211201866978e-06,
+      "loss": 0.7295,
+      "step": 1744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.889294049008168e-06,
+      "loss": 0.7214,
+      "step": 1748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.886376896149359e-06,
+      "loss": 0.5946,
+      "step": 1752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.883459743290549e-06,
+      "loss": 0.6668,
+      "step": 1756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.880542590431739e-06,
+      "loss": 0.6054,
+      "step": 1760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.877625437572929e-06,
+      "loss": 0.6904,
+      "step": 1764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.874708284714119e-06,
+      "loss": 0.7803,
+      "step": 1768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.871791131855309e-06,
+      "loss": 0.6766,
+      "step": 1772
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.868873978996499e-06,
+      "loss": 0.6957,
+      "step": 1776
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.86595682613769e-06,
+      "loss": 0.8129,
+      "step": 1780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.86303967327888e-06,
+      "loss": 0.7652,
+      "step": 1784
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.860122520420071e-06,
+      "loss": 0.4505,
+      "step": 1788
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.85720536756126e-06,
+      "loss": 0.6318,
+      "step": 1792
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.854288214702451e-06,
+      "loss": 0.7018,
+      "step": 1796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.8513710618436415e-06,
+      "loss": 0.8348,
+      "step": 1800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.848453908984831e-06,
+      "loss": 0.5523,
+      "step": 1804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.845536756126021e-06,
+      "loss": 0.5858,
+      "step": 1808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.842619603267211e-06,
+      "loss": 0.3905,
+      "step": 1812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.839702450408402e-06,
+      "loss": 0.7015,
+      "step": 1816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.836785297549592e-06,
+      "loss": 0.4228,
+      "step": 1820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.833868144690782e-06,
+      "loss": 0.7709,
+      "step": 1824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.830950991831972e-06,
+      "loss": 0.5833,
+      "step": 1828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.828033838973163e-06,
+      "loss": 0.6342,
+      "step": 1832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.825116686114353e-06,
+      "loss": 0.843,
+      "step": 1836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.8221995332555425e-06,
+      "loss": 0.5876,
+      "step": 1840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.819282380396733e-06,
+      "loss": 0.5432,
+      "step": 1844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.816365227537923e-06,
+      "loss": 0.7204,
+      "step": 1848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8134480746791136e-06,
+      "loss": 0.4466,
+      "step": 1852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8105309218203036e-06,
+      "loss": 0.623,
+      "step": 1856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.807613768961494e-06,
+      "loss": 0.3623,
+      "step": 1860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.804696616102684e-06,
+      "loss": 0.6491,
+      "step": 1864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.8017794632438742e-06,
+      "loss": 0.6469,
+      "step": 1868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7988623103850646e-06,
+      "loss": 0.5412,
+      "step": 1872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.7959451575262545e-06,
+      "loss": 0.995,
+      "step": 1876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.793028004667445e-06,
+      "loss": 0.6152,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.790110851808635e-06,
+      "loss": 0.7216,
+      "step": 1884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.787193698949825e-06,
+      "loss": 0.5137,
+      "step": 1888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.784276546091015e-06,
+      "loss": 0.6703,
+      "step": 1892
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7813593932322055e-06,
+      "loss": 0.5394,
+      "step": 1896
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7784422403733963e-06,
+      "loss": 0.6228,
+      "step": 1900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7755250875145862e-06,
+      "loss": 0.6231,
+      "step": 1904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7726079346557766e-06,
+      "loss": 0.6658,
+      "step": 1908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7696907817969665e-06,
+      "loss": 0.5279,
+      "step": 1912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.766773628938157e-06,
+      "loss": 0.4709,
+      "step": 1916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.763856476079347e-06,
+      "loss": 0.6442,
+      "step": 1920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.760939323220537e-06,
+      "loss": 0.6787,
+      "step": 1924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.758022170361727e-06,
+      "loss": 0.6453,
+      "step": 1928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7551050175029175e-06,
+      "loss": 0.7487,
+      "step": 1932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.752187864644108e-06,
+      "loss": 0.6078,
+      "step": 1936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.749270711785298e-06,
+      "loss": 0.5252,
+      "step": 1940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.746353558926488e-06,
+      "loss": 0.4936,
+      "step": 1944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.743436406067678e-06,
+      "loss": 0.4545,
+      "step": 1948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7405192532088685e-06,
+      "loss": 0.7937,
+      "step": 1952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7376021003500584e-06,
+      "loss": 0.8007,
+      "step": 1956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.7346849474912488e-06,
+      "loss": 0.4401,
+      "step": 1960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.731767794632439e-06,
+      "loss": 0.8051,
+      "step": 1964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.728850641773629e-06,
+      "loss": 0.7178,
+      "step": 1968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.72593348891482e-06,
+      "loss": 0.5673,
+      "step": 1972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7230163360560094e-06,
+      "loss": 0.8238,
+      "step": 1976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7200991831972e-06,
+      "loss": 0.646,
+      "step": 1980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.7171820303383897e-06,
+      "loss": 0.5166,
+      "step": 1984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7142648774795804e-06,
+      "loss": 0.5725,
+      "step": 1988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7113477246207704e-06,
+      "loss": 0.8298,
+      "step": 1992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.7084305717619607e-06,
+      "loss": 0.5858,
+      "step": 1996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.705513418903151e-06,
+      "loss": 0.5432,
+      "step": 2000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.702596266044341e-06,
+      "loss": 0.5454,
+      "step": 2004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.6996791131855314e-06,
+      "loss": 0.6,
+      "step": 2008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6967619603267213e-06,
+      "loss": 0.5889,
+      "step": 2012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6938448074679117e-06,
+      "loss": 0.7542,
+      "step": 2016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6909276546091016e-06,
+      "loss": 0.7573,
+      "step": 2020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.688010501750292e-06,
+      "loss": 0.7429,
+      "step": 2024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.685093348891482e-06,
+      "loss": 0.6015,
+      "step": 2028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.6821761960326723e-06,
+      "loss": 0.6009,
+      "step": 2032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6792590431738627e-06,
+      "loss": 0.4447,
+      "step": 2036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6763418903150526e-06,
+      "loss": 0.667,
+      "step": 2040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.673424737456243e-06,
+      "loss": 0.594,
+      "step": 2044
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.670507584597433e-06,
+      "loss": 0.6368,
+      "step": 2048
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6675904317386237e-06,
+      "loss": 0.5883,
+      "step": 2052
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.6646732788798132e-06,
+      "loss": 0.7004,
+      "step": 2056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.661756126021004e-06,
+      "loss": 0.6169,
+      "step": 2060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6588389731621944e-06,
+      "loss": 0.6757,
+      "step": 2064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6559218203033843e-06,
+      "loss": 0.6134,
+      "step": 2068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6530046674445747e-06,
+      "loss": 0.6381,
+      "step": 2072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6500875145857646e-06,
+      "loss": 0.715,
+      "step": 2076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.647170361726955e-06,
+      "loss": 0.6319,
+      "step": 2080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.644253208868145e-06,
+      "loss": 0.4187,
+      "step": 2084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6413360560093353e-06,
+      "loss": 0.5733,
+      "step": 2088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.638418903150525e-06,
+      "loss": 0.6943,
+      "step": 2092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6355017502917156e-06,
+      "loss": 0.4808,
+      "step": 2096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.632584597432906e-06,
+      "loss": 0.4813,
+      "step": 2100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.629667444574096e-06,
+      "loss": 0.5777,
+      "step": 2104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.6267502917152862e-06,
+      "loss": 0.5967,
+      "step": 2108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.623833138856476e-06,
+      "loss": 0.5644,
+      "step": 2112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.6209159859976665e-06,
+      "loss": 0.6584,
+      "step": 2116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.6179988331388565e-06,
+      "loss": 0.5162,
+      "step": 2120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.615081680280047e-06,
+      "loss": 0.6585,
+      "step": 2124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.6121645274212368e-06,
+      "loss": 0.7716,
+      "step": 2128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6092473745624276e-06,
+      "loss": 0.5249,
+      "step": 2132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.606330221703618e-06,
+      "loss": 0.722,
+      "step": 2136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.603413068844808e-06,
+      "loss": 0.6634,
+      "step": 2140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6004959159859982e-06,
+      "loss": 0.6547,
+      "step": 2144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.597578763127188e-06,
+      "loss": 0.7804,
+      "step": 2148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5946616102683785e-06,
+      "loss": 0.5373,
+      "step": 2152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5917444574095685e-06,
+      "loss": 0.6122,
+      "step": 2156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.588827304550759e-06,
+      "loss": 0.5015,
+      "step": 2160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.585910151691949e-06,
+      "loss": 0.6872,
+      "step": 2164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.582992998833139e-06,
+      "loss": 0.3367,
+      "step": 2168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5800758459743295e-06,
+      "loss": 0.4729,
+      "step": 2172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5771586931155194e-06,
+      "loss": 0.7766,
+      "step": 2176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5742415402567098e-06,
+      "loss": 0.7324,
+      "step": 2180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5713243873978997e-06,
+      "loss": 0.5244,
+      "step": 2184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.56840723453909e-06,
+      "loss": 0.6321,
+      "step": 2188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.56549008168028e-06,
+      "loss": 0.6109,
+      "step": 2192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.5625729288214704e-06,
+      "loss": 0.5534,
+      "step": 2196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.559655775962661e-06,
+      "loss": 0.5453,
+      "step": 2200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5567386231038507e-06,
+      "loss": 0.4379,
+      "step": 2204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5538214702450415e-06,
+      "loss": 0.5268,
+      "step": 2208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5509043173862314e-06,
+      "loss": 0.7081,
+      "step": 2212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5479871645274218e-06,
+      "loss": 0.5149,
+      "step": 2216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.5450700116686117e-06,
+      "loss": 0.4048,
+      "step": 2220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.542152858809802e-06,
+      "loss": 0.7552,
+      "step": 2224
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.539235705950992e-06,
+      "loss": 0.4624,
+      "step": 2228
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5363185530921824e-06,
+      "loss": 0.5872,
+      "step": 2232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5334014002333727e-06,
+      "loss": 0.5265,
+      "step": 2236
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.5304842473745627e-06,
+      "loss": 0.4802,
+      "step": 2240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.527567094515753e-06,
+      "loss": 0.6172,
+      "step": 2244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.524649941656943e-06,
+      "loss": 0.5788,
+      "step": 2248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5217327887981333e-06,
+      "loss": 0.5426,
+      "step": 2252
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5188156359393233e-06,
+      "loss": 0.6427,
+      "step": 2256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5158984830805136e-06,
+      "loss": 0.6474,
+      "step": 2260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.512981330221704e-06,
+      "loss": 0.5518,
+      "step": 2264
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.510064177362894e-06,
+      "loss": 0.6028,
+      "step": 2268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5071470245040843e-06,
+      "loss": 0.4178,
+      "step": 2272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5042298716452742e-06,
+      "loss": 0.761,
+      "step": 2276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.501312718786465e-06,
+      "loss": 0.7076,
+      "step": 2280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4983955659276545e-06,
+      "loss": 0.5466,
+      "step": 2284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4954784130688453e-06,
+      "loss": 0.4539,
+      "step": 2288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4925612602100353e-06,
+      "loss": 0.6436,
+      "step": 2292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4896441073512256e-06,
+      "loss": 0.471,
+      "step": 2296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.486726954492416e-06,
+      "loss": 0.6848,
+      "step": 2300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.483809801633606e-06,
+      "loss": 0.6043,
+      "step": 2304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4808926487747963e-06,
+      "loss": 0.6458,
+      "step": 2308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4779754959159862e-06,
+      "loss": 0.6336,
+      "step": 2312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4750583430571766e-06,
+      "loss": 0.5843,
+      "step": 2316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4721411901983665e-06,
+      "loss": 0.4364,
+      "step": 2320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.469224037339557e-06,
+      "loss": 0.5796,
+      "step": 2324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.466306884480747e-06,
+      "loss": 0.5754,
+      "step": 2328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.463389731621937e-06,
+      "loss": 0.6848,
+      "step": 2332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4604725787631276e-06,
+      "loss": 0.6489,
+      "step": 2336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4575554259043175e-06,
+      "loss": 0.6255,
+      "step": 2340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.454638273045508e-06,
+      "loss": 0.4827,
+      "step": 2344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.451721120186698e-06,
+      "loss": 0.5178,
+      "step": 2348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.448803967327888e-06,
+      "loss": 0.5294,
+      "step": 2352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.445886814469078e-06,
+      "loss": 0.4968,
+      "step": 2356
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.442969661610269e-06,
+      "loss": 0.7301,
+      "step": 2360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4400525087514592e-06,
+      "loss": 0.5503,
+      "step": 2364
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.437135355892649e-06,
+      "loss": 0.4724,
+      "step": 2368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4342182030338395e-06,
+      "loss": 0.5471,
+      "step": 2372
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4313010501750295e-06,
+      "loss": 0.5414,
+      "step": 2376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.42838389731622e-06,
+      "loss": 0.4228,
+      "step": 2380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.4254667444574098e-06,
+      "loss": 0.6098,
+      "step": 2384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4225495915986e-06,
+      "loss": 0.5935,
+      "step": 2388
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.41963243873979e-06,
+      "loss": 0.5567,
+      "step": 2392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4167152858809804e-06,
+      "loss": 0.4576,
+      "step": 2396
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.413798133022171e-06,
+      "loss": 0.5125,
+      "step": 2400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.4108809801633607e-06,
+      "loss": 0.4188,
+      "step": 2404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.407963827304551e-06,
+      "loss": 0.6133,
+      "step": 2408
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.405046674445741e-06,
+      "loss": 0.5203,
+      "step": 2412
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.4021295215869314e-06,
+      "loss": 0.5936,
+      "step": 2416
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3992123687281213e-06,
+      "loss": 0.5783,
+      "step": 2420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3962952158693117e-06,
+      "loss": 0.5501,
+      "step": 2424
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.3933780630105017e-06,
+      "loss": 0.6859,
+      "step": 2428
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.390460910151692e-06,
+      "loss": 0.5361,
+      "step": 2432
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.387543757292883e-06,
+      "loss": 0.3481,
+      "step": 2436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3846266044340727e-06,
+      "loss": 0.5475,
+      "step": 2440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.381709451575263e-06,
+      "loss": 0.4606,
+      "step": 2444
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.378792298716453e-06,
+      "loss": 0.3753,
+      "step": 2448
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3758751458576434e-06,
+      "loss": 0.5286,
+      "step": 2452
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.3729579929988333e-06,
+      "loss": 0.5214,
+      "step": 2456
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.3700408401400237e-06,
+      "loss": 0.4971,
+      "step": 2460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.367123687281214e-06,
+      "loss": 0.5731,
+      "step": 2464
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.364206534422404e-06,
+      "loss": 0.6563,
+      "step": 2468
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.3612893815635944e-06,
+      "loss": 0.3885,
+      "step": 2472
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.3583722287047843e-06,
+      "loss": 0.7304,
+      "step": 2476
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.3554550758459747e-06,
+      "loss": 0.6601,
+      "step": 2480
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.3525379229871646e-06,
+      "loss": 0.4629,
+      "step": 2484
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.349620770128355e-06,
+      "loss": 0.5769,
+      "step": 2488
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.346703617269545e-06,
+      "loss": 0.5616,
+      "step": 2492
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.3437864644107353e-06,
+      "loss": 0.4976,
+      "step": 2496
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.340869311551926e-06,
+      "loss": 0.6104,
+      "step": 2500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.3379521586931156e-06,
+      "loss": 0.4875,
+      "step": 2504
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.3350350058343064e-06,
+      "loss": 0.3093,
+      "step": 2508
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.332117852975496e-06,
+      "loss": 0.3887,
+      "step": 2512
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.3292007001166867e-06,
+      "loss": 0.6688,
+      "step": 2516
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.3262835472578766e-06,
+      "loss": 0.479,
+      "step": 2520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.323366394399067e-06,
+      "loss": 0.616,
+      "step": 2524
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.320449241540257e-06,
+      "loss": 0.8513,
+      "step": 2528
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.3175320886814473e-06,
+      "loss": 0.4224,
+      "step": 2532
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.3146149358226376e-06,
+      "loss": 0.4577,
+      "step": 2536
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.3116977829638276e-06,
+      "loss": 0.4286,
+      "step": 2540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.308780630105018e-06,
+      "loss": 0.5298,
+      "step": 2544
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.305863477246208e-06,
+      "loss": 0.4531,
+      "step": 2548
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.3029463243873982e-06,
+      "loss": 0.561,
+      "step": 2552
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.300029171528588e-06,
+      "loss": 0.4931,
+      "step": 2556
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.2971120186697785e-06,
+      "loss": 0.6341,
+      "step": 2560
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.294194865810969e-06,
+      "loss": 0.5096,
+      "step": 2564
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.291277712952159e-06,
+      "loss": 0.3863,
+      "step": 2568
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.288360560093349e-06,
+      "loss": 0.6633,
+      "step": 2572
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.285443407234539e-06,
+      "loss": 0.7305,
+      "step": 2576
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.2825262543757295e-06,
+      "loss": 0.627,
+      "step": 2580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.2796091015169194e-06,
+      "loss": 0.4865,
+      "step": 2584
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.27669194865811e-06,
+      "loss": 0.507,
+      "step": 2588
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.2737747957992997e-06,
+      "loss": 0.5226,
+      "step": 2592
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.2708576429404905e-06,
+      "loss": 0.5523,
+      "step": 2596
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.267940490081681e-06,
+      "loss": 0.5195,
+      "step": 2600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.265023337222871e-06,
+      "loss": 0.3034,
+      "step": 2604
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.262106184364061e-06,
+      "loss": 0.7056,
+      "step": 2608
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.259189031505251e-06,
+      "loss": 0.6256,
+      "step": 2612
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.2562718786464415e-06,
+      "loss": 0.4434,
+      "step": 2616
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.2533547257876314e-06,
+      "loss": 0.3871,
+      "step": 2620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.2504375729288218e-06,
+      "loss": 0.6902,
+      "step": 2624
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.2475204200700117e-06,
+      "loss": 0.3734,
+      "step": 2628
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.244603267211202e-06,
+      "loss": 0.4771,
+      "step": 2632
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.2416861143523924e-06,
+      "loss": 0.5457,
+      "step": 2636
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.2387689614935824e-06,
+      "loss": 0.5787,
+      "step": 2640
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.2358518086347727e-06,
+      "loss": 0.7326,
+      "step": 2644
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.2329346557759627e-06,
+      "loss": 0.4606,
+      "step": 2648
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.230017502917153e-06,
+      "loss": 0.5832,
+      "step": 2652
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.227100350058343e-06,
+      "loss": 0.4674,
+      "step": 2656
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.2241831971995333e-06,
+      "loss": 0.3263,
+      "step": 2660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.221266044340724e-06,
+      "loss": 0.433,
+      "step": 2664
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.218348891481914e-06,
+      "loss": 0.562,
+      "step": 2668
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.2154317386231044e-06,
+      "loss": 0.5465,
+      "step": 2672
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.2125145857642944e-06,
+      "loss": 0.5806,
+      "step": 2676
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.2095974329054847e-06,
+      "loss": 0.5941,
+      "step": 2680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.2066802800466747e-06,
+      "loss": 0.5672,
+      "step": 2684
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.203763127187865e-06,
+      "loss": 0.4044,
+      "step": 2688
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.200845974329055e-06,
+      "loss": 0.6372,
+      "step": 2692
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1979288214702453e-06,
+      "loss": 0.5306,
+      "step": 2696
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1950116686114357e-06,
+      "loss": 0.6751,
+      "step": 2700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1920945157526256e-06,
+      "loss": 0.3999,
+      "step": 2704
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.189177362893816e-06,
+      "loss": 0.6431,
+      "step": 2708
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.186260210035006e-06,
+      "loss": 0.7637,
+      "step": 2712
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.1833430571761963e-06,
+      "loss": 0.4698,
+      "step": 2716
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.1804259043173862e-06,
+      "loss": 0.5558,
+      "step": 2720
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.1775087514585766e-06,
+      "loss": 0.5202,
+      "step": 2724
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.1745915985997665e-06,
+      "loss": 0.4455,
+      "step": 2728
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.171674445740957e-06,
+      "loss": 0.6568,
+      "step": 2732
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.1687572928821477e-06,
+      "loss": 0.6879,
+      "step": 2736
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.165840140023337e-06,
+      "loss": 0.474,
+      "step": 2740
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.162922987164528e-06,
+      "loss": 0.5385,
+      "step": 2744
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.160005834305718e-06,
+      "loss": 0.6955,
+      "step": 2748
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.1570886814469083e-06,
+      "loss": 0.501,
+      "step": 2752
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.1541715285880982e-06,
+      "loss": 0.5093,
+      "step": 2756
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.1512543757292886e-06,
+      "loss": 0.3698,
+      "step": 2760
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.148337222870479e-06,
+      "loss": 0.4271,
+      "step": 2764
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.145420070011669e-06,
+      "loss": 0.5814,
+      "step": 2768
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.1425029171528592e-06,
+      "loss": 0.3312,
+      "step": 2772
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.139585764294049e-06,
+      "loss": 0.5727,
+      "step": 2776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.1366686114352395e-06,
+      "loss": 0.6723,
+      "step": 2780
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.1337514585764295e-06,
+      "loss": 0.5772,
+      "step": 2784
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.13083430571762e-06,
+      "loss": 0.2121,
+      "step": 2788
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.1279171528588098e-06,
+      "loss": 0.5148,
+      "step": 2792
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.125e-06,
+      "loss": 0.402,
+      "step": 2796
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.1220828471411905e-06,
+      "loss": 0.4757,
+      "step": 2800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.1191656942823805e-06,
+      "loss": 0.5157,
+      "step": 2804
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.1162485414235712e-06,
+      "loss": 0.5417,
+      "step": 2808
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.1133313885647608e-06,
+      "loss": 0.3401,
+      "step": 2812
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.1104142357059515e-06,
+      "loss": 0.6862,
+      "step": 2816
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.107497082847141e-06,
+      "loss": 0.516,
+      "step": 2820
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.104579929988332e-06,
+      "loss": 0.5201,
+      "step": 2824
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.1016627771295218e-06,
+      "loss": 0.4436,
+      "step": 2828
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.098745624270712e-06,
+      "loss": 0.5983,
+      "step": 2832
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0958284714119025e-06,
+      "loss": 0.5076,
+      "step": 2836
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0929113185530924e-06,
+      "loss": 0.4357,
+      "step": 2840
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.089994165694283e-06,
+      "loss": 0.5015,
+      "step": 2844
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.0870770128354727e-06,
+      "loss": 0.5186,
+      "step": 2848
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.084159859976663e-06,
+      "loss": 0.5653,
+      "step": 2852
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.081242707117853e-06,
+      "loss": 0.6339,
+      "step": 2856
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.0783255542590434e-06,
+      "loss": 0.5534,
+      "step": 2860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.0754084014002338e-06,
+      "loss": 0.5144,
+      "step": 2864
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.0724912485414237e-06,
+      "loss": 0.393,
+      "step": 2868
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.069574095682614e-06,
+      "loss": 0.4453,
+      "step": 2872
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.066656942823804e-06,
+      "loss": 0.5278,
+      "step": 2876
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.0637397899649944e-06,
+      "loss": 0.7833,
+      "step": 2880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.0608226371061843e-06,
+      "loss": 0.4581,
+      "step": 2884
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.0579054842473747e-06,
+      "loss": 0.3942,
+      "step": 2888
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.0549883313885646e-06,
+      "loss": 0.6298,
+      "step": 2892
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.0520711785297554e-06,
+      "loss": 0.6276,
+      "step": 2896
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.0491540256709458e-06,
+      "loss": 0.737,
+      "step": 2900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.0462368728121357e-06,
+      "loss": 0.5208,
+      "step": 2904
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.043319719953326e-06,
+      "loss": 0.5953,
+      "step": 2908
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.040402567094516e-06,
+      "loss": 0.5373,
+      "step": 2912
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.0374854142357064e-06,
+      "loss": 0.3865,
+      "step": 2916
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.0345682613768963e-06,
+      "loss": 0.5451,
+      "step": 2920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.0316511085180867e-06,
+      "loss": 0.4434,
+      "step": 2924
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.0287339556592766e-06,
+      "loss": 0.6256,
+      "step": 2928
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.025816802800467e-06,
+      "loss": 0.4372,
+      "step": 2932
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.0228996499416573e-06,
+      "loss": 0.4225,
+      "step": 2936
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.0199824970828473e-06,
+      "loss": 0.6176,
+      "step": 2940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.0170653442240376e-06,
+      "loss": 0.5814,
+      "step": 2944
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.0141481913652276e-06,
+      "loss": 0.6453,
+      "step": 2948
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.011231038506418e-06,
+      "loss": 0.599,
+      "step": 2952
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.008313885647608e-06,
+      "loss": 0.4661,
+      "step": 2956
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.0053967327887982e-06,
+      "loss": 0.3218,
+      "step": 2960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.003208868144691e-06,
+      "loss": 0.6456,
+      "step": 2964
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.0002917152858813e-06,
+      "loss": 0.2891,
+      "step": 2968
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9973745624270716e-06,
+      "loss": 0.4821,
+      "step": 2972
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9944574095682616e-06,
+      "loss": 0.733,
+      "step": 2976
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.991540256709452e-06,
+      "loss": 0.2193,
+      "step": 2980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.988623103850642e-06,
+      "loss": 0.4148,
+      "step": 2984
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9857059509918322e-06,
+      "loss": 0.5474,
+      "step": 2988
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.982788798133022e-06,
+      "loss": 0.4005,
+      "step": 2992
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.9798716452742125e-06,
+      "loss": 0.5799,
+      "step": 2996
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.9769544924154025e-06,
+      "loss": 0.4967,
+      "step": 3000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.974037339556593e-06,
+      "loss": 0.4672,
+      "step": 3004
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.9711201866977836e-06,
+      "loss": 0.4892,
+      "step": 3008
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.9682030338389735e-06,
+      "loss": 0.554,
+      "step": 3012
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.965285880980164e-06,
+      "loss": 0.3553,
+      "step": 3016
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.962368728121354e-06,
+      "loss": 0.5259,
+      "step": 3020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.959451575262544e-06,
+      "loss": 0.4461,
+      "step": 3024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.956534422403734e-06,
+      "loss": 0.4768,
+      "step": 3028
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.9536172695449245e-06,
+      "loss": 0.4537,
+      "step": 3032
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.9507001166861144e-06,
+      "loss": 0.5611,
+      "step": 3036
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.947782963827305e-06,
+      "loss": 0.5684,
+      "step": 3040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.944865810968495e-06,
+      "loss": 0.2357,
+      "step": 3044
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.941948658109685e-06,
+      "loss": 0.4909,
+      "step": 3048
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.9390315052508755e-06,
+      "loss": 0.5225,
+      "step": 3052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.9361143523920654e-06,
+      "loss": 0.391,
+      "step": 3056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.9331971995332558e-06,
+      "loss": 0.4058,
+      "step": 3060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.9302800466744457e-06,
+      "loss": 0.446,
+      "step": 3064
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.927362893815636e-06,
+      "loss": 0.3136,
+      "step": 3068
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.924445740956827e-06,
+      "loss": 0.4259,
+      "step": 3072
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.9215285880980164e-06,
+      "loss": 0.4293,
+      "step": 3076
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.918611435239207e-06,
+      "loss": 0.6366,
+      "step": 3080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.9156942823803967e-06,
+      "loss": 0.5103,
+      "step": 3084
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.9127771295215875e-06,
+      "loss": 0.4994,
+      "step": 3088
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.9098599766627774e-06,
+      "loss": 0.5369,
+      "step": 3092
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.9069428238039678e-06,
+      "loss": 0.4632,
+      "step": 3096
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.9040256709451577e-06,
+      "loss": 0.3315,
+      "step": 3100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.901108518086348e-06,
+      "loss": 0.5776,
+      "step": 3104
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8981913652275384e-06,
+      "loss": 0.4545,
+      "step": 3108
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8952742123687284e-06,
+      "loss": 0.6122,
+      "step": 3112
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8923570595099187e-06,
+      "loss": 0.2894,
+      "step": 3116
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8894399066511087e-06,
+      "loss": 0.5543,
+      "step": 3120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.886522753792299e-06,
+      "loss": 0.524,
+      "step": 3124
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.883605600933489e-06,
+      "loss": 0.5279,
+      "step": 3128
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8806884480746793e-06,
+      "loss": 0.3323,
+      "step": 3132
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8777712952158693e-06,
+      "loss": 0.5169,
+      "step": 3136
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8748541423570596e-06,
+      "loss": 0.4117,
+      "step": 3140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.87193698949825e-06,
+      "loss": 0.2958,
+      "step": 3144
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.86901983663944e-06,
+      "loss": 0.5434,
+      "step": 3148
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8661026837806307e-06,
+      "loss": 0.3813,
+      "step": 3152
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8631855309218202e-06,
+      "loss": 0.4985,
+      "step": 3156
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.860268378063011e-06,
+      "loss": 0.5145,
+      "step": 3160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8573512252042005e-06,
+      "loss": 0.4176,
+      "step": 3164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8544340723453913e-06,
+      "loss": 0.4034,
+      "step": 3168
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8515169194865817e-06,
+      "loss": 0.514,
+      "step": 3172
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8485997666277716e-06,
+      "loss": 0.4951,
+      "step": 3176
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.845682613768962e-06,
+      "loss": 0.3946,
+      "step": 3180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.842765460910152e-06,
+      "loss": 0.4,
+      "step": 3184
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8398483080513423e-06,
+      "loss": 0.5394,
+      "step": 3188
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8369311551925322e-06,
+      "loss": 0.6328,
+      "step": 3192
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8340140023337226e-06,
+      "loss": 0.4449,
+      "step": 3196
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8310968494749125e-06,
+      "loss": 0.4787,
+      "step": 3200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.828179696616103e-06,
+      "loss": 0.3408,
+      "step": 3204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8252625437572932e-06,
+      "loss": 0.4688,
+      "step": 3208
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.822345390898483e-06,
+      "loss": 0.557,
+      "step": 3212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8194282380396735e-06,
+      "loss": 0.5299,
+      "step": 3216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8165110851808635e-06,
+      "loss": 0.468,
+      "step": 3220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.813593932322054e-06,
+      "loss": 0.3655,
+      "step": 3224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8106767794632438e-06,
+      "loss": 0.2575,
+      "step": 3228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.807759626604434e-06,
+      "loss": 0.5705,
+      "step": 3232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.804842473745624e-06,
+      "loss": 0.4812,
+      "step": 3236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.801925320886815e-06,
+      "loss": 0.5761,
+      "step": 3240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7990081680280052e-06,
+      "loss": 0.6039,
+      "step": 3244
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.796091015169195e-06,
+      "loss": 0.3454,
+      "step": 3248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7931738623103855e-06,
+      "loss": 0.5554,
+      "step": 3252
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7902567094515755e-06,
+      "loss": 0.4531,
+      "step": 3256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.787339556592766e-06,
+      "loss": 0.3332,
+      "step": 3260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7844224037339558e-06,
+      "loss": 0.3159,
+      "step": 3264
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.781505250875146e-06,
+      "loss": 0.498,
+      "step": 3268
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7785880980163365e-06,
+      "loss": 0.4386,
+      "step": 3272
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7756709451575264e-06,
+      "loss": 0.4108,
+      "step": 3276
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.772753792298717e-06,
+      "loss": 0.3739,
+      "step": 3280
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7698366394399067e-06,
+      "loss": 0.3242,
+      "step": 3284
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.766919486581097e-06,
+      "loss": 0.5226,
+      "step": 3288
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.764002333722287e-06,
+      "loss": 0.655,
+      "step": 3292
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7610851808634774e-06,
+      "loss": 0.3379,
+      "step": 3296
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7581680280046673e-06,
+      "loss": 0.5461,
+      "step": 3300
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7552508751458577e-06,
+      "loss": 0.4614,
+      "step": 3304
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7523337222870485e-06,
+      "loss": 0.6258,
+      "step": 3308
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.749416569428238e-06,
+      "loss": 0.3237,
+      "step": 3312
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7464994165694288e-06,
+      "loss": 0.4256,
+      "step": 3316
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7435822637106187e-06,
+      "loss": 0.4114,
+      "step": 3320
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.740665110851809e-06,
+      "loss": 0.7195,
+      "step": 3324
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.737747957992999e-06,
+      "loss": 0.5899,
+      "step": 3328
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7348308051341894e-06,
+      "loss": 0.4853,
+      "step": 3332
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7319136522753793e-06,
+      "loss": 0.4043,
+      "step": 3336
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7289964994165697e-06,
+      "loss": 0.5738,
+      "step": 3340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.72607934655776e-06,
+      "loss": 0.5704,
+      "step": 3344
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.72316219369895e-06,
+      "loss": 0.6312,
+      "step": 3348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7202450408401404e-06,
+      "loss": 0.3233,
+      "step": 3352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7173278879813303e-06,
+      "loss": 0.5598,
+      "step": 3356
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7144107351225207e-06,
+      "loss": 0.3501,
+      "step": 3360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7114935822637106e-06,
+      "loss": 0.3945,
+      "step": 3364
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.708576429404901e-06,
+      "loss": 0.4801,
+      "step": 3368
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7056592765460913e-06,
+      "loss": 0.5775,
+      "step": 3372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.7027421236872813e-06,
+      "loss": 0.3939,
+      "step": 3376
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.699824970828472e-06,
+      "loss": 0.4818,
+      "step": 3380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6969078179696616e-06,
+      "loss": 0.4698,
+      "step": 3384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6939906651108523e-06,
+      "loss": 0.3852,
+      "step": 3388
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.691073512252042e-06,
+      "loss": 0.4516,
+      "step": 3392
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6881563593932326e-06,
+      "loss": 0.3766,
+      "step": 3396
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6852392065344226e-06,
+      "loss": 0.315,
+      "step": 3400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.682322053675613e-06,
+      "loss": 0.4293,
+      "step": 3404
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6794049008168033e-06,
+      "loss": 0.5251,
+      "step": 3408
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6764877479579932e-06,
+      "loss": 0.4517,
+      "step": 3412
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6735705950991836e-06,
+      "loss": 0.568,
+      "step": 3416
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6706534422403735e-06,
+      "loss": 0.5349,
+      "step": 3420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.667736289381564e-06,
+      "loss": 0.4316,
+      "step": 3424
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.664819136522754e-06,
+      "loss": 0.372,
+      "step": 3428
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.661901983663944e-06,
+      "loss": 0.4546,
+      "step": 3432
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.658984830805134e-06,
+      "loss": 0.3791,
+      "step": 3436
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6560676779463245e-06,
+      "loss": 0.4281,
+      "step": 3440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.653150525087515e-06,
+      "loss": 0.3564,
+      "step": 3444
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.650233372228705e-06,
+      "loss": 0.443,
+      "step": 3448
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.647316219369895e-06,
+      "loss": 0.6713,
+      "step": 3452
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.644399066511085e-06,
+      "loss": 0.2832,
+      "step": 3456
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.641481913652276e-06,
+      "loss": 0.5267,
+      "step": 3460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6385647607934654e-06,
+      "loss": 0.688,
+      "step": 3464
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.635647607934656e-06,
+      "loss": 0.6487,
+      "step": 3468
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6327304550758466e-06,
+      "loss": 0.6653,
+      "step": 3472
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6298133022170365e-06,
+      "loss": 0.4218,
+      "step": 3476
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.626896149358227e-06,
+      "loss": 0.5679,
+      "step": 3480
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.623978996499417e-06,
+      "loss": 0.3429,
+      "step": 3484
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.621061843640607e-06,
+      "loss": 0.4428,
+      "step": 3488
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.618144690781797e-06,
+      "loss": 0.4073,
+      "step": 3492
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6152275379229875e-06,
+      "loss": 0.476,
+      "step": 3496
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6123103850641774e-06,
+      "loss": 0.3964,
+      "step": 3500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6093932322053678e-06,
+      "loss": 0.4971,
+      "step": 3504
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.606476079346558e-06,
+      "loss": 0.5309,
+      "step": 3508
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.603558926487748e-06,
+      "loss": 0.7094,
+      "step": 3512
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6006417736289384e-06,
+      "loss": 0.4345,
+      "step": 3516
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5977246207701284e-06,
+      "loss": 0.5559,
+      "step": 3520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5948074679113187e-06,
+      "loss": 0.519,
+      "step": 3524
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5918903150525087e-06,
+      "loss": 0.4054,
+      "step": 3528
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.588973162193699e-06,
+      "loss": 0.4334,
+      "step": 3532
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.586056009334889e-06,
+      "loss": 0.3535,
+      "step": 3536
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5831388564760793e-06,
+      "loss": 0.6168,
+      "step": 3540
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.58022170361727e-06,
+      "loss": 0.3484,
+      "step": 3544
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.57730455075846e-06,
+      "loss": 0.3373,
+      "step": 3548
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5743873978996504e-06,
+      "loss": 0.5002,
+      "step": 3552
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5714702450408404e-06,
+      "loss": 0.4713,
+      "step": 3556
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5685530921820307e-06,
+      "loss": 0.4236,
+      "step": 3560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5656359393232207e-06,
+      "loss": 0.49,
+      "step": 3564
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.562718786464411e-06,
+      "loss": 0.3107,
+      "step": 3568
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5598016336056014e-06,
+      "loss": 0.6111,
+      "step": 3572
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5568844807467913e-06,
+      "loss": 0.4386,
+      "step": 3576
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5539673278879817e-06,
+      "loss": 0.434,
+      "step": 3580
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5510501750291716e-06,
+      "loss": 0.3085,
+      "step": 3584
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.548133022170362e-06,
+      "loss": 0.6368,
+      "step": 3588
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.545215869311552e-06,
+      "loss": 0.3797,
+      "step": 3592
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5422987164527423e-06,
+      "loss": 0.5329,
+      "step": 3596
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5393815635939322e-06,
+      "loss": 0.5015,
+      "step": 3600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5364644107351226e-06,
+      "loss": 0.333,
+      "step": 3604
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5335472578763134e-06,
+      "loss": 0.349,
+      "step": 3608
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.530630105017503e-06,
+      "loss": 0.5365,
+      "step": 3612
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5277129521586937e-06,
+      "loss": 0.6777,
+      "step": 3616
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.524795799299883e-06,
+      "loss": 0.5432,
+      "step": 3620
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.521878646441074e-06,
+      "loss": 0.2763,
+      "step": 3624
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.518961493582264e-06,
+      "loss": 0.5183,
+      "step": 3628
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5160443407234543e-06,
+      "loss": 0.4486,
+      "step": 3632
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.513127187864644e-06,
+      "loss": 0.3521,
+      "step": 3636
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5102100350058346e-06,
+      "loss": 0.5112,
+      "step": 3640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.507292882147025e-06,
+      "loss": 0.378,
+      "step": 3644
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.504375729288215e-06,
+      "loss": 0.4282,
+      "step": 3648
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5014585764294052e-06,
+      "loss": 0.5283,
+      "step": 3652
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4985414235705956e-06,
+      "loss": 0.5269,
+      "step": 3656
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4956242707117855e-06,
+      "loss": 0.3568,
+      "step": 3660
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.492707117852976e-06,
+      "loss": 0.3911,
+      "step": 3664
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.489789964994166e-06,
+      "loss": 0.554,
+      "step": 3668
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.486872812135356e-06,
+      "loss": 0.6132,
+      "step": 3672
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.483955659276546e-06,
+      "loss": 0.5473,
+      "step": 3676
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4810385064177365e-06,
+      "loss": 0.472,
+      "step": 3680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.4781213535589264e-06,
+      "loss": 0.4616,
+      "step": 3684
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.4752042007001172e-06,
+      "loss": 0.5357,
+      "step": 3688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.472287047841307e-06,
+      "loss": 0.5629,
+      "step": 3692
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.4693698949824975e-06,
+      "loss": 0.5284,
+      "step": 3696
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.4664527421236875e-06,
+      "loss": 0.3988,
+      "step": 3700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.463535589264878e-06,
+      "loss": 0.7007,
+      "step": 3704
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.4606184364060678e-06,
+      "loss": 0.2642,
+      "step": 3708
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.457701283547258e-06,
+      "loss": 0.6179,
+      "step": 3712
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.454784130688448e-06,
+      "loss": 0.5595,
+      "step": 3716
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.4518669778296384e-06,
+      "loss": 0.4074,
+      "step": 3720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.448949824970829e-06,
+      "loss": 0.3635,
+      "step": 3724
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.446032672112019e-06,
+      "loss": 0.3888,
+      "step": 3728
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.443115519253209e-06,
+      "loss": 0.2755,
+      "step": 3732
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4401983663943995e-06,
+      "loss": 0.4511,
+      "step": 3736
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4372812135355894e-06,
+      "loss": 0.5516,
+      "step": 3740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4343640606767798e-06,
+      "loss": 0.4041,
+      "step": 3744
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.4314469078179697e-06,
+      "loss": 0.1592,
+      "step": 3748
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.42852975495916e-06,
+      "loss": 0.5583,
+      "step": 3752
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.4256126021003504e-06,
+      "loss": 0.445,
+      "step": 3756
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.4226954492415404e-06,
+      "loss": 0.3201,
+      "step": 3760
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.4197782963827307e-06,
+      "loss": 0.521,
+      "step": 3764
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.416861143523921e-06,
+      "loss": 0.4229,
+      "step": 3768
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.413943990665111e-06,
+      "loss": 0.4319,
+      "step": 3772
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.4110268378063014e-06,
+      "loss": 0.4708,
+      "step": 3776
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.4081096849474913e-06,
+      "loss": 0.5419,
+      "step": 3780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.4051925320886817e-06,
+      "loss": 0.4454,
+      "step": 3784
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.402275379229872e-06,
+      "loss": 0.4804,
+      "step": 3788
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.399358226371062e-06,
+      "loss": 0.5686,
+      "step": 3792
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3964410735122523e-06,
+      "loss": 0.5354,
+      "step": 3796
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3935239206534423e-06,
+      "loss": 0.5101,
+      "step": 3800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3906067677946326e-06,
+      "loss": 0.3946,
+      "step": 3804
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.387689614935823e-06,
+      "loss": 0.4951,
+      "step": 3808
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.384772462077013e-06,
+      "loss": 0.4952,
+      "step": 3812
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3818553092182033e-06,
+      "loss": 0.5854,
+      "step": 3816
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3789381563593932e-06,
+      "loss": 0.4074,
+      "step": 3820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.3760210035005836e-06,
+      "loss": 0.4058,
+      "step": 3824
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.373103850641774e-06,
+      "loss": 0.4952,
+      "step": 3828
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.370186697782964e-06,
+      "loss": 0.5502,
+      "step": 3832
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.3672695449241543e-06,
+      "loss": 0.4379,
+      "step": 3836
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.364352392065344e-06,
+      "loss": 0.4695,
+      "step": 3840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.3614352392065346e-06,
+      "loss": 0.5127,
+      "step": 3844
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.3585180863477245e-06,
+      "loss": 0.4037,
+      "step": 3848
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.355600933488915e-06,
+      "loss": 0.3387,
+      "step": 3852
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.3526837806301052e-06,
+      "loss": 0.5302,
+      "step": 3856
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.3497666277712956e-06,
+      "loss": 0.4549,
+      "step": 3860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.3468494749124855e-06,
+      "loss": 0.4871,
+      "step": 3864
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.343932322053676e-06,
+      "loss": 0.4167,
+      "step": 3868
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.341015169194866e-06,
+      "loss": 0.4498,
+      "step": 3872
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.338098016336056e-06,
+      "loss": 0.3591,
+      "step": 3876
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.335180863477246e-06,
+      "loss": 0.3389,
+      "step": 3880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3322637106184365e-06,
+      "loss": 0.4688,
+      "step": 3884
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.329346557759627e-06,
+      "loss": 0.4151,
+      "step": 3888
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.3264294049008172e-06,
+      "loss": 0.5132,
+      "step": 3892
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.323512252042007e-06,
+      "loss": 0.3682,
+      "step": 3896
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3205950991831975e-06,
+      "loss": 0.3668,
+      "step": 3900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3176779463243875e-06,
+      "loss": 0.2637,
+      "step": 3904
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.314760793465578e-06,
+      "loss": 0.5291,
+      "step": 3908
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.3118436406067678e-06,
+      "loss": 0.5459,
+      "step": 3912
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.308926487747958e-06,
+      "loss": 0.5774,
+      "step": 3916
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.306009334889148e-06,
+      "loss": 0.5955,
+      "step": 3920
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.303092182030339e-06,
+      "loss": 0.2941,
+      "step": 3924
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.300175029171529e-06,
+      "loss": 0.4735,
+      "step": 3928
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.297257876312719e-06,
+      "loss": 0.27,
+      "step": 3932
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.294340723453909e-06,
+      "loss": 0.496,
+      "step": 3936
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2914235705950995e-06,
+      "loss": 0.279,
+      "step": 3940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2885064177362894e-06,
+      "loss": 0.3848,
+      "step": 3944
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2855892648774798e-06,
+      "loss": 0.4868,
+      "step": 3948
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2826721120186697e-06,
+      "loss": 0.5328,
+      "step": 3952
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2797549591598605e-06,
+      "loss": 0.3838,
+      "step": 3956
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2768378063010504e-06,
+      "loss": 0.4603,
+      "step": 3960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.2739206534422408e-06,
+      "loss": 0.515,
+      "step": 3964
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.2710035005834307e-06,
+      "loss": 0.4371,
+      "step": 3968
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.268086347724621e-06,
+      "loss": 0.5561,
+      "step": 3972
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.265169194865811e-06,
+      "loss": 0.3882,
+      "step": 3976
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.2622520420070014e-06,
+      "loss": 0.4774,
+      "step": 3980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.2593348891481913e-06,
+      "loss": 0.6546,
+      "step": 3984
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2564177362893817e-06,
+      "loss": 0.4335,
+      "step": 3988
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.253500583430572e-06,
+      "loss": 0.3352,
+      "step": 3992
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2505834305717624e-06,
+      "loss": 0.2946,
+      "step": 3996
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2476662777129523e-06,
+      "loss": 0.2657,
+      "step": 4000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2447491248541427e-06,
+      "loss": 0.4859,
+      "step": 4004
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.2418319719953326e-06,
+      "loss": 0.3047,
+      "step": 4008
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.238914819136523e-06,
+      "loss": 0.5828,
+      "step": 4012
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.235997666277713e-06,
+      "loss": 0.4141,
+      "step": 4016
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.2330805134189033e-06,
+      "loss": 0.4643,
+      "step": 4020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.2301633605600937e-06,
+      "loss": 0.5517,
+      "step": 4024
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.2272462077012836e-06,
+      "loss": 0.551,
+      "step": 4028
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.224329054842474e-06,
+      "loss": 0.4237,
+      "step": 4032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.2214119019836643e-06,
+      "loss": 0.4345,
+      "step": 4036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.2184947491248543e-06,
+      "loss": 0.428,
+      "step": 4040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.2155775962660446e-06,
+      "loss": 0.3503,
+      "step": 4044
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.2126604434072346e-06,
+      "loss": 0.4281,
+      "step": 4048
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.209743290548425e-06,
+      "loss": 0.5451,
+      "step": 4052
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.2068261376896153e-06,
+      "loss": 0.6199,
+      "step": 4056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.2039089848308052e-06,
+      "loss": 0.6851,
+      "step": 4060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.2009918319719956e-06,
+      "loss": 0.3541,
+      "step": 4064
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1980746791131855e-06,
+      "loss": 0.4333,
+      "step": 4068
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.195157526254376e-06,
+      "loss": 0.413,
+      "step": 4072
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1922403733955663e-06,
+      "loss": 0.4385,
+      "step": 4076
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.189323220536756e-06,
+      "loss": 0.3163,
+      "step": 4080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1864060676779466e-06,
+      "loss": 0.4678,
+      "step": 4084
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.183488914819137e-06,
+      "loss": 0.2889,
+      "step": 4088
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.180571761960327e-06,
+      "loss": 0.3339,
+      "step": 4092
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1776546091015172e-06,
+      "loss": 0.4381,
+      "step": 4096
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.174737456242707e-06,
+      "loss": 0.4926,
+      "step": 4100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1718203033838975e-06,
+      "loss": 0.4147,
+      "step": 4104
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1689031505250875e-06,
+      "loss": 0.525,
+      "step": 4108
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.165985997666278e-06,
+      "loss": 0.6573,
+      "step": 4112
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.163068844807468e-06,
+      "loss": 0.4188,
+      "step": 4116
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1601516919486586e-06,
+      "loss": 0.2622,
+      "step": 4120
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.1572345390898485e-06,
+      "loss": 0.4533,
+      "step": 4124
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.154317386231039e-06,
+      "loss": 0.5007,
+      "step": 4128
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.151400233372229e-06,
+      "loss": 0.5307,
+      "step": 4132
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.148483080513419e-06,
+      "loss": 0.4071,
+      "step": 4136
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.145565927654609e-06,
+      "loss": 0.5252,
+      "step": 4140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1426487747957995e-06,
+      "loss": 0.3672,
+      "step": 4144
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.1397316219369894e-06,
+      "loss": 0.3608,
+      "step": 4148
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1368144690781798e-06,
+      "loss": 0.4581,
+      "step": 4152
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.13389731621937e-06,
+      "loss": 0.4953,
+      "step": 4156
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1309801633605605e-06,
+      "loss": 0.515,
+      "step": 4160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1280630105017504e-06,
+      "loss": 0.4272,
+      "step": 4164
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1251458576429408e-06,
+      "loss": 0.5713,
+      "step": 4168
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.1222287047841307e-06,
+      "loss": 0.3837,
+      "step": 4172
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.119311551925321e-06,
+      "loss": 0.4367,
+      "step": 4176
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.116394399066511e-06,
+      "loss": 0.3561,
+      "step": 4180
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1134772462077014e-06,
+      "loss": 0.2825,
+      "step": 4184
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.1105600933488917e-06,
+      "loss": 0.2891,
+      "step": 4188
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.107642940490082e-06,
+      "loss": 0.4617,
+      "step": 4192
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.104725787631272e-06,
+      "loss": 0.4423,
+      "step": 4196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.1018086347724624e-06,
+      "loss": 0.2344,
+      "step": 4200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0988914819136523e-06,
+      "loss": 0.5355,
+      "step": 4204
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0959743290548427e-06,
+      "loss": 0.427,
+      "step": 4208
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0930571761960326e-06,
+      "loss": 0.3997,
+      "step": 4212
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.090140023337223e-06,
+      "loss": 0.3945,
+      "step": 4216
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0872228704784134e-06,
+      "loss": 0.3998,
+      "step": 4220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0843057176196037e-06,
+      "loss": 0.4695,
+      "step": 4224
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0813885647607937e-06,
+      "loss": 0.3051,
+      "step": 4228
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.078471411901984e-06,
+      "loss": 0.4195,
+      "step": 4232
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.075554259043174e-06,
+      "loss": 0.4064,
+      "step": 4236
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0726371061843643e-06,
+      "loss": 0.4709,
+      "step": 4240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0697199533255543e-06,
+      "loss": 0.6638,
+      "step": 4244
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0668028004667446e-06,
+      "loss": 0.4436,
+      "step": 4248
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0638856476079346e-06,
+      "loss": 0.3434,
+      "step": 4252
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.060968494749125e-06,
+      "loss": 0.2871,
+      "step": 4256
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0580513418903153e-06,
+      "loss": 0.2696,
+      "step": 4260
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0551341890315057e-06,
+      "loss": 0.373,
+      "step": 4264
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0522170361726956e-06,
+      "loss": 0.4656,
+      "step": 4268
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.049299883313886e-06,
+      "loss": 0.3027,
+      "step": 4272
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.046382730455076e-06,
+      "loss": 0.7287,
+      "step": 4276
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0434655775962663e-06,
+      "loss": 0.3299,
+      "step": 4280
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.040548424737456e-06,
+      "loss": 0.3556,
+      "step": 4284
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0376312718786466e-06,
+      "loss": 0.3275,
+      "step": 4288
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.034714119019837e-06,
+      "loss": 0.334,
+      "step": 4292
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.031796966161027e-06,
+      "loss": 0.4437,
+      "step": 4296
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0288798133022172e-06,
+      "loss": 0.4776,
+      "step": 4300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0259626604434076e-06,
+      "loss": 0.497,
+      "step": 4304
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0230455075845975e-06,
+      "loss": 0.6054,
+      "step": 4308
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.020128354725788e-06,
+      "loss": 0.3877,
+      "step": 4312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.017211201866978e-06,
+      "loss": 0.3442,
+      "step": 4316
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.014294049008168e-06,
+      "loss": 0.5557,
+      "step": 4320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0113768961493586e-06,
+      "loss": 0.5137,
+      "step": 4324
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0084597432905485e-06,
+      "loss": 0.6162,
+      "step": 4328
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.005542590431739e-06,
+      "loss": 0.4181,
+      "step": 4332
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.002625437572929e-06,
+      "loss": 0.3354,
+      "step": 4336
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.999708284714119e-06,
+      "loss": 0.3924,
+      "step": 4340
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9967911318553095e-06,
+      "loss": 0.3781,
+      "step": 4344
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9938739789964995e-06,
+      "loss": 0.4388,
+      "step": 4348
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.99095682613769e-06,
+      "loss": 0.4512,
+      "step": 4352
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.98803967327888e-06,
+      "loss": 0.4171,
+      "step": 4356
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.98512252042007e-06,
+      "loss": 0.3768,
+      "step": 4360
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9822053675612605e-06,
+      "loss": 0.3517,
+      "step": 4364
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9792882147024504e-06,
+      "loss": 0.412,
+      "step": 4368
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9763710618436408e-06,
+      "loss": 0.2887,
+      "step": 4372
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9734539089848307e-06,
+      "loss": 0.3032,
+      "step": 4376
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.970536756126021e-06,
+      "loss": 0.3046,
+      "step": 4380
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9676196032672114e-06,
+      "loss": 0.3144,
+      "step": 4384
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.964702450408402e-06,
+      "loss": 0.533,
+      "step": 4388
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9617852975495917e-06,
+      "loss": 0.3993,
+      "step": 4392
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.958868144690782e-06,
+      "loss": 0.3253,
+      "step": 4396
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.955950991831972e-06,
+      "loss": 0.4888,
+      "step": 4400
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9530338389731624e-06,
+      "loss": 0.5357,
+      "step": 4404
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9501166861143524e-06,
+      "loss": 0.3057,
+      "step": 4408
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9471995332555427e-06,
+      "loss": 0.3553,
+      "step": 4412
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9442823803967327e-06,
+      "loss": 0.3878,
+      "step": 4416
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9413652275379234e-06,
+      "loss": 0.315,
+      "step": 4420
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9384480746791134e-06,
+      "loss": 0.3791,
+      "step": 4424
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9355309218203037e-06,
+      "loss": 0.4092,
+      "step": 4428
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9326137689614937e-06,
+      "loss": 0.3911,
+      "step": 4432
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.929696616102684e-06,
+      "loss": 0.3135,
+      "step": 4436
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.926779463243874e-06,
+      "loss": 0.3574,
+      "step": 4440
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9238623103850643e-06,
+      "loss": 0.3147,
+      "step": 4444
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9209451575262543e-06,
+      "loss": 0.5802,
+      "step": 4448
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9180280046674446e-06,
+      "loss": 0.5169,
+      "step": 4452
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.915110851808635e-06,
+      "loss": 0.4599,
+      "step": 4456
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9121936989498254e-06,
+      "loss": 0.3763,
+      "step": 4460
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9092765460910153e-06,
+      "loss": 0.4633,
+      "step": 4464
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9063593932322055e-06,
+      "loss": 0.462,
+      "step": 4468
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9034422403733956e-06,
+      "loss": 0.4908,
+      "step": 4472
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.900525087514586e-06,
+      "loss": 0.3367,
+      "step": 4476
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8976079346557761e-06,
+      "loss": 0.4497,
+      "step": 4480
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8946907817969663e-06,
+      "loss": 0.3253,
+      "step": 4484
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8917736289381566e-06,
+      "loss": 0.5464,
+      "step": 4488
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8888564760793468e-06,
+      "loss": 0.3004,
+      "step": 4492
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.885939323220537e-06,
+      "loss": 0.3876,
+      "step": 4496
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.883022170361727e-06,
+      "loss": 0.4443,
+      "step": 4500
+    }
+  ],
+  "logging_steps": 4,
+  "max_steps": 7056,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 34569317253120.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}