diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,20954 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 17429,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1474469305794606e-07,
+      "loss": 1.1705,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.737234652897304e-07,
+      "loss": 1.0623,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1474469305794607e-06,
+      "loss": 1.0881,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.721170395869191e-06,
+      "loss": 1.1118,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2948938611589215e-06,
+      "loss": 1.1156,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.868617326448652e-06,
+      "loss": 1.0686,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.442340791738382e-06,
+      "loss": 1.1194,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.016064257028113e-06,
+      "loss": 1.1242,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.589787722317843e-06,
+      "loss": 1.0856,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.163511187607573e-06,
+      "loss": 1.083,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.737234652897304e-06,
+      "loss": 0.9754,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.310958118187034e-06,
+      "loss": 1.1019,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.884681583476764e-06,
+      "loss": 1.0793,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4584050487664955e-06,
+      "loss": 1.1459,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.032128514056226e-06,
+      "loss": 1.0241,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.605851979345956e-06,
+      "loss": 1.0542,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.179575444635686e-06,
+      "loss": 1.0993,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.753298909925416e-06,
+      "loss": 1.0939,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0327022375215146e-05,
+      "loss": 1.0606,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0900745840504876e-05,
+      "loss": 1.0591,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1474469305794608e-05,
+      "loss": 1.0768,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2048192771084338e-05,
+      "loss": 1.0601,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2621916236374069e-05,
+      "loss": 1.0676,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3195639701663797e-05,
+      "loss": 1.0468,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3769363166953527e-05,
+      "loss": 1.0915,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.434308663224326e-05,
+      "loss": 1.0004,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4916810097532991e-05,
+      "loss": 1.0573,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.549053356282272e-05,
+      "loss": 1.0377,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.606425702811245e-05,
+      "loss": 1.0571,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.663798049340218e-05,
+      "loss": 1.0273,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.721170395869191e-05,
+      "loss": 0.9969,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7785427423981642e-05,
+      "loss": 0.9618,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8359150889271372e-05,
+      "loss": 1.0193,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8932874354561102e-05,
+      "loss": 0.9667,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9506597819850832e-05,
+      "loss": 0.9475,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0080321285140562e-05,
+      "loss": 1.0302,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0654044750430293e-05,
+      "loss": 1.0012,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1227768215720023e-05,
+      "loss": 0.9659,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1801491681009753e-05,
+      "loss": 1.0251,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2375215146299486e-05,
+      "loss": 0.9829,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2948938611589217e-05,
+      "loss": 0.9823,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3522662076878947e-05,
+      "loss": 0.9471,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4096385542168677e-05,
+      "loss": 0.9049,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4670109007458407e-05,
+      "loss": 0.9992,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5243832472748137e-05,
+      "loss": 0.9511,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5817555938037867e-05,
+      "loss": 0.9359,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6391279403327594e-05,
+      "loss": 0.9657,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6965002868617328e-05,
+      "loss": 1.0469,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7538726333907055e-05,
+      "loss": 0.9768,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8112449799196788e-05,
+      "loss": 1.0059,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.868617326448652e-05,
+      "loss": 0.982,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.925989672977625e-05,
+      "loss": 0.941,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9833620195065982e-05,
+      "loss": 0.9555,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.040734366035571e-05,
+      "loss": 1.0077,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.098106712564544e-05,
+      "loss": 1.0177,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.155479059093517e-05,
+      "loss": 1.0121,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.21285140562249e-05,
+      "loss": 0.9461,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2702237521514636e-05,
+      "loss": 0.9811,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.327596098680436e-05,
+      "loss": 0.9625,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3849684452094096e-05,
+      "loss": 0.9899,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.442340791738382e-05,
+      "loss": 0.9944,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.499713138267356e-05,
+      "loss": 1.0502,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5570854847963284e-05,
+      "loss": 0.9586,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.614457831325301e-05,
+      "loss": 0.9985,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6718301778542744e-05,
+      "loss": 1.0447,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.729202524383247e-05,
+      "loss": 1.0216,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7865748709122204e-05,
+      "loss": 1.0232,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.843947217441193e-05,
+      "loss": 0.9247,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9013195639701665e-05,
+      "loss": 0.9981,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.958691910499139e-05,
+      "loss": 0.9576,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0160642570281125e-05,
+      "loss": 0.9793,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.073436603557086e-05,
+      "loss": 0.9646,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1308089500860585e-05,
+      "loss": 0.9432,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.188181296615032e-05,
+      "loss": 1.0045,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2455536431440046e-05,
+      "loss": 0.9872,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.302925989672978e-05,
+      "loss": 0.9436,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3602983362019506e-05,
+      "loss": 1.0294,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.417670682730924e-05,
+      "loss": 0.9198,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.475043029259897e-05,
+      "loss": 0.9835,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.53241537578887e-05,
+      "loss": 0.9656,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.589787722317843e-05,
+      "loss": 0.9951,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.647160068846816e-05,
+      "loss": 0.9385,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7045324153757894e-05,
+      "loss": 0.9489,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.9516,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8192771084337354e-05,
+      "loss": 0.9873,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.876649454962709e-05,
+      "loss": 1.01,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9340218014916814e-05,
+      "loss": 0.9336,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991394148020654e-05,
+      "loss": 0.8962,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.0487664945496275e-05,
+      "loss": 0.9802,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1061388410786e-05,
+      "loss": 0.9845,
+      "step": 445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1635111876075735e-05,
+      "loss": 0.9281,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.220883534136547e-05,
+      "loss": 0.9978,
+      "step": 455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.278255880665519e-05,
+      "loss": 0.957,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.335628227194492e-05,
+      "loss": 0.8935,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.3930005737234656e-05,
+      "loss": 0.9418,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.450372920252439e-05,
+      "loss": 1.0086,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.507745266781411e-05,
+      "loss": 0.9691,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.565117613310384e-05,
+      "loss": 0.9682,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.6224899598393576e-05,
+      "loss": 0.9255,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.679862306368331e-05,
+      "loss": 1.0181,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.737234652897304e-05,
+      "loss": 0.9934,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.794606999426276e-05,
+      "loss": 0.9871,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.85197934595525e-05,
+      "loss": 0.9117,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.909351692484223e-05,
+      "loss": 1.0144,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9667240390131964e-05,
+      "loss": 0.9958,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.02409638554217e-05,
+      "loss": 0.9979,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.081468732071142e-05,
+      "loss": 0.9258,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.138841078600115e-05,
+      "loss": 0.9732,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.196213425129088e-05,
+      "loss": 0.9912,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.253585771658062e-05,
+      "loss": 0.906,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.310958118187034e-05,
+      "loss": 0.9822,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.368330464716007e-05,
+      "loss": 0.9325,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.42570281124498e-05,
+      "loss": 0.8918,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.483075157773954e-05,
+      "loss": 0.961,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.540447504302927e-05,
+      "loss": 0.9948,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.597819850831899e-05,
+      "loss": 0.9441,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.655192197360873e-05,
+      "loss": 0.9356,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.712564543889846e-05,
+      "loss": 1.0858,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.769936890418819e-05,
+      "loss": 0.9111,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.827309236947793e-05,
+      "loss": 0.9252,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.884681583476765e-05,
+      "loss": 1.0092,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.942053930005738e-05,
+      "loss": 0.9494,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.999426276534711e-05,
+      "loss": 0.939,
+      "step": 610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.056798623063683e-05,
+      "loss": 0.9906,
+      "step": 615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.114170969592657e-05,
+      "loss": 1.0128,
+      "step": 620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.17154331612163e-05,
+      "loss": 1.0403,
+      "step": 625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.228915662650602e-05,
+      "loss": 0.9104,
+      "step": 630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.286288009179575e-05,
+      "loss": 1.0033,
+      "step": 635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.343660355708549e-05,
+      "loss": 0.9468,
+      "step": 640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.401032702237521e-05,
+      "loss": 0.9881,
+      "step": 645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.458405048766494e-05,
+      "loss": 0.9598,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.515777395295467e-05,
+      "loss": 0.9252,
+      "step": 655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.573149741824441e-05,
+      "loss": 0.8959,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.630522088353414e-05,
+      "loss": 0.983,
+      "step": 665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.687894434882386e-05,
+      "loss": 1.0582,
+      "step": 670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.74526678141136e-05,
+      "loss": 0.9976,
+      "step": 675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.802639127940333e-05,
+      "loss": 1.0383,
+      "step": 680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.860011474469306e-05,
+      "loss": 0.9541,
+      "step": 685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.917383820998278e-05,
+      "loss": 1.0002,
+      "step": 690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.974756167527252e-05,
+      "loss": 0.998,
+      "step": 695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.032128514056225e-05,
+      "loss": 0.9839,
+      "step": 700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.089500860585198e-05,
+      "loss": 0.9201,
+      "step": 705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.146873207114172e-05,
+      "loss": 1.0262,
+      "step": 710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.204245553643144e-05,
+      "loss": 0.9774,
+      "step": 715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.261617900172117e-05,
+      "loss": 0.9659,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.31899024670109e-05,
+      "loss": 0.9591,
+      "step": 725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.376362593230064e-05,
+      "loss": 0.9,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.433734939759037e-05,
+      "loss": 1.0239,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.491107286288009e-05,
+      "loss": 1.0009,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.548479632816982e-05,
+      "loss": 0.9494,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.605851979345956e-05,
+      "loss": 0.9904,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.663224325874929e-05,
+      "loss": 1.0126,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.720596672403901e-05,
+      "loss": 1.0262,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.777969018932875e-05,
+      "loss": 1.0356,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.835341365461848e-05,
+      "loss": 0.957,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.892713711990821e-05,
+      "loss": 1.0135,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.950086058519795e-05,
+      "loss": 0.94,
+      "step": 780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.007458405048767e-05,
+      "loss": 0.9753,
+      "step": 785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.06483075157774e-05,
+      "loss": 0.9569,
+      "step": 790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.122203098106713e-05,
+      "loss": 0.9542,
+      "step": 795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.179575444635687e-05,
+      "loss": 0.9389,
+      "step": 800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.23694779116466e-05,
+      "loss": 0.9925,
+      "step": 805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.294320137693632e-05,
+      "loss": 0.9401,
+      "step": 810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.351692484222605e-05,
+      "loss": 1.0296,
+      "step": 815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.409064830751579e-05,
+      "loss": 0.9979,
+      "step": 820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.466437177280552e-05,
+      "loss": 0.9868,
+      "step": 825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.523809523809524e-05,
+      "loss": 0.9528,
+      "step": 830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.581181870338497e-05,
+      "loss": 0.9407,
+      "step": 835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.638554216867471e-05,
+      "loss": 0.93,
+      "step": 840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.695926563396444e-05,
+      "loss": 0.9299,
+      "step": 845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.753298909925417e-05,
+      "loss": 0.9063,
+      "step": 850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.81067125645439e-05,
+      "loss": 1.0166,
+      "step": 855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.868043602983363e-05,
+      "loss": 1.0016,
+      "step": 860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.925415949512336e-05,
+      "loss": 0.9345,
+      "step": 865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.982788296041308e-05,
+      "loss": 0.9582,
+      "step": 870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010040160642570282,
+      "loss": 1.0,
+      "step": 875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010097532989099255,
+      "loss": 0.9176,
+      "step": 880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010154905335628228,
+      "loss": 0.9909,
+      "step": 885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000102122776821572,
+      "loss": 0.9756,
+      "step": 890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010269650028686174,
+      "loss": 1.0205,
+      "step": 895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010327022375215147,
+      "loss": 1.0203,
+      "step": 900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010384394721744119,
+      "loss": 1.0004,
+      "step": 905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010441767068273094,
+      "loss": 0.9371,
+      "step": 910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010499139414802066,
+      "loss": 0.9558,
+      "step": 915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010556511761331038,
+      "loss": 1.0647,
+      "step": 920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010613884107860012,
+      "loss": 1.0336,
+      "step": 925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010671256454388984,
+      "loss": 0.96,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010728628800917956,
+      "loss": 0.9684,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010786001147446931,
+      "loss": 1.0326,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010843373493975903,
+      "loss": 1.0274,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00010900745840504878,
+      "loss": 1.0132,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001095811818703385,
+      "loss": 0.9982,
+      "step": 955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011015490533562822,
+      "loss": 0.9481,
+      "step": 960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011072862880091797,
+      "loss": 0.9013,
+      "step": 965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011130235226620769,
+      "loss": 0.9961,
+      "step": 970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011187607573149743,
+      "loss": 0.9674,
+      "step": 975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011244979919678715,
+      "loss": 0.9243,
+      "step": 980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011302352266207687,
+      "loss": 0.9551,
+      "step": 985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011359724612736662,
+      "loss": 1.0446,
+      "step": 990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011417096959265634,
+      "loss": 0.9859,
+      "step": 995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011474469305794609,
+      "loss": 0.9351,
+      "step": 1000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001153184165232358,
+      "loss": 0.9987,
+      "step": 1005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011589213998852553,
+      "loss": 1.0441,
+      "step": 1010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011646586345381527,
+      "loss": 0.9965,
+      "step": 1015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000117039586919105,
+      "loss": 1.0476,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011761331038439474,
+      "loss": 0.978,
+      "step": 1025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011818703384968446,
+      "loss": 0.9991,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011876075731497418,
+      "loss": 0.9742,
+      "step": 1035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011933448078026393,
+      "loss": 0.9523,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00011990820424555365,
+      "loss": 0.9462,
+      "step": 1045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001204819277108434,
+      "loss": 0.9774,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012105565117613311,
+      "loss": 0.9817,
+      "step": 1055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012162937464142283,
+      "loss": 0.9062,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012220309810671257,
+      "loss": 0.9564,
+      "step": 1065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001227768215720023,
+      "loss": 0.9818,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000123350545037292,
+      "loss": 0.9531,
+      "step": 1075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012392426850258177,
+      "loss": 0.9669,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012449799196787148,
+      "loss": 0.9073,
+      "step": 1085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012507171543316124,
+      "loss": 1.0106,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012564543889845094,
+      "loss": 0.951,
+      "step": 1095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012621916236374068,
+      "loss": 1.0042,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001267928858290304,
+      "loss": 1.0158,
+      "step": 1105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012736660929432014,
+      "loss": 1.0665,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012794033275960988,
+      "loss": 0.9769,
+      "step": 1115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001285140562248996,
+      "loss": 0.9897,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012908777969018932,
+      "loss": 1.0277,
+      "step": 1125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00012966150315547908,
+      "loss": 0.999,
+      "step": 1130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013023522662076878,
+      "loss": 0.9439,
+      "step": 1135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013080895008605854,
+      "loss": 1.0495,
+      "step": 1140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013138267355134825,
+      "loss": 0.9381,
+      "step": 1145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013195639701663798,
+      "loss": 0.9482,
+      "step": 1150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013253012048192772,
+      "loss": 0.9986,
+      "step": 1155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013310384394721745,
+      "loss": 1.0784,
+      "step": 1160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013367756741250719,
+      "loss": 0.961,
+      "step": 1165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013425129087779692,
+      "loss": 0.9253,
+      "step": 1170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013482501434308663,
+      "loss": 0.9941,
+      "step": 1175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013539873780837639,
+      "loss": 0.9823,
+      "step": 1180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001359724612736661,
+      "loss": 0.9685,
+      "step": 1185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013654618473895585,
+      "loss": 0.9845,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013711990820424556,
+      "loss": 0.9238,
+      "step": 1195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001376936316695353,
+      "loss": 0.971,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013826735513482503,
+      "loss": 0.9483,
+      "step": 1205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013884107860011476,
+      "loss": 0.933,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013941480206540447,
+      "loss": 1.0371,
+      "step": 1215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00013998852553069423,
+      "loss": 0.9498,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014056224899598393,
+      "loss": 0.9524,
+      "step": 1225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014113597246127367,
+      "loss": 0.9567,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001417096959265634,
+      "loss": 0.9739,
+      "step": 1235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014228341939185313,
+      "loss": 0.9551,
+      "step": 1240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014285714285714287,
+      "loss": 0.9942,
+      "step": 1245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001434308663224326,
+      "loss": 0.9211,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014400458978772233,
+      "loss": 1.0062,
+      "step": 1255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014457831325301204,
+      "loss": 0.9997,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014515203671830177,
+      "loss": 0.9647,
+      "step": 1265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001457257601835915,
+      "loss": 1.0235,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014629948364888124,
+      "loss": 1.0429,
+      "step": 1275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014687320711417098,
+      "loss": 0.9401,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001474469305794607,
+      "loss": 0.9661,
+      "step": 1285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014802065404475042,
+      "loss": 0.9528,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014859437751004018,
+      "loss": 0.9837,
+      "step": 1295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014916810097532988,
+      "loss": 1.0043,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014974182444061964,
+      "loss": 1.0775,
+      "step": 1305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015031554790590935,
+      "loss": 0.9867,
+      "step": 1310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015088927137119908,
+      "loss": 0.9846,
+      "step": 1315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015146299483648882,
+      "loss": 0.9796,
+      "step": 1320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015203671830177855,
+      "loss": 1.053,
+      "step": 1325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015261044176706828,
+      "loss": 0.944,
+      "step": 1330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015318416523235802,
+      "loss": 0.9963,
+      "step": 1335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015375788869764772,
+      "loss": 0.9996,
+      "step": 1340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015433161216293748,
+      "loss": 1.0571,
+      "step": 1345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001549053356282272,
+      "loss": 0.9715,
+      "step": 1350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015547905909351695,
+      "loss": 0.9664,
+      "step": 1355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015605278255880666,
+      "loss": 1.0288,
+      "step": 1360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001566265060240964,
+      "loss": 0.9653,
+      "step": 1365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015720022948938613,
+      "loss": 0.9467,
+      "step": 1370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015777395295467586,
+      "loss": 0.9892,
+      "step": 1375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015834767641996557,
+      "loss": 1.0086,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015892139988525533,
+      "loss": 1.0353,
+      "step": 1385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00015949512335054503,
+      "loss": 0.9849,
+      "step": 1390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001600688468158348,
+      "loss": 0.9907,
+      "step": 1395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001606425702811245,
+      "loss": 1.0342,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016121629374641423,
+      "loss": 1.0026,
+      "step": 1405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016179001721170397,
+      "loss": 0.9064,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001623637406769937,
+      "loss": 0.9871,
+      "step": 1415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016293746414228343,
+      "loss": 1.0059,
+      "step": 1420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016351118760757317,
+      "loss": 0.9875,
+      "step": 1425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016408491107286287,
+      "loss": 0.9565,
+      "step": 1430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016465863453815263,
+      "loss": 0.945,
+      "step": 1435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016523235800344234,
+      "loss": 0.9648,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001658060814687321,
+      "loss": 1.0576,
+      "step": 1445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001663798049340218,
+      "loss": 0.9933,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016695352839931154,
+      "loss": 1.0253,
+      "step": 1455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016752725186460127,
+      "loss": 0.9739,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000168100975329891,
+      "loss": 1.0493,
+      "step": 1465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016867469879518074,
+      "loss": 1.0177,
+      "step": 1470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016924842226047048,
+      "loss": 1.0153,
+      "step": 1475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00016982214572576018,
+      "loss": 0.9422,
+      "step": 1480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017039586919104992,
+      "loss": 1.0439,
+      "step": 1485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017096959265633965,
+      "loss": 0.9329,
+      "step": 1490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017154331612162938,
+      "loss": 0.9185,
+      "step": 1495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017211703958691912,
+      "loss": 0.9594,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017269076305220885,
+      "loss": 0.9732,
+      "step": 1505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017326448651749858,
+      "loss": 1.0135,
+      "step": 1510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001738382099827883,
+      "loss": 1.0635,
+      "step": 1515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017441193344807802,
+      "loss": 1.0554,
+      "step": 1520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017498565691336776,
+      "loss": 1.0567,
+      "step": 1525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001755593803786575,
+      "loss": 1.0248,
+      "step": 1530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017613310384394722,
+      "loss": 1.0039,
+      "step": 1535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017670682730923696,
+      "loss": 0.9948,
+      "step": 1540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017728055077452666,
+      "loss": 1.0919,
+      "step": 1545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017785427423981642,
+      "loss": 1.0281,
+      "step": 1550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00017842799770510613,
+      "loss": 0.9439,
+      "step": 1555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001790017211703959,
+      "loss": 1.1149,
+      "step": 1560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001795754446356856,
+      "loss": 1.0071,
+      "step": 1565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018014916810097533,
+      "loss": 1.1358,
+      "step": 1570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018072289156626507,
+      "loss": 1.389,
+      "step": 1575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001812966150315548,
+      "loss": 1.8714,
+      "step": 1580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018187033849684453,
+      "loss": 2.5047,
+      "step": 1585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018244406196213427,
+      "loss": 2.6965,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018301778542742397,
+      "loss": 3.029,
+      "step": 1595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018359150889271373,
+      "loss": 3.8631,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018416523235800344,
+      "loss": 3.9482,
+      "step": 1605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001847389558232932,
+      "loss": 4.1312,
+      "step": 1610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001853126792885829,
+      "loss": 3.8592,
+      "step": 1615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018588640275387264,
+      "loss": 3.7334,
+      "step": 1620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018646012621916237,
+      "loss": 3.9385,
+      "step": 1625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001870338496844521,
+      "loss": 3.7088,
+      "step": 1630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018760757314974184,
+      "loss": 3.9882,
+      "step": 1635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018818129661503157,
+      "loss": 3.6985,
+      "step": 1640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018875502008032128,
+      "loss": 3.9489,
+      "step": 1645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018932874354561104,
+      "loss": 3.7395,
+      "step": 1650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00018990246701090075,
+      "loss": 3.8121,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019047619047619048,
+      "loss": 3.6156,
+      "step": 1660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019104991394148021,
+      "loss": 3.6947,
+      "step": 1665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019162363740676995,
+      "loss": 3.7272,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019219736087205968,
+      "loss": 3.4795,
+      "step": 1675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019277108433734942,
+      "loss": 3.5508,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019334480780263912,
+      "loss": 3.4919,
+      "step": 1685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019391853126792888,
+      "loss": 3.4661,
+      "step": 1690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001944922547332186,
+      "loss": 3.7751,
+      "step": 1695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019506597819850835,
+      "loss": 3.5772,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019563970166379806,
+      "loss": 4.0097,
+      "step": 1705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001962134251290878,
+      "loss": 4.098,
+      "step": 1710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019678714859437752,
+      "loss": 3.6938,
+      "step": 1715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019736087205966726,
+      "loss": 4.0025,
+      "step": 1720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000197934595524957,
+      "loss": 3.8794,
+      "step": 1725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019850831899024672,
+      "loss": 4.074,
+      "step": 1730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019908204245553643,
+      "loss": 3.9138,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019965576592082616,
+      "loss": 3.8634,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999999919775815,
+      "loss": 4.0523,
+      "step": 1745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999990172538815,
+      "loss": 4.1861,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999971119306908,
+      "loss": 4.1269,
+      "step": 1755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999994203808154,
+      "loss": 4.0752,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999902928891875,
+      "loss": 4.4602,
+      "step": 1765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999853791777126,
+      "loss": 4.5566,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999794626786573,
+      "loss": 4.4079,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999725433979544,
+      "loss": 4.2193,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999646213425426,
+      "loss": 4.1227,
+      "step": 1785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999556965203663,
+      "loss": 4.1697,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999457689403753,
+      "loss": 4.1337,
+      "step": 1795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999934838612525,
+      "loss": 3.9798,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999922905547776,
+      "loss": 3.9761,
+      "step": 1805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999099697580954,
+      "loss": 3.7486,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998960312564548,
+      "loss": 3.9724,
+      "step": 1815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999881090056832,
+      "loss": 3.9185,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199986514617421,
+      "loss": 3.6141,
+      "step": 1825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019998481996245772,
+      "loss": 3.9031,
+      "step": 1830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998302504249278,
+      "loss": 3.765,
+      "step": 1835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998112985932613,
+      "loss": 3.6993,
+      "step": 1840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997913441485826,
+      "loss": 3.7082,
+      "step": 1845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997703871109021,
+      "loss": 3.9119,
+      "step": 1850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997484275012357,
+      "loss": 3.915,
+      "step": 1855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997254653416043,
+      "loss": 3.9133,
+      "step": 1860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997015006550342,
+      "loss": 3.8751,
+      "step": 1865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999676533465558,
+      "loss": 4.5563,
+      "step": 1870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996505637982122,
+      "loss": 4.1213,
+      "step": 1875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019996235916790392,
+      "loss": 3.7805,
+      "step": 1880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999595617135087,
+      "loss": 4.1168,
+      "step": 1885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019995666401944085,
+      "loss": 4.1894,
+      "step": 1890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999536660886062,
+      "loss": 3.9056,
+      "step": 1895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019995056792401105,
+      "loss": 3.721,
+      "step": 1900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019994736952876226,
+      "loss": 3.7291,
+      "step": 1905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999440709060672,
+      "loss": 3.9405,
+      "step": 1910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999406720592337,
+      "loss": 4.2803,
+      "step": 1915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019993717299167014,
+      "loss": 4.0277,
+      "step": 1920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019993357370688542,
+      "loss": 3.6311,
+      "step": 1925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019992987420848891,
+      "loss": 3.598,
+      "step": 1930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019992607450019048,
+      "loss": 4.0497,
+      "step": 1935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019992217458580043,
+      "loss": 4.165,
+      "step": 1940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019991817446922964,
+      "loss": 3.7754,
+      "step": 1945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019991407415448947,
+      "loss": 3.8028,
+      "step": 1950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999098736456917,
+      "loss": 3.8109,
+      "step": 1955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019990557294704856,
+      "loss": 3.9275,
+      "step": 1960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019990117206287287,
+      "loss": 3.8176,
+      "step": 1965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001998966709975778,
+      "loss": 3.817,
+      "step": 1970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019989206975567708,
+      "loss": 3.6848,
+      "step": 1975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001998873683417848,
+      "loss": 3.7611,
+      "step": 1980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019988256676061554,
+      "loss": 3.6979,
+      "step": 1985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019987766501698438,
+      "loss": 3.7871,
+      "step": 1990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019987266311580676,
+      "loss": 3.9897,
+      "step": 1995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019986756106209864,
+      "loss": 3.9645,
+      "step": 2000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998623588609763,
+      "loss": 3.8128,
+      "step": 2005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998570565176566,
+      "loss": 4.0431,
+      "step": 2010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998516540374567,
+      "loss": 4.1457,
+      "step": 2015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019984615142579424,
+      "loss": 3.8534,
+      "step": 2020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019984054868818724,
+      "loss": 3.7402,
+      "step": 2025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998348458302541,
+      "loss": 4.2085,
+      "step": 2030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001998290428577137,
+      "loss": 4.1647,
+      "step": 2035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019982313977638528,
+      "loss": 3.9169,
+      "step": 2040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019981713659218846,
+      "loss": 3.7777,
+      "step": 2045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019981103331114322,
+      "loss": 3.7297,
+      "step": 2050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019980482993936995,
+      "loss": 3.9071,
+      "step": 2055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019979852648308947,
+      "loss": 3.863,
+      "step": 2060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997921229486228,
+      "loss": 3.8292,
+      "step": 2065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019978561934239149,
+      "loss": 3.9396,
+      "step": 2070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997790156709173,
+      "loss": 3.9623,
+      "step": 2075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019977231194082248,
+      "loss": 3.9638,
+      "step": 2080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019976550815882952,
+      "loss": 4.0928,
+      "step": 2085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019975860433176128,
+      "loss": 4.1301,
+      "step": 2090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997516004665409,
+      "loss": 3.9469,
+      "step": 2095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997444965701919,
+      "loss": 3.8513,
+      "step": 2100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019973729264983808,
+      "loss": 3.9729,
+      "step": 2105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019972998871270353,
+      "loss": 3.9228,
+      "step": 2110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997225847661127,
+      "loss": 3.9859,
+      "step": 2115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019971508081749023,
+      "loss": 3.7867,
+      "step": 2120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001997074768743611,
+      "loss": 4.0779,
+      "step": 2125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019969977294435057,
+      "loss": 4.2388,
+      "step": 2130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996919690351842,
+      "loss": 3.9913,
+      "step": 2135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001996840651546877,
+      "loss": 3.9429,
+      "step": 2140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019967606131078718,
+      "loss": 3.7879,
+      "step": 2145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019966795751150885,
+      "loss": 3.6277,
+      "step": 2150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019965975376497918,
+      "loss": 3.7193,
+      "step": 2155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000199651450079425,
+      "loss": 4.1212,
+      "step": 2160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019964304646317323,
+      "loss": 4.3442,
+      "step": 2165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019963454292465103,
+      "loss": 4.1223,
+      "step": 2170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019962593947238578,
+      "loss": 3.9412,
+      "step": 2175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000199617236115005,
+      "loss": 4.104,
+      "step": 2180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019960843286123648,
+      "loss": 4.1006,
+      "step": 2185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995995297199081,
+      "loss": 4.1352,
+      "step": 2190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000199590526699948,
+      "loss": 4.1343,
+      "step": 2195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995814238103844,
+      "loss": 3.9857,
+      "step": 2200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019957222106034572,
+      "loss": 4.125,
+      "step": 2205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019956291845906046,
+      "loss": 4.1851,
+      "step": 2210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019955351601585731,
+      "loss": 3.9971,
+      "step": 2215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995440137401651,
+      "loss": 4.009,
+      "step": 2220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019953441164151264,
+      "loss": 3.8323,
+      "step": 2225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019952470972952902,
+      "loss": 3.6549,
+      "step": 2230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995149080139433,
+      "loss": 3.6998,
+      "step": 2235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995050065045847,
+      "loss": 3.9666,
+      "step": 2240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949500521138243,
+      "loss": 4.2325,
+      "step": 2245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948490414436584,
+      "loss": 4.2426,
+      "step": 2250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947470331366427,
+      "loss": 4.0235,
+      "step": 2255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019946440272950716,
+      "loss": 3.9708,
+      "step": 2260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019945400240222396,
+      "loss": 3.9884,
+      "step": 2265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019944350234224416,
+      "loss": 3.9914,
+      "step": 2270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994329025600972,
+      "loss": 3.9556,
+      "step": 2275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019942220306641258,
+      "loss": 3.8435,
+      "step": 2280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994114038719198,
+      "loss": 3.8276,
+      "step": 2285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019940050498744828,
+      "loss": 3.797,
+      "step": 2290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019938950642392746,
+      "loss": 3.8611,
+      "step": 2295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019937840819238677,
+      "loss": 3.9601,
+      "step": 2300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019936721030395547,
+      "loss": 4.0263,
+      "step": 2305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019935591276986286,
+      "loss": 4.1813,
+      "step": 2310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019934451560143815,
+      "loss": 4.0571,
+      "step": 2315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993330188101104,
+      "loss": 3.9657,
+      "step": 2320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019932142240740866,
+      "loss": 4.0322,
+      "step": 2325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001993097264049618,
+      "loss": 3.8831,
+      "step": 2330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001992979308144986,
+      "loss": 3.817,
+      "step": 2335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019928603564784773,
+      "loss": 3.8604,
+      "step": 2340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001992740409169377,
+      "loss": 3.9535,
+      "step": 2345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019926194663379677,
+      "loss": 4.0421,
+      "step": 2350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019924975281055324,
+      "loss": 4.0244,
+      "step": 2355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019923745945943502,
+      "loss": 3.9142,
+      "step": 2360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992250665927699,
+      "loss": 3.7527,
+      "step": 2365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001992125742229855,
+      "loss": 3.6311,
+      "step": 2370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019919998236260923,
+      "loss": 3.6715,
+      "step": 2375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019918729102426816,
+      "loss": 4.1127,
+      "step": 2380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019917450022068927,
+      "loss": 4.0694,
+      "step": 2385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019916160996469914,
+      "loss": 4.0454,
+      "step": 2390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991486202692242,
+      "loss": 3.9685,
+      "step": 2395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019913553114729053,
+      "loss": 4.0583,
+      "step": 2400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001991223426120239,
+      "loss": 3.898,
+      "step": 2405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019910905467664987,
+      "loss": 3.7952,
+      "step": 2410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019909566735449354,
+      "loss": 3.7981,
+      "step": 2415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019908218065897978,
+      "loss": 3.8415,
+      "step": 2420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019906859460363307,
+      "loss": 3.7588,
+      "step": 2425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019905490920207755,
+      "loss": 3.7436,
+      "step": 2430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019904112446803699,
+      "loss": 3.7062,
+      "step": 2435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001990272404153347,
+      "loss": 3.8589,
+      "step": 2440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019901325705789366,
+      "loss": 3.8012,
+      "step": 2445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989991744097364,
+      "loss": 3.8096,
+      "step": 2450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989849924849851,
+      "loss": 3.7199,
+      "step": 2455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019897071129786132,
+      "loss": 3.7549,
+      "step": 2460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019895633086268637,
+      "loss": 3.822,
+      "step": 2465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001989418511938809,
+      "loss": 3.808,
+      "step": 2470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019892727230596519,
+      "loss": 3.7525,
+      "step": 2475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019891259421355895,
+      "loss": 3.7883,
+      "step": 2480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988978169313815,
+      "loss": 3.7333,
+      "step": 2485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019888294047425143,
+      "loss": 3.727,
+      "step": 2490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019886796485708692,
+      "loss": 3.7949,
+      "step": 2495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019885289009490556,
+      "loss": 3.7852,
+      "step": 2500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001988377162028243,
+      "loss": 3.7526,
+      "step": 2505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019882244319605966,
+      "loss": 3.6963,
+      "step": 2510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019880707108992738,
+      "loss": 3.5769,
+      "step": 2515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001987915998998426,
+      "loss": 3.8725,
+      "step": 2520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019877602964131995,
+      "loss": 3.8018,
+      "step": 2525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987603603299733,
+      "loss": 3.9893,
+      "step": 2530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019874459198151583,
+      "loss": 3.9219,
+      "step": 2535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001987287246117601,
+      "loss": 3.8392,
+      "step": 2540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019871275823661795,
+      "loss": 3.8697,
+      "step": 2545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019869669287210046,
+      "loss": 3.9055,
+      "step": 2550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019868052853431808,
+      "loss": 3.8216,
+      "step": 2555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019866426523948037,
+      "loss": 3.7507,
+      "step": 2560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019864790300389625,
+      "loss": 3.8071,
+      "step": 2565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019863144184397376,
+      "loss": 3.7521,
+      "step": 2570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001986148817762203,
+      "loss": 3.6759,
+      "step": 2575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001985982228172422,
+      "loss": 3.7664,
+      "step": 2580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001985814649837452,
+      "loss": 3.835,
+      "step": 2585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000198564608292534,
+      "loss": 4.0195,
+      "step": 2590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019854765276051264,
+      "loss": 3.8679,
+      "step": 2595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019853059840468408,
+      "loss": 3.9831,
+      "step": 2600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001985134452421505,
+      "loss": 3.8295,
+      "step": 2605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019849619329011315,
+      "loss": 3.8679,
+      "step": 2610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001984788425658723,
+      "loss": 3.7208,
+      "step": 2615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019846139308682729,
+      "loss": 3.906,
+      "step": 2620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001984438448704765,
+      "loss": 3.8421,
+      "step": 2625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001984261979344173,
+      "loss": 3.8865,
+      "step": 2630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019840845229634612,
+      "loss": 3.8744,
+      "step": 2635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019839060797405833,
+      "loss": 3.6772,
+      "step": 2640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001983726649854482,
+      "loss": 3.5666,
+      "step": 2645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001983546233485091,
+      "loss": 3.625,
+      "step": 2650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001983364830813331,
+      "loss": 3.9214,
+      "step": 2655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019831824420211137,
+      "loss": 3.8779,
+      "step": 2660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019829990672913387,
+      "loss": 3.693,
+      "step": 2665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001982814706807895,
+      "loss": 3.5565,
+      "step": 2670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019826293607556593,
+      "loss": 3.5689,
+      "step": 2675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019824430293204973,
+      "loss": 3.8407,
+      "step": 2680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019822557126892627,
+      "loss": 3.9356,
+      "step": 2685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019820674110497966,
+      "loss": 3.947,
+      "step": 2690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019818781245909285,
+      "loss": 4.0446,
+      "step": 2695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019816878535024754,
+      "loss": 4.0271,
+      "step": 2700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019814965979752413,
+      "loss": 3.9889,
+      "step": 2705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001981304358201018,
+      "loss": 3.849,
+      "step": 2710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019811111343725842,
+      "loss": 3.8934,
+      "step": 2715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019809169266837043,
+      "loss": 3.6373,
+      "step": 2720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001980721735329131,
+      "loss": 3.8044,
+      "step": 2725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001980525560504602,
+      "loss": 3.9025,
+      "step": 2730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019803284024068427,
+      "loss": 3.9514,
+      "step": 2735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019801302612335628,
+      "loss": 4.0759,
+      "step": 2740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019799311371834595,
+      "loss": 4.0086,
+      "step": 2745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019797310304562143,
+      "loss": 3.9306,
+      "step": 2750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019795299412524945,
+      "loss": 3.7961,
+      "step": 2755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019793278697739533,
+      "loss": 3.8123,
+      "step": 2760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019791248162232285,
+      "loss": 3.8225,
+      "step": 2765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019789207808039425,
+      "loss": 3.8726,
+      "step": 2770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001978715763720702,
+      "loss": 3.9609,
+      "step": 2775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019785097651790992,
+      "loss": 4.164,
+      "step": 2780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019783027853857097,
+      "loss": 4.1816,
+      "step": 2785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019780948245480933,
+      "loss": 4.0187,
+      "step": 2790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019778858828747934,
+      "loss": 3.8095,
+      "step": 2795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019776759605753377,
+      "loss": 3.9314,
+      "step": 2800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001977465057860236,
+      "loss": 3.9116,
+      "step": 2805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019772531749409828,
+      "loss": 3.8993,
+      "step": 2810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001977040312030054,
+      "loss": 3.9434,
+      "step": 2815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019768264693409098,
+      "loss": 4.136,
+      "step": 2820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019766116470879913,
+      "loss": 4.0453,
+      "step": 2825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019763958454867235,
+      "loss": 3.9698,
+      "step": 2830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019761790647535124,
+      "loss": 3.9778,
+      "step": 2835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019759613051057462,
+      "loss": 3.8023,
+      "step": 2840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019757425667617945,
+      "loss": 3.8562,
+      "step": 2845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019755228499410092,
+      "loss": 3.7574,
+      "step": 2850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019753021548637222,
+      "loss": 3.9049,
+      "step": 2855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019750804817512477,
+      "loss": 4.0198,
+      "step": 2860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001974857830825879,
+      "loss": 4.0176,
+      "step": 2865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001974634202310892,
+      "loss": 3.9994,
+      "step": 2870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019744095964305413,
+      "loss": 3.8939,
+      "step": 2875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019741840134100623,
+      "loss": 3.9773,
+      "step": 2880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000197395745347567,
+      "loss": 3.8911,
+      "step": 2885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019737299168545597,
+      "loss": 3.9289,
+      "step": 2890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019735014037749053,
+      "loss": 3.8102,
+      "step": 2895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000197327191446586,
+      "loss": 3.7984,
+      "step": 2900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019730414491575564,
+      "loss": 4.0121,
+      "step": 2905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019728100080811057,
+      "loss": 3.9728,
+      "step": 2910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019725775914685977,
+      "loss": 3.9335,
+      "step": 2915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019723441995531,
+      "loss": 3.8701,
+      "step": 2920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019721098325686584,
+      "loss": 3.8718,
+      "step": 2925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001971874490750297,
+      "loss": 3.8089,
+      "step": 2930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001971638174334017,
+      "loss": 3.9738,
+      "step": 2935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001971400883556797,
+      "loss": 4.1796,
+      "step": 2940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019711626186565929,
+      "loss": 4.2039,
+      "step": 2945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001970923379872337,
+      "loss": 4.0764,
+      "step": 2950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019706831674439382,
+      "loss": 3.9673,
+      "step": 2955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019704419816122826,
+      "loss": 3.9997,
+      "step": 2960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019701998226192319,
+      "loss": 3.9886,
+      "step": 2965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019699566907076236,
+      "loss": 3.975,
+      "step": 2970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019697125861212707,
+      "loss": 3.8967,
+      "step": 2975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019694675091049617,
+      "loss": 3.9401,
+      "step": 2980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001969221459904461,
+      "loss": 3.8775,
+      "step": 2985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001968974438766507,
+      "loss": 4.0269,
+      "step": 2990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001968726445938813,
+      "loss": 4.1525,
+      "step": 2995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019684774816700664,
+      "loss": 4.1041,
+      "step": 3000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019682275462099298,
+      "loss": 3.9867,
+      "step": 3005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019679766398090383,
+      "loss": 3.9841,
+      "step": 3010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019677247627190026,
+      "loss": 4.1187,
+      "step": 3015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019674719151924043,
+      "loss": 4.0399,
+      "step": 3020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019672180974828,
+      "loss": 4.1619,
+      "step": 3025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001966963309844719,
+      "loss": 4.0161,
+      "step": 3030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019667075525336622,
+      "loss": 4.1612,
+      "step": 3035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019664508258061044,
+      "loss": 4.1662,
+      "step": 3040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001966193129919491,
+      "loss": 4.1021,
+      "step": 3045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019659344651322405,
+      "loss": 4.0468,
+      "step": 3050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019656748317037424,
+      "loss": 4.0429,
+      "step": 3055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019654142298943574,
+      "loss": 4.1886,
+      "step": 3060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019651526599654182,
+      "loss": 4.1133,
+      "step": 3065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001964890122179227,
+      "loss": 3.9713,
+      "step": 3070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019646266167990578,
+      "loss": 3.9225,
+      "step": 3075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019643621440891543,
+      "loss": 3.8559,
+      "step": 3080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019640967043147302,
+      "loss": 3.9145,
+      "step": 3085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001963830297741969,
+      "loss": 4.2188,
+      "step": 3090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001963562924638024,
+      "loss": 4.0723,
+      "step": 3095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019632945852710173,
+      "loss": 4.0658,
+      "step": 3100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019630252799100409,
+      "loss": 4.0304,
+      "step": 3105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001962755008825154,
+      "loss": 3.9481,
+      "step": 3110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019624837722873856,
+      "loss": 3.9624,
+      "step": 3115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019622115705687318,
+      "loss": 3.8568,
+      "step": 3120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019619384039421575,
+      "loss": 3.9446,
+      "step": 3125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019616642726815947,
+      "loss": 3.8899,
+      "step": 3130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001961389177061943,
+      "loss": 3.9193,
+      "step": 3135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019611131173590687,
+      "loss": 4.005,
+      "step": 3140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001960836093849805,
+      "loss": 3.972,
+      "step": 3145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019605581068119518,
+      "loss": 3.9895,
+      "step": 3150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019602791565242754,
+      "loss": 3.8762,
+      "step": 3155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019599992432665073,
+      "loss": 3.8782,
+      "step": 3160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019597183673193452,
+      "loss": 3.7442,
+      "step": 3165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019594365289644529,
+      "loss": 3.87,
+      "step": 3170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019591537284844573,
+      "loss": 3.7133,
+      "step": 3175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019588699661629523,
+      "loss": 3.6664,
+      "step": 3180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001958585242284495,
+      "loss": 3.6013,
+      "step": 3185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019582995571346072,
+      "loss": 3.4943,
+      "step": 3190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001958012910999775,
+      "loss": 3.4814,
+      "step": 3195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001957725304167447,
+      "loss": 3.5685,
+      "step": 3200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019574367369260364,
+      "loss": 3.8341,
+      "step": 3205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019571472095649192,
+      "loss": 3.8938,
+      "step": 3210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019568567223744339,
+      "loss": 3.8136,
+      "step": 3215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019565652756458818,
+      "loss": 3.6728,
+      "step": 3220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019562728696715263,
+      "loss": 3.7193,
+      "step": 3225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019559795047445927,
+      "loss": 3.7583,
+      "step": 3230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001955685181159268,
+      "loss": 3.7125,
+      "step": 3235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001955389899210701,
+      "loss": 3.7783,
+      "step": 3240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019550936591950006,
+      "loss": 3.704,
+      "step": 3245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001954796461409237,
+      "loss": 3.8314,
+      "step": 3250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001954498306151441,
+      "loss": 3.8665,
+      "step": 3255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001954199193720603,
+      "loss": 4.0657,
+      "step": 3260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019538991244166738,
+      "loss": 3.8345,
+      "step": 3265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019535980985405639,
+      "loss": 3.7016,
+      "step": 3270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019532961163941422,
+      "loss": 3.8221,
+      "step": 3275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019529931782802376,
+      "loss": 3.9244,
+      "step": 3280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019526892845026365,
+      "loss": 3.9155,
+      "step": 3285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019523844353660849,
+      "loss": 3.9713,
+      "step": 3290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001952078631176286,
+      "loss": 3.784,
+      "step": 3295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019517718722399002,
+      "loss": 4.0003,
+      "step": 3300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019514641588645471,
+      "loss": 4.0561,
+      "step": 3305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001951155491358802,
+      "loss": 4.0978,
+      "step": 3310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019508458700321973,
+      "loss": 4.0968,
+      "step": 3315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019505352951952221,
+      "loss": 3.9513,
+      "step": 3320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019502237671593212,
+      "loss": 3.8549,
+      "step": 3325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001949911286236896,
+      "loss": 3.8865,
+      "step": 3330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019495978527413028,
+      "loss": 4.0474,
+      "step": 3335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019492834669868536,
+      "loss": 4.0087,
+      "step": 3340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019489681292888148,
+      "loss": 4.0238,
+      "step": 3345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019486518399634083,
+      "loss": 3.9486,
+      "step": 3350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019483345993278093,
+      "loss": 3.8412,
+      "step": 3355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019480164077001475,
+      "loss": 3.9113,
+      "step": 3360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019476972653995062,
+      "loss": 3.7728,
+      "step": 3365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019473771727459224,
+      "loss": 3.9642,
+      "step": 3370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019470561300603852,
+      "loss": 3.9977,
+      "step": 3375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019467341376648372,
+      "loss": 3.8893,
+      "step": 3380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019464111958821727,
+      "loss": 3.8033,
+      "step": 3385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001946087305036239,
+      "loss": 3.6589,
+      "step": 3390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019457624654518343,
+      "loss": 3.5628,
+      "step": 3395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019454366774547083,
+      "loss": 3.5085,
+      "step": 3400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019451099413715626,
+      "loss": 3.5625,
+      "step": 3405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001944782257530048,
+      "loss": 3.747,
+      "step": 3410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019444536262587669,
+      "loss": 3.8478,
+      "step": 3415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019441240478872718,
+      "loss": 3.8523,
+      "step": 3420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001943793522746064,
+      "loss": 3.8604,
+      "step": 3425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019434620511665958,
+      "loss": 3.9138,
+      "step": 3430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001943129633481267,
+      "loss": 3.7584,
+      "step": 3435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019427962700234268,
+      "loss": 3.8157,
+      "step": 3440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019424619611273727,
+      "loss": 3.9208,
+      "step": 3445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001942126707128351,
+      "loss": 3.8077,
+      "step": 3450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019417905083625545,
+      "loss": 3.9654,
+      "step": 3455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019414533651671242,
+      "loss": 3.8997,
+      "step": 3460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019411152778801486,
+      "loss": 3.7784,
+      "step": 3465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019407762468406619,
+      "loss": 3.7086,
+      "step": 3470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019404362723886452,
+      "loss": 3.7156,
+      "step": 3475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019400953548650258,
+      "loss": 3.7288,
+      "step": 3480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019397534946116762,
+      "loss": 3.9152,
+      "step": 3485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019394106919714155,
+      "loss": 4.0103,
+      "step": 3490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001939066947288006,
+      "loss": 3.9918,
+      "step": 3495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001938722260906156,
+      "loss": 3.8941,
+      "step": 3500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019383766331715178,
+      "loss": 3.8285,
+      "step": 3505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019380300644306878,
+      "loss": 3.7736,
+      "step": 3510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019376825550312057,
+      "loss": 3.7274,
+      "step": 3515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019373341053215547,
+      "loss": 3.7859,
+      "step": 3520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001936984715651161,
+      "loss": 3.7412,
+      "step": 3525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019366343863703932,
+      "loss": 3.6731,
+      "step": 3530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001936283117830563,
+      "loss": 3.7049,
+      "step": 3535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019359309103839225,
+      "loss": 3.7449,
+      "step": 3540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001935577764383666,
+      "loss": 3.6854,
+      "step": 3545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019352236801839298,
+      "loss": 3.7409,
+      "step": 3550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000193486865813979,
+      "loss": 3.7686,
+      "step": 3555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019345126986072635,
+      "loss": 3.6891,
+      "step": 3560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001934155801943307,
+      "loss": 3.7046,
+      "step": 3565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001933797968505818,
+      "loss": 3.6589,
+      "step": 3570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001933439198653632,
+      "loss": 3.7049,
+      "step": 3575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019330794927465247,
+      "loss": 3.5971,
+      "step": 3580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019327188511452094,
+      "loss": 3.6787,
+      "step": 3585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019323572742113387,
+      "loss": 3.9087,
+      "step": 3590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001931994762307503,
+      "loss": 3.7466,
+      "step": 3595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019316313157972297,
+      "loss": 3.7719,
+      "step": 3600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019312669350449836,
+      "loss": 3.7041,
+      "step": 3605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001930901620416167,
+      "loss": 3.6416,
+      "step": 3610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001930535372277118,
+      "loss": 3.7735,
+      "step": 3615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001930168190995111,
+      "loss": 3.6501,
+      "step": 3620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019298000769383565,
+      "loss": 3.5963,
+      "step": 3625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019294310304759994,
+      "loss": 3.6151,
+      "step": 3630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019290610519781212,
+      "loss": 3.5001,
+      "step": 3635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019286901418157367,
+      "loss": 3.622,
+      "step": 3640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019283183003607955,
+      "loss": 3.4281,
+      "step": 3645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001927945527986181,
+      "loss": 3.5798,
+      "step": 3650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000192757182506571,
+      "loss": 3.6014,
+      "step": 3655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019271971919741332,
+      "loss": 3.8189,
+      "step": 3660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001926821629087133,
+      "loss": 3.6664,
+      "step": 3665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001926445136781325,
+      "loss": 3.7436,
+      "step": 3670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019260677154342564,
+      "loss": 3.7009,
+      "step": 3675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001925689365424406,
+      "loss": 3.7745,
+      "step": 3680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019253100871311843,
+      "loss": 3.637,
+      "step": 3685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019249298809349323,
+      "loss": 3.649,
+      "step": 3690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019245487472169216,
+      "loss": 3.6066,
+      "step": 3695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001924166686359354,
+      "loss": 3.6039,
+      "step": 3700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019237836987453613,
+      "loss": 3.484,
+      "step": 3705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019233997847590035,
+      "loss": 3.7251,
+      "step": 3710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019230149447852714,
+      "loss": 3.6939,
+      "step": 3715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019226291792100826,
+      "loss": 3.7109,
+      "step": 3720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001922242488420284,
+      "loss": 3.687,
+      "step": 3725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019218548728036503,
+      "loss": 3.6643,
+      "step": 3730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019214663327488828,
+      "loss": 3.8078,
+      "step": 3735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019210768686456106,
+      "loss": 3.6718,
+      "step": 3740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019206864808843892,
+      "loss": 3.748,
+      "step": 3745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019202951698566999,
+      "loss": 3.6618,
+      "step": 3750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001919902935954951,
+      "loss": 3.6457,
+      "step": 3755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019195097795724747,
+      "loss": 3.9062,
+      "step": 3760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000191911570110353,
+      "loss": 3.8633,
+      "step": 3765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001918720700943299,
+      "loss": 3.8972,
+      "step": 3770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001918324779487889,
+      "loss": 3.869,
+      "step": 3775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001917927937134331,
+      "loss": 3.8311,
+      "step": 3780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019175301742805793,
+      "loss": 3.6895,
+      "step": 3785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019171314913255113,
+      "loss": 3.7514,
+      "step": 3790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019167318886689273,
+      "loss": 3.7494,
+      "step": 3795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019163313667115497,
+      "loss": 3.7804,
+      "step": 3800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019159299258550227,
+      "loss": 3.7613,
+      "step": 3805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001915527566501912,
+      "loss": 3.7294,
+      "step": 3810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001915124289055705,
+      "loss": 3.7854,
+      "step": 3815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019147200939208088,
+      "loss": 3.7894,
+      "step": 3820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001914314981502551,
+      "loss": 3.8074,
+      "step": 3825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000191390895220718,
+      "loss": 3.8708,
+      "step": 3830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001913502006441862,
+      "loss": 3.7719,
+      "step": 3835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019130941446146837,
+      "loss": 3.7068,
+      "step": 3840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019126853671346496,
+      "loss": 3.609,
+      "step": 3845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019122756744116828,
+      "loss": 3.4527,
+      "step": 3850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001911865066856624,
+      "loss": 3.6164,
+      "step": 3855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019114535448812311,
+      "loss": 3.7123,
+      "step": 3860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019110411088981802,
+      "loss": 3.7241,
+      "step": 3865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001910627759321062,
+      "loss": 3.7562,
+      "step": 3870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019102134965643847,
+      "loss": 3.726,
+      "step": 3875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001909798321043572,
+      "loss": 3.7099,
+      "step": 3880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019093822331749634,
+      "loss": 3.7649,
+      "step": 3885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019089652333758114,
+      "loss": 3.7754,
+      "step": 3890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019085473220642856,
+      "loss": 3.7369,
+      "step": 3895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001908128499659468,
+      "loss": 3.6229,
+      "step": 3900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019077087665813545,
+      "loss": 3.6502,
+      "step": 3905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001907288123250854,
+      "loss": 3.6347,
+      "step": 3910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019068665700897896,
+      "loss": 3.7046,
+      "step": 3915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001906444107520895,
+      "loss": 3.6719,
+      "step": 3920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019060207359678164,
+      "loss": 3.6832,
+      "step": 3925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019055964558551124,
+      "loss": 3.7177,
+      "step": 3930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019051712676082522,
+      "loss": 3.741,
+      "step": 3935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019047451716536147,
+      "loss": 3.65,
+      "step": 3940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000190431816841849,
+      "loss": 3.7518,
+      "step": 3945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019038902583310785,
+      "loss": 3.7061,
+      "step": 3950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019034614418204893,
+      "loss": 3.8555,
+      "step": 3955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000190303171931674,
+      "loss": 3.9051,
+      "step": 3960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019026010912507577,
+      "loss": 3.7699,
+      "step": 3965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019021695580543772,
+      "loss": 3.584,
+      "step": 3970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019017371201603407,
+      "loss": 3.6451,
+      "step": 3975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019013037780022982,
+      "loss": 3.5583,
+      "step": 3980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019008695320148062,
+      "loss": 3.5669,
+      "step": 3985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001900434382633327,
+      "loss": 3.3989,
+      "step": 3990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018999983302942302,
+      "loss": 3.4922,
+      "step": 3995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018995613754347893,
+      "loss": 3.4463,
+      "step": 4000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018991235184931843,
+      "loss": 3.4361,
+      "step": 4005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018986847599084986,
+      "loss": 3.5724,
+      "step": 4010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018982451001207205,
+      "loss": 3.5376,
+      "step": 4015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018978045395707418,
+      "loss": 3.6229,
+      "step": 4020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018973630787003575,
+      "loss": 3.6468,
+      "step": 4025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001896920717952266,
+      "loss": 3.4721,
+      "step": 4030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018964774577700667,
+      "loss": 3.5094,
+      "step": 4035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018960332985982627,
+      "loss": 3.5052,
+      "step": 4040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001895588240882258,
+      "loss": 3.5885,
+      "step": 4045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001895142285068357,
+      "loss": 3.5499,
+      "step": 4050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018946954316037648,
+      "loss": 3.5257,
+      "step": 4055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001894247680936588,
+      "loss": 3.6446,
+      "step": 4060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018937990335158312,
+      "loss": 3.6611,
+      "step": 4065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018933494897913997,
+      "loss": 3.7228,
+      "step": 4070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018928990502140963,
+      "loss": 3.6856,
+      "step": 4075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018924477152356233,
+      "loss": 3.6701,
+      "step": 4080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018919954853085803,
+      "loss": 3.6045,
+      "step": 4085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001891542360886464,
+      "loss": 3.5751,
+      "step": 4090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018910883424236695,
+      "loss": 3.3857,
+      "step": 4095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001890633430375487,
+      "loss": 3.498,
+      "step": 4100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018901776251981032,
+      "loss": 3.4611,
+      "step": 4105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001889720927348601,
+      "loss": 3.6429,
+      "step": 4110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018892633372849575,
+      "loss": 3.6317,
+      "step": 4115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018888048554660454,
+      "loss": 3.6851,
+      "step": 4120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018883454823516313,
+      "loss": 3.5615,
+      "step": 4125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001887885218402375,
+      "loss": 3.5977,
+      "step": 4130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018874240640798316,
+      "loss": 3.5784,
+      "step": 4135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001886962019846446,
+      "loss": 3.6911,
+      "step": 4140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018864990861655584,
+      "loss": 3.6698,
+      "step": 4145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001886035263501399,
+      "loss": 3.6395,
+      "step": 4150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018855705523190908,
+      "loss": 3.6592,
+      "step": 4155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001885104953084647,
+      "loss": 3.5498,
+      "step": 4160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018846384662649714,
+      "loss": 3.6721,
+      "step": 4165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018841710923278582,
+      "loss": 3.5488,
+      "step": 4170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018837028317419908,
+      "loss": 3.539,
+      "step": 4175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018832336849769425,
+      "loss": 3.6718,
+      "step": 4180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001882763652503174,
+      "loss": 3.6152,
+      "step": 4185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018822927347920355,
+      "loss": 3.6229,
+      "step": 4190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018818209323157638,
+      "loss": 3.5397,
+      "step": 4195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001881348245547484,
+      "loss": 3.518,
+      "step": 4200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018808746749612073,
+      "loss": 3.6075,
+      "step": 4205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001880400221031831,
+      "loss": 3.597,
+      "step": 4210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018799248842351393,
+      "loss": 3.56,
+      "step": 4215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018794486650478004,
+      "loss": 3.5637,
+      "step": 4220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001878971563947368,
+      "loss": 3.5983,
+      "step": 4225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018784935814122804,
+      "loss": 3.6155,
+      "step": 4230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001878014717921859,
+      "loss": 3.6629,
+      "step": 4235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000187753497395631,
+      "loss": 3.6388,
+      "step": 4240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018770543499967217,
+      "loss": 3.6751,
+      "step": 4245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018765728465250644,
+      "loss": 3.715,
+      "step": 4250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018760904640241906,
+      "loss": 3.6934,
+      "step": 4255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018756072029778352,
+      "loss": 3.6022,
+      "step": 4260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018751230638706131,
+      "loss": 3.6856,
+      "step": 4265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00018746380471880203,
+      "loss": 3.5337,
+      "step": 4270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018741521534164325,
+      "loss": 3.6901,
+      "step": 4275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018736653830431048,
+      "loss": 3.6717,
+      "step": 4280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001873177736556172,
+      "loss": 3.6712,
+      "step": 4285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018726892144446474,
+      "loss": 3.5113,
+      "step": 4290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001872199817198421,
+      "loss": 3.6584,
+      "step": 4295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018717095453082627,
+      "loss": 3.7496,
+      "step": 4300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018712183992658174,
+      "loss": 3.751,
+      "step": 4305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018707263795636077,
+      "loss": 3.8571,
+      "step": 4310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001870233486695032,
+      "loss": 3.8003,
+      "step": 4315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001869739721154365,
+      "loss": 3.7638,
+      "step": 4320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018692450834367546,
+      "loss": 3.5772,
+      "step": 4325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018687495740382258,
+      "loss": 3.5074,
+      "step": 4330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018682531934556757,
+      "loss": 3.4393,
+      "step": 4335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018677559421868766,
+      "loss": 3.5662,
+      "step": 4340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018672578207304727,
+      "loss": 3.5174,
+      "step": 4345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018667588295859816,
+      "loss": 3.5552,
+      "step": 4350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001866258969253792,
+      "loss": 3.6251,
+      "step": 4355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018657582402351663,
+      "loss": 3.7857,
+      "step": 4360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018652566430322356,
+      "loss": 3.6927,
+      "step": 4365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001864754178148003,
+      "loss": 3.7058,
+      "step": 4370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018642508460863416,
+      "loss": 3.5257,
+      "step": 4375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018637466473519937,
+      "loss": 3.5633,
+      "step": 4380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001863241582450571,
+      "loss": 3.4822,
+      "step": 4385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018627356518885536,
+      "loss": 3.4217,
+      "step": 4390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000186222885617329,
+      "loss": 3.4629,
+      "step": 4395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018617211958129958,
+      "loss": 3.5844,
+      "step": 4400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018612126713167542,
+      "loss": 3.6745,
+      "step": 4405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001860703283194515,
+      "loss": 3.7324,
+      "step": 4410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018601930319570934,
+      "loss": 3.7076,
+      "step": 4415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000185968191811617,
+      "loss": 3.6037,
+      "step": 4420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018591699421842916,
+      "loss": 3.5516,
+      "step": 4425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018586571046748685,
+      "loss": 3.4967,
+      "step": 4430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018581434061021754,
+      "loss": 3.6028,
+      "step": 4435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018576288469813505,
+      "loss": 3.358,
+      "step": 4440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018571134278283946,
+      "loss": 3.3776,
+      "step": 4445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001856597149160171,
+      "loss": 3.4153,
+      "step": 4450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018560800114944063,
+      "loss": 3.5318,
+      "step": 4455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018555620153496856,
+      "loss": 3.5411,
+      "step": 4460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018550431612454578,
+      "loss": 3.6165,
+      "step": 4465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018545234497020302,
+      "loss": 3.7134,
+      "step": 4470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018540028812405717,
+      "loss": 3.6751,
+      "step": 4475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018534814563831082,
+      "loss": 3.6107,
+      "step": 4480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018529591756525268,
+      "loss": 3.5181,
+      "step": 4485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018524360395725715,
+      "loss": 3.4012,
+      "step": 4490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001851912048667844,
+      "loss": 3.3784,
+      "step": 4495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018513872034638037,
+      "loss": 3.3335,
+      "step": 4500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018508615044867668,
+      "loss": 3.4279,
+      "step": 4505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018503349522639052,
+      "loss": 3.5395,
+      "step": 4510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018498075473232469,
+      "loss": 3.4997,
+      "step": 4515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018492792901936742,
+      "loss": 3.4724,
+      "step": 4520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001848750181404925,
+      "loss": 3.4462,
+      "step": 4525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018482202214875908,
+      "loss": 3.5919,
+      "step": 4530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018476894109731166,
+      "loss": 3.6629,
+      "step": 4535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018471577503938,
+      "loss": 3.6637,
+      "step": 4540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018466252402827915,
+      "loss": 3.7109,
+      "step": 4545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018460918811740937,
+      "loss": 3.6368,
+      "step": 4550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018455576736025602,
+      "loss": 3.6998,
+      "step": 4555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018450226181038955,
+      "loss": 3.7358,
+      "step": 4560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018444867152146545,
+      "loss": 3.6636,
+      "step": 4565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018439499654722421,
+      "loss": 3.6544,
+      "step": 4570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018434123694149117,
+      "loss": 3.6887,
+      "step": 4575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001842873927581766,
+      "loss": 3.659,
+      "step": 4580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018423346405127555,
+      "loss": 3.6723,
+      "step": 4585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001841794508748679,
+      "loss": 3.6484,
+      "step": 4590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018412535328311814,
+      "loss": 3.6946,
+      "step": 4595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018407117133027544,
+      "loss": 3.5223,
+      "step": 4600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001840169050706736,
+      "loss": 3.6291,
+      "step": 4605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018396255455873102,
+      "loss": 3.4695,
+      "step": 4610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001839081198489504,
+      "loss": 3.4071,
+      "step": 4615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001838536009959191,
+      "loss": 3.5346,
+      "step": 4620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018379899805430862,
+      "loss": 3.5617,
+      "step": 4625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018374431107887502,
+      "loss": 3.5673,
+      "step": 4630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018368954012445846,
+      "loss": 3.5891,
+      "step": 4635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018363468524598342,
+      "loss": 3.5041,
+      "step": 4640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001835797464984585,
+      "loss": 3.4957,
+      "step": 4645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018352472393697632,
+      "loss": 3.4831,
+      "step": 4650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001834696176167137,
+      "loss": 3.4898,
+      "step": 4655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018341442759293137,
+      "loss": 3.6039,
+      "step": 4660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018335915392097402,
+      "loss": 3.5844,
+      "step": 4665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018330379665627014,
+      "loss": 3.7121,
+      "step": 4670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018324835585433225,
+      "loss": 3.641,
+      "step": 4675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018319283157075636,
+      "loss": 3.5295,
+      "step": 4680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018313722386122247,
+      "loss": 3.5396,
+      "step": 4685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018308153278149406,
+      "loss": 3.5559,
+      "step": 4690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018302575838741826,
+      "loss": 3.5536,
+      "step": 4695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001829699007349258,
+      "loss": 3.7527,
+      "step": 4700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001829139598800308,
+      "loss": 3.5316,
+      "step": 4705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018285793587883092,
+      "loss": 3.5608,
+      "step": 4710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018280182878750717,
+      "loss": 3.5152,
+      "step": 4715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001827456386623238,
+      "loss": 3.434,
+      "step": 4720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018268936555962845,
+      "loss": 3.3267,
+      "step": 4725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001826330095358519,
+      "loss": 3.4513,
+      "step": 4730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018257657064750808,
+      "loss": 3.5652,
+      "step": 4735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018252004895119404,
+      "loss": 3.5403,
+      "step": 4740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018246344450358986,
+      "loss": 3.4439,
+      "step": 4745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018240675736145865,
+      "loss": 3.474,
+      "step": 4750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001823499875816464,
+      "loss": 3.5473,
+      "step": 4755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018229313522108187,
+      "loss": 3.6774,
+      "step": 4760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018223620033677685,
+      "loss": 3.6312,
+      "step": 4765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018217918298582572,
+      "loss": 3.6326,
+      "step": 4770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001821220832254056,
+      "loss": 3.6744,
+      "step": 4775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018206490111277629,
+      "loss": 3.7397,
+      "step": 4780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018200763670528011,
+      "loss": 3.6019,
+      "step": 4785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018195029006034193,
+      "loss": 3.7168,
+      "step": 4790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018189286123546916,
+      "loss": 3.6102,
+      "step": 4795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018183535028825149,
+      "loss": 3.6562,
+      "step": 4800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018177775727636105,
+      "loss": 3.5175,
+      "step": 4805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018172008225755224,
+      "loss": 3.4547,
+      "step": 4810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018166232528966169,
+      "loss": 3.5996,
+      "step": 4815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001816044864306082,
+      "loss": 3.5837,
+      "step": 4820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018154656573839275,
+      "loss": 3.5342,
+      "step": 4825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018148856327109832,
+      "loss": 3.4871,
+      "step": 4830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018143047908688993,
+      "loss": 3.459,
+      "step": 4835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018137231324401448,
+      "loss": 3.4399,
+      "step": 4840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018131406580080084,
+      "loss": 3.5712,
+      "step": 4845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018125573681565969,
+      "loss": 3.5813,
+      "step": 4850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018119732634708346,
+      "loss": 3.5923,
+      "step": 4855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001811388344536463,
+      "loss": 3.6825,
+      "step": 4860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018108026119400397,
+      "loss": 3.6596,
+      "step": 4865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018102160662689394,
+      "loss": 3.6993,
+      "step": 4870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001809628708111351,
+      "loss": 3.707,
+      "step": 4875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018090405380562786,
+      "loss": 3.6086,
+      "step": 4880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018084515566935402,
+      "loss": 3.5868,
+      "step": 4885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018078617646137682,
+      "loss": 3.5422,
+      "step": 4890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018072711624084068,
+      "loss": 3.5596,
+      "step": 4895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018066797506697136,
+      "loss": 3.4794,
+      "step": 4900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001806087529990758,
+      "loss": 3.6109,
+      "step": 4905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018054945009654194,
+      "loss": 3.5043,
+      "step": 4910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018049006641883888,
+      "loss": 3.5935,
+      "step": 4915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018043060202551674,
+      "loss": 3.469,
+      "step": 4920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018037105697620655,
+      "loss": 3.5094,
+      "step": 4925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001803114313306202,
+      "loss": 3.4461,
+      "step": 4930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018025172514855043,
+      "loss": 3.4925,
+      "step": 4935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001801919384898707,
+      "loss": 3.4768,
+      "step": 4940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018013207141453523,
+      "loss": 3.5027,
+      "step": 4945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018007212398257888,
+      "loss": 3.6509,
+      "step": 4950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018001209625411705,
+      "loss": 3.5809,
+      "step": 4955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017995198828934568,
+      "loss": 3.5605,
+      "step": 4960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017989180014854115,
+      "loss": 3.3905,
+      "step": 4965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001798315318920603,
+      "loss": 3.4068,
+      "step": 4970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017977118358034023,
+      "loss": 3.4062,
+      "step": 4975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001797107552738984,
+      "loss": 3.316,
+      "step": 4980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017965024703333246,
+      "loss": 3.3665,
+      "step": 4985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001795896589193202,
+      "loss": 3.4396,
+      "step": 4990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017952899099261943,
+      "loss": 3.4862,
+      "step": 4995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017946824331406823,
+      "loss": 3.5381,
+      "step": 5000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017940741594458444,
+      "loss": 3.6569,
+      "step": 5005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017934650894516584,
+      "loss": 3.6969,
+      "step": 5010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017928552237689015,
+      "loss": 3.6113,
+      "step": 5015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017922445630091485,
+      "loss": 3.5132,
+      "step": 5020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001791633107784771,
+      "loss": 3.613,
+      "step": 5025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001791020858708938,
+      "loss": 3.4321,
+      "step": 5030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017904078163956142,
+      "loss": 3.4922,
+      "step": 5035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017897939814595596,
+      "loss": 3.4979,
+      "step": 5040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017891793545163297,
+      "loss": 3.4183,
+      "step": 5045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017885639361822728,
+      "loss": 3.4688,
+      "step": 5050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017879477270745328,
+      "loss": 3.4716,
+      "step": 5055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001787330727811045,
+      "loss": 3.4939,
+      "step": 5060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017867129390105384,
+      "loss": 3.5855,
+      "step": 5065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001786094361292532,
+      "loss": 3.6108,
+      "step": 5070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017854749952773372,
+      "loss": 3.568,
+      "step": 5075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017848548415860563,
+      "loss": 3.6623,
+      "step": 5080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017842339008405803,
+      "loss": 3.6116,
+      "step": 5085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017836121736635894,
+      "loss": 3.5552,
+      "step": 5090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017829896606785543,
+      "loss": 3.5208,
+      "step": 5095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017823663625097312,
+      "loss": 3.5107,
+      "step": 5100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017817422797821656,
+      "loss": 3.4409,
+      "step": 5105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001781117413121689,
+      "loss": 3.4277,
+      "step": 5110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017804917631549189,
+      "loss": 3.6021,
+      "step": 5115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017798653305092584,
+      "loss": 3.5987,
+      "step": 5120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017792381158128956,
+      "loss": 3.6167,
+      "step": 5125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017786101196948034,
+      "loss": 3.5973,
+      "step": 5130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017779813427847368,
+      "loss": 3.6569,
+      "step": 5135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00017773517857132355,
+      "loss": 3.5712,
+      "step": 5140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017767214491116198,
+      "loss": 3.623,
+      "step": 5145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017760903336119937,
+      "loss": 3.4881,
+      "step": 5150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017754584398472405,
+      "loss": 3.4635,
+      "step": 5155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001774825768451025,
+      "loss": 3.4988,
+      "step": 5160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017741923200577917,
+      "loss": 3.5065,
+      "step": 5165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017735580953027636,
+      "loss": 3.4935,
+      "step": 5170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017729230948219428,
+      "loss": 3.6191,
+      "step": 5175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017722873192521096,
+      "loss": 3.5619,
+      "step": 5180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017716507692308207,
+      "loss": 3.5122,
+      "step": 5185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000177101344539641,
+      "loss": 3.4819,
+      "step": 5190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017703753483879874,
+      "loss": 3.5116,
+      "step": 5195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001769736478845438,
+      "loss": 3.4395,
+      "step": 5200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017690968374094217,
+      "loss": 3.4563,
+      "step": 5205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017684564247213722,
+      "loss": 3.5256,
+      "step": 5210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017678152414234968,
+      "loss": 3.6109,
+      "step": 5215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017671732881587756,
+      "loss": 3.5264,
+      "step": 5220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001766530565570961,
+      "loss": 3.6489,
+      "step": 5225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017658870743045757,
+      "loss": 3.5348,
+      "step": 5230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017652428150049152,
+      "loss": 3.6791,
+      "step": 5235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001764597788318044,
+      "loss": 3.5172,
+      "step": 5240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017639519948907961,
+      "loss": 3.4979,
+      "step": 5245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017633054353707745,
+      "loss": 3.4882,
+      "step": 5250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001762658110406351,
+      "loss": 3.4439,
+      "step": 5255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017620100206466635,
+      "loss": 3.4567,
+      "step": 5260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017613611667416192,
+      "loss": 3.3919,
+      "step": 5265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017607115493418896,
+      "loss": 3.4757,
+      "step": 5270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017600611690989123,
+      "loss": 3.5503,
+      "step": 5275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017594100266648906,
+      "loss": 3.5627,
+      "step": 5280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001758758122692791,
+      "loss": 3.5196,
+      "step": 5285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017581054578363445,
+      "loss": 3.6055,
+      "step": 5290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017574520327500451,
+      "loss": 3.5199,
+      "step": 5295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001756797848089149,
+      "loss": 3.58,
+      "step": 5300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017561429045096733,
+      "loss": 3.5646,
+      "step": 5305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017554872026683978,
+      "loss": 3.5007,
+      "step": 5310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00017548307432228608,
+      "loss": 3.6639,
+      "step": 5315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017541735268313623,
+      "loss": 3.5978,
+      "step": 5320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000175351555415296,
+      "loss": 3.5732,
+      "step": 5325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017528568258474704,
+      "loss": 3.5469,
+      "step": 5330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017521973425754675,
+      "loss": 3.586,
+      "step": 5335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017515371049982824,
+      "loss": 3.4719,
+      "step": 5340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017508761137780037,
+      "loss": 3.4752,
+      "step": 5345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017502143695774741,
+      "loss": 3.4714,
+      "step": 5350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017495518730602924,
+      "loss": 3.5341,
+      "step": 5355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017488886248908118,
+      "loss": 3.5975,
+      "step": 5360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001748224625734139,
+      "loss": 3.6289,
+      "step": 5365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017475598762561333,
+      "loss": 3.7205,
+      "step": 5370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017468943771234075,
+      "loss": 3.7765,
+      "step": 5375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017462281290033256,
+      "loss": 3.8245,
+      "step": 5380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017455611325640024,
+      "loss": 3.6582,
+      "step": 5385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017448933884743037,
+      "loss": 3.5739,
+      "step": 5390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001744224897403845,
+      "loss": 3.5564,
+      "step": 5395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017435556600229902,
+      "loss": 3.5791,
+      "step": 5400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001742885677002852,
+      "loss": 3.6856,
+      "step": 5405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017422149490152914,
+      "loss": 3.6104,
+      "step": 5410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017415434767329154,
+      "loss": 3.5441,
+      "step": 5415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001740871260829078,
+      "loss": 3.4595,
+      "step": 5420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001740198301977879,
+      "loss": 3.5485,
+      "step": 5425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001739524600854163,
+      "loss": 3.4794,
+      "step": 5430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001738850158133519,
+      "loss": 3.6365,
+      "step": 5435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017381749744922796,
+      "loss": 3.7492,
+      "step": 5440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017374990506075207,
+      "loss": 3.733,
+      "step": 5445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017368223871570596,
+      "loss": 3.7354,
+      "step": 5450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001736144984819457,
+      "loss": 3.6032,
+      "step": 5455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017354668442740126,
+      "loss": 3.6179,
+      "step": 5460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017347879662007676,
+      "loss": 3.6409,
+      "step": 5465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017341083512805025,
+      "loss": 3.6133,
+      "step": 5470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017334280001947362,
+      "loss": 3.5418,
+      "step": 5475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017327469136257272,
+      "loss": 3.5113,
+      "step": 5480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000173206509225647,
+      "loss": 3.4768,
+      "step": 5485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00017313825367706967,
+      "loss": 3.5098,
+      "step": 5490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017306992478528753,
+      "loss": 3.5288,
+      "step": 5495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000173001522618821,
+      "loss": 3.5987,
+      "step": 5500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017293304724626385,
+      "loss": 3.4316,
+      "step": 5505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001728644987362834,
+      "loss": 3.3949,
+      "step": 5510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017279587715762022,
+      "loss": 3.4527,
+      "step": 5515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001727271825790882,
+      "loss": 3.4952,
+      "step": 5520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001726584150695744,
+      "loss": 3.5837,
+      "step": 5525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017258957469803906,
+      "loss": 3.6553,
+      "step": 5530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001725206615335154,
+      "loss": 3.6109,
+      "step": 5535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017245167564510974,
+      "loss": 3.6551,
+      "step": 5540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017238261710200128,
+      "loss": 3.6372,
+      "step": 5545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001723134859734421,
+      "loss": 3.545,
+      "step": 5550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017224428232875703,
+      "loss": 3.5932,
+      "step": 5555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017217500623734365,
+      "loss": 3.5913,
+      "step": 5560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017210565776867216,
+      "loss": 3.5502,
+      "step": 5565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017203623699228537,
+      "loss": 3.5324,
+      "step": 5570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001719667439777986,
+      "loss": 3.6524,
+      "step": 5575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017189717879489958,
+      "loss": 3.4114,
+      "step": 5580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017182754151334842,
+      "loss": 3.5108,
+      "step": 5585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017175783220297762,
+      "loss": 3.6321,
+      "step": 5590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017168805093369173,
+      "loss": 3.6101,
+      "step": 5595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017161819777546767,
+      "loss": 3.5566,
+      "step": 5600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001715482727983542,
+      "loss": 3.6725,
+      "step": 5605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017147827607247242,
+      "loss": 3.6074,
+      "step": 5610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017140820766801506,
+      "loss": 3.6478,
+      "step": 5615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017133806765524693,
+      "loss": 3.4643,
+      "step": 5620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001712678561045046,
+      "loss": 3.4981,
+      "step": 5625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017119757308619639,
+      "loss": 3.6574,
+      "step": 5630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017112721867080217,
+      "loss": 3.5729,
+      "step": 5635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001710567929288736,
+      "loss": 3.7408,
+      "step": 5640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017098629593103378,
+      "loss": 3.7945,
+      "step": 5645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017091572774797714,
+      "loss": 3.6832,
+      "step": 5650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017084508845046975,
+      "loss": 3.5869,
+      "step": 5655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00017077437810934882,
+      "loss": 3.5659,
+      "step": 5660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001707035967955228,
+      "loss": 3.6252,
+      "step": 5665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017063274457997137,
+      "loss": 3.5674,
+      "step": 5670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017056182153374526,
+      "loss": 3.6366,
+      "step": 5675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017049082772796633,
+      "loss": 3.4651,
+      "step": 5680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017041976323382726,
+      "loss": 3.6629,
+      "step": 5685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017034862812259174,
+      "loss": 3.6828,
+      "step": 5690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017027742246559417,
+      "loss": 3.6665,
+      "step": 5695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017020614633423976,
+      "loss": 3.731,
+      "step": 5700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017013479980000436,
+      "loss": 3.5837,
+      "step": 5705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00017006338293443446,
+      "loss": 3.4745,
+      "step": 5710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016999189580914708,
+      "loss": 3.558,
+      "step": 5715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016992033849582962,
+      "loss": 3.4107,
+      "step": 5720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016984871106623988,
+      "loss": 3.4014,
+      "step": 5725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016977701359220613,
+      "loss": 3.4296,
+      "step": 5730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016970524614562664,
+      "loss": 3.5029,
+      "step": 5735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016963340879847002,
+      "loss": 3.5958,
+      "step": 5740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001695615016227749,
+      "loss": 3.6557,
+      "step": 5745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016948952469065,
+      "loss": 3.7016,
+      "step": 5750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016941747807427387,
+      "loss": 3.6829,
+      "step": 5755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016934536184589512,
+      "loss": 3.565,
+      "step": 5760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016927317607783198,
+      "loss": 3.5091,
+      "step": 5765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016920092084247255,
+      "loss": 3.456,
+      "step": 5770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001691285962122745,
+      "loss": 3.4439,
+      "step": 5775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016905620225976517,
+      "loss": 3.4839,
+      "step": 5780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016898373905754137,
+      "loss": 3.3974,
+      "step": 5785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016891120667826933,
+      "loss": 3.4783,
+      "step": 5790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016883860519468472,
+      "loss": 3.5309,
+      "step": 5795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001687659346795925,
+      "loss": 3.4802,
+      "step": 5800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016869319520586675,
+      "loss": 3.5528,
+      "step": 5805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016862038684645078,
+      "loss": 3.558,
+      "step": 5810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016854750967435704,
+      "loss": 3.5805,
+      "step": 5815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001684745637626669,
+      "loss": 3.5233,
+      "step": 5820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016840154918453063,
+      "loss": 3.576,
+      "step": 5825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016832846601316749,
+      "loss": 3.5552,
+      "step": 5830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00016825531432186543,
+      "loss": 3.5863,
+      "step": 5835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016818209418398107,
+      "loss": 3.5365,
+      "step": 5840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001681088056729398,
+      "loss": 3.6114,
+      "step": 5845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016803544886223547,
+      "loss": 3.6611,
+      "step": 5850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016796202382543047,
+      "loss": 3.5575,
+      "step": 5855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016788853063615556,
+      "loss": 3.608,
+      "step": 5860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001678149693681099,
+      "loss": 3.5907,
+      "step": 5865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001677413400950609,
+      "loss": 3.5429,
+      "step": 5870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016766764289084414,
+      "loss": 3.5161,
+      "step": 5875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016759387782936335,
+      "loss": 3.532,
+      "step": 5880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016752004498459032,
+      "loss": 3.5109,
+      "step": 5885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016744614443056475,
+      "loss": 3.626,
+      "step": 5890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016737217624139433,
+      "loss": 3.494,
+      "step": 5895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001672981404912545,
+      "loss": 3.4936,
+      "step": 5900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016722403725438845,
+      "loss": 3.5505,
+      "step": 5905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016714986660510715,
+      "loss": 3.5738,
+      "step": 5910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000167075628617789,
+      "loss": 3.527,
+      "step": 5915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016700132336688005,
+      "loss": 3.5489,
+      "step": 5920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001669269509268938,
+      "loss": 3.5462,
+      "step": 5925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016685251137241113,
+      "loss": 3.5018,
+      "step": 5930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001667780047780801,
+      "loss": 3.5255,
+      "step": 5935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016670343121861613,
+      "loss": 3.4632,
+      "step": 5940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016662879076880178,
+      "loss": 3.5284,
+      "step": 5945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016655408350348664,
+      "loss": 3.5107,
+      "step": 5950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001664793094975873,
+      "loss": 3.5045,
+      "step": 5955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016640446882608737,
+      "loss": 3.5215,
+      "step": 5960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016632956156403716,
+      "loss": 3.6234,
+      "step": 5965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016625458778655387,
+      "loss": 3.6275,
+      "step": 5970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016617954756882144,
+      "loss": 3.7022,
+      "step": 5975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016610444098609026,
+      "loss": 3.7182,
+      "step": 5980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016602926811367744,
+      "loss": 3.6733,
+      "step": 5985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016595402902696646,
+      "loss": 3.4904,
+      "step": 5990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001658787238014073,
+      "loss": 3.566,
+      "step": 5995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016580335251251623,
+      "loss": 3.6039,
+      "step": 6000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016572791523587562,
+      "loss": 3.6402,
+      "step": 6005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00016565241204713428,
+      "loss": 3.7669,
+      "step": 6010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001655768430220069,
+      "loss": 3.6709,
+      "step": 6015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001655012082362743,
+      "loss": 3.5203,
+      "step": 6020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016542550776578322,
+      "loss": 3.5563,
+      "step": 6025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016534974168644625,
+      "loss": 3.5158,
+      "step": 6030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001652739100742417,
+      "loss": 3.4622,
+      "step": 6035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016519801300521385,
+      "loss": 3.4893,
+      "step": 6040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001651220505554723,
+      "loss": 3.6005,
+      "step": 6045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016504602280119243,
+      "loss": 3.4028,
+      "step": 6050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001649699298186151,
+      "loss": 3.5378,
+      "step": 6055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001648937716840464,
+      "loss": 3.5115,
+      "step": 6060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016481754847385793,
+      "loss": 3.4949,
+      "step": 6065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016474126026448652,
+      "loss": 3.4338,
+      "step": 6070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016466490713243416,
+      "loss": 3.4798,
+      "step": 6075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016458848915426792,
+      "loss": 3.5758,
+      "step": 6080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016451200640661993,
+      "loss": 3.5334,
+      "step": 6085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016443545896618723,
+      "loss": 3.5144,
+      "step": 6090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001643588469097318,
+      "loss": 3.5127,
+      "step": 6095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016428217031408038,
+      "loss": 3.4371,
+      "step": 6100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001642054292561244,
+      "loss": 3.5499,
+      "step": 6105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016412862381282004,
+      "loss": 3.4309,
+      "step": 6110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016405175406118786,
+      "loss": 3.4376,
+      "step": 6115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016397482007831312,
+      "loss": 3.4657,
+      "step": 6120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016389782194134534,
+      "loss": 3.4929,
+      "step": 6125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016382075972749843,
+      "loss": 3.6048,
+      "step": 6130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016374363351405054,
+      "loss": 3.6179,
+      "step": 6135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016366644337834405,
+      "loss": 3.6143,
+      "step": 6140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016358918939778536,
+      "loss": 3.5434,
+      "step": 6145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016351187164984494,
+      "loss": 3.543,
+      "step": 6150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016343449021205726,
+      "loss": 3.4592,
+      "step": 6155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016335704516202051,
+      "loss": 3.5511,
+      "step": 6160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016327953657739678,
+      "loss": 3.5785,
+      "step": 6165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001632019645359119,
+      "loss": 3.6696,
+      "step": 6170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016312432911535528,
+      "loss": 3.509,
+      "step": 6175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00016304663039357986,
+      "loss": 3.5554,
+      "step": 6180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001629688684485021,
+      "loss": 3.5759,
+      "step": 6185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016289104335810185,
+      "loss": 3.7073,
+      "step": 6190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016281315520042233,
+      "loss": 3.6145,
+      "step": 6195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001627352040535699,
+      "loss": 3.6244,
+      "step": 6200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016265718999571415,
+      "loss": 3.5074,
+      "step": 6205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001625791131050878,
+      "loss": 3.4655,
+      "step": 6210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001625009734599865,
+      "loss": 3.5015,
+      "step": 6215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016242277113876887,
+      "loss": 3.5042,
+      "step": 6220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016234450621985635,
+      "loss": 3.5255,
+      "step": 6225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016226617878173317,
+      "loss": 3.4764,
+      "step": 6230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016218778890294636,
+      "loss": 3.5077,
+      "step": 6235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016210933666210533,
+      "loss": 3.4549,
+      "step": 6240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001620308221378822,
+      "loss": 3.429,
+      "step": 6245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016195224540901156,
+      "loss": 3.4301,
+      "step": 6250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016187360655429034,
+      "loss": 3.4431,
+      "step": 6255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001617949056525777,
+      "loss": 3.5251,
+      "step": 6260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001617161427827951,
+      "loss": 3.4229,
+      "step": 6265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001616373180239261,
+      "loss": 3.3444,
+      "step": 6270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001615584314550164,
+      "loss": 3.4663,
+      "step": 6275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016147948315517357,
+      "loss": 3.6094,
+      "step": 6280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016140047320356723,
+      "loss": 3.469,
+      "step": 6285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016132140167942862,
+      "loss": 3.437,
+      "step": 6290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001612422686620509,
+      "loss": 3.3836,
+      "step": 6295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001611630742307889,
+      "loss": 3.3962,
+      "step": 6300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016108381846505885,
+      "loss": 3.4229,
+      "step": 6305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001610045014443387,
+      "loss": 3.3483,
+      "step": 6310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016092512324816772,
+      "loss": 3.4173,
+      "step": 6315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016084568395614648,
+      "loss": 3.3849,
+      "step": 6320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016076618364793696,
+      "loss": 3.4033,
+      "step": 6325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001606866224032622,
+      "loss": 3.3968,
+      "step": 6330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001606070003019064,
+      "loss": 3.37,
+      "step": 6335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016052731742371485,
+      "loss": 3.4304,
+      "step": 6340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016044757384859365,
+      "loss": 3.5981,
+      "step": 6345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001603677696565098,
+      "loss": 3.4604,
+      "step": 6350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016028790492749118,
+      "loss": 3.522,
+      "step": 6355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00016020797974162636,
+      "loss": 3.5371,
+      "step": 6360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001601279941790644,
+      "loss": 3.4421,
+      "step": 6365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016004794832001507,
+      "loss": 3.53,
+      "step": 6370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001599678422447485,
+      "loss": 3.4144,
+      "step": 6375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015988767603359526,
+      "loss": 3.4577,
+      "step": 6380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015980744976694622,
+      "loss": 3.5133,
+      "step": 6385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015972716352525242,
+      "loss": 3.437,
+      "step": 6390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001596468173890251,
+      "loss": 3.5091,
+      "step": 6395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001595664114388356,
+      "loss": 3.5598,
+      "step": 6400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015948594575531508,
+      "loss": 3.4805,
+      "step": 6405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015940542041915478,
+      "loss": 3.5829,
+      "step": 6410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015932483551110572,
+      "loss": 3.4433,
+      "step": 6415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015924419111197852,
+      "loss": 3.5017,
+      "step": 6420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015916348730264367,
+      "loss": 3.4387,
+      "step": 6425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015908272416403105,
+      "loss": 3.506,
+      "step": 6430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015900190177713016,
+      "loss": 3.4489,
+      "step": 6435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015892102022298986,
+      "loss": 3.4474,
+      "step": 6440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001588400795827184,
+      "loss": 3.4771,
+      "step": 6445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015875907993748314,
+      "loss": 3.3982,
+      "step": 6450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001586780213685108,
+      "loss": 3.4813,
+      "step": 6455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015859690395708702,
+      "loss": 3.4482,
+      "step": 6460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015851572778455657,
+      "loss": 3.4593,
+      "step": 6465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015843449293232307,
+      "loss": 3.5203,
+      "step": 6470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015835319948184903,
+      "loss": 3.438,
+      "step": 6475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001582718475146557,
+      "loss": 3.442,
+      "step": 6480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000158190437112323,
+      "loss": 3.5365,
+      "step": 6485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015810896835648952,
+      "loss": 3.5282,
+      "step": 6490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015802744132885227,
+      "loss": 3.432,
+      "step": 6495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001579458561111667,
+      "loss": 3.4429,
+      "step": 6500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001578642127852467,
+      "loss": 3.5468,
+      "step": 6505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015778251143296437,
+      "loss": 3.5394,
+      "step": 6510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015770075213625,
+      "loss": 3.4705,
+      "step": 6515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000157618934977092,
+      "loss": 3.4472,
+      "step": 6520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015753706003753678,
+      "loss": 3.5572,
+      "step": 6525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015745512739968878,
+      "loss": 3.4419,
+      "step": 6530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015737313714571017,
+      "loss": 3.4559,
+      "step": 6535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015729108935782094,
+      "loss": 3.6062,
+      "step": 6540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015720898411829889,
+      "loss": 3.5413,
+      "step": 6545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015712682150947923,
+      "loss": 3.6413,
+      "step": 6550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001570446016137549,
+      "loss": 3.4494,
+      "step": 6555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015696232451357616,
+      "loss": 3.5807,
+      "step": 6560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001568799902914506,
+      "loss": 3.4453,
+      "step": 6565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015679759902994332,
+      "loss": 3.4969,
+      "step": 6570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001567151508116763,
+      "loss": 3.4427,
+      "step": 6575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015663264571932892,
+      "loss": 3.421,
+      "step": 6580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001565500838356374,
+      "loss": 3.4833,
+      "step": 6585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015646746524339497,
+      "loss": 3.4319,
+      "step": 6590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015638479002545182,
+      "loss": 3.4571,
+      "step": 6595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015630205826471478,
+      "loss": 3.4472,
+      "step": 6600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015621927004414747,
+      "loss": 3.5269,
+      "step": 6605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001561364254467701,
+      "loss": 3.5947,
+      "step": 6610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015605352455565937,
+      "loss": 3.5058,
+      "step": 6615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001559705674539486,
+      "loss": 3.4868,
+      "step": 6620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001558875542248272,
+      "loss": 3.5223,
+      "step": 6625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001558044849515411,
+      "loss": 3.5227,
+      "step": 6630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015572135971739242,
+      "loss": 3.5201,
+      "step": 6635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001556381786057392,
+      "loss": 3.6066,
+      "step": 6640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015555494169999578,
+      "loss": 3.6072,
+      "step": 6645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015547164908363224,
+      "loss": 3.5514,
+      "step": 6650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015538830084017456,
+      "loss": 3.5122,
+      "step": 6655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015530489705320463,
+      "loss": 3.555,
+      "step": 6660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001552214378063599,
+      "loss": 3.5747,
+      "step": 6665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001551379231833335,
+      "loss": 3.5918,
+      "step": 6670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015505435326787414,
+      "loss": 3.5799,
+      "step": 6675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015497072814378584,
+      "loss": 3.6473,
+      "step": 6680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001548870478949281,
+      "loss": 3.5756,
+      "step": 6685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015480331260521565,
+      "loss": 3.5637,
+      "step": 6690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015471952235861843,
+      "loss": 3.5847,
+      "step": 6695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001546356772391615,
+      "loss": 3.5189,
+      "step": 6700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001545517773309249,
+      "loss": 3.453,
+      "step": 6705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015446782271804366,
+      "loss": 3.4587,
+      "step": 6710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015438381348470767,
+      "loss": 3.4759,
+      "step": 6715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015429974971516156,
+      "loss": 3.4924,
+      "step": 6720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001542156314937047,
+      "loss": 3.4918,
+      "step": 6725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000154131458904691,
+      "loss": 3.4925,
+      "step": 6730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015404723203252894,
+      "loss": 3.5319,
+      "step": 6735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001539629509616814,
+      "loss": 3.528,
+      "step": 6740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015387861577666559,
+      "loss": 3.5666,
+      "step": 6745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015379422656205307,
+      "loss": 3.4085,
+      "step": 6750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015370978340246955,
+      "loss": 3.5032,
+      "step": 6755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015362528638259478,
+      "loss": 3.4526,
+      "step": 6760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001535407355871626,
+      "loss": 3.3519,
+      "step": 6765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015345613110096068,
+      "loss": 3.4045,
+      "step": 6770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015337147300883066,
+      "loss": 3.4666,
+      "step": 6775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001532867613956678,
+      "loss": 3.4439,
+      "step": 6780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001532019963464211,
+      "loss": 3.4539,
+      "step": 6785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015311717794609325,
+      "loss": 3.4985,
+      "step": 6790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001530323062797402,
+      "loss": 3.5674,
+      "step": 6795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015294738143247148,
+      "loss": 3.5435,
+      "step": 6800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015286240348944997,
+      "loss": 3.5603,
+      "step": 6805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015277737253589164,
+      "loss": 3.4159,
+      "step": 6810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015269228865706584,
+      "loss": 3.4943,
+      "step": 6815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001526071519382948,
+      "loss": 3.4371,
+      "step": 6820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015252196246495382,
+      "loss": 3.5355,
+      "step": 6825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015243672032247112,
+      "loss": 3.464,
+      "step": 6830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015235142559632766,
+      "loss": 3.5677,
+      "step": 6835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015226607837205727,
+      "loss": 3.4668,
+      "step": 6840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015218067873524625,
+      "loss": 3.4798,
+      "step": 6845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015209522677153364,
+      "loss": 3.4834,
+      "step": 6850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015200972256661075,
+      "loss": 3.5855,
+      "step": 6855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015192416620622145,
+      "loss": 3.4574,
+      "step": 6860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015183855777616188,
+      "loss": 3.528,
+      "step": 6865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001517528973622803,
+      "loss": 3.5396,
+      "step": 6870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00015166718505047722,
+      "loss": 3.5505,
+      "step": 6875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001515814209267051,
+      "loss": 3.5786,
+      "step": 6880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015149560507696837,
+      "loss": 3.6595,
+      "step": 6885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015140973758732347,
+      "loss": 3.5752,
+      "step": 6890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001513238185438784,
+      "loss": 3.4613,
+      "step": 6895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015123784803279302,
+      "loss": 3.475,
+      "step": 6900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015115182614027872,
+      "loss": 3.4469,
+      "step": 6905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015106575295259847,
+      "loss": 3.489,
+      "step": 6910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015097962855606663,
+      "loss": 3.4058,
+      "step": 6915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015089345303704902,
+      "loss": 3.5599,
+      "step": 6920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015080722648196253,
+      "loss": 3.4849,
+      "step": 6925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001507209489772754,
+      "loss": 3.4747,
+      "step": 6930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001506346206095069,
+      "loss": 3.5436,
+      "step": 6935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001505482414652273,
+      "loss": 3.5455,
+      "step": 6940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015046181163105786,
+      "loss": 3.5934,
+      "step": 6945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015037533119367053,
+      "loss": 3.5427,
+      "step": 6950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001502888002397881,
+      "loss": 3.6125,
+      "step": 6955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015020221885618407,
+      "loss": 3.5157,
+      "step": 6960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00015011558712968234,
+      "loss": 3.4037,
+      "step": 6965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001500289051471575,
+      "loss": 3.3257,
+      "step": 6970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001499421729955344,
+      "loss": 3.3597,
+      "step": 6975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001498553907617882,
+      "loss": 3.3435,
+      "step": 6980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014976855853294436,
+      "loss": 3.4584,
+      "step": 6985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014968167639607845,
+      "loss": 3.423,
+      "step": 6990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014959474443831597,
+      "loss": 3.5043,
+      "step": 6995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014950776274683266,
+      "loss": 3.3893,
+      "step": 7000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014942073140885377,
+      "loss": 3.3906,
+      "step": 7005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001493336505116546,
+      "loss": 3.5513,
+      "step": 7010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014924652014256014,
+      "loss": 3.441,
+      "step": 7015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001491593403889448,
+      "loss": 3.3788,
+      "step": 7020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014907211133823273,
+      "loss": 3.3586,
+      "step": 7025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001489848330778973,
+      "loss": 3.3128,
+      "step": 7030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001488975056954615,
+      "loss": 3.3312,
+      "step": 7035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014881012927849728,
+      "loss": 3.4789,
+      "step": 7040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000148722703914626,
+      "loss": 3.4802,
+      "step": 7045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014863522969151796,
+      "loss": 3.459,
+      "step": 7050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00014854770669689253,
+      "loss": 3.426,
+      "step": 7055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014846013501851796,
+      "loss": 3.3598,
+      "step": 7060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014837251474421133,
+      "loss": 3.4684,
+      "step": 7065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014828484596183844,
+      "loss": 3.2775,
+      "step": 7070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001481971287593138,
+      "loss": 3.269,
+      "step": 7075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001481093632246003,
+      "loss": 3.3682,
+      "step": 7080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014802154944570952,
+      "loss": 3.359,
+      "step": 7085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014793368751070125,
+      "loss": 3.4297,
+      "step": 7090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014784577750768363,
+      "loss": 3.3985,
+      "step": 7095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001477578195248131,
+      "loss": 3.4944,
+      "step": 7100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000147669813650294,
+      "loss": 3.4472,
+      "step": 7105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001475817599723789,
+      "loss": 3.5856,
+      "step": 7110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014749365857936824,
+      "loss": 3.6229,
+      "step": 7115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014740550955961022,
+      "loss": 3.5199,
+      "step": 7120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001473173130015009,
+      "loss": 3.5306,
+      "step": 7125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014722906899348402,
+      "loss": 3.5089,
+      "step": 7130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014714077762405085,
+      "loss": 3.3489,
+      "step": 7135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014705243898174017,
+      "loss": 3.5303,
+      "step": 7140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014696405315513814,
+      "loss": 3.4564,
+      "step": 7145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014687562023287833,
+      "loss": 3.577,
+      "step": 7150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014678714030364143,
+      "loss": 3.5409,
+      "step": 7155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014669861345615532,
+      "loss": 3.5197,
+      "step": 7160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014661003977919492,
+      "loss": 3.7331,
+      "step": 7165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001465214193615821,
+      "loss": 3.5235,
+      "step": 7170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014643275229218563,
+      "loss": 3.5341,
+      "step": 7175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014634403865992107,
+      "loss": 3.4939,
+      "step": 7180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001462552785537506,
+      "loss": 3.5044,
+      "step": 7185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014616647206268306,
+      "loss": 3.5047,
+      "step": 7190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001460776192757738,
+      "loss": 3.5562,
+      "step": 7195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014598872028212463,
+      "loss": 3.4772,
+      "step": 7200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014589977517088365,
+      "loss": 3.4511,
+      "step": 7205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001458107840312452,
+      "loss": 3.4308,
+      "step": 7210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014572174695244976,
+      "loss": 3.4673,
+      "step": 7215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000145632664023784,
+      "loss": 3.3752,
+      "step": 7220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014554353533458042,
+      "loss": 3.3931,
+      "step": 7225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00014545436097421744,
+      "loss": 3.3601,
+      "step": 7230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001453651410321194,
+      "loss": 3.3539,
+      "step": 7235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014527587559775616,
+      "loss": 3.3549,
+      "step": 7240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001451865647606434,
+      "loss": 3.4563,
+      "step": 7245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014509720861034212,
+      "loss": 3.421,
+      "step": 7250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014500780723645897,
+      "loss": 3.4295,
+      "step": 7255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014491836072864578,
+      "loss": 3.4677,
+      "step": 7260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001448288691765997,
+      "loss": 3.4615,
+      "step": 7265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001447393326700631,
+      "loss": 3.4632,
+      "step": 7270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001446497512988234,
+      "loss": 3.4683,
+      "step": 7275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014456012515271294,
+      "loss": 3.4657,
+      "step": 7280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001444704543216091,
+      "loss": 3.4245,
+      "step": 7285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001443807388954339,
+      "loss": 3.4695,
+      "step": 7290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014429097896415425,
+      "loss": 3.52,
+      "step": 7295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014420117461778155,
+      "loss": 3.4714,
+      "step": 7300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014411132594637185,
+      "loss": 3.5128,
+      "step": 7305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001440214330400256,
+      "loss": 3.5545,
+      "step": 7310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014393149598888752,
+      "loss": 3.5879,
+      "step": 7315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001438415148831468,
+      "loss": 3.535,
+      "step": 7320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014375148981303663,
+      "loss": 3.4474,
+      "step": 7325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014366142086883436,
+      "loss": 3.4278,
+      "step": 7330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014357130814086135,
+      "loss": 3.4087,
+      "step": 7335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014348115171948283,
+      "loss": 3.3881,
+      "step": 7340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014339095169510786,
+      "loss": 3.4754,
+      "step": 7345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014330070815818922,
+      "loss": 3.4787,
+      "step": 7350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014321042119922337,
+      "loss": 3.4813,
+      "step": 7355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014312009090875025,
+      "loss": 3.4928,
+      "step": 7360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014302971737735324,
+      "loss": 3.5872,
+      "step": 7365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001429393006956592,
+      "loss": 3.6154,
+      "step": 7370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001428488409543381,
+      "loss": 3.4776,
+      "step": 7375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001427583382441032,
+      "loss": 3.4298,
+      "step": 7380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014266779265571087,
+      "loss": 3.4607,
+      "step": 7385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014257720427996037,
+      "loss": 3.4419,
+      "step": 7390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014248657320769392,
+      "loss": 3.4743,
+      "step": 7395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014239589952979662,
+      "loss": 3.4788,
+      "step": 7400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00014230518333719616,
+      "loss": 3.4183,
+      "step": 7405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014221442472086304,
+      "loss": 3.3945,
+      "step": 7410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001421236237718101,
+      "loss": 3.5529,
+      "step": 7415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014203278058109282,
+      "loss": 3.4581,
+      "step": 7420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001419418952398089,
+      "loss": 3.5005,
+      "step": 7425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014185096783909837,
+      "loss": 3.4987,
+      "step": 7430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014175999847014346,
+      "loss": 3.4392,
+      "step": 7435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014166898722416845,
+      "loss": 3.4617,
+      "step": 7440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014157793419243962,
+      "loss": 3.4424,
+      "step": 7445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014148683946626516,
+      "loss": 3.3479,
+      "step": 7450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014139570313699502,
+      "loss": 3.4125,
+      "step": 7455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014130452529602096,
+      "loss": 3.3344,
+      "step": 7460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014121330603477633,
+      "loss": 3.4292,
+      "step": 7465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014112204544473598,
+      "loss": 3.4597,
+      "step": 7470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014103074361741623,
+      "loss": 3.4574,
+      "step": 7475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014093940064437477,
+      "loss": 3.4205,
+      "step": 7480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001408480166172106,
+      "loss": 3.4575,
+      "step": 7485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014075659162756372,
+      "loss": 3.4016,
+      "step": 7490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014066512576711536,
+      "loss": 3.4171,
+      "step": 7495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001405736191275877,
+      "loss": 3.4309,
+      "step": 7500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014048207180074383,
+      "loss": 3.4368,
+      "step": 7505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014039048387838756,
+      "loss": 3.5356,
+      "step": 7510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014029885545236348,
+      "loss": 3.4471,
+      "step": 7515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014020718661455678,
+      "loss": 3.3656,
+      "step": 7520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001401154774568932,
+      "loss": 3.4545,
+      "step": 7525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00014002372807133887,
+      "loss": 3.4414,
+      "step": 7530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013993193854990027,
+      "loss": 3.4883,
+      "step": 7535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013984010898462416,
+      "loss": 3.5103,
+      "step": 7540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013974823946759742,
+      "loss": 3.497,
+      "step": 7545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000139656330090947,
+      "loss": 3.3844,
+      "step": 7550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013956438094683986,
+      "loss": 3.4489,
+      "step": 7555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013947239212748277,
+      "loss": 3.4537,
+      "step": 7560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013938036372512235,
+      "loss": 3.5462,
+      "step": 7565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001392882958320449,
+      "loss": 3.4655,
+      "step": 7570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013919618854057626,
+      "loss": 3.4888,
+      "step": 7575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00013910404194308188,
+      "loss": 3.4834,
+      "step": 7580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013901185613196654,
+      "loss": 3.5,
+      "step": 7585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013891963119967439,
+      "loss": 3.5461,
+      "step": 7590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013882736723868884,
+      "loss": 3.4856,
+      "step": 7595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013873506434153228,
+      "loss": 3.5023,
+      "step": 7600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001386427226007664,
+      "loss": 3.5596,
+      "step": 7605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013855034210899161,
+      "loss": 3.6012,
+      "step": 7610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013845792295884735,
+      "loss": 3.5641,
+      "step": 7615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001383654652430117,
+      "loss": 3.5753,
+      "step": 7620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013827296905420143,
+      "loss": 3.4504,
+      "step": 7625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013818043448517202,
+      "loss": 3.4865,
+      "step": 7630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013808786162871728,
+      "loss": 3.4065,
+      "step": 7635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013799525057766948,
+      "loss": 3.4177,
+      "step": 7640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013790260142489922,
+      "loss": 3.3931,
+      "step": 7645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013780991426331522,
+      "loss": 3.417,
+      "step": 7650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013771718918586444,
+      "loss": 3.3492,
+      "step": 7655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013762442628553179,
+      "loss": 3.3848,
+      "step": 7660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013753162565534004,
+      "loss": 3.4446,
+      "step": 7665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013743878738834998,
+      "loss": 3.3916,
+      "step": 7670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013734591157765994,
+      "loss": 3.3724,
+      "step": 7675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000137252998316406,
+      "loss": 3.3976,
+      "step": 7680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013716004769776189,
+      "loss": 3.4344,
+      "step": 7685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013706705981493853,
+      "loss": 3.3135,
+      "step": 7690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013697403476118454,
+      "loss": 3.433,
+      "step": 7695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013688097262978555,
+      "loss": 3.5037,
+      "step": 7700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001367878735140645,
+      "loss": 3.5432,
+      "step": 7705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013669473750738142,
+      "loss": 3.4425,
+      "step": 7710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013660156470313327,
+      "loss": 3.4169,
+      "step": 7715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013650835519475395,
+      "loss": 3.4031,
+      "step": 7720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001364151090757142,
+      "loss": 3.396,
+      "step": 7725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001363218264395214,
+      "loss": 3.3657,
+      "step": 7730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013622850737971963,
+      "loss": 3.3624,
+      "step": 7735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013613515198988938,
+      "loss": 3.3581,
+      "step": 7740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001360417603636477,
+      "loss": 3.4531,
+      "step": 7745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001359483325946479,
+      "loss": 3.408,
+      "step": 7750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00013585486877657957,
+      "loss": 3.6096,
+      "step": 7755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013576136900316844,
+      "loss": 3.4814,
+      "step": 7760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013566783336817627,
+      "loss": 3.5047,
+      "step": 7765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013557426196540083,
+      "loss": 3.5586,
+      "step": 7770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013548065488867573,
+      "loss": 3.6723,
+      "step": 7775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013538701223187033,
+      "loss": 3.5402,
+      "step": 7780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001352933340888897,
+      "loss": 3.5497,
+      "step": 7785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001351996205536745,
+      "loss": 3.3999,
+      "step": 7790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001351058717202009,
+      "loss": 3.5724,
+      "step": 7795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013501208768248042,
+      "loss": 3.452,
+      "step": 7800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001349182685345599,
+      "loss": 3.4706,
+      "step": 7805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013482441437052134,
+      "loss": 3.4409,
+      "step": 7810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013473052528448201,
+      "loss": 3.4,
+      "step": 7815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013463660137059407,
+      "loss": 3.3028,
+      "step": 7820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001345426427230446,
+      "loss": 3.4394,
+      "step": 7825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001344486494360555,
+      "loss": 3.4557,
+      "step": 7830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013435462160388351,
+      "loss": 3.451,
+      "step": 7835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013426055932081997,
+      "loss": 3.4238,
+      "step": 7840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013416646268119074,
+      "loss": 3.4676,
+      "step": 7845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013407233177935608,
+      "loss": 3.4631,
+      "step": 7850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013397816670971072,
+      "loss": 3.4243,
+      "step": 7855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013388396756668354,
+      "loss": 3.4205,
+      "step": 7860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013378973444473776,
+      "loss": 3.3703,
+      "step": 7865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001336954674383705,
+      "loss": 3.3447,
+      "step": 7870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013360116664211293,
+      "loss": 3.3976,
+      "step": 7875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013350683215053013,
+      "loss": 3.4977,
+      "step": 7880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013341246405822088,
+      "loss": 3.4343,
+      "step": 7885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013331806245981775,
+      "loss": 3.4122,
+      "step": 7890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001332236274499869,
+      "loss": 3.4713,
+      "step": 7895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013312915912342793,
+      "loss": 3.5454,
+      "step": 7900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001330346575748739,
+      "loss": 3.4461,
+      "step": 7905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013294012289909114,
+      "loss": 3.5587,
+      "step": 7910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013284555519087933,
+      "loss": 3.4918,
+      "step": 7915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001327509545450711,
+      "loss": 3.49,
+      "step": 7920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001326563210565322,
+      "loss": 3.4059,
+      "step": 7925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00013256165482016137,
+      "loss": 3.515,
+      "step": 7930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013246695593089,
+      "loss": 3.4244,
+      "step": 7935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013237222448368247,
+      "loss": 3.544,
+      "step": 7940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013227746057353562,
+      "loss": 3.5153,
+      "step": 7945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001321826642954789,
+      "loss": 3.6766,
+      "step": 7950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013208783574457432,
+      "loss": 3.5054,
+      "step": 7955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013199297501591603,
+      "loss": 3.5346,
+      "step": 7960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013189808220463072,
+      "loss": 3.5431,
+      "step": 7965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013180315740587701,
+      "loss": 3.5907,
+      "step": 7970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013170820071484572,
+      "loss": 3.6089,
+      "step": 7975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001316132122267597,
+      "loss": 3.635,
+      "step": 7980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013151819203687356,
+      "loss": 3.5179,
+      "step": 7985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013142314024047375,
+      "loss": 3.4676,
+      "step": 7990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013132805693287844,
+      "loss": 3.4573,
+      "step": 7995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001312329422094374,
+      "loss": 3.5012,
+      "step": 8000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001311377961655319,
+      "loss": 3.5817,
+      "step": 8005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013104261889657453,
+      "loss": 3.4574,
+      "step": 8010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013094741049800936,
+      "loss": 3.4982,
+      "step": 8015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013085217106531153,
+      "loss": 3.4739,
+      "step": 8020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013075690069398738,
+      "loss": 3.4639,
+      "step": 8025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013066159947957426,
+      "loss": 3.4945,
+      "step": 8030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001305662675176404,
+      "loss": 3.4611,
+      "step": 8035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013047090490378495,
+      "loss": 3.4568,
+      "step": 8040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013037551173363774,
+      "loss": 3.4884,
+      "step": 8045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013028008810285924,
+      "loss": 3.5121,
+      "step": 8050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013018463410714048,
+      "loss": 3.4505,
+      "step": 8055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00013008914984220294,
+      "loss": 3.4589,
+      "step": 8060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012999363540379852,
+      "loss": 3.4695,
+      "step": 8065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012989809088770923,
+      "loss": 3.5258,
+      "step": 8070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012980251638974733,
+      "loss": 3.5527,
+      "step": 8075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001297069120057552,
+      "loss": 3.4829,
+      "step": 8080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001296112778316051,
+      "loss": 3.4205,
+      "step": 8085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012951561396319918,
+      "loss": 3.4917,
+      "step": 8090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012941992049646936,
+      "loss": 3.435,
+      "step": 8095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012932419752737735,
+      "loss": 3.4664,
+      "step": 8100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012922844515191425,
+      "loss": 3.4601,
+      "step": 8105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012913266346610086,
+      "loss": 3.3784,
+      "step": 8110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001290368525659872,
+      "loss": 3.4292,
+      "step": 8115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012894101254765268,
+      "loss": 3.4623,
+      "step": 8120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012884514350720586,
+      "loss": 3.4684,
+      "step": 8125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012874924554078448,
+      "loss": 3.4219,
+      "step": 8130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012865331874455517,
+      "loss": 3.4366,
+      "step": 8135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001285573632147136,
+      "loss": 3.564,
+      "step": 8140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012846137904748414,
+      "loss": 3.5688,
+      "step": 8145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012836536633911995,
+      "loss": 3.524,
+      "step": 8150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001282693251859028,
+      "loss": 3.4872,
+      "step": 8155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012817325568414297,
+      "loss": 3.4709,
+      "step": 8160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012807715793017918,
+      "loss": 3.4763,
+      "step": 8165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012798103202037842,
+      "loss": 3.5505,
+      "step": 8170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012788487805113602,
+      "loss": 3.4407,
+      "step": 8175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001277886961188754,
+      "loss": 3.426,
+      "step": 8180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012769248632004795,
+      "loss": 3.4404,
+      "step": 8185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001275962487511332,
+      "loss": 3.4364,
+      "step": 8190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012749998350863827,
+      "loss": 3.4853,
+      "step": 8195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001274036906890982,
+      "loss": 3.4901,
+      "step": 8200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012730737038907567,
+      "loss": 3.4876,
+      "step": 8205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012721102270516087,
+      "loss": 3.4468,
+      "step": 8210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012711464773397152,
+      "loss": 3.5149,
+      "step": 8215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001270182455721526,
+      "loss": 3.4695,
+      "step": 8220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012692181631637642,
+      "loss": 3.553,
+      "step": 8225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012682536006334248,
+      "loss": 3.4484,
+      "step": 8230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012672887690977732,
+      "loss": 3.4058,
+      "step": 8235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012663236695243448,
+      "loss": 3.3824,
+      "step": 8240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001265358302880943,
+      "loss": 3.4013,
+      "step": 8245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012643926701356404,
+      "loss": 3.3883,
+      "step": 8250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012634267722567752,
+      "loss": 3.4381,
+      "step": 8255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012624606102129516,
+      "loss": 3.4436,
+      "step": 8260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012614941849730405,
+      "loss": 3.367,
+      "step": 8265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012605274975061736,
+      "loss": 3.4176,
+      "step": 8270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012595605487817482,
+      "loss": 3.4536,
+      "step": 8275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012585933397694224,
+      "loss": 3.4219,
+      "step": 8280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012576258714391155,
+      "loss": 3.3925,
+      "step": 8285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012566581447610072,
+      "loss": 3.3326,
+      "step": 8290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001255690160705536,
+      "loss": 3.3995,
+      "step": 8295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001254721920243398,
+      "loss": 3.3727,
+      "step": 8300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012537534243455472,
+      "loss": 3.3997,
+      "step": 8305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012527846739831934,
+      "loss": 3.3836,
+      "step": 8310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012518156701278019,
+      "loss": 3.3737,
+      "step": 8315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001250846413751092,
+      "loss": 3.4188,
+      "step": 8320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012498769058250355,
+      "loss": 3.4227,
+      "step": 8325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012489071473218574,
+      "loss": 3.4729,
+      "step": 8330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001247937139214034,
+      "loss": 3.4225,
+      "step": 8335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012469668824742914,
+      "loss": 3.5231,
+      "step": 8340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012459963780756054,
+      "loss": 3.4537,
+      "step": 8345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012450256269911996,
+      "loss": 3.4557,
+      "step": 8350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001244054630194546,
+      "loss": 3.4949,
+      "step": 8355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012430833886593613,
+      "loss": 3.3971,
+      "step": 8360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012421119033596102,
+      "loss": 3.4856,
+      "step": 8365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001241140175269499,
+      "loss": 3.5281,
+      "step": 8370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012401682053634792,
+      "loss": 3.5432,
+      "step": 8375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012391959946162447,
+      "loss": 3.5991,
+      "step": 8380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012382235440027307,
+      "loss": 3.5185,
+      "step": 8385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001237250854498112,
+      "loss": 3.4634,
+      "step": 8390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012362779270778048,
+      "loss": 3.445,
+      "step": 8395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012353047627174625,
+      "loss": 3.4523,
+      "step": 8400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012343313623929764,
+      "loss": 3.3927,
+      "step": 8405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012333577270804745,
+      "loss": 3.4183,
+      "step": 8410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001232383857756321,
+      "loss": 3.4643,
+      "step": 8415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012314097553971137,
+      "loss": 3.475,
+      "step": 8420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012304354209796846,
+      "loss": 3.458,
+      "step": 8425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012294608554810988,
+      "loss": 3.4628,
+      "step": 8430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012284860598786525,
+      "loss": 3.4764,
+      "step": 8435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001227511035149873,
+      "loss": 3.5294,
+      "step": 8440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012265357822725172,
+      "loss": 3.6069,
+      "step": 8445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012255603022245712,
+      "loss": 3.4769,
+      "step": 8450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001224584595984248,
+      "loss": 3.5978,
+      "step": 8455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012236086645299888,
+      "loss": 3.4736,
+      "step": 8460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012226325088404588,
+      "loss": 3.5129,
+      "step": 8465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012216561298945502,
+      "loss": 3.5887,
+      "step": 8470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012206795286713774,
+      "loss": 3.5517,
+      "step": 8475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012197027061502781,
+      "loss": 3.4093,
+      "step": 8480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012187256633108129,
+      "loss": 3.4541,
+      "step": 8485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012177484011327618,
+      "loss": 3.5046,
+      "step": 8490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012167709205961256,
+      "loss": 3.4509,
+      "step": 8495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012157932226811246,
+      "loss": 3.4786,
+      "step": 8500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012148153083681954,
+      "loss": 3.443,
+      "step": 8505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012138371786379938,
+      "loss": 3.393,
+      "step": 8510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012128588344713899,
+      "loss": 3.4577,
+      "step": 8515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001211880276849469,
+      "loss": 3.4403,
+      "step": 8520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012109015067535321,
+      "loss": 3.4695,
+      "step": 8525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012099225251650907,
+      "loss": 3.4281,
+      "step": 8530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012089433330658705,
+      "loss": 3.5161,
+      "step": 8535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012079639314378075,
+      "loss": 3.5009,
+      "step": 8540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012069843212630474,
+      "loss": 3.4722,
+      "step": 8545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012060045035239465,
+      "loss": 3.4772,
+      "step": 8550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012050244792030667,
+      "loss": 3.4992,
+      "step": 8555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012040442492831798,
+      "loss": 3.4334,
+      "step": 8560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012030638147472623,
+      "loss": 3.3973,
+      "step": 8565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012020831765784957,
+      "loss": 3.4616,
+      "step": 8570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012011023357602668,
+      "loss": 3.5077,
+      "step": 8575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012001212932761645,
+      "loss": 3.3947,
+      "step": 8580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011991400501099805,
+      "loss": 3.3951,
+      "step": 8585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011981586072457078,
+      "loss": 3.454,
+      "step": 8590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011971769656675391,
+      "loss": 3.5133,
+      "step": 8595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011961951263598677,
+      "loss": 3.4537,
+      "step": 8600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011952130903072832,
+      "loss": 3.3742,
+      "step": 8605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011942308584945741,
+      "loss": 3.5245,
+      "step": 8610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011932484319067245,
+      "loss": 3.5371,
+      "step": 8615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011922658115289141,
+      "loss": 3.4723,
+      "step": 8620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011912829983465168,
+      "loss": 3.4782,
+      "step": 8625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011902999933450997,
+      "loss": 3.5778,
+      "step": 8630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001189316797510423,
+      "loss": 3.567,
+      "step": 8635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011883334118284369,
+      "loss": 3.4798,
+      "step": 8640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011873498372852828,
+      "loss": 3.5305,
+      "step": 8645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001186366074867292,
+      "loss": 3.4151,
+      "step": 8650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011853821255609836,
+      "loss": 3.4176,
+      "step": 8655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011843979903530638,
+      "loss": 3.4367,
+      "step": 8660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011834136702304257,
+      "loss": 3.5757,
+      "step": 8665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011824291661801479,
+      "loss": 3.3508,
+      "step": 8670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011814444791894934,
+      "loss": 3.4016,
+      "step": 8675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001180459610245908,
+      "loss": 3.4411,
+      "step": 8680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011794745603370212,
+      "loss": 3.4093,
+      "step": 8685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011784893304506424,
+      "loss": 3.3866,
+      "step": 8690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001177503921574763,
+      "loss": 3.41,
+      "step": 8695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011765183346975528,
+      "loss": 3.448,
+      "step": 8700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001175532570807361,
+      "loss": 3.4959,
+      "step": 8705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011745466308927136,
+      "loss": 3.5446,
+      "step": 8710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011735605159423131,
+      "loss": 3.4133,
+      "step": 8715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011725742269450382,
+      "loss": 3.3382,
+      "step": 8720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011715877648899413,
+      "loss": 3.4214,
+      "step": 8725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001170601130766249,
+      "loss": 3.3563,
+      "step": 8730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011696143255633607,
+      "loss": 3.4294,
+      "step": 8735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001168627350270846,
+      "loss": 3.4487,
+      "step": 8740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011676402058784463,
+      "loss": 3.3951,
+      "step": 8745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011666528933760725,
+      "loss": 3.3864,
+      "step": 8750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011656654137538032,
+      "loss": 3.4799,
+      "step": 8755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001164677768001886,
+      "loss": 3.3755,
+      "step": 8760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011636899571107333,
+      "loss": 3.4174,
+      "step": 8765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011627019820709246,
+      "loss": 3.4656,
+      "step": 8770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011617138438732036,
+      "loss": 3.5169,
+      "step": 8775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011607255435084772,
+      "loss": 3.4928,
+      "step": 8780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011597370819678157,
+      "loss": 3.4291,
+      "step": 8785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011587484602424499,
+      "loss": 3.346,
+      "step": 8790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011577596793237722,
+      "loss": 3.5374,
+      "step": 8795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011567707402033345,
+      "loss": 3.5087,
+      "step": 8800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011557816438728467,
+      "loss": 3.5382,
+      "step": 8805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011547923913241774,
+      "loss": 3.5291,
+      "step": 8810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011538029835493507,
+      "loss": 3.5019,
+      "step": 8815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011528134215405473,
+      "loss": 3.4719,
+      "step": 8820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011518237062901023,
+      "loss": 3.4307,
+      "step": 8825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011508338387905038,
+      "loss": 3.4941,
+      "step": 8830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001149843820034394,
+      "loss": 3.4934,
+      "step": 8835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011488536510145651,
+      "loss": 3.4862,
+      "step": 8840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011478633327239614,
+      "loss": 3.5151,
+      "step": 8845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001146872866155676,
+      "loss": 3.4967,
+      "step": 8850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011458822523029509,
+      "loss": 3.4283,
+      "step": 8855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011448914921591765,
+      "loss": 3.5573,
+      "step": 8860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011439005867178884,
+      "loss": 3.4325,
+      "step": 8865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011429095369727696,
+      "loss": 3.3635,
+      "step": 8870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011419183439176464,
+      "loss": 3.3917,
+      "step": 8875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011409270085464898,
+      "loss": 3.4305,
+      "step": 8880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001139935531853413,
+      "loss": 3.4148,
+      "step": 8885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001138943914832671,
+      "loss": 3.4412,
+      "step": 8890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011379521584786599,
+      "loss": 3.4597,
+      "step": 8895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001136960263785915,
+      "loss": 3.4042,
+      "step": 8900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011359682317491098,
+      "loss": 3.4872,
+      "step": 8905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011349760633630575,
+      "loss": 3.4709,
+      "step": 8910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011339837596227061,
+      "loss": 3.3989,
+      "step": 8915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011329913215231401,
+      "loss": 3.388,
+      "step": 8920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011319987500595785,
+      "loss": 3.5418,
+      "step": 8925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011310060462273744,
+      "loss": 3.4708,
+      "step": 8930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011300132110220134,
+      "loss": 3.4527,
+      "step": 8935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001129020245439113,
+      "loss": 3.4219,
+      "step": 8940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011280271504744208,
+      "loss": 3.5115,
+      "step": 8945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011270339271238153,
+      "loss": 3.467,
+      "step": 8950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011260405763833029,
+      "loss": 3.4677,
+      "step": 8955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011250470992490176,
+      "loss": 3.4673,
+      "step": 8960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001124053496717221,
+      "loss": 3.4545,
+      "step": 8965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011230597697842998,
+      "loss": 3.3728,
+      "step": 8970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001122065919446765,
+      "loss": 3.4272,
+      "step": 8975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011210719467012529,
+      "loss": 3.433,
+      "step": 8980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001120077852544521,
+      "loss": 3.3992,
+      "step": 8985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011190836379734495,
+      "loss": 3.3951,
+      "step": 8990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011180893039850388,
+      "loss": 3.4045,
+      "step": 8995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011170948515764088,
+      "loss": 3.4196,
+      "step": 9000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011161002817447996,
+      "loss": 3.3977,
+      "step": 9005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011151055954875673,
+      "loss": 3.5185,
+      "step": 9010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011141107938021858,
+      "loss": 3.4191,
+      "step": 9015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011131158776862445,
+      "loss": 3.4764,
+      "step": 9020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001112120848137447,
+      "loss": 3.497,
+      "step": 9025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001111125706153612,
+      "loss": 3.4183,
+      "step": 9030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011101304527326695,
+      "loss": 3.4361,
+      "step": 9035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011091350888726619,
+      "loss": 3.3706,
+      "step": 9040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001108139615571743,
+      "loss": 3.4838,
+      "step": 9045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011071440338281745,
+      "loss": 3.4956,
+      "step": 9050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011061483446403289,
+      "loss": 3.4855,
+      "step": 9055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011051525490066852,
+      "loss": 3.4375,
+      "step": 9060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011041566479258294,
+      "loss": 3.4705,
+      "step": 9065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001103160642396454,
+      "loss": 3.4337,
+      "step": 9070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011021645334173547,
+      "loss": 3.4517,
+      "step": 9075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011011683219874323,
+      "loss": 3.4052,
+      "step": 9080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011001720091056897,
+      "loss": 3.4145,
+      "step": 9085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010991755957712318,
+      "loss": 3.4563,
+      "step": 9090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010981790829832641,
+      "loss": 3.4243,
+      "step": 9095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010971824717410917,
+      "loss": 3.4745,
+      "step": 9100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010961857630441187,
+      "loss": 3.3461,
+      "step": 9105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010951889578918471,
+      "loss": 3.3334,
+      "step": 9110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010941920572838747,
+      "loss": 3.3496,
+      "step": 9115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010931950622198965,
+      "loss": 3.3052,
+      "step": 9120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010921979736997006,
+      "loss": 3.3457,
+      "step": 9125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000109120079272317,
+      "loss": 3.3322,
+      "step": 9130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010902035202902798,
+      "loss": 3.3435,
+      "step": 9135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010892061574010972,
+      "loss": 3.4383,
+      "step": 9140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010882087050557803,
+      "loss": 3.4166,
+      "step": 9145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00010872111642545759,
+      "loss": 3.5438,
+      "step": 9150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010862135359978205,
+      "loss": 3.5156,
+      "step": 9155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010852158212859378,
+      "loss": 3.5473,
+      "step": 9160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010842180211194384,
+      "loss": 3.5342,
+      "step": 9165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010832201364989186,
+      "loss": 3.4987,
+      "step": 9170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010822221684250593,
+      "loss": 3.5329,
+      "step": 9175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010812241178986254,
+      "loss": 3.5798,
+      "step": 9180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010802259859204635,
+      "loss": 3.5865,
+      "step": 9185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010792277734915033,
+      "loss": 3.541,
+      "step": 9190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001078229481612754,
+      "loss": 3.5058,
+      "step": 9195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010772311112853053,
+      "loss": 3.5591,
+      "step": 9200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010762326635103251,
+      "loss": 3.5342,
+      "step": 9205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010752341392890587,
+      "loss": 3.4278,
+      "step": 9210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010742355396228287,
+      "loss": 3.5376,
+      "step": 9215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010732368655130333,
+      "loss": 3.3675,
+      "step": 9220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010722381179611449,
+      "loss": 3.3711,
+      "step": 9225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000107123929796871,
+      "loss": 3.3635,
+      "step": 9230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001070240406537347,
+      "loss": 3.4032,
+      "step": 9235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010692414446687471,
+      "loss": 3.7505,
+      "step": 9240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001068242413364671,
+      "loss": 3.6249,
+      "step": 9245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010672433136269499,
+      "loss": 3.3861,
+      "step": 9250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010662441464574833,
+      "loss": 3.571,
+      "step": 9255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010652449128582376,
+      "loss": 3.4986,
+      "step": 9260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010642456138312473,
+      "loss": 3.4416,
+      "step": 9265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010632462503786114,
+      "loss": 3.4873,
+      "step": 9270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010622468235024936,
+      "loss": 3.5095,
+      "step": 9275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010612473342051219,
+      "loss": 3.4704,
+      "step": 9280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010602477834887858,
+      "loss": 3.5554,
+      "step": 9285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010592481723558374,
+      "loss": 3.4456,
+      "step": 9290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010582485018086891,
+      "loss": 3.4669,
+      "step": 9295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010572487728498127,
+      "loss": 3.4011,
+      "step": 9300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010562489864817382,
+      "loss": 3.4686,
+      "step": 9305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010552491437070537,
+      "loss": 3.5753,
+      "step": 9310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010542492455284043,
+      "loss": 3.4919,
+      "step": 9315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00010532492929484898,
+      "loss": 3.5018,
+      "step": 9320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010522492869700648,
+      "loss": 3.4408,
+      "step": 9325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010512492285959382,
+      "loss": 3.4225,
+      "step": 9330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010502491188289695,
+      "loss": 3.4192,
+      "step": 9335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010492489586720724,
+      "loss": 3.4308,
+      "step": 9340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010482487491282089,
+      "loss": 3.4666,
+      "step": 9345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010472484912003913,
+      "loss": 3.557,
+      "step": 9350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010462481858916812,
+      "loss": 3.4161,
+      "step": 9355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001045247834205186,
+      "loss": 3.4066,
+      "step": 9360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010442474371440618,
+      "loss": 3.502,
+      "step": 9365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010432469957115083,
+      "loss": 3.5101,
+      "step": 9370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010422465109107702,
+      "loss": 3.4485,
+      "step": 9375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010412459837451367,
+      "loss": 3.447,
+      "step": 9380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010402454152179377,
+      "loss": 3.4666,
+      "step": 9385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010392448063325463,
+      "loss": 3.4728,
+      "step": 9390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010382441580923752,
+      "loss": 3.4636,
+      "step": 9395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010372434715008763,
+      "loss": 3.4724,
+      "step": 9400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010362427475615413,
+      "loss": 3.5011,
+      "step": 9405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010352419872778971,
+      "loss": 3.3862,
+      "step": 9410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010342411916535093,
+      "loss": 3.3817,
+      "step": 9415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010332403616919779,
+      "loss": 3.4104,
+      "step": 9420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010322394983969368,
+      "loss": 3.3997,
+      "step": 9425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001031238602772055,
+      "loss": 3.4299,
+      "step": 9430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010302376758210319,
+      "loss": 3.4613,
+      "step": 9435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010292367185475997,
+      "loss": 3.4249,
+      "step": 9440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010282357319555207,
+      "loss": 3.4632,
+      "step": 9445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010272347170485863,
+      "loss": 3.4284,
+      "step": 9450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010262336748306165,
+      "loss": 3.4313,
+      "step": 9455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001025232606305459,
+      "loss": 3.4485,
+      "step": 9460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010242315124769872,
+      "loss": 3.414,
+      "step": 9465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010232303943491004,
+      "loss": 3.4192,
+      "step": 9470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010222292529257217,
+      "loss": 3.3931,
+      "step": 9475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010212280892107988,
+      "loss": 3.4524,
+      "step": 9480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010202269042083001,
+      "loss": 3.4416,
+      "step": 9485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010192256989222169,
+      "loss": 3.398,
+      "step": 9490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00010182244743565594,
+      "loss": 3.4301,
+      "step": 9495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001017223231515358,
+      "loss": 3.4463,
+      "step": 9500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010162219714026617,
+      "loss": 3.4039,
+      "step": 9505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001015220695022536,
+      "loss": 3.3085,
+      "step": 9510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010142194033790633,
+      "loss": 3.3191,
+      "step": 9515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001013218097476341,
+      "loss": 3.3856,
+      "step": 9520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010122167783184806,
+      "loss": 3.313,
+      "step": 9525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010112154469096078,
+      "loss": 3.4163,
+      "step": 9530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010102141042538597,
+      "loss": 3.3564,
+      "step": 9535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001009212751355385,
+      "loss": 3.3791,
+      "step": 9540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010082113892183423,
+      "loss": 3.3274,
+      "step": 9545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010072100188469002,
+      "loss": 3.3159,
+      "step": 9550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010062086412452352,
+      "loss": 3.4423,
+      "step": 9555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010052072574175306,
+      "loss": 3.4804,
+      "step": 9560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010042058683679769,
+      "loss": 3.5206,
+      "step": 9565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010032044751007685,
+      "loss": 3.5033,
+      "step": 9570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010022030786201058,
+      "loss": 3.499,
+      "step": 9575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010012016799301907,
+      "loss": 3.5253,
+      "step": 9580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00010002002800352281,
+      "loss": 3.4599,
+      "step": 9585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.991988799394245e-05,
+      "loss": 3.5653,
+      "step": 9590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.981974806469858e-05,
+      "loss": 3.536,
+      "step": 9595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.971960831621173e-05,
+      "loss": 3.4623,
+      "step": 9600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.961946884890232e-05,
+      "loss": 3.4403,
+      "step": 9605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.951932976319041e-05,
+      "loss": 3.3946,
+      "step": 9610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.941919115949565e-05,
+      "loss": 3.4717,
+      "step": 9615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.931905313823733e-05,
+      "loss": 3.425,
+      "step": 9620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.921891579983404e-05,
+      "loss": 3.4353,
+      "step": 9625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.911877924470373e-05,
+      "loss": 3.4593,
+      "step": 9630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.901864357326358e-05,
+      "loss": 3.453,
+      "step": 9635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.891850888592987e-05,
+      "loss": 3.4532,
+      "step": 9640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.881837528311787e-05,
+      "loss": 3.5019,
+      "step": 9645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.871824286524175e-05,
+      "loss": 3.459,
+      "step": 9650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.861811173271459e-05,
+      "loss": 3.487,
+      "step": 9655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.851798198594809e-05,
+      "loss": 3.4984,
+      "step": 9660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.841785372535254e-05,
+      "loss": 3.4206,
+      "step": 9665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.831772705133685e-05,
+      "loss": 3.4782,
+      "step": 9670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.821760206430825e-05,
+      "loss": 3.5127,
+      "step": 9675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.811747886467226e-05,
+      "loss": 3.4766,
+      "step": 9680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.801735755283273e-05,
+      "loss": 3.511,
+      "step": 9685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.791723822919149e-05,
+      "loss": 3.5174,
+      "step": 9690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.781712099414842e-05,
+      "loss": 3.3848,
+      "step": 9695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.771700594810128e-05,
+      "loss": 3.3986,
+      "step": 9700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.761689319144573e-05,
+      "loss": 3.4746,
+      "step": 9705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.751678282457501e-05,
+      "loss": 3.3683,
+      "step": 9710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.741667494788003e-05,
+      "loss": 3.4235,
+      "step": 9715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.731656966174924e-05,
+      "loss": 3.5468,
+      "step": 9720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.721646706656839e-05,
+      "loss": 3.5306,
+      "step": 9725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.71163672627206e-05,
+      "loss": 3.4396,
+      "step": 9730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.70162703505862e-05,
+      "loss": 3.4224,
+      "step": 9735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.69161764305426e-05,
+      "loss": 3.3931,
+      "step": 9740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.681608560296413e-05,
+      "loss": 3.4806,
+      "step": 9745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.671599796822223e-05,
+      "loss": 3.4404,
+      "step": 9750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.661591362668491e-05,
+      "loss": 3.3803,
+      "step": 9755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.651583267871697e-05,
+      "loss": 3.4107,
+      "step": 9760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.641575522467984e-05,
+      "loss": 3.3617,
+      "step": 9765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.631568136493142e-05,
+      "loss": 3.3925,
+      "step": 9770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.621561119982598e-05,
+      "loss": 3.3395,
+      "step": 9775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.61155448297141e-05,
+      "loss": 3.33,
+      "step": 9780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.60154823549426e-05,
+      "loss": 3.409,
+      "step": 9785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.591542387585434e-05,
+      "loss": 3.3876,
+      "step": 9790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.581536949278814e-05,
+      "loss": 3.4272,
+      "step": 9795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.571531930607884e-05,
+      "loss": 3.4503,
+      "step": 9800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.561527341605691e-05,
+      "loss": 3.4269,
+      "step": 9805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.551523192304863e-05,
+      "loss": 3.3646,
+      "step": 9810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.541519492737586e-05,
+      "loss": 3.3592,
+      "step": 9815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.531516252935588e-05,
+      "loss": 3.4481,
+      "step": 9820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.521513482930144e-05,
+      "loss": 3.4373,
+      "step": 9825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.511511192752049e-05,
+      "loss": 3.4068,
+      "step": 9830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.501509392431627e-05,
+      "loss": 3.3923,
+      "step": 9835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.491508091998707e-05,
+      "loss": 3.4492,
+      "step": 9840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.481507301482604e-05,
+      "loss": 3.4388,
+      "step": 9845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.471507030912151e-05,
+      "loss": 3.3932,
+      "step": 9850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.46150729031563e-05,
+      "loss": 3.4571,
+      "step": 9855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.451508089720803e-05,
+      "loss": 3.4526,
+      "step": 9860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.441509439154895e-05,
+      "loss": 3.436,
+      "step": 9865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.431511348644575e-05,
+      "loss": 3.4363,
+      "step": 9870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.421513828215946e-05,
+      "loss": 3.4935,
+      "step": 9875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.41151688789455e-05,
+      "loss": 3.4785,
+      "step": 9880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.401520537705339e-05,
+      "loss": 3.4572,
+      "step": 9885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.391524787672676e-05,
+      "loss": 3.4368,
+      "step": 9890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.381529647820314e-05,
+      "loss": 3.4845,
+      "step": 9895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.371535128171416e-05,
+      "loss": 3.3886,
+      "step": 9900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.361541238748496e-05,
+      "loss": 3.3162,
+      "step": 9905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.351547989573453e-05,
+      "loss": 3.3372,
+      "step": 9910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.341555390667542e-05,
+      "loss": 3.4391,
+      "step": 9915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.331563452051362e-05,
+      "loss": 3.4464,
+      "step": 9920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.321572183744849e-05,
+      "loss": 3.3738,
+      "step": 9925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.311581595767273e-05,
+      "loss": 3.403,
+      "step": 9930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.301591698137217e-05,
+      "loss": 3.4138,
+      "step": 9935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.29160250087257e-05,
+      "loss": 3.4459,
+      "step": 9940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.281614013990526e-05,
+      "loss": 3.417,
+      "step": 9945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.271626247507561e-05,
+      "loss": 3.401,
+      "step": 9950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.261639211439427e-05,
+      "loss": 3.4619,
+      "step": 9955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.251652915801144e-05,
+      "loss": 3.4527,
+      "step": 9960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.241667370607e-05,
+      "loss": 3.3666,
+      "step": 9965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.231682585870514e-05,
+      "loss": 3.3911,
+      "step": 9970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.221698571604453e-05,
+      "loss": 3.425,
+      "step": 9975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.211715337820811e-05,
+      "loss": 3.4106,
+      "step": 9980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.201732894530797e-05,
+      "loss": 3.4019,
+      "step": 9985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.191751251744823e-05,
+      "loss": 3.4174,
+      "step": 9990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.181770419472509e-05,
+      "loss": 3.3719,
+      "step": 9995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.171790407722656e-05,
+      "loss": 3.3884,
+      "step": 10000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.161811226503233e-05,
+      "loss": 3.3333,
+      "step": 10005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.151832885821396e-05,
+      "loss": 3.4037,
+      "step": 10010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.141855395683444e-05,
+      "loss": 3.4492,
+      "step": 10015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.131878766094822e-05,
+      "loss": 3.3164,
+      "step": 10020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.121903007060121e-05,
+      "loss": 3.3646,
+      "step": 10025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.111928128583054e-05,
+      "loss": 3.4143,
+      "step": 10030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.101954140666451e-05,
+      "loss": 3.3719,
+      "step": 10035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.091981053312247e-05,
+      "loss": 3.3316,
+      "step": 10040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.082008876521481e-05,
+      "loss": 3.358,
+      "step": 10045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.072037620294275e-05,
+      "loss": 3.5402,
+      "step": 10050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.06206729462982e-05,
+      "loss": 3.4011,
+      "step": 10055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.052097909526388e-05,
+      "loss": 3.4199,
+      "step": 10060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.042129474981297e-05,
+      "loss": 3.3874,
+      "step": 10065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.032162000990914e-05,
+      "loss": 3.4635,
+      "step": 10070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.02219549755065e-05,
+      "loss": 3.3611,
+      "step": 10075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.012229974654932e-05,
+      "loss": 3.3858,
+      "step": 10080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.002265442297212e-05,
+      "loss": 3.501,
+      "step": 10085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.99230191046994e-05,
+      "loss": 3.4578,
+      "step": 10090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.982339389164575e-05,
+      "loss": 3.4302,
+      "step": 10095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.972377888371555e-05,
+      "loss": 3.3904,
+      "step": 10100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.962417418080285e-05,
+      "loss": 3.4465,
+      "step": 10105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.952457988279161e-05,
+      "loss": 3.4748,
+      "step": 10110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.942499608955516e-05,
+      "loss": 3.5204,
+      "step": 10115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.93254229009563e-05,
+      "loss": 3.4629,
+      "step": 10120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.922586041684732e-05,
+      "loss": 3.4275,
+      "step": 10125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.912630873706967e-05,
+      "loss": 3.4544,
+      "step": 10130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.902676796145403e-05,
+      "loss": 3.4336,
+      "step": 10135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.892723818982001e-05,
+      "loss": 3.4666,
+      "step": 10140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.882771952197642e-05,
+      "loss": 3.3364,
+      "step": 10145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.872821205772074e-05,
+      "loss": 3.4488,
+      "step": 10150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.862871589683924e-05,
+      "loss": 3.4661,
+      "step": 10155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.8529231139107e-05,
+      "loss": 3.5639,
+      "step": 10160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.842975788428748e-05,
+      "loss": 3.5045,
+      "step": 10165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.833029623213267e-05,
+      "loss": 3.4536,
+      "step": 10170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.823084628238298e-05,
+      "loss": 3.3744,
+      "step": 10175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.813140813476704e-05,
+      "loss": 3.5622,
+      "step": 10180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.803198188900161e-05,
+      "loss": 3.4703,
+      "step": 10185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.79325676447916e-05,
+      "loss": 3.4271,
+      "step": 10190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.783316550182982e-05,
+      "loss": 3.3598,
+      "step": 10195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.773377555979699e-05,
+      "loss": 3.4217,
+      "step": 10200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.763439791836145e-05,
+      "loss": 3.3361,
+      "step": 10205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.753503267717948e-05,
+      "loss": 3.317,
+      "step": 10210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.743567993589466e-05,
+      "loss": 3.352,
+      "step": 10215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.733633979413817e-05,
+      "loss": 3.2942,
+      "step": 10220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.723701235152854e-05,
+      "loss": 3.4149,
+      "step": 10225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.713769770767155e-05,
+      "loss": 3.406,
+      "step": 10230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.703839596216012e-05,
+      "loss": 3.3384,
+      "step": 10235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.69391072145743e-05,
+      "loss": 3.4647,
+      "step": 10240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.683983156448104e-05,
+      "loss": 3.3693,
+      "step": 10245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.67405691114342e-05,
+      "loss": 3.3358,
+      "step": 10250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.664131995497439e-05,
+      "loss": 3.3255,
+      "step": 10255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.654208419462893e-05,
+      "loss": 3.4213,
+      "step": 10260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.644286192991158e-05,
+      "loss": 3.297,
+      "step": 10265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.634365326032265e-05,
+      "loss": 3.3733,
+      "step": 10270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.62444582853489e-05,
+      "loss": 3.4788,
+      "step": 10275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.614527710446322e-05,
+      "loss": 3.3682,
+      "step": 10280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.604610981712471e-05,
+      "loss": 3.372,
+      "step": 10285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.594695652277858e-05,
+      "loss": 3.4457,
+      "step": 10290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.584781732085598e-05,
+      "loss": 3.4072,
+      "step": 10295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.574869231077383e-05,
+      "loss": 3.3953,
+      "step": 10300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.564958159193506e-05,
+      "loss": 3.5424,
+      "step": 10305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.555048526372805e-05,
+      "loss": 3.4545,
+      "step": 10310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.545140342552676e-05,
+      "loss": 3.511,
+      "step": 10315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.53523361766908e-05,
+      "loss": 3.5115,
+      "step": 10320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.525328361656494e-05,
+      "loss": 3.5528,
+      "step": 10325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.515424584447935e-05,
+      "loss": 3.5314,
+      "step": 10330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.505522295974929e-05,
+      "loss": 3.4791,
+      "step": 10335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.495621506167519e-05,
+      "loss": 3.5547,
+      "step": 10340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.485722224954237e-05,
+      "loss": 3.5492,
+      "step": 10345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.475824462262096e-05,
+      "loss": 3.5793,
+      "step": 10350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.465928228016608e-05,
+      "loss": 3.5645,
+      "step": 10355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.456033532141735e-05,
+      "loss": 3.6315,
+      "step": 10360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.44614038455989e-05,
+      "loss": 3.6001,
+      "step": 10365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.436248795191961e-05,
+      "loss": 3.5612,
+      "step": 10370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.426358773957243e-05,
+      "loss": 3.4017,
+      "step": 10375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.416470330773471e-05,
+      "loss": 3.534,
+      "step": 10380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.406583475556807e-05,
+      "loss": 3.5057,
+      "step": 10385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.396698218221807e-05,
+      "loss": 3.4671,
+      "step": 10390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.386814568681429e-05,
+      "loss": 3.4825,
+      "step": 10395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.376932536847014e-05,
+      "loss": 3.5642,
+      "step": 10400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.367052132628294e-05,
+      "loss": 3.507,
+      "step": 10405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.35717336593336e-05,
+      "loss": 3.4765,
+      "step": 10410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.347296246668653e-05,
+      "loss": 3.5383,
+      "step": 10415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.33742078473898e-05,
+      "loss": 3.4187,
+      "step": 10420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.327546990047471e-05,
+      "loss": 3.5604,
+      "step": 10425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.317674872495589e-05,
+      "loss": 3.4808,
+      "step": 10430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.30780444198312e-05,
+      "loss": 3.5421,
+      "step": 10435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.29793570840815e-05,
+      "loss": 3.4582,
+      "step": 10440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.288068681667065e-05,
+      "loss": 3.497,
+      "step": 10445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.278203371654549e-05,
+      "loss": 3.4858,
+      "step": 10450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.268339788263551e-05,
+      "loss": 3.4417,
+      "step": 10455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.2584779413853e-05,
+      "loss": 3.467,
+      "step": 10460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.248617840909268e-05,
+      "loss": 3.4081,
+      "step": 10465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.238759496723199e-05,
+      "loss": 3.475,
+      "step": 10470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.228902918713053e-05,
+      "loss": 3.4069,
+      "step": 10475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.21904811676303e-05,
+      "loss": 3.4962,
+      "step": 10480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.209195100755551e-05,
+      "loss": 3.4025,
+      "step": 10485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.199343880571241e-05,
+      "loss": 3.3879,
+      "step": 10490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.189494466088923e-05,
+      "loss": 3.5702,
+      "step": 10495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.179646867185617e-05,
+      "loss": 3.4021,
+      "step": 10500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.169801093736515e-05,
+      "loss": 3.4315,
+      "step": 10505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.159957155614974e-05,
+      "loss": 3.4108,
+      "step": 10510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.15011506269253e-05,
+      "loss": 3.4287,
+      "step": 10515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.140274824838849e-05,
+      "loss": 3.386,
+      "step": 10520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.130436451921743e-05,
+      "loss": 3.4984,
+      "step": 10525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.120599953807153e-05,
+      "loss": 3.5098,
+      "step": 10530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.110765340359145e-05,
+      "loss": 3.445,
+      "step": 10535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.10093262143989e-05,
+      "loss": 3.4813,
+      "step": 10540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.09110180690966e-05,
+      "loss": 3.4614,
+      "step": 10545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.08127290662682e-05,
+      "loss": 3.4539,
+      "step": 10550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.071445930447815e-05,
+      "loss": 3.5426,
+      "step": 10555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.061620888227145e-05,
+      "loss": 3.3623,
+      "step": 10560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.051797789817403e-05,
+      "loss": 3.3788,
+      "step": 10565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.041976645069207e-05,
+      "loss": 3.4262,
+      "step": 10570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.032157463831216e-05,
+      "loss": 3.4663,
+      "step": 10575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.022340255950138e-05,
+      "loss": 3.3835,
+      "step": 10580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.012525031270685e-05,
+      "loss": 3.4929,
+      "step": 10585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.002711799635588e-05,
+      "loss": 3.383,
+      "step": 10590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.992900570885572e-05,
+      "loss": 3.391,
+      "step": 10595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.983091354859369e-05,
+      "loss": 3.4463,
+      "step": 10600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.97328416139368e-05,
+      "loss": 3.5025,
+      "step": 10605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.963479000323171e-05,
+      "loss": 3.4358,
+      "step": 10610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.953675881480493e-05,
+      "loss": 3.3594,
+      "step": 10615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.94387481469623e-05,
+      "loss": 3.5577,
+      "step": 10620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.934075809798908e-05,
+      "loss": 3.4044,
+      "step": 10625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.924278876615004e-05,
+      "loss": 3.4446,
+      "step": 10630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.914484024968893e-05,
+      "loss": 3.4229,
+      "step": 10635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.90469126468288e-05,
+      "loss": 3.4395,
+      "step": 10640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.894900605577161e-05,
+      "loss": 3.4185,
+      "step": 10645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.885112057469839e-05,
+      "loss": 3.4847,
+      "step": 10650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.87532563017689e-05,
+      "loss": 3.5373,
+      "step": 10655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.865541333512157e-05,
+      "loss": 3.4387,
+      "step": 10660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.855759177287368e-05,
+      "loss": 3.5397,
+      "step": 10665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.84597917131208e-05,
+      "loss": 3.4661,
+      "step": 10670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.836201325393706e-05,
+      "loss": 3.5664,
+      "step": 10675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.826425649337501e-05,
+      "loss": 3.3854,
+      "step": 10680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.816652152946528e-05,
+      "loss": 3.4263,
+      "step": 10685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.806880846021669e-05,
+      "loss": 3.4713,
+      "step": 10690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.797111738361618e-05,
+      "loss": 3.4629,
+      "step": 10695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.787344839762855e-05,
+      "loss": 3.503,
+      "step": 10700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.777580160019649e-05,
+      "loss": 3.4162,
+      "step": 10705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.767817708924038e-05,
+      "loss": 3.4464,
+      "step": 10710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.758057496265839e-05,
+      "loss": 3.5447,
+      "step": 10715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.748299531832609e-05,
+      "loss": 3.5309,
+      "step": 10720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.738543825409652e-05,
+      "loss": 3.4894,
+      "step": 10725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.728790386780025e-05,
+      "loss": 3.4639,
+      "step": 10730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.71903922572449e-05,
+      "loss": 3.391,
+      "step": 10735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.70929035202153e-05,
+      "loss": 3.4486,
+      "step": 10740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.699543775447345e-05,
+      "loss": 3.4808,
+      "step": 10745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.689799505775822e-05,
+      "loss": 3.4177,
+      "step": 10750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.68005755277853e-05,
+      "loss": 3.4379,
+      "step": 10755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.67031792622473e-05,
+      "loss": 3.4123,
+      "step": 10760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.660580635881338e-05,
+      "loss": 3.5498,
+      "step": 10765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.65084569151293e-05,
+      "loss": 3.4582,
+      "step": 10770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.641113102881726e-05,
+      "loss": 3.4057,
+      "step": 10775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.631382879747597e-05,
+      "loss": 3.4094,
+      "step": 10780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.621655031868026e-05,
+      "loss": 3.3613,
+      "step": 10785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.61192956899812e-05,
+      "loss": 3.417,
+      "step": 10790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.6022065008906e-05,
+      "loss": 3.4533,
+      "step": 10795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.592485837295777e-05,
+      "loss": 3.47,
+      "step": 10800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.582767587961552e-05,
+      "loss": 3.3907,
+      "step": 10805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.573051762633414e-05,
+      "loss": 3.3771,
+      "step": 10810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.563338371054412e-05,
+      "loss": 3.4655,
+      "step": 10815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.553627422965148e-05,
+      "loss": 3.3781,
+      "step": 10820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.543918928103795e-05,
+      "loss": 3.4229,
+      "step": 10825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.534212896206051e-05,
+      "loss": 3.4061,
+      "step": 10830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.524509337005141e-05,
+      "loss": 3.3877,
+      "step": 10835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.514808260231818e-05,
+      "loss": 3.3792,
+      "step": 10840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.505109675614346e-05,
+      "loss": 3.473,
+      "step": 10845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.495413592878484e-05,
+      "loss": 3.391,
+      "step": 10850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.485720021747486e-05,
+      "loss": 3.4303,
+      "step": 10855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.476028971942093e-05,
+      "loss": 3.4607,
+      "step": 10860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.466340453180505e-05,
+      "loss": 3.4583,
+      "step": 10865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.456654475178389e-05,
+      "loss": 3.5336,
+      "step": 10870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.446971047648873e-05,
+      "loss": 3.3367,
+      "step": 10875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.437290180302512e-05,
+      "loss": 3.3926,
+      "step": 10880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.427611882847301e-05,
+      "loss": 3.4691,
+      "step": 10885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.41793616498867e-05,
+      "loss": 3.3938,
+      "step": 10890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.40826303642944e-05,
+      "loss": 3.4652,
+      "step": 10895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.398592506869849e-05,
+      "loss": 3.4003,
+      "step": 10900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.388924586007523e-05,
+      "loss": 3.38,
+      "step": 10905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.379259283537479e-05,
+      "loss": 3.412,
+      "step": 10910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.369596609152105e-05,
+      "loss": 3.403,
+      "step": 10915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.359936572541142e-05,
+      "loss": 3.4365,
+      "step": 10920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.350279183391712e-05,
+      "loss": 3.4292,
+      "step": 10925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.340624451388257e-05,
+      "loss": 3.3731,
+      "step": 10930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.330972386212558e-05,
+      "loss": 3.3804,
+      "step": 10935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.321322997543743e-05,
+      "loss": 3.3717,
+      "step": 10940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.311676295058232e-05,
+      "loss": 3.3671,
+      "step": 10945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.302032288429756e-05,
+      "loss": 3.4532,
+      "step": 10950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.292390987329356e-05,
+      "loss": 3.431,
+      "step": 10955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.282752401425343e-05,
+      "loss": 3.4105,
+      "step": 10960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.273116540383319e-05,
+      "loss": 3.4186,
+      "step": 10965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.263483413866135e-05,
+      "loss": 3.4403,
+      "step": 10970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.253853031533928e-05,
+      "loss": 3.4506,
+      "step": 10975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.244225403044056e-05,
+      "loss": 3.4044,
+      "step": 10980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.234600538051124e-05,
+      "loss": 3.3981,
+      "step": 10985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.22497844620698e-05,
+      "loss": 3.4186,
+      "step": 10990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.215359137160673e-05,
+      "loss": 3.4113,
+      "step": 10995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.205742620558464e-05,
+      "loss": 3.4839,
+      "step": 11000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.196128906043822e-05,
+      "loss": 3.4613,
+      "step": 11005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.1865180032574e-05,
+      "loss": 3.5436,
+      "step": 11010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.176909921837033e-05,
+      "loss": 3.3653,
+      "step": 11015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.167304671417729e-05,
+      "loss": 3.4557,
+      "step": 11020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.157702261631653e-05,
+      "loss": 3.4844,
+      "step": 11025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.148102702108122e-05,
+      "loss": 3.3948,
+      "step": 11030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.138506002473591e-05,
+      "loss": 3.4026,
+      "step": 11035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.128912172351664e-05,
+      "loss": 3.5242,
+      "step": 11040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.119321221363047e-05,
+      "loss": 3.4407,
+      "step": 11045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.109733159125566e-05,
+      "loss": 3.3768,
+      "step": 11050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.100147995254156e-05,
+      "loss": 3.4979,
+      "step": 11055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.09056573936084e-05,
+      "loss": 3.4611,
+      "step": 11060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.080986401054721e-05,
+      "loss": 3.393,
+      "step": 11065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.071409989941989e-05,
+      "loss": 3.4694,
+      "step": 11070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.061836515625886e-05,
+      "loss": 3.4676,
+      "step": 11075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.052265987706708e-05,
+      "loss": 3.4752,
+      "step": 11080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.042698415781813e-05,
+      "loss": 3.5297,
+      "step": 11085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.033133809445577e-05,
+      "loss": 3.4373,
+      "step": 11090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.02357217828941e-05,
+      "loss": 3.3685,
+      "step": 11095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.014013531901733e-05,
+      "loss": 3.4258,
+      "step": 11100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.004457879867986e-05,
+      "loss": 3.4328,
+      "step": 11105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.994905231770593e-05,
+      "loss": 3.5577,
+      "step": 11110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.985355597188971e-05,
+      "loss": 3.4142,
+      "step": 11115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.975808985699518e-05,
+      "loss": 3.4338,
+      "step": 11120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.966265406875597e-05,
+      "loss": 3.4854,
+      "step": 11125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.956724870287524e-05,
+      "loss": 3.4536,
+      "step": 11130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.94718738550258e-05,
+      "loss": 3.2955,
+      "step": 11135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.93765296208497e-05,
+      "loss": 3.4547,
+      "step": 11140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.928121609595835e-05,
+      "loss": 3.4441,
+      "step": 11145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.918593337593238e-05,
+      "loss": 3.537,
+      "step": 11150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.909068155632153e-05,
+      "loss": 3.5565,
+      "step": 11155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.899546073264454e-05,
+      "loss": 3.429,
+      "step": 11160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.890027100038901e-05,
+      "loss": 3.4186,
+      "step": 11165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.880511245501149e-05,
+      "loss": 3.4329,
+      "step": 11170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.870998519193717e-05,
+      "loss": 3.4657,
+      "step": 11175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.861488930655979e-05,
+      "loss": 3.442,
+      "step": 11180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.851982489424187e-05,
+      "loss": 3.486,
+      "step": 11185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.842479205031411e-05,
+      "loss": 3.4021,
+      "step": 11190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.832979087007565e-05,
+      "loss": 3.4643,
+      "step": 11195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.823482144879398e-05,
+      "loss": 3.4441,
+      "step": 11200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.813988388170456e-05,
+      "loss": 3.4551,
+      "step": 11205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.804497826401105e-05,
+      "loss": 3.4461,
+      "step": 11210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.795010469088495e-05,
+      "loss": 3.4256,
+      "step": 11215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.785526325746576e-05,
+      "loss": 3.4731,
+      "step": 11220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.776045405886066e-05,
+      "loss": 3.3837,
+      "step": 11225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.766567719014449e-05,
+      "loss": 3.4065,
+      "step": 11230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.75709327463598e-05,
+      "loss": 3.4959,
+      "step": 11235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.747622082251643e-05,
+      "loss": 3.4477,
+      "step": 11240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.738154151359172e-05,
+      "loss": 3.4453,
+      "step": 11245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.728689491453039e-05,
+      "loss": 3.4632,
+      "step": 11250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.719228112024417e-05,
+      "loss": 3.4852,
+      "step": 11255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.709770022561198e-05,
+      "loss": 3.4135,
+      "step": 11260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.700315232547981e-05,
+      "loss": 3.4759,
+      "step": 11265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.690863751466048e-05,
+      "loss": 3.4847,
+      "step": 11270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.681415588793367e-05,
+      "loss": 3.4493,
+      "step": 11275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.67197075400457e-05,
+      "loss": 3.4629,
+      "step": 11280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.662529256570969e-05,
+      "loss": 3.4686,
+      "step": 11285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.653091105960512e-05,
+      "loss": 3.445,
+      "step": 11290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.643656311637796e-05,
+      "loss": 3.4797,
+      "step": 11295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.634224883064059e-05,
+      "loss": 3.4727,
+      "step": 11300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.624796829697158e-05,
+      "loss": 3.3793,
+      "step": 11305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.615372160991561e-05,
+      "loss": 3.4989,
+      "step": 11310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.605950886398353e-05,
+      "loss": 3.4154,
+      "step": 11315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.596533015365207e-05,
+      "loss": 3.4803,
+      "step": 11320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.587118557336382e-05,
+      "loss": 3.5228,
+      "step": 11325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.577707521752725e-05,
+      "loss": 3.4514,
+      "step": 11330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.56829991805164e-05,
+      "loss": 3.51,
+      "step": 11335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.558895755667091e-05,
+      "loss": 3.5296,
+      "step": 11340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.549495044029592e-05,
+      "loss": 3.4889,
+      "step": 11345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.540097792566202e-05,
+      "loss": 3.4241,
+      "step": 11350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.530704010700504e-05,
+      "loss": 3.4034,
+      "step": 11355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.521313707852601e-05,
+      "loss": 3.3991,
+      "step": 11360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.511926893439115e-05,
+      "loss": 3.5144,
+      "step": 11365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.502543576873163e-05,
+      "loss": 3.4728,
+      "step": 11370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.493163767564352e-05,
+      "loss": 3.3865,
+      "step": 11375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.483787474918779e-05,
+      "loss": 3.4376,
+      "step": 11380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.474414708339013e-05,
+      "loss": 3.4098,
+      "step": 11385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.465045477224079e-05,
+      "loss": 3.4165,
+      "step": 11390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.455679790969473e-05,
+      "loss": 3.4513,
+      "step": 11395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.446317658967119e-05,
+      "loss": 3.4158,
+      "step": 11400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.436959090605383e-05,
+      "loss": 3.3591,
+      "step": 11405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.42760409526906e-05,
+      "loss": 3.4458,
+      "step": 11410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.418252682339361e-05,
+      "loss": 3.4662,
+      "step": 11415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.408904861193906e-05,
+      "loss": 3.4544,
+      "step": 11420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.399560641206706e-05,
+      "loss": 3.3752,
+      "step": 11425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.39022003174817e-05,
+      "loss": 3.4972,
+      "step": 11430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.380883042185084e-05,
+      "loss": 3.4263,
+      "step": 11435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.371549681880593e-05,
+      "loss": 3.4885,
+      "step": 11440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.362219960194223e-05,
+      "loss": 3.4345,
+      "step": 11445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.352893886481829e-05,
+      "loss": 3.3657,
+      "step": 11450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.343571470095625e-05,
+      "loss": 3.3585,
+      "step": 11455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.334252720384153e-05,
+      "loss": 3.4423,
+      "step": 11460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.32493764669227e-05,
+      "loss": 3.3574,
+      "step": 11465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.315626258361158e-05,
+      "loss": 3.3472,
+      "step": 11470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.306318564728294e-05,
+      "loss": 3.4906,
+      "step": 11475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.297014575127455e-05,
+      "loss": 3.4221,
+      "step": 11480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.287714298888709e-05,
+      "loss": 3.4076,
+      "step": 11485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.27841774533838e-05,
+      "loss": 3.3956,
+      "step": 11490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.26912492379909e-05,
+      "loss": 3.3515,
+      "step": 11495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.259835843589688e-05,
+      "loss": 3.3825,
+      "step": 11500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.250550514025287e-05,
+      "loss": 3.3199,
+      "step": 11505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.24126894441724e-05,
+      "loss": 3.3946,
+      "step": 11510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.231991144073126e-05,
+      "loss": 3.4146,
+      "step": 11515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.222717122296739e-05,
+      "loss": 3.3709,
+      "step": 11520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.213446888388093e-05,
+      "loss": 3.2866,
+      "step": 11525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.204180451643399e-05,
+      "loss": 3.3664,
+      "step": 11530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.194917821355062e-05,
+      "loss": 3.4697,
+      "step": 11535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.18565900681166e-05,
+      "loss": 3.4193,
+      "step": 11540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.176404017297965e-05,
+      "loss": 3.4256,
+      "step": 11545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.167152862094893e-05,
+      "loss": 3.4797,
+      "step": 11550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.157905550479525e-05,
+      "loss": 3.4303,
+      "step": 11555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.148662091725087e-05,
+      "loss": 3.3707,
+      "step": 11560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.139422495100939e-05,
+      "loss": 3.4022,
+      "step": 11565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.13018676987257e-05,
+      "loss": 3.4739,
+      "step": 11570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.120954925301587e-05,
+      "loss": 3.4261,
+      "step": 11575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.111726970645703e-05,
+      "loss": 3.4282,
+      "step": 11580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.102502915158733e-05,
+      "loss": 3.2973,
+      "step": 11585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.093282768090574e-05,
+      "loss": 3.4744,
+      "step": 11590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.084066538687222e-05,
+      "loss": 3.4135,
+      "step": 11595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.074854236190723e-05,
+      "loss": 3.4237,
+      "step": 11600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.065645869839196e-05,
+      "loss": 3.4466,
+      "step": 11605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0564414488668165e-05,
+      "loss": 3.475,
+      "step": 11610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0472409825037926e-05,
+      "loss": 3.3962,
+      "step": 11615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.038044479976375e-05,
+      "loss": 3.2949,
+      "step": 11620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0288519505068375e-05,
+      "loss": 3.4998,
+      "step": 11625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.01966340331347e-05,
+      "loss": 3.4259,
+      "step": 11630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.010478847610565e-05,
+      "loss": 3.4416,
+      "step": 11635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.0012982926084195e-05,
+      "loss": 3.4436,
+      "step": 11640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.992121747513315e-05,
+      "loss": 3.4195,
+      "step": 11645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.982949221527506e-05,
+      "loss": 3.4703,
+      "step": 11650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.973780723849225e-05,
+      "loss": 3.4013,
+      "step": 11655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9646162636726634e-05,
+      "loss": 3.4248,
+      "step": 11660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.955455850187962e-05,
+      "loss": 3.3009,
+      "step": 11665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.946299492581201e-05,
+      "loss": 3.4256,
+      "step": 11670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9371472000344006e-05,
+      "loss": 3.392,
+      "step": 11675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9279989817255e-05,
+      "loss": 3.4602,
+      "step": 11680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9188548468283475e-05,
+      "loss": 3.371,
+      "step": 11685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9097148045127095e-05,
+      "loss": 3.4414,
+      "step": 11690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9005788639442394e-05,
+      "loss": 3.3925,
+      "step": 11695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8914470342844694e-05,
+      "loss": 3.4386,
+      "step": 11700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8823193246908346e-05,
+      "loss": 3.3821,
+      "step": 11705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.873195744316611e-05,
+      "loss": 3.4906,
+      "step": 11710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.86407630231095e-05,
+      "loss": 3.3811,
+      "step": 11715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8549610078188446e-05,
+      "loss": 3.5176,
+      "step": 11720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.845849869981137e-05,
+      "loss": 3.4476,
+      "step": 11725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.836742897934497e-05,
+      "loss": 3.4053,
+      "step": 11730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.827640100811409e-05,
+      "loss": 3.466,
+      "step": 11735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8185414877401876e-05,
+      "loss": 3.4372,
+      "step": 11740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.80944706784494e-05,
+      "loss": 3.3834,
+      "step": 11745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8003568502455676e-05,
+      "loss": 3.3965,
+      "step": 11750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.7912708440577635e-05,
+      "loss": 3.397,
+      "step": 11755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.782189058392995e-05,
+      "loss": 3.5251,
+      "step": 11760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.773111502358492e-05,
+      "loss": 3.5267,
+      "step": 11765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.764038185057259e-05,
+      "loss": 3.3989,
+      "step": 11770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.754969115588034e-05,
+      "loss": 3.4337,
+      "step": 11775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7459043030452966e-05,
+      "loss": 3.469,
+      "step": 11780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.736843756519259e-05,
+      "loss": 3.4211,
+      "step": 11785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.727787485095866e-05,
+      "loss": 3.4247,
+      "step": 11790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.718735497856762e-05,
+      "loss": 3.4502,
+      "step": 11795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.709687803879301e-05,
+      "loss": 3.4649,
+      "step": 11800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.700644412236531e-05,
+      "loss": 3.385,
+      "step": 11805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.691605331997185e-05,
+      "loss": 3.4822,
+      "step": 11810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.682570572225671e-05,
+      "loss": 3.4928,
+      "step": 11815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.67354014198207e-05,
+      "loss": 3.4464,
+      "step": 11820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.664514050322122e-05,
+      "loss": 3.3819,
+      "step": 11825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6554923062971966e-05,
+      "loss": 3.3406,
+      "step": 11830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.646474918954334e-05,
+      "loss": 3.4917,
+      "step": 11835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.637461897336185e-05,
+      "loss": 3.5118,
+      "step": 11840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.628453250481026e-05,
+      "loss": 3.367,
+      "step": 11845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.6194489874227504e-05,
+      "loss": 3.4495,
+      "step": 11850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.610449117190855e-05,
+      "loss": 3.5048,
+      "step": 11855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.601453648810426e-05,
+      "loss": 3.4683,
+      "step": 11860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5924625913021386e-05,
+      "loss": 3.4911,
+      "step": 11865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.583475953682251e-05,
+      "loss": 3.4188,
+      "step": 11870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5744937449625854e-05,
+      "loss": 3.4387,
+      "step": 11875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.565515974150508e-05,
+      "loss": 3.4269,
+      "step": 11880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.556542650248959e-05,
+      "loss": 3.4387,
+      "step": 11885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.547573782256403e-05,
+      "loss": 3.403,
+      "step": 11890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.538609379166845e-05,
+      "loss": 3.4218,
+      "step": 11895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.529649449969804e-05,
+      "loss": 3.4166,
+      "step": 11900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5206940036503194e-05,
+      "loss": 3.3965,
+      "step": 11905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.511743049188931e-05,
+      "loss": 3.4845,
+      "step": 11910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5027965955616743e-05,
+      "loss": 3.4237,
+      "step": 11915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.49385465174008e-05,
+      "loss": 3.4743,
+      "step": 11920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.48491722669115e-05,
+      "loss": 3.4059,
+      "step": 11925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.47598432937734e-05,
+      "loss": 3.3823,
+      "step": 11930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.467055968756595e-05,
+      "loss": 3.3649,
+      "step": 11935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4581321537822875e-05,
+      "loss": 3.414,
+      "step": 11940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4492128934032416e-05,
+      "loss": 3.4893,
+      "step": 11945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.440298196563711e-05,
+      "loss": 3.504,
+      "step": 11950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.431388072203373e-05,
+      "loss": 3.3406,
+      "step": 11955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4224825292573154e-05,
+      "loss": 3.4897,
+      "step": 11960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.4135815766560486e-05,
+      "loss": 3.4618,
+      "step": 11965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.40468522332546e-05,
+      "loss": 3.4703,
+      "step": 11970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.395793478186838e-05,
+      "loss": 3.3449,
+      "step": 11975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.386906350156833e-05,
+      "loss": 3.3473,
+      "step": 11980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.378023848147487e-05,
+      "loss": 3.42,
+      "step": 11985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.36914598106619e-05,
+      "loss": 3.3635,
+      "step": 11990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3602727578156895e-05,
+      "loss": 3.3623,
+      "step": 11995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.35140418729407e-05,
+      "loss": 3.4905,
+      "step": 12000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3425402783947564e-05,
+      "loss": 3.5037,
+      "step": 12005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.3336810400064904e-05,
+      "loss": 3.4495,
+      "step": 12010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.324826481013345e-05,
+      "loss": 3.3432,
+      "step": 12015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.315976610294689e-05,
+      "loss": 3.4057,
+      "step": 12020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.307131436725191e-05,
+      "loss": 3.3668,
+      "step": 12025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.298290969174812e-05,
+      "loss": 3.4558,
+      "step": 12030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2894552165087916e-05,
+      "loss": 3.3819,
+      "step": 12035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2806241875876426e-05,
+      "loss": 3.3791,
+      "step": 12040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.271797891267142e-05,
+      "loss": 3.33,
+      "step": 12045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.262976336398318e-05,
+      "loss": 3.3616,
+      "step": 12050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.254159531827445e-05,
+      "loss": 3.3364,
+      "step": 12055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.245347486396033e-05,
+      "loss": 3.5023,
+      "step": 12060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.236540208940827e-05,
+      "loss": 3.3692,
+      "step": 12065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2277377082937806e-05,
+      "loss": 3.425,
+      "step": 12070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2189399932820616e-05,
+      "loss": 3.4607,
+      "step": 12075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.210147072728038e-05,
+      "loss": 3.329,
+      "step": 12080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.2013589554492714e-05,
+      "loss": 3.3323,
+      "step": 12085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.192575650258503e-05,
+      "loss": 3.4905,
+      "step": 12090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1837971659636545e-05,
+      "loss": 3.4058,
+      "step": 12095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.175023511367807e-05,
+      "loss": 3.4603,
+      "step": 12100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1662546952692015e-05,
+      "loss": 3.5054,
+      "step": 12105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.1574907264612224e-05,
+      "loss": 3.4549,
+      "step": 12110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.148731613732407e-05,
+      "loss": 3.423,
+      "step": 12115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.139977365866406e-05,
+      "loss": 3.3626,
+      "step": 12120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.131227991642001e-05,
+      "loss": 3.4474,
+      "step": 12125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.122483499833084e-05,
+      "loss": 3.4126,
+      "step": 12130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1137438992086506e-05,
+      "loss": 3.4001,
+      "step": 12135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.1050091985327884e-05,
+      "loss": 3.462,
+      "step": 12140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.096279406564686e-05,
+      "loss": 3.3863,
+      "step": 12145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.087554532058586e-05,
+      "loss": 3.385,
+      "step": 12150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.078834583763817e-05,
+      "loss": 3.4652,
+      "step": 12155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0701195704247595e-05,
+      "loss": 3.3696,
+      "step": 12160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.061409500780854e-05,
+      "loss": 3.4675,
+      "step": 12165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.052704383566577e-05,
+      "loss": 3.4232,
+      "step": 12170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.044004227511436e-05,
+      "loss": 3.4355,
+      "step": 12175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0353090413399705e-05,
+      "loss": 3.3166,
+      "step": 12180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.02661883377173e-05,
+      "loss": 3.4568,
+      "step": 12185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.017933613521273e-05,
+      "loss": 3.3627,
+      "step": 12190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.009253389298165e-05,
+      "loss": 3.4372,
+      "step": 12195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.0005781698069474e-05,
+      "loss": 3.4063,
+      "step": 12200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.991907963747148e-05,
+      "loss": 3.3947,
+      "step": 12205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.983242779813276e-05,
+      "loss": 3.4777,
+      "step": 12210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9745826266947934e-05,
+      "loss": 3.4502,
+      "step": 12215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.965927513076123e-05,
+      "loss": 3.4368,
+      "step": 12220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.957277447636629e-05,
+      "loss": 3.4979,
+      "step": 12225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.94863243905062e-05,
+      "loss": 3.4374,
+      "step": 12230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.939992495987327e-05,
+      "loss": 3.2904,
+      "step": 12235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.931357627110902e-05,
+      "loss": 3.3877,
+      "step": 12240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.9227278410804225e-05,
+      "loss": 3.4203,
+      "step": 12245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.914103146549844e-05,
+      "loss": 3.4669,
+      "step": 12250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.905483552168032e-05,
+      "loss": 3.4255,
+      "step": 12255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.896869066578741e-05,
+      "loss": 3.4052,
+      "step": 12260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.888259698420594e-05,
+      "loss": 3.2722,
+      "step": 12265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.879655456327083e-05,
+      "loss": 3.3265,
+      "step": 12270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8710563489265624e-05,
+      "loss": 3.4165,
+      "step": 12275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.862462384842237e-05,
+      "loss": 3.434,
+      "step": 12280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.853873572692151e-05,
+      "loss": 3.4655,
+      "step": 12285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.845289921089182e-05,
+      "loss": 3.4223,
+      "step": 12290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8367114386410486e-05,
+      "loss": 3.4676,
+      "step": 12295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8281381339502565e-05,
+      "loss": 3.4124,
+      "step": 12300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8195700156141386e-05,
+      "loss": 3.3337,
+      "step": 12305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8110070922248284e-05,
+      "loss": 3.3714,
+      "step": 12310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.802449372369242e-05,
+      "loss": 3.3965,
+      "step": 12315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.79389686462908e-05,
+      "loss": 3.3495,
+      "step": 12320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.785349577580817e-05,
+      "loss": 3.4524,
+      "step": 12325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.77680751979569e-05,
+      "loss": 3.4715,
+      "step": 12330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.768270699839691e-05,
+      "loss": 3.3759,
+      "step": 12335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.759739126273569e-05,
+      "loss": 3.4551,
+      "step": 12340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.751212807652806e-05,
+      "loss": 3.4238,
+      "step": 12345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.742691752527606e-05,
+      "loss": 3.447,
+      "step": 12350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7341759694429014e-05,
+      "loss": 3.4534,
+      "step": 12355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.725665466938346e-05,
+      "loss": 3.3696,
+      "step": 12360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.717160253548287e-05,
+      "loss": 3.4005,
+      "step": 12365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.708660337801773e-05,
+      "loss": 3.4423,
+      "step": 12370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.700165728222538e-05,
+      "loss": 3.4279,
+      "step": 12375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6916764333289934e-05,
+      "loss": 3.3313,
+      "step": 12380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6831924616342217e-05,
+      "loss": 3.4458,
+      "step": 12385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.674713821645975e-05,
+      "loss": 3.4747,
+      "step": 12390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6662405218666525e-05,
+      "loss": 3.3762,
+      "step": 12395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.657772570793289e-05,
+      "loss": 3.4377,
+      "step": 12400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.649309976917574e-05,
+      "loss": 3.4851,
+      "step": 12405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6408527487258124e-05,
+      "loss": 3.3974,
+      "step": 12410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.6324008946989314e-05,
+      "loss": 3.459,
+      "step": 12415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.62395442331247e-05,
+      "loss": 3.4655,
+      "step": 12420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.615513343036567e-05,
+      "loss": 3.4901,
+      "step": 12425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.607077662335959e-05,
+      "loss": 3.4094,
+      "step": 12430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.59864738966996e-05,
+      "loss": 3.3699,
+      "step": 12435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.590222533492473e-05,
+      "loss": 3.4407,
+      "step": 12440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.581803102251966e-05,
+      "loss": 3.4523,
+      "step": 12445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.573389104391449e-05,
+      "loss": 3.4302,
+      "step": 12450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.564980548348511e-05,
+      "loss": 3.4404,
+      "step": 12455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.556577442555265e-05,
+      "loss": 3.4195,
+      "step": 12460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5481797954383674e-05,
+      "loss": 3.5431,
+      "step": 12465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5397876154189956e-05,
+      "loss": 3.3908,
+      "step": 12470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5314009109128464e-05,
+      "loss": 3.3795,
+      "step": 12475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5230196903301266e-05,
+      "loss": 3.4046,
+      "step": 12480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.51464396207554e-05,
+      "loss": 3.439,
+      "step": 12485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.506273734548292e-05,
+      "loss": 3.3901,
+      "step": 12490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4979090161420645e-05,
+      "loss": 3.4594,
+      "step": 12495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.489549815245008e-05,
+      "loss": 3.4072,
+      "step": 12500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4811961402397554e-05,
+      "loss": 3.4482,
+      "step": 12505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.472847999503389e-05,
+      "loss": 3.3897,
+      "step": 12510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4645054014074426e-05,
+      "loss": 3.3642,
+      "step": 12515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.456168354317892e-05,
+      "loss": 3.4536,
+      "step": 12520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.4478368665951476e-05,
+      "loss": 3.4515,
+      "step": 12525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.43951094659404e-05,
+      "loss": 3.4916,
+      "step": 12530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.431190602663827e-05,
+      "loss": 3.4085,
+      "step": 12535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.422875843148165e-05,
+      "loss": 3.4196,
+      "step": 12540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.414566676385118e-05,
+      "loss": 3.5058,
+      "step": 12545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.406263110707125e-05,
+      "loss": 3.5372,
+      "step": 12550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.39796515444103e-05,
+      "loss": 3.4284,
+      "step": 12555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3896728159080424e-05,
+      "loss": 3.3515,
+      "step": 12560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.381386103423735e-05,
+      "loss": 3.3939,
+      "step": 12565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.373105025298041e-05,
+      "loss": 3.4603,
+      "step": 12570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.364829589835245e-05,
+      "loss": 3.4805,
+      "step": 12575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.356559805333971e-05,
+      "loss": 3.3778,
+      "step": 12580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.348295680087181e-05,
+      "loss": 3.4858,
+      "step": 12585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.340037222382156e-05,
+      "loss": 3.5436,
+      "step": 12590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3317844405005e-05,
+      "loss": 3.4001,
+      "step": 12595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.323537342718111e-05,
+      "loss": 3.4307,
+      "step": 12600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.315295937305207e-05,
+      "loss": 3.4018,
+      "step": 12605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.307060232526283e-05,
+      "loss": 3.4222,
+      "step": 12610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2988302366401254e-05,
+      "loss": 3.3999,
+      "step": 12615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2906059578997896e-05,
+      "loss": 3.4324,
+      "step": 12620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2823874045526026e-05,
+      "loss": 3.5599,
+      "step": 12625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.274174584840143e-05,
+      "loss": 3.4724,
+      "step": 12630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.265967506998253e-05,
+      "loss": 3.4982,
+      "step": 12635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.257766179257005e-05,
+      "loss": 3.4577,
+      "step": 12640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2495706098407085e-05,
+      "loss": 3.5103,
+      "step": 12645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2413808069678996e-05,
+      "loss": 3.4502,
+      "step": 12650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2331967788513295e-05,
+      "loss": 3.5103,
+      "step": 12655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.225018533697962e-05,
+      "loss": 3.4122,
+      "step": 12660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.216846079708958e-05,
+      "loss": 3.4944,
+      "step": 12665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2086794250796734e-05,
+      "loss": 3.3778,
+      "step": 12670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2005185779996484e-05,
+      "loss": 3.4573,
+      "step": 12675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1923635466525936e-05,
+      "loss": 3.4365,
+      "step": 12680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1842143392164004e-05,
+      "loss": 3.4388,
+      "step": 12685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.17607096386311e-05,
+      "loss": 3.4097,
+      "step": 12690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.167933428758916e-05,
+      "loss": 3.4335,
+      "step": 12695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.159801742064158e-05,
+      "loss": 3.3369,
+      "step": 12700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.151675911933308e-05,
+      "loss": 3.3873,
+      "step": 12705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.143555946514964e-05,
+      "loss": 3.4301,
+      "step": 12710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.135441853951857e-05,
+      "loss": 3.4915,
+      "step": 12715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1273336423808065e-05,
+      "loss": 3.4126,
+      "step": 12720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.119231319932747e-05,
+      "loss": 3.4103,
+      "step": 12725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1111348947327034e-05,
+      "loss": 3.4326,
+      "step": 12730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.1030443748997974e-05,
+      "loss": 3.4294,
+      "step": 12735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.094959768547214e-05,
+      "loss": 3.5006,
+      "step": 12740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.086881083782216e-05,
+      "loss": 3.4587,
+      "step": 12745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.078808328706127e-05,
+      "loss": 3.4071,
+      "step": 12750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.070741511414323e-05,
+      "loss": 3.4122,
+      "step": 12755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.062680639996225e-05,
+      "loss": 3.4174,
+      "step": 12760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.054625722535301e-05,
+      "loss": 3.4378,
+      "step": 12765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0465767671090304e-05,
+      "loss": 3.4344,
+      "step": 12770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.038533781788924e-05,
+      "loss": 3.4297,
+      "step": 12775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.030496774640514e-05,
+      "loss": 3.4925,
+      "step": 12780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.022465753723323e-05,
+      "loss": 3.4498,
+      "step": 12785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.014440727090879e-05,
+      "loss": 3.3666,
+      "step": 12790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.0064217027906945e-05,
+      "loss": 3.4716,
+      "step": 12795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.998408688864267e-05,
+      "loss": 3.428,
+      "step": 12800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.990401693347065e-05,
+      "loss": 3.4751,
+      "step": 12805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.982400724268516e-05,
+      "loss": 3.4432,
+      "step": 12810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.974405789652022e-05,
+      "loss": 3.4092,
+      "step": 12815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.96641689751491e-05,
+      "loss": 3.4849,
+      "step": 12820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.95843405586846e-05,
+      "loss": 3.5055,
+      "step": 12825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.950457272717889e-05,
+      "loss": 3.3467,
+      "step": 12830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9424865560623305e-05,
+      "loss": 3.4317,
+      "step": 12835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9345219138948365e-05,
+      "loss": 3.3582,
+      "step": 12840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9265633542023684e-05,
+      "loss": 3.4635,
+      "step": 12845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9186108849657885e-05,
+      "loss": 3.359,
+      "step": 12850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.91066451415985e-05,
+      "loss": 3.4359,
+      "step": 12855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9027242497531865e-05,
+      "loss": 3.4464,
+      "step": 12860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8947900997083255e-05,
+      "loss": 3.3677,
+      "step": 12865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8868620719816395e-05,
+      "loss": 3.4575,
+      "step": 12870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.878940174523371e-05,
+      "loss": 3.4131,
+      "step": 12875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8710244152776264e-05,
+      "loss": 3.4459,
+      "step": 12880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8631148021823406e-05,
+      "loss": 3.4743,
+      "step": 12885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8552113431692925e-05,
+      "loss": 3.4034,
+      "step": 12890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.847314046164089e-05,
+      "loss": 3.3817,
+      "step": 12895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8394229190861567e-05,
+      "loss": 3.3604,
+      "step": 12900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.831537969848731e-05,
+      "loss": 3.5609,
+      "step": 12905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.823659206358865e-05,
+      "loss": 3.3017,
+      "step": 12910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.8157866365174e-05,
+      "loss": 3.3909,
+      "step": 12915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.807920268218961e-05,
+      "loss": 3.4834,
+      "step": 12920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.800060109351957e-05,
+      "loss": 3.3578,
+      "step": 12925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.792206167798582e-05,
+      "loss": 3.4263,
+      "step": 12930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.784358451434783e-05,
+      "loss": 3.4707,
+      "step": 12935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776516968130266e-05,
+      "loss": 3.4061,
+      "step": 12940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768681725748488e-05,
+      "loss": 3.3753,
+      "step": 12945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.760852732146649e-05,
+      "loss": 3.5493,
+      "step": 12950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.753029995175677e-05,
+      "loss": 3.336,
+      "step": 12955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7452135226802385e-05,
+      "loss": 3.4704,
+      "step": 12960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7374033224987084e-05,
+      "loss": 3.5301,
+      "step": 12965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.729599402463162e-05,
+      "loss": 3.4371,
+      "step": 12970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7218017703993994e-05,
+      "loss": 3.3675,
+      "step": 12975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.714010434126899e-05,
+      "loss": 3.3916,
+      "step": 12980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.706225401458831e-05,
+      "loss": 3.4445,
+      "step": 12985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6984466802020436e-05,
+      "loss": 3.4208,
+      "step": 12990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.690674278157056e-05,
+      "loss": 3.4449,
+      "step": 12995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6829082031180496e-05,
+      "loss": 3.4187,
+      "step": 13000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6751484628728594e-05,
+      "loss": 3.3511,
+      "step": 13005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6673950652029766e-05,
+      "loss": 3.4735,
+      "step": 13010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.659648017883526e-05,
+      "loss": 3.5323,
+      "step": 13015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.651907328683254e-05,
+      "loss": 3.4325,
+      "step": 13020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6441730053645506e-05,
+      "loss": 3.3847,
+      "step": 13025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6364450556834097e-05,
+      "loss": 3.4291,
+      "step": 13030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.628723487389437e-05,
+      "loss": 3.4373,
+      "step": 13035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.621008308225837e-05,
+      "loss": 3.3928,
+      "step": 13040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.61329952592941e-05,
+      "loss": 3.3976,
+      "step": 13045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.605597148230541e-05,
+      "loss": 3.3787,
+      "step": 13050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.597901182853185e-05,
+      "loss": 3.4422,
+      "step": 13055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.590211637514884e-05,
+      "loss": 3.4278,
+      "step": 13060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.582528519926729e-05,
+      "loss": 3.4047,
+      "step": 13065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.574851837793357e-05,
+      "loss": 3.4911,
+      "step": 13070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.567181598812973e-05,
+      "loss": 3.4252,
+      "step": 13075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.559517810677308e-05,
+      "loss": 3.4642,
+      "step": 13080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.551860481071624e-05,
+      "loss": 3.3949,
+      "step": 13085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.544209617674707e-05,
+      "loss": 3.3708,
+      "step": 13090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.536565228158864e-05,
+      "loss": 3.4549,
+      "step": 13095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.528927320189903e-05,
+      "loss": 3.5864,
+      "step": 13100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.521295901427132e-05,
+      "loss": 3.3629,
+      "step": 13105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.5136709795233626e-05,
+      "loss": 3.4732,
+      "step": 13110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.506052562124883e-05,
+      "loss": 3.4384,
+      "step": 13115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.498440656871449e-05,
+      "loss": 3.5115,
+      "step": 13120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4908352713963077e-05,
+      "loss": 3.4633,
+      "step": 13125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.483236413326151e-05,
+      "loss": 3.5703,
+      "step": 13130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.475644090281133e-05,
+      "loss": 3.5147,
+      "step": 13135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.468058309874851e-05,
+      "loss": 3.444,
+      "step": 13140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.460479079714343e-05,
+      "loss": 3.4569,
+      "step": 13145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.452906407400074e-05,
+      "loss": 3.4465,
+      "step": 13150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.4453403005259444e-05,
+      "loss": 3.3716,
+      "step": 13155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.43778076667926e-05,
+      "loss": 3.4576,
+      "step": 13160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.43022781344074e-05,
+      "loss": 3.4954,
+      "step": 13165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4226814483844946e-05,
+      "loss": 3.3581,
+      "step": 13170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4151416790780456e-05,
+      "loss": 3.4964,
+      "step": 13175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.4076085130822866e-05,
+      "loss": 3.463,
+      "step": 13180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.400081957951492e-05,
+      "loss": 3.4221,
+      "step": 13185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.392562021233311e-05,
+      "loss": 3.3691,
+      "step": 13190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.38504871046875e-05,
+      "loss": 3.4659,
+      "step": 13195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3775420331921736e-05,
+      "loss": 3.3342,
+      "step": 13200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3700419969312994e-05,
+      "loss": 3.3964,
+      "step": 13205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.362548609207177e-05,
+      "loss": 3.4668,
+      "step": 13210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.355061877534192e-05,
+      "loss": 3.3761,
+      "step": 13215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3475818094200585e-05,
+      "loss": 3.466,
+      "step": 13220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.340108412365803e-05,
+      "loss": 3.3765,
+      "step": 13225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.332641693865766e-05,
+      "loss": 3.4859,
+      "step": 13230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.3251816614075884e-05,
+      "loss": 3.3735,
+      "step": 13235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.317728322472209e-05,
+      "loss": 3.3621,
+      "step": 13240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.310281684533852e-05,
+      "loss": 3.3748,
+      "step": 13245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.302841755060018e-05,
+      "loss": 3.5278,
+      "step": 13250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2954085415114946e-05,
+      "loss": 3.5288,
+      "step": 13255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2879820513423184e-05,
+      "loss": 3.4456,
+      "step": 13260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2805622919997934e-05,
+      "loss": 3.4631,
+      "step": 13265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.273149270924468e-05,
+      "loss": 3.5066,
+      "step": 13270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2657429955501394e-05,
+      "loss": 3.4205,
+      "step": 13275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.258343473303832e-05,
+      "loss": 3.4313,
+      "step": 13280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2509507116058134e-05,
+      "loss": 3.3916,
+      "step": 13285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.243564717869552e-05,
+      "loss": 3.4521,
+      "step": 13290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2361854995017416e-05,
+      "loss": 3.4749,
+      "step": 13295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.228813063902276e-05,
+      "loss": 3.4534,
+      "step": 13300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2214474184642574e-05,
+      "loss": 3.453,
+      "step": 13305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2140885705739674e-05,
+      "loss": 3.515,
+      "step": 13310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.2067365276108754e-05,
+      "loss": 3.3841,
+      "step": 13315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.199391296947627e-05,
+      "loss": 3.4814,
+      "step": 13320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.192052885950034e-05,
+      "loss": 3.3602,
+      "step": 13325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1847213019770716e-05,
+      "loss": 3.484,
+      "step": 13330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1773965523808754e-05,
+      "loss": 3.4659,
+      "step": 13335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1700786445067135e-05,
+      "loss": 3.3604,
+      "step": 13340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.162767585692997e-05,
+      "loss": 3.4001,
+      "step": 13345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.155463383271282e-05,
+      "loss": 3.5798,
+      "step": 13350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.148166044566233e-05,
+      "loss": 3.3826,
+      "step": 13355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.14087557689564e-05,
+      "loss": 3.4659,
+      "step": 13360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.133591987570399e-05,
+      "loss": 3.4531,
+      "step": 13365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1263152838945095e-05,
+      "loss": 3.4191,
+      "step": 13370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1190454731650675e-05,
+      "loss": 3.3692,
+      "step": 13375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.111782562672251e-05,
+      "loss": 3.4698,
+      "step": 13380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.104526559699333e-05,
+      "loss": 3.3677,
+      "step": 13385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0972774715226406e-05,
+      "loss": 3.4207,
+      "step": 13390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.090035305411575e-05,
+      "loss": 3.4267,
+      "step": 13395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0828000686286027e-05,
+      "loss": 3.3806,
+      "step": 13400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.075571768429233e-05,
+      "loss": 3.3986,
+      "step": 13405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.06835041206202e-05,
+      "loss": 3.4173,
+      "step": 13410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0611360067685576e-05,
+      "loss": 3.3747,
+      "step": 13415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0539285597834675e-05,
+      "loss": 3.3857,
+      "step": 13420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0467280783343944e-05,
+      "loss": 3.4542,
+      "step": 13425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0395345696419918e-05,
+      "loss": 3.4552,
+      "step": 13430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0323480409199378e-05,
+      "loss": 3.5176,
+      "step": 13435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.0251684993748886e-05,
+      "loss": 3.481,
+      "step": 13440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.017995952206506e-05,
+      "loss": 3.4387,
+      "step": 13445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.010830406607441e-05,
+      "loss": 3.4442,
+      "step": 13450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.003671869763317e-05,
+      "loss": 3.4148,
+      "step": 13455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9965203488527317e-05,
+      "loss": 3.5284,
+      "step": 13460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9893758510472436e-05,
+      "loss": 3.4433,
+      "step": 13465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.982238383511373e-05,
+      "loss": 3.4068,
+      "step": 13470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.975107953402585e-05,
+      "loss": 3.4897,
+      "step": 13475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.967984567871297e-05,
+      "loss": 3.4903,
+      "step": 13480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.960868234060855e-05,
+      "loss": 3.5319,
+      "step": 13485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9537589591075298e-05,
+      "loss": 3.4193,
+      "step": 13490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9466567501405185e-05,
+      "loss": 3.3527,
+      "step": 13495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.939561614281936e-05,
+      "loss": 3.3385,
+      "step": 13500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9324735586468e-05,
+      "loss": 3.4873,
+      "step": 13505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9253925903430267e-05,
+      "loss": 3.4693,
+      "step": 13510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9183187164714288e-05,
+      "loss": 3.4874,
+      "step": 13515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9112519441257e-05,
+      "loss": 3.4882,
+      "step": 13520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.9041922803924158e-05,
+      "loss": 3.385,
+      "step": 13525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8971397323510275e-05,
+      "loss": 3.41,
+      "step": 13530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.890094307073845e-05,
+      "loss": 3.4329,
+      "step": 13535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.883056011626032e-05,
+      "loss": 3.4824,
+      "step": 13540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8760248530656063e-05,
+      "loss": 3.4619,
+      "step": 13545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8690008384434363e-05,
+      "loss": 3.4815,
+      "step": 13550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.861983974803215e-05,
+      "loss": 3.3843,
+      "step": 13555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8549742691814705e-05,
+      "loss": 3.3387,
+      "step": 13560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8479717286075502e-05,
+      "loss": 3.3733,
+      "step": 13565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8409763601036188e-05,
+      "loss": 3.4172,
+      "step": 13570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8339881706846427e-05,
+      "loss": 3.5058,
+      "step": 13575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8270071673584008e-05,
+      "loss": 3.4697,
+      "step": 13580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.82003335712546e-05,
+      "loss": 3.397,
+      "step": 13585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8130667469791626e-05,
+      "loss": 3.3922,
+      "step": 13590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8061073439056507e-05,
+      "loss": 3.4303,
+      "step": 13595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.799155154883826e-05,
+      "loss": 3.3739,
+      "step": 13600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7922101868853577e-05,
+      "loss": 3.3829,
+      "step": 13605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.785272446874677e-05,
+      "loss": 3.5141,
+      "step": 13610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.778341941808965e-05,
+      "loss": 3.5435,
+      "step": 13615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.771418678638147e-05,
+      "loss": 3.4333,
+      "step": 13620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7645026643048855e-05,
+      "loss": 3.525,
+      "step": 13625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7575939057445786e-05,
+      "loss": 3.4927,
+      "step": 13630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.750692409885347e-05,
+      "loss": 3.4071,
+      "step": 13635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7437981836480166e-05,
+      "loss": 3.3944,
+      "step": 13640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.736911233946141e-05,
+      "loss": 3.4253,
+      "step": 13645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.730031567685968e-05,
+      "loss": 3.404,
+      "step": 13650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.723159191766439e-05,
+      "loss": 3.4327,
+      "step": 13655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.716294113079192e-05,
+      "loss": 3.3446,
+      "step": 13660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7094363385085398e-05,
+      "loss": 3.481,
+      "step": 13665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.7025858749314758e-05,
+      "loss": 3.4406,
+      "step": 13670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6957427292176572e-05,
+      "loss": 3.4653,
+      "step": 13675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6889069082294114e-05,
+      "loss": 3.4196,
+      "step": 13680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6820784188217164e-05,
+      "loss": 3.4672,
+      "step": 13685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.675257267842185e-05,
+      "loss": 3.4957,
+      "step": 13690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.668443462131094e-05,
+      "loss": 3.344,
+      "step": 13695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6616370085213394e-05,
+      "loss": 3.3648,
+      "step": 13700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6548379138384483e-05,
+      "loss": 3.3936,
+      "step": 13705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.648046184900568e-05,
+      "loss": 3.4392,
+      "step": 13710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6412618285184587e-05,
+      "loss": 3.4115,
+      "step": 13715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6344848514954856e-05,
+      "loss": 3.4271,
+      "step": 13720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6277152606276234e-05,
+      "loss": 3.416,
+      "step": 13725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.6209530627034295e-05,
+      "loss": 3.4698,
+      "step": 13730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.614198264504053e-05,
+      "loss": 3.4536,
+      "step": 13735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.607450872803213e-05,
+      "loss": 3.403,
+      "step": 13740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.600710894367219e-05,
+      "loss": 3.4069,
+      "step": 13745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5939783359549306e-05,
+      "loss": 3.4021,
+      "step": 13750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5872532043177743e-05,
+      "loss": 3.3974,
+      "step": 13755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.580535506199727e-05,
+      "loss": 3.3361,
+      "step": 13760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5738252483373117e-05,
+      "loss": 3.4486,
+      "step": 13765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.567122437459586e-05,
+      "loss": 3.4401,
+      "step": 13770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5604270802881503e-05,
+      "loss": 3.2842,
+      "step": 13775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5537391835371217e-05,
+      "loss": 3.4002,
+      "step": 13780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5470587539131362e-05,
+      "loss": 3.4572,
+      "step": 13785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5403857981153457e-05,
+      "loss": 3.4028,
+      "step": 13790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5337203228354035e-05,
+      "loss": 3.4156,
+      "step": 13795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.527062334757464e-05,
+      "loss": 3.3197,
+      "step": 13800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.5204118405581724e-05,
+      "loss": 3.4788,
+      "step": 13805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.513768846906659e-05,
+      "loss": 3.4155,
+      "step": 13810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.507133360464533e-05,
+      "loss": 3.3215,
+      "step": 13815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.500505387885872e-05,
+      "loss": 3.4519,
+      "step": 13820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.493884935817228e-05,
+      "loss": 3.3837,
+      "step": 13825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.487272010897601e-05,
+      "loss": 3.3798,
+      "step": 13830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4806666197584483e-05,
+      "loss": 3.4635,
+      "step": 13835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.474068769023671e-05,
+      "loss": 3.3723,
+      "step": 13840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4674784653096083e-05,
+      "loss": 3.3135,
+      "step": 13845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.460895715225028e-05,
+      "loss": 3.4588,
+      "step": 13850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4543205253711355e-05,
+      "loss": 3.3511,
+      "step": 13855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.447752902341538e-05,
+      "loss": 3.3753,
+      "step": 13860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.441192852722265e-05,
+      "loss": 3.395,
+      "step": 13865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4346403830917464e-05,
+      "loss": 3.4534,
+      "step": 13870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4280955000208184e-05,
+      "loss": 3.4165,
+      "step": 13875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.421558210072702e-05,
+      "loss": 3.4198,
+      "step": 13880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4150285198030066e-05,
+      "loss": 3.4805,
+      "step": 13885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4085064357597197e-05,
+      "loss": 3.4468,
+      "step": 13890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4019919644832023e-05,
+      "loss": 3.425,
+      "step": 13895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.395485112506177e-05,
+      "loss": 3.3638,
+      "step": 13900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3889858863537396e-05,
+      "loss": 3.5091,
+      "step": 13905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.382494292543319e-05,
+      "loss": 3.4622,
+      "step": 13910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.376010337584701e-05,
+      "loss": 3.4499,
+      "step": 13915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.369534027980015e-05,
+      "loss": 3.4294,
+      "step": 13920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.363065370223716e-05,
+      "loss": 3.3564,
+      "step": 13925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3566043708025874e-05,
+      "loss": 3.3961,
+      "step": 13930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3501510361957367e-05,
+      "loss": 3.4948,
+      "step": 13935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3437053728745807e-05,
+      "loss": 3.4406,
+      "step": 13940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.337267387302844e-05,
+      "loss": 3.4303,
+      "step": 13945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3308370859365523e-05,
+      "loss": 3.4901,
+      "step": 13950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.324414475224034e-05,
+      "loss": 3.3908,
+      "step": 13955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.317999561605888e-05,
+      "loss": 3.4473,
+      "step": 13960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.311592351515004e-05,
+      "loss": 3.3866,
+      "step": 13965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3051928513765542e-05,
+      "loss": 3.4441,
+      "step": 13970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2988010676079674e-05,
+      "loss": 3.4323,
+      "step": 13975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.292417006618939e-05,
+      "loss": 3.3951,
+      "step": 13980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2860406748114195e-05,
+      "loss": 3.4342,
+      "step": 13985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.279672078579609e-05,
+      "loss": 3.4855,
+      "step": 13990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2733112243099507e-05,
+      "loss": 3.5198,
+      "step": 13995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2669581183811196e-05,
+      "loss": 3.3637,
+      "step": 14000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2606127671640333e-05,
+      "loss": 3.4896,
+      "step": 14005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.254275177021816e-05,
+      "loss": 3.4494,
+      "step": 14010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.247945354309817e-05,
+      "loss": 3.3785,
+      "step": 14015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2416233053756032e-05,
+      "loss": 3.3874,
+      "step": 14020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2353090365589348e-05,
+      "loss": 3.355,
+      "step": 14025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2290025541917768e-05,
+      "loss": 3.3877,
+      "step": 14030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2227038645982833e-05,
+      "loss": 3.3963,
+      "step": 14035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2164129740947935e-05,
+      "loss": 3.4826,
+      "step": 14040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.210129888989827e-05,
+      "loss": 3.375,
+      "step": 14045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.2038546155840735e-05,
+      "loss": 3.5578,
+      "step": 14050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1975871601703977e-05,
+      "loss": 3.4041,
+      "step": 14055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.191327529033812e-05,
+      "loss": 3.4386,
+      "step": 14060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1850757284514877e-05,
+      "loss": 3.4653,
+      "step": 14065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.178831764692749e-05,
+      "loss": 3.467,
+      "step": 14070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1725956440190542e-05,
+      "loss": 3.4013,
+      "step": 14075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1663673726840006e-05,
+      "loss": 3.4173,
+      "step": 14080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.160146956933311e-05,
+      "loss": 3.4734,
+      "step": 14085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1539344030048337e-05,
+      "loss": 3.3504,
+      "step": 14090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1477297171285282e-05,
+      "loss": 3.3729,
+      "step": 14095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.141532905526472e-05,
+      "loss": 3.482,
+      "step": 14100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1353439744128434e-05,
+      "loss": 3.5459,
+      "step": 14105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1291629299939097e-05,
+      "loss": 3.4397,
+      "step": 14110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1229897784680365e-05,
+      "loss": 3.461,
+      "step": 14115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.116824526025679e-05,
+      "loss": 3.5427,
+      "step": 14120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1106671788493636e-05,
+      "loss": 3.4621,
+      "step": 14125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.104517743113693e-05,
+      "loss": 3.3904,
+      "step": 14130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0983762249853344e-05,
+      "loss": 3.4069,
+      "step": 14135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.092242630623016e-05,
+      "loss": 3.3439,
+      "step": 14140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.086116966177516e-05,
+      "loss": 3.4757,
+      "step": 14145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.079999237791672e-05,
+      "loss": 3.4535,
+      "step": 14150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0738894516003536e-05,
+      "loss": 3.4249,
+      "step": 14155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.067787613730462e-05,
+      "loss": 3.4831,
+      "step": 14160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0616937303009408e-05,
+      "loss": 3.3863,
+      "step": 14165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.055607807422748e-05,
+      "loss": 3.4013,
+      "step": 14170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0495298511988602e-05,
+      "loss": 3.4698,
+      "step": 14175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0434598677242656e-05,
+      "loss": 3.4213,
+      "step": 14180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.037397863085957e-05,
+      "loss": 3.4132,
+      "step": 14185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0313438433629263e-05,
+      "loss": 3.4241,
+      "step": 14190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0252978146261557e-05,
+      "loss": 3.3805,
+      "step": 14195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.0192597829386217e-05,
+      "loss": 3.4466,
+      "step": 14200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0132297543552757e-05,
+      "loss": 3.3628,
+      "step": 14205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0072077349230357e-05,
+      "loss": 3.4779,
+      "step": 14210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0011937306808048e-05,
+      "loss": 3.3846,
+      "step": 14215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9951877476594382e-05,
+      "loss": 3.4497,
+      "step": 14220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9891897918817472e-05,
+      "loss": 3.4313,
+      "step": 14225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9831998693624964e-05,
+      "loss": 3.418,
+      "step": 14230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.977217986108393e-05,
+      "loss": 3.5431,
+      "step": 14235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9712441481180833e-05,
+      "loss": 3.4099,
+      "step": 14240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9652783613821435e-05,
+      "loss": 3.4722,
+      "step": 14245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9593206318830815e-05,
+      "loss": 3.4351,
+      "step": 14250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9533709655953235e-05,
+      "loss": 3.3911,
+      "step": 14255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9474293684851984e-05,
+      "loss": 3.4257,
+      "step": 14260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9414958465109635e-05,
+      "loss": 3.4322,
+      "step": 14265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9355704056227632e-05,
+      "loss": 3.3928,
+      "step": 14270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9296530517626445e-05,
+      "loss": 3.4135,
+      "step": 14275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9237437908645417e-05,
+      "loss": 3.4219,
+      "step": 14280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.917842628854275e-05,
+      "loss": 3.4789,
+      "step": 14285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9119495716495417e-05,
+      "loss": 3.506,
+      "step": 14290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9060646251599157e-05,
+      "loss": 3.4243,
+      "step": 14295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.900187795286834e-05,
+      "loss": 3.42,
+      "step": 14300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8943190879235972e-05,
+      "loss": 3.4216,
+      "step": 14305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8884585089553498e-05,
+      "loss": 3.4604,
+      "step": 14310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8826060642591005e-05,
+      "loss": 3.4383,
+      "step": 14315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8767617597036925e-05,
+      "loss": 3.4727,
+      "step": 14320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8709256011498076e-05,
+      "loss": 3.483,
+      "step": 14325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.865097594449958e-05,
+      "loss": 3.5216,
+      "step": 14330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8592777454484835e-05,
+      "loss": 3.445,
+      "step": 14335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8534660599815368e-05,
+      "loss": 3.4595,
+      "step": 14340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8476625438770944e-05,
+      "loss": 3.5099,
+      "step": 14345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8418672029549355e-05,
+      "loss": 3.3344,
+      "step": 14350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.836080043026638e-05,
+      "loss": 3.4814,
+      "step": 14355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8303010698955804e-05,
+      "loss": 3.5051,
+      "step": 14360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8245302893569295e-05,
+      "loss": 3.5077,
+      "step": 14365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.818767707197636e-05,
+      "loss": 3.3886,
+      "step": 14370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8130133291964323e-05,
+      "loss": 3.4814,
+      "step": 14375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.80726716112382e-05,
+      "loss": 3.4338,
+      "step": 14380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.80152920874207e-05,
+      "loss": 3.4751,
+      "step": 14385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7957994778052112e-05,
+      "loss": 3.4174,
+      "step": 14390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7900779740590344e-05,
+      "loss": 3.4163,
+      "step": 14395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.784364703241076e-05,
+      "loss": 3.4301,
+      "step": 14400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.778659671080616e-05,
+      "loss": 3.3914,
+      "step": 14405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7729628832986722e-05,
+      "loss": 3.4694,
+      "step": 14410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7672743456079976e-05,
+      "loss": 3.4456,
+      "step": 14415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.761594063713068e-05,
+      "loss": 3.4836,
+      "step": 14420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.75592204331009e-05,
+      "loss": 3.4763,
+      "step": 14425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7502582900869702e-05,
+      "loss": 3.3957,
+      "step": 14430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.744602809723337e-05,
+      "loss": 3.36,
+      "step": 14435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7389556078905144e-05,
+      "loss": 3.4606,
+      "step": 14440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7333166902515363e-05,
+      "loss": 3.3859,
+      "step": 14445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.727686062461118e-05,
+      "loss": 3.464,
+      "step": 14450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.722063730165665e-05,
+      "loss": 3.4611,
+      "step": 14455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7164496990032665e-05,
+      "loss": 3.4874,
+      "step": 14460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7108439746036842e-05,
+      "loss": 3.419,
+      "step": 14465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7052465625883494e-05,
+      "loss": 3.4372,
+      "step": 14470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699657468570367e-05,
+      "loss": 3.4152,
+      "step": 14475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694076698154484e-05,
+      "loss": 3.4146,
+      "step": 14480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6885042569371146e-05,
+      "loss": 3.417,
+      "step": 14485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.68294015050631e-05,
+      "loss": 3.4524,
+      "step": 14490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.677384384441776e-05,
+      "loss": 3.4683,
+      "step": 14495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6718369643148435e-05,
+      "loss": 3.3752,
+      "step": 14500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6662978956884778e-05,
+      "loss": 3.4058,
+      "step": 14505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.66076718411727e-05,
+      "loss": 3.4853,
+      "step": 14510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6552448351474304e-05,
+      "loss": 3.3861,
+      "step": 14515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.649730854316779e-05,
+      "loss": 3.5503,
+      "step": 14520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.644225247154756e-05,
+      "loss": 3.4072,
+      "step": 14525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6387280191823896e-05,
+      "loss": 3.4415,
+      "step": 14530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6332391759123123e-05,
+      "loss": 3.3839,
+      "step": 14535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6277587228487533e-05,
+      "loss": 3.433,
+      "step": 14540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6222866654875213e-05,
+      "loss": 3.5072,
+      "step": 14545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6168230093160062e-05,
+      "loss": 3.4123,
+      "step": 14550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.611367759813176e-05,
+      "loss": 3.3892,
+      "step": 14555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6059209224495676e-05,
+      "loss": 3.4171,
+      "step": 14560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6004825026872806e-05,
+      "loss": 3.4167,
+      "step": 14565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5950525059799714e-05,
+      "loss": 3.399,
+      "step": 14570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5896309377728624e-05,
+      "loss": 3.4173,
+      "step": 14575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5842178035027044e-05,
+      "loss": 3.4675,
+      "step": 14580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5788131085978032e-05,
+      "loss": 3.3305,
+      "step": 14585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.573416858478003e-05,
+      "loss": 3.4459,
+      "step": 14590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.568029058554672e-05,
+      "loss": 3.4384,
+      "step": 14595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5626497142307084e-05,
+      "loss": 3.4213,
+      "step": 14600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5572788309005315e-05,
+      "loss": 3.4246,
+      "step": 14605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5519164139500743e-05,
+      "loss": 3.4161,
+      "step": 14610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5465624687567816e-05,
+      "loss": 3.479,
+      "step": 14615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5412170006895986e-05,
+      "loss": 3.3769,
+      "step": 14620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5358800151089803e-05,
+      "loss": 3.428,
+      "step": 14625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5305515173668594e-05,
+      "loss": 3.3479,
+      "step": 14630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5252315128066663e-05,
+      "loss": 3.3877,
+      "step": 14635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.519920006763319e-05,
+      "loss": 3.4366,
+      "step": 14640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5146170045632035e-05,
+      "loss": 3.418,
+      "step": 14645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5093225115241838e-05,
+      "loss": 3.4155,
+      "step": 14650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.5040365329555895e-05,
+      "loss": 3.4507,
+      "step": 14655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4987590741582102e-05,
+      "loss": 3.464,
+      "step": 14660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.493490140424293e-05,
+      "loss": 3.4153,
+      "step": 14665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4882297370375387e-05,
+      "loss": 3.4218,
+      "step": 14670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4829778692730944e-05,
+      "loss": 3.3416,
+      "step": 14675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4777345423975375e-05,
+      "loss": 3.3855,
+      "step": 14680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4724997616688907e-05,
+      "loss": 3.4864,
+      "step": 14685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4672735323366061e-05,
+      "loss": 3.4367,
+      "step": 14690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4620558596415578e-05,
+      "loss": 3.4543,
+      "step": 14695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4568467488160386e-05,
+      "loss": 3.4182,
+      "step": 14700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4516462050837564e-05,
+      "loss": 3.3946,
+      "step": 14705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4464542336598274e-05,
+      "loss": 3.4149,
+      "step": 14710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4412708397507724e-05,
+      "loss": 3.4045,
+      "step": 14715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4360960285545133e-05,
+      "loss": 3.3899,
+      "step": 14720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4309298052603626e-05,
+      "loss": 3.4212,
+      "step": 14725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4257721750490127e-05,
+      "loss": 3.3206,
+      "step": 14730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4206231430925553e-05,
+      "loss": 3.4083,
+      "step": 14735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4154827145544492e-05,
+      "loss": 3.3937,
+      "step": 14740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.410350894589525e-05,
+      "loss": 3.4312,
+      "step": 14745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4052276883439864e-05,
+      "loss": 3.5616,
+      "step": 14750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4001131009553936e-05,
+      "loss": 3.3679,
+      "step": 14755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3950071375526685e-05,
+      "loss": 3.3709,
+      "step": 14760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3899098032560787e-05,
+      "loss": 3.4623,
+      "step": 14765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3848211031772473e-05,
+      "loss": 3.4438,
+      "step": 14770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3797410424191337e-05,
+      "loss": 3.407,
+      "step": 14775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3746696260760295e-05,
+      "loss": 3.4802,
+      "step": 14780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3696068592335676e-05,
+      "loss": 3.4063,
+      "step": 14785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3645527469686992e-05,
+      "loss": 3.3345,
+      "step": 14790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3595072943497011e-05,
+      "loss": 3.4694,
+      "step": 14795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3544705064361629e-05,
+      "loss": 3.392,
+      "step": 14800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3494423882789874e-05,
+      "loss": 3.3785,
+      "step": 14805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3444229449203827e-05,
+      "loss": 3.423,
+      "step": 14810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3394121813938554e-05,
+      "loss": 3.3379,
+      "step": 14815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3344101027242161e-05,
+      "loss": 3.4533,
+      "step": 14820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3294167139275593e-05,
+      "loss": 3.4545,
+      "step": 14825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3244320200112592e-05,
+      "loss": 3.4008,
+      "step": 14830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3194560259739863e-05,
+      "loss": 3.3769,
+      "step": 14835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3144887368056757e-05,
+      "loss": 3.4203,
+      "step": 14840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3095301574875363e-05,
+      "loss": 3.3891,
+      "step": 14845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.3045802929920414e-05,
+      "loss": 3.4377,
+      "step": 14850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2996391482829273e-05,
+      "loss": 3.3731,
+      "step": 14855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2947067283151837e-05,
+      "loss": 3.4025,
+      "step": 14860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.289783038035055e-05,
+      "loss": 3.4156,
+      "step": 14865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2848680823800275e-05,
+      "loss": 3.3655,
+      "step": 14870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2799618662788315e-05,
+      "loss": 3.4224,
+      "step": 14875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2750643946514252e-05,
+      "loss": 3.4031,
+      "step": 14880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2701756724090108e-05,
+      "loss": 3.4192,
+      "step": 14885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2652957044540082e-05,
+      "loss": 3.4428,
+      "step": 14890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2604244956800593e-05,
+      "loss": 3.4066,
+      "step": 14895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2555620509720233e-05,
+      "loss": 3.387,
+      "step": 14900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2507083752059723e-05,
+      "loss": 3.4369,
+      "step": 14905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2458634732491781e-05,
+      "loss": 3.3398,
+      "step": 14910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2410273499601266e-05,
+      "loss": 3.3277,
+      "step": 14915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2362000101884885e-05,
+      "loss": 3.3688,
+      "step": 14920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2313814587751316e-05,
+      "loss": 3.3822,
+      "step": 14925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2265717005521115e-05,
+      "loss": 3.4371,
+      "step": 14930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2217707403426627e-05,
+      "loss": 3.4282,
+      "step": 14935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2169785829612001e-05,
+      "loss": 3.4913,
+      "step": 14940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2121952332133091e-05,
+      "loss": 3.4664,
+      "step": 14945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2074206958957447e-05,
+      "loss": 3.43,
+      "step": 14950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2026549757964212e-05,
+      "loss": 3.4028,
+      "step": 14955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1978980776944137e-05,
+      "loss": 3.3877,
+      "step": 14960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1931500063599543e-05,
+      "loss": 3.4267,
+      "step": 14965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1884107665544164e-05,
+      "loss": 3.4878,
+      "step": 14970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1836803630303206e-05,
+      "loss": 3.5178,
+      "step": 14975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1789588005313257e-05,
+      "loss": 3.3866,
+      "step": 14980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1742460837922265e-05,
+      "loss": 3.4579,
+      "step": 14985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1695422175389447e-05,
+      "loss": 3.3527,
+      "step": 14990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1648472064885286e-05,
+      "loss": 3.4532,
+      "step": 14995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.160161055349146e-05,
+      "loss": 3.4171,
+      "step": 15000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1554837688200793e-05,
+      "loss": 3.3658,
+      "step": 15005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1508153515917196e-05,
+      "loss": 3.3988,
+      "step": 15010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1461558083455704e-05,
+      "loss": 3.3761,
+      "step": 15015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1415051437542302e-05,
+      "loss": 3.3766,
+      "step": 15020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1368633624813974e-05,
+      "loss": 3.4008,
+      "step": 15025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1322304691818575e-05,
+      "loss": 3.3589,
+      "step": 15030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1276064685014886e-05,
+      "loss": 3.4871,
+      "step": 15035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1229913650772472e-05,
+      "loss": 3.4084,
+      "step": 15040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1183851635371734e-05,
+      "loss": 3.4025,
+      "step": 15045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1137878685003722e-05,
+      "loss": 3.4326,
+      "step": 15050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1091994845770226e-05,
+      "loss": 3.4232,
+      "step": 15055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.104620016368364e-05,
+      "loss": 3.4361,
+      "step": 15060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1000494684667017e-05,
+      "loss": 3.4489,
+      "step": 15065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0954878454553908e-05,
+      "loss": 3.4315,
+      "step": 15070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0909351519088352e-05,
+      "loss": 3.3761,
+      "step": 15075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0863913923924862e-05,
+      "loss": 3.4033,
+      "step": 15080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.081856571462837e-05,
+      "loss": 3.4356,
+      "step": 15085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0773306936674133e-05,
+      "loss": 3.4922,
+      "step": 15090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0728137635447821e-05,
+      "loss": 3.4293,
+      "step": 15095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0683057856245259e-05,
+      "loss": 3.4008,
+      "step": 15100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0638067644272532e-05,
+      "loss": 3.4822,
+      "step": 15105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.059316704464598e-05,
+      "loss": 3.4208,
+      "step": 15110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0548356102391999e-05,
+      "loss": 3.4093,
+      "step": 15115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0503634862447099e-05,
+      "loss": 3.4012,
+      "step": 15120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0459003369657849e-05,
+      "loss": 3.4201,
+      "step": 15125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0414461668780806e-05,
+      "loss": 3.4019,
+      "step": 15130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0370009804482483e-05,
+      "loss": 3.4056,
+      "step": 15135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.032564782133929e-05,
+      "loss": 3.4937,
+      "step": 15140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0281375763837598e-05,
+      "loss": 3.4876,
+      "step": 15145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0237193676373435e-05,
+      "loss": 3.4868,
+      "step": 15150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.019310160325273e-05,
+      "loss": 3.3569,
+      "step": 15155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0149099588691135e-05,
+      "loss": 3.3968,
+      "step": 15160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0105187676813954e-05,
+      "loss": 3.3505,
+      "step": 15165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.006136591165614e-05,
+      "loss": 3.474,
+      "step": 15170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.0017634337162275e-05,
+      "loss": 3.451,
+      "step": 15175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.973992997186465e-06,
+      "loss": 3.4255,
+      "step": 15180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.930441935492363e-06,
+      "loss": 3.3469,
+      "step": 15185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.88698119575302e-06,
+      "loss": 3.4609,
+      "step": 15190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.843610821551053e-06,
+      "loss": 3.3845,
+      "step": 15195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.800330856378303e-06,
+      "loss": 3.4614,
+      "step": 15200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.757141343636e-06,
+      "loss": 3.4441,
+      "step": 15205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.714042326634743e-06,
+      "loss": 3.3735,
+      "step": 15210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.671033848594301e-06,
+      "loss": 3.4226,
+      "step": 15215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.628115952643657e-06,
+      "loss": 3.3962,
+      "step": 15220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.585288681820992e-06,
+      "loss": 3.4856,
+      "step": 15225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.542552079073586e-06,
+      "loss": 3.4181,
+      "step": 15230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.499906187257768e-06,
+      "loss": 3.3866,
+      "step": 15235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.457351049138974e-06,
+      "loss": 3.3888,
+      "step": 15240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.414886707391613e-06,
+      "loss": 3.4324,
+      "step": 15245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.372513204598954e-06,
+      "loss": 3.3369,
+      "step": 15250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.330230583253263e-06,
+      "loss": 3.4252,
+      "step": 15255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.288038885755679e-06,
+      "loss": 3.494,
+      "step": 15260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.245938154416112e-06,
+      "loss": 3.4341,
+      "step": 15265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.203928431453269e-06,
+      "loss": 3.4506,
+      "step": 15270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.162009758994593e-06,
+      "loss": 3.4266,
+      "step": 15275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.12018217907622e-06,
+      "loss": 3.4195,
+      "step": 15280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.078445733642926e-06,
+      "loss": 3.4822,
+      "step": 15285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.036800464548157e-06,
+      "loss": 3.3651,
+      "step": 15290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.995246413553871e-06,
+      "loss": 3.4373,
+      "step": 15295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.953783622330515e-06,
+      "loss": 3.4492,
+      "step": 15300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.912412132457116e-06,
+      "loss": 3.4483,
+      "step": 15305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.871131985421089e-06,
+      "loss": 3.4806,
+      "step": 15310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.829943222618242e-06,
+      "loss": 3.4217,
+      "step": 15315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.788845885352782e-06,
+      "loss": 3.4207,
+      "step": 15320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.747840014837194e-06,
+      "loss": 3.3894,
+      "step": 15325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.706925652192255e-06,
+      "loss": 3.4067,
+      "step": 15330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.666102838446976e-06,
+      "loss": 3.4235,
+      "step": 15335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.625371614538591e-06,
+      "loss": 3.3815,
+      "step": 15340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.584732021312469e-06,
+      "loss": 3.4519,
+      "step": 15345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.544184099522024e-06,
+      "loss": 3.4089,
+      "step": 15350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.50372788982886e-06,
+      "loss": 3.3804,
+      "step": 15355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.46336343280254e-06,
+      "loss": 3.4243,
+      "step": 15360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.423090768920628e-06,
+      "loss": 3.4613,
+      "step": 15365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.38290993856865e-06,
+      "loss": 3.3849,
+      "step": 15370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.342820982040011e-06,
+      "loss": 3.3897,
+      "step": 15375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.30282393953603e-06,
+      "loss": 3.4186,
+      "step": 15380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.262918851165813e-06,
+      "loss": 3.3917,
+      "step": 15385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.223105756946292e-06,
+      "loss": 3.3931,
+      "step": 15390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.183384696802132e-06,
+      "loss": 3.4408,
+      "step": 15395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.143755710565648e-06,
+      "loss": 3.3533,
+      "step": 15400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.10421883797694e-06,
+      "loss": 3.4292,
+      "step": 15405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.064774118683638e-06,
+      "loss": 3.4905,
+      "step": 15410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.025421592241012e-06,
+      "loss": 3.4325,
+      "step": 15415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.98616129811185e-06,
+      "loss": 3.4235,
+      "step": 15420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.94699327566647e-06,
+      "loss": 3.4667,
+      "step": 15425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.907917564182631e-06,
+      "loss": 3.3734,
+      "step": 15430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.86893420284559e-06,
+      "loss": 3.3776,
+      "step": 15435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.830043230747918e-06,
+      "loss": 3.514,
+      "step": 15440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.791244686889588e-06,
+      "loss": 3.4158,
+      "step": 15445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.752538610177817e-06,
+      "loss": 3.4004,
+      "step": 15450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.713925039427206e-06,
+      "loss": 3.4145,
+      "step": 15455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.67540401335951e-06,
+      "loss": 3.4869,
+      "step": 15460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.636975570603689e-06,
+      "loss": 3.3582,
+      "step": 15465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.5986397496958796e-06,
+      "loss": 3.5646,
+      "step": 15470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.560396589079322e-06,
+      "loss": 3.3347,
+      "step": 15475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.522246127104348e-06,
+      "loss": 3.4199,
+      "step": 15480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.484188402028336e-06,
+      "loss": 3.3738,
+      "step": 15485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.446223452015644e-06,
+      "loss": 3.426,
+      "step": 15490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.40835131513764e-06,
+      "loss": 3.395,
+      "step": 15495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.3705720293725245e-06,
+      "loss": 3.3783,
+      "step": 15500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.332885632605513e-06,
+      "loss": 3.4279,
+      "step": 15505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.295292162628575e-06,
+      "loss": 3.3332,
+      "step": 15510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.257791657140545e-06,
+      "loss": 3.4532,
+      "step": 15515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.220384153746995e-06,
+      "loss": 3.3776,
+      "step": 15520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.183069689960265e-06,
+      "loss": 3.4649,
+      "step": 15525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.145848303199365e-06,
+      "loss": 3.3873,
+      "step": 15530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.108720030790028e-06,
+      "loss": 3.4305,
+      "step": 15535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.071684909964526e-06,
+      "loss": 3.4011,
+      "step": 15540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.034742977861786e-06,
+      "loss": 3.4082,
+      "step": 15545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.99789427152725e-06,
+      "loss": 3.3679,
+      "step": 15550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.9611388279128835e-06,
+      "loss": 3.4245,
+      "step": 15555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.9244766838771235e-06,
+      "loss": 3.3979,
+      "step": 15560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.887907876184862e-06,
+      "loss": 3.3892,
+      "step": 15565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.851432441507377e-06,
+      "loss": 3.4322,
+      "step": 15570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8150504164223085e-06,
+      "loss": 3.4381,
+      "step": 15575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.778761837413627e-06,
+      "loss": 3.4225,
+      "step": 15580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.742566740871625e-06,
+      "loss": 3.4122,
+      "step": 15585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.706465163092823e-06,
+      "loss": 3.4566,
+      "step": 15590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.67045714027994e-06,
+      "loss": 3.3956,
+      "step": 15595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.634542708541935e-06,
+      "loss": 3.4254,
+      "step": 15600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.5987219038938455e-06,
+      "loss": 3.469,
+      "step": 15605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.562994762256869e-06,
+      "loss": 3.38,
+      "step": 15610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.527361319458292e-06,
+      "loss": 3.4685,
+      "step": 15615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.491821611231364e-06,
+      "loss": 3.4602,
+      "step": 15620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.456375673215409e-06,
+      "loss": 3.3888,
+      "step": 15625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.421023540955684e-06,
+      "loss": 3.4289,
+      "step": 15630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.3857652499033974e-06,
+      "loss": 3.4247,
+      "step": 15635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.350600835415632e-06,
+      "loss": 3.3407,
+      "step": 15640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.31553033275536e-06,
+      "loss": 3.4866,
+      "step": 15645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.2805537770913356e-06,
+      "loss": 3.4535,
+      "step": 15650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.245671203498149e-06,
+      "loss": 3.3811,
+      "step": 15655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.210882646956084e-06,
+      "loss": 3.4042,
+      "step": 15660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.176188142351247e-06,
+      "loss": 3.5016,
+      "step": 15665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.141587724475317e-06,
+      "loss": 3.4632,
+      "step": 15670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.107081428025674e-06,
+      "loss": 3.3803,
+      "step": 15675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.072669287605326e-06,
+      "loss": 3.412,
+      "step": 15680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.038351337722836e-06,
+      "loss": 3.4406,
+      "step": 15685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.004127612792332e-06,
+      "loss": 3.4092,
+      "step": 15690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.969998147133415e-06,
+      "loss": 3.4037,
+      "step": 15695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.935962974971221e-06,
+      "loss": 3.3613,
+      "step": 15700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.9020221304362686e-06,
+      "loss": 3.4796,
+      "step": 15705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.868175647564522e-06,
+      "loss": 3.3717,
+      "step": 15710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.834423560297353e-06,
+      "loss": 3.3071,
+      "step": 15715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.800765902481364e-06,
+      "loss": 3.4349,
+      "step": 15720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.767202707868558e-06,
+      "loss": 3.3897,
+      "step": 15725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.733734010116188e-06,
+      "loss": 3.4839,
+      "step": 15730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.700359842786729e-06,
+      "loss": 3.4399,
+      "step": 15735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667080239347889e-06,
+      "loss": 3.423,
+      "step": 15740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.633895233172504e-06,
+      "loss": 3.4389,
+      "step": 15745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.600804857538588e-06,
+      "loss": 3.4346,
+      "step": 15750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.567809145629244e-06,
+      "loss": 3.5118,
+      "step": 15755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.534908130532623e-06,
+      "loss": 3.4787,
+      "step": 15760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.50210184524198e-06,
+      "loss": 3.3663,
+      "step": 15765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.469390322655498e-06,
+      "loss": 3.4279,
+      "step": 15770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.436773595576361e-06,
+      "loss": 3.3763,
+      "step": 15775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.404251696712714e-06,
+      "loss": 3.3387,
+      "step": 15780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.371824658677594e-06,
+      "loss": 3.4136,
+      "step": 15785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.339492513988897e-06,
+      "loss": 3.5241,
+      "step": 15790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.307255295069369e-06,
+      "loss": 3.4415,
+      "step": 15795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.275113034246571e-06,
+      "loss": 3.4787,
+      "step": 15800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.243065763752819e-06,
+      "loss": 3.3971,
+      "step": 15805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2111135157252076e-06,
+      "loss": 3.3735,
+      "step": 15810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.179256322205539e-06,
+      "loss": 3.4765,
+      "step": 15815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.147494215140236e-06,
+      "loss": 3.5324,
+      "step": 15820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.115827226380421e-06,
+      "loss": 3.3728,
+      "step": 15825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.084255387681836e-06,
+      "loss": 3.329,
+      "step": 15830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.052778730704788e-06,
+      "loss": 3.3808,
+      "step": 15835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.021397287014129e-06,
+      "loss": 3.5177,
+      "step": 15840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.990111088079263e-06,
+      "loss": 3.41,
+      "step": 15845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.958920165274039e-06,
+      "loss": 3.3683,
+      "step": 15850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.92782454987678e-06,
+      "loss": 3.4698,
+      "step": 15855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.896824273070255e-06,
+      "loss": 3.4777,
+      "step": 15860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.865919365941629e-06,
+      "loss": 3.3998,
+      "step": 15865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8351098594823674e-06,
+      "loss": 3.3848,
+      "step": 15870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.804395784588334e-06,
+      "loss": 3.3568,
+      "step": 15875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.77377717205969e-06,
+      "loss": 3.3915,
+      "step": 15880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7432540526008205e-06,
+      "loss": 3.4137,
+      "step": 15885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.712826456820385e-06,
+      "loss": 3.4409,
+      "step": 15890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.682494415231253e-06,
+      "loss": 3.4081,
+      "step": 15895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.652257958250461e-06,
+      "loss": 3.4192,
+      "step": 15900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6221171161991874e-06,
+      "loss": 3.4142,
+      "step": 15905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.592071919302743e-06,
+      "loss": 3.4613,
+      "step": 15910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.562122397690538e-06,
+      "loss": 3.4465,
+      "step": 15915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.532268581395982e-06,
+      "loss": 3.4302,
+      "step": 15920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.502510500356571e-06,
+      "loss": 3.3493,
+      "step": 15925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.472848184413769e-06,
+      "loss": 3.3985,
+      "step": 15930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.443281663313026e-06,
+      "loss": 3.3603,
+      "step": 15935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.413810966703702e-06,
+      "loss": 3.388,
+      "step": 15940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3844361241390795e-06,
+      "loss": 3.476,
+      "step": 15945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.355157165076318e-06,
+      "loss": 3.3577,
+      "step": 15950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.325974118876408e-06,
+      "loss": 3.4231,
+      "step": 15955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.296887014804207e-06,
+      "loss": 3.4153,
+      "step": 15960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.267895882028328e-06,
+      "loss": 3.4403,
+      "step": 15965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.239000749621092e-06,
+      "loss": 3.3955,
+      "step": 15970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.210201646558653e-06,
+      "loss": 3.4898,
+      "step": 15975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.181498601720801e-06,
+      "loss": 3.4349,
+      "step": 15980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.15289164389101e-06,
+      "loss": 3.4316,
+      "step": 15985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.124380801756411e-06,
+      "loss": 3.396,
+      "step": 15990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.095966103907723e-06,
+      "loss": 3.3852,
+      "step": 15995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0676475788392845e-06,
+      "loss": 3.4381,
+      "step": 16000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.039425254948958e-06,
+      "loss": 3.4296,
+      "step": 16005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.011299160538185e-06,
+      "loss": 3.4722,
+      "step": 16010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.983269323811856e-06,
+      "loss": 3.4337,
+      "step": 16015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.955335772878343e-06,
+      "loss": 3.4559,
+      "step": 16020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.927498535749486e-06,
+      "loss": 3.3807,
+      "step": 16025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.89975764034054e-06,
+      "loss": 3.4179,
+      "step": 16030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.872113114470122e-06,
+      "loss": 3.4281,
+      "step": 16035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.844564985860222e-06,
+      "loss": 3.4428,
+      "step": 16040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.817113282136176e-06,
+      "loss": 3.4547,
+      "step": 16045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7897580308265954e-06,
+      "loss": 3.4602,
+      "step": 16050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.762499259363417e-06,
+      "loss": 3.4494,
+      "step": 16055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.735336995081795e-06,
+      "loss": 3.4163,
+      "step": 16060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7082712652200867e-06,
+      "loss": 3.4102,
+      "step": 16065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6813020969198585e-06,
+      "loss": 3.4305,
+      "step": 16070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.654429517225877e-06,
+      "loss": 3.3537,
+      "step": 16075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.62765355308603e-06,
+      "loss": 3.4169,
+      "step": 16080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.600974231351306e-06,
+      "loss": 3.4335,
+      "step": 16085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.574391578775771e-06,
+      "loss": 3.3909,
+      "step": 16090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.547905622016601e-06,
+      "loss": 3.4554,
+      "step": 16095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5215163876339274e-06,
+      "loss": 3.396,
+      "step": 16100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.495223902090983e-06,
+      "loss": 3.3914,
+      "step": 16105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4690281917539203e-06,
+      "loss": 3.4643,
+      "step": 16110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.442929282891827e-06,
+      "loss": 3.3935,
+      "step": 16115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.416927201676767e-06,
+      "loss": 3.3379,
+      "step": 16120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3910219741836944e-06,
+      "loss": 3.3195,
+      "step": 16125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.365213626390418e-06,
+      "loss": 3.451,
+      "step": 16130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.339502184177612e-06,
+      "loss": 3.4756,
+      "step": 16135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.3138876733287638e-06,
+      "loss": 3.4397,
+      "step": 16140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.28837011953016e-06,
+      "loss": 3.4613,
+      "step": 16145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.262949548370853e-06,
+      "loss": 3.4659,
+      "step": 16150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.237625985342674e-06,
+      "loss": 3.3994,
+      "step": 16155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.212399455840154e-06,
+      "loss": 3.4041,
+      "step": 16160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.187269985160457e-06,
+      "loss": 3.3417,
+      "step": 16165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1622375985035367e-06,
+      "loss": 3.4132,
+      "step": 16170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.137302320971891e-06,
+      "loss": 3.4417,
+      "step": 16175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.112464177570662e-06,
+      "loss": 3.3599,
+      "step": 16180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.087723193207648e-06,
+      "loss": 3.3959,
+      "step": 16185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0630793926931132e-06,
+      "loss": 3.4788,
+      "step": 16190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.038532800739935e-06,
+      "loss": 3.5079,
+      "step": 16195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.014083441963478e-06,
+      "loss": 3.4389,
+      "step": 16200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9897313408816407e-06,
+      "loss": 3.4252,
+      "step": 16205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9654765219147563e-06,
+      "loss": 3.4051,
+      "step": 16210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.941319009385579e-06,
+      "loss": 3.4969,
+      "step": 16215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9172588275193534e-06,
+      "loss": 3.3736,
+      "step": 16220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8932960004436795e-06,
+      "loss": 3.4316,
+      "step": 16225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.869430552188501e-06,
+      "loss": 3.3857,
+      "step": 16230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8456625066861973e-06,
+      "loss": 3.4781,
+      "step": 16235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8219918877713804e-06,
+      "loss": 3.466,
+      "step": 16240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7984187191810063e-06,
+      "loss": 3.4648,
+      "step": 16245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7749430245542997e-06,
+      "loss": 3.3612,
+      "step": 16250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.751564827432751e-06,
+      "loss": 3.4354,
+      "step": 16255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7282841512600632e-06,
+      "loss": 3.872,
+      "step": 16260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.705101019382139e-06,
+      "loss": 3.4008,
+      "step": 16265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.682015455047093e-06,
+      "loss": 3.4858,
+      "step": 16270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.659027481405163e-06,
+      "loss": 3.3736,
+      "step": 16275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.636137121508753e-06,
+      "loss": 3.4891,
+      "step": 16280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6133443983123785e-06,
+      "loss": 3.4677,
+      "step": 16285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5906493346726126e-06,
+      "loss": 3.4149,
+      "step": 16290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5680519533481052e-06,
+      "loss": 3.4942,
+      "step": 16295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.5455522769995966e-06,
+      "loss": 3.5232,
+      "step": 16300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.523150328189783e-06,
+      "loss": 3.3462,
+      "step": 16305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.500846129383416e-06,
+      "loss": 3.4967,
+      "step": 16310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.478639702947172e-06,
+      "loss": 3.403,
+      "step": 16315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.4565310711497146e-06,
+      "loss": 3.5001,
+      "step": 16320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.434520256161632e-06,
+      "loss": 3.4267,
+      "step": 16325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.412607280055401e-06,
+      "loss": 3.3484,
+      "step": 16330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.390792164805433e-06,
+      "loss": 3.3782,
+      "step": 16335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3690749322879624e-06,
+      "loss": 3.388,
+      "step": 16340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.347455604281057e-06,
+      "loss": 3.4352,
+      "step": 16345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3259342024646524e-06,
+      "loss": 3.4887,
+      "step": 16350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.304510748420463e-06,
+      "loss": 3.3867,
+      "step": 16355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2831852636319594e-06,
+      "loss": 3.446,
+      "step": 16360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2619577694843907e-06,
+      "loss": 3.38,
+      "step": 16365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.240828287264729e-06,
+      "loss": 3.3642,
+      "step": 16370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.219796838161681e-06,
+      "loss": 3.4133,
+      "step": 16375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1988634432656197e-06,
+      "loss": 3.371,
+      "step": 16380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1780281235686206e-06,
+      "loss": 3.4616,
+      "step": 16385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1572908999643705e-06,
+      "loss": 3.3266,
+      "step": 16390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.13665179324819e-06,
+      "loss": 3.4578,
+      "step": 16395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.116110824117046e-06,
+      "loss": 3.414,
+      "step": 16400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0956680131694604e-06,
+      "loss": 3.3896,
+      "step": 16405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.075323380905536e-06,
+      "loss": 3.5391,
+      "step": 16410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0550769477269084e-06,
+      "loss": 3.4138,
+      "step": 16415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0349287339367364e-06,
+      "loss": 3.359,
+      "step": 16420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0148787597397136e-06,
+      "loss": 3.425,
+      "step": 16425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.99492704524199e-06,
+      "loss": 3.3849,
+      "step": 16430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9750736104511947e-06,
+      "loss": 3.421,
+      "step": 16435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.955318475276391e-06,
+      "loss": 3.3681,
+      "step": 16440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.935661659528054e-06,
+      "loss": 3.4099,
+      "step": 16445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9161031829181275e-06,
+      "loss": 3.4122,
+      "step": 16450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8966430650598554e-06,
+      "loss": 3.3337,
+      "step": 16455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8772813254679166e-06,
+      "loss": 3.3977,
+      "step": 16460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.85801798355828e-06,
+      "loss": 3.48,
+      "step": 16465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8388530586482932e-06,
+      "loss": 3.3735,
+      "step": 16470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8197865699565497e-06,
+      "loss": 3.5132,
+      "step": 16475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8008185366030217e-06,
+      "loss": 3.3535,
+      "step": 16480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7819489776088493e-06,
+      "loss": 3.4122,
+      "step": 16485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7631779118964852e-06,
+      "loss": 3.3775,
+      "step": 16490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7445053582895944e-06,
+      "loss": 3.4757,
+      "step": 16495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7259313355130647e-06,
+      "loss": 3.4438,
+      "step": 16500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7074558621929526e-06,
+      "loss": 3.4568,
+      "step": 16505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6890789568565156e-06,
+      "loss": 3.3492,
+      "step": 16510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.670800637932146e-06,
+      "loss": 3.4443,
+      "step": 16515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6526209237493928e-06,
+      "loss": 3.29,
+      "step": 16520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.634539832538895e-06,
+      "loss": 3.4434,
+      "step": 16525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6165573824324488e-06,
+      "loss": 3.3871,
+      "step": 16530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5986735914628625e-06,
+      "loss": 3.4267,
+      "step": 16535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5808884775640464e-06,
+      "loss": 3.5371,
+      "step": 16540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5632020585709673e-06,
+      "loss": 3.4673,
+      "step": 16545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5456143522195931e-06,
+      "loss": 3.4332,
+      "step": 16550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5281253761469161e-06,
+      "loss": 3.4395,
+      "step": 16555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5107351478909293e-06,
+      "loss": 3.368,
+      "step": 16560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.493443684890583e-06,
+      "loss": 3.4064,
+      "step": 16565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4762510044857957e-06,
+      "loss": 3.3378,
+      "step": 16570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4591571239174317e-06,
+      "loss": 3.4045,
+      "step": 16575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4421620603272789e-06,
+      "loss": 3.4136,
+      "step": 16580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4252658307580048e-06,
+      "loss": 3.3964,
+      "step": 16585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4084684521531887e-06,
+      "loss": 3.3881,
+      "step": 16590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3917699413573014e-06,
+      "loss": 3.3623,
+      "step": 16595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.375170315115637e-06,
+      "loss": 3.4391,
+      "step": 16600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3586695900743352e-06,
+      "loss": 3.4098,
+      "step": 16605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3422677827803599e-06,
+      "loss": 3.4488,
+      "step": 16610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3259649096814763e-06,
+      "loss": 3.478,
+      "step": 16615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3097609871262295e-06,
+      "loss": 3.4537,
+      "step": 16620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.293656031363988e-06,
+      "loss": 3.3566,
+      "step": 16625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2776500585448215e-06,
+      "loss": 3.4279,
+      "step": 16630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2617430847195356e-06,
+      "loss": 3.4323,
+      "step": 16635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2459351258396812e-06,
+      "loss": 3.429,
+      "step": 16640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2302261977575447e-06,
+      "loss": 3.3681,
+      "step": 16645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2146163162260581e-06,
+      "loss": 3.3966,
+      "step": 16650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1991054968988336e-06,
+      "loss": 3.4434,
+      "step": 16655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.183693755330173e-06,
+      "loss": 3.4042,
+      "step": 16660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1683811069749916e-06,
+      "loss": 3.4279,
+      "step": 16665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1531675671888619e-06,
+      "loss": 3.4383,
+      "step": 16670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1380531512279469e-06,
+      "loss": 3.4612,
+      "step": 16675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1230378742490222e-06,
+      "loss": 3.4785,
+      "step": 16680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1081217513094212e-06,
+      "loss": 3.4413,
+      "step": 16685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0933047973670896e-06,
+      "loss": 3.5041,
+      "step": 16690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0785870272804977e-06,
+      "loss": 3.3471,
+      "step": 16695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0639684558086504e-06,
+      "loss": 3.5865,
+      "step": 16700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0494490976110883e-06,
+      "loss": 3.4429,
+      "step": 16705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.035028967247864e-06,
+      "loss": 3.4391,
+      "step": 16710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0207080791794998e-06,
+      "loss": 3.3926,
+      "step": 16715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.006486447767019e-06,
+      "loss": 3.4304,
+      "step": 16720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.923640872719131e-07,
+      "loss": 3.3861,
+      "step": 16725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.78341011856121e-07,
+      "loss": 3.4625,
+      "step": 16730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.644172355819936e-07,
+      "loss": 3.5085,
+      "step": 16735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.505927724123509e-07,
+      "loss": 3.4166,
+      "step": 16740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.368676362103701e-07,
+      "loss": 3.4101,
+      "step": 16745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.232418407396636e-07,
+      "loss": 3.3577,
+      "step": 16750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.097153996642238e-07,
+      "loss": 3.457,
+      "step": 16755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.962883265483668e-07,
+      "loss": 3.3854,
+      "step": 16760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.829606348567999e-07,
+      "loss": 3.4381,
+      "step": 16765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.697323379545653e-07,
+      "loss": 3.3898,
+      "step": 16770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.566034491070407e-07,
+      "loss": 3.4394,
+      "step": 16775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.435739814798949e-07,
+      "loss": 3.554,
+      "step": 16780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.30643948139087e-07,
+      "loss": 3.4869,
+      "step": 16785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.178133620509232e-07,
+      "loss": 3.3997,
+      "step": 16790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.050822360819221e-07,
+      "loss": 3.4091,
+      "step": 16795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.924505829988716e-07,
+      "loss": 3.3729,
+      "step": 16800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.79918415468861e-07,
+      "loss": 3.4036,
+      "step": 16805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.674857460591379e-07,
+      "loss": 3.3937,
+      "step": 16810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.551525872372289e-07,
+      "loss": 3.4239,
+      "step": 16815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.429189513708524e-07,
+      "loss": 3.3967,
+      "step": 16820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.307848507279169e-07,
+      "loss": 3.3652,
+      "step": 16825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.187502974765448e-07,
+      "loss": 3.4464,
+      "step": 16830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.068153036849934e-07,
+      "loss": 3.4788,
+      "step": 16835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.949798813217001e-07,
+      "loss": 3.4489,
+      "step": 16840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.83244042255271e-07,
+      "loss": 3.4216,
+      "step": 16845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.716077982544256e-07,
+      "loss": 3.4615,
+      "step": 16850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.600711609880072e-07,
+      "loss": 3.3639,
+      "step": 16855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.486341420249842e-07,
+      "loss": 3.469,
+      "step": 16860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.372967528344264e-07,
+      "loss": 3.3457,
+      "step": 16865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.260590047854952e-07,
+      "loss": 3.4112,
+      "step": 16870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.149209091474318e-07,
+      "loss": 3.3981,
+      "step": 16875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.038824770895457e-07,
+      "loss": 3.3743,
+      "step": 16880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.929437196811827e-07,
+      "loss": 3.3627,
+      "step": 16885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.821046478917791e-07,
+      "loss": 3.3199,
+      "step": 16890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.713652725907626e-07,
+      "loss": 3.4238,
+      "step": 16895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.607256045475961e-07,
+      "loss": 3.4091,
+      "step": 16900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.501856544317896e-07,
+      "loss": 3.3896,
+      "step": 16905
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.397454328128104e-07,
+      "loss": 3.4281,
+      "step": 16910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.294049501601283e-07,
+      "loss": 3.4183,
+      "step": 16915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.191642168432154e-07,
+      "loss": 3.4636,
+      "step": 16920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.090232431315123e-07,
+      "loss": 3.3741,
+      "step": 16925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.989820391943845e-07,
+      "loss": 3.4348,
+      "step": 16930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.890406151011884e-07,
+      "loss": 3.4067,
+      "step": 16935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.79198980821216e-07,
+      "loss": 3.3656,
+      "step": 16940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.694571462236619e-07,
+      "loss": 3.3766,
+      "step": 16945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5981512107766687e-07,
+      "loss": 3.4242,
+      "step": 16950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5027291505227443e-07,
+      "loss": 3.3524,
+      "step": 16955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.408305377164301e-07,
+      "loss": 3.4701,
+      "step": 16960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.314879985389708e-07,
+      "loss": 3.4688,
+      "step": 16965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.222453068886245e-07,
+      "loss": 3.4002,
+      "step": 16970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.13102472033966e-07,
+      "loss": 3.4864,
+      "step": 16975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0405950314347243e-07,
+      "loss": 3.3569,
+      "step": 16980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.951164092854343e-07,
+      "loss": 3.4588,
+      "step": 16985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.862731994280111e-07,
+      "loss": 3.4159,
+      "step": 16990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.775298824391982e-07,
+      "loss": 3.4006,
+      "step": 16995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.688864670868153e-07,
+      "loss": 3.3577,
+      "step": 17000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6034296203848463e-07,
+      "loss": 3.5087,
+      "step": 17005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.51899375861664e-07,
+      "loss": 3.372,
+      "step": 17010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.435557170236026e-07,
+      "loss": 3.3523,
+      "step": 17015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3531199389132963e-07,
+      "loss": 3.3763,
+      "step": 17020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.271682147316879e-07,
+      "loss": 3.3921,
+      "step": 17025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1912438771125594e-07,
+      "loss": 3.4268,
+      "step": 17030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.111805208964036e-07,
+      "loss": 3.4141,
+      "step": 17035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0333662225328074e-07,
+      "loss": 3.5221,
+      "step": 17040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.955926996477398e-07,
+      "loss": 3.4738,
+      "step": 17045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8794876084541346e-07,
+      "loss": 3.3299,
+      "step": 17050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8040481351166993e-07,
+      "loss": 3.4663,
+      "step": 17055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7296086521158003e-07,
+      "loss": 3.3776,
+      "step": 17060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6561692340997255e-07,
+      "loss": 3.4761,
+      "step": 17065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.583729954713454e-07,
+      "loss": 3.4914,
+      "step": 17070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.512290886599433e-07,
+      "loss": 3.4293,
+      "step": 17075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.441852101396802e-07,
+      "loss": 3.3301,
+      "step": 17080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3724136697418353e-07,
+      "loss": 3.3773,
+      "step": 17085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.303975661267499e-07,
+      "loss": 3.3801,
+      "step": 17090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2365381446035617e-07,
+      "loss": 3.3833,
+      "step": 17095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.170101187376594e-07,
+      "loss": 3.463,
+      "step": 17100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.104664856209637e-07,
+      "loss": 3.3564,
+      "step": 17105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0402292167224225e-07,
+      "loss": 3.4219,
+      "step": 17110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.976794333531151e-07,
+      "loss": 3.4198,
+      "step": 17115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9143602702484942e-07,
+      "loss": 3.4459,
+      "step": 17120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8529270894833694e-07,
+      "loss": 3.4669,
+      "step": 17125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7924948528412755e-07,
+      "loss": 3.3376,
+      "step": 17130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.733063620923625e-07,
+      "loss": 3.4309,
+      "step": 17135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6746334533284115e-07,
+      "loss": 3.4079,
+      "step": 17140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6172044086492088e-07,
+      "loss": 3.46,
+      "step": 17145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5607765444762834e-07,
+      "loss": 3.4364,
+      "step": 17150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5053499173955933e-07,
+      "loss": 3.4168,
+      "step": 17155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4509245829888996e-07,
+      "loss": 3.5191,
+      "step": 17160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3975005958341003e-07,
+      "loss": 3.504,
+      "step": 17165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3450780095051186e-07,
+      "loss": 3.3598,
+      "step": 17170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2936568765711254e-07,
+      "loss": 3.5448,
+      "step": 17175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2432372485975395e-07,
+      "loss": 3.3861,
+      "step": 17180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.193819176145361e-07,
+      "loss": 3.4468,
+      "step": 17185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1454027087708375e-07,
+      "loss": 3.2909,
+      "step": 17190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0979878950263534e-07,
+      "loss": 3.3866,
+      "step": 17195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0515747824595413e-07,
+      "loss": 3.3499,
+      "step": 17200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0061634176136148e-07,
+      "loss": 3.4299,
+      "step": 17205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.617538460270358e-08,
+      "loss": 3.3898,
+      "step": 17210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.183461122339587e-08,
+      "loss": 3.3819,
+      "step": 17215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.759402597637855e-08,
+      "loss": 3.4828,
+      "step": 17220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.345363311410559e-08,
+      "loss": 3.4786,
+      "step": 17225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.941343678857794e-08,
+      "loss": 3.5254,
+      "step": 17230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.547344105132137e-08,
+      "loss": 3.4258,
+      "step": 17235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.16336498533643e-08,
+      "loss": 3.4967,
+      "step": 17240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.789406704527102e-08,
+      "loss": 3.4317,
+      "step": 17245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.425469637708625e-08,
+      "loss": 3.4728,
+      "step": 17250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.071554149837955e-08,
+      "loss": 3.4544,
+      "step": 17255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.727660595823414e-08,
+      "loss": 3.4559,
+      "step": 17260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.39378932052248e-08,
+      "loss": 3.3827,
+      "step": 17265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.069940658740668e-08,
+      "loss": 3.4338,
+      "step": 17270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7561149352348675e-08,
+      "loss": 3.3445,
+      "step": 17275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4523124647100065e-08,
+      "loss": 3.3746,
+      "step": 17280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.158533551820165e-08,
+      "loss": 3.4412,
+      "step": 17285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.874778491167463e-08,
+      "loss": 3.4129,
+      "step": 17290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6010475673009524e-08,
+      "loss": 3.3887,
+      "step": 17295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.337341054721055e-08,
+      "loss": 3.3326,
+      "step": 17300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0836592178717926e-08,
+      "loss": 3.3864,
+      "step": 17305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.840002311145229e-08,
+      "loss": 3.4259,
+      "step": 17310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6063705788825776e-08,
+      "loss": 3.4227,
+      "step": 17315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3827642553686523e-08,
+      "loss": 3.372,
+      "step": 17320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.169183564837418e-08,
+      "loss": 3.365,
+      "step": 17325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9656287214686598e-08,
+      "loss": 3.4321,
+      "step": 17330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.772099929385762e-08,
+      "loss": 3.4251,
+      "step": 17335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.588597382661261e-08,
+      "loss": 3.4773,
+      "step": 17340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4151212653112922e-08,
+      "loss": 3.4041,
+      "step": 17345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2516717512989219e-08,
+      "loss": 3.4205,
+      "step": 17350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0982490045308157e-08,
+      "loss": 3.2826,
+      "step": 17355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.548531788605707e-09,
+      "loss": 3.3906,
+      "step": 17360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.21484418084273e-09,
+      "loss": 3.4246,
+      "step": 17365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.98142855946049e-09,
+      "loss": 3.3791,
+      "step": 17370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.848286161314054e-09,
+      "loss": 3.4087,
+      "step": 17375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.81541812273889e-09,
+      "loss": 3.3641,
+      "step": 17380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.882825479495367e-09,
+      "loss": 3.4101,
+      "step": 17385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.050509166779847e-09,
+      "loss": 3.444,
+      "step": 17390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3184700192357966e-09,
+      "loss": 3.3518,
+      "step": 17395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6867087709759866e-09,
+      "loss": 3.5253,
+      "step": 17400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1552260555047767e-09,
+      "loss": 3.3462,
+      "step": 17405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.240224058180367e-10,
+      "loss": 3.4175,
+      "step": 17410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.93098254314328e-10,
+      "loss": 3.5075,
+      "step": 17415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.624539328615171e-10,
+      "loss": 3.4466,
+      "step": 17420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.208967271906005e-11,
+      "loss": 3.3339,
+      "step": 17425
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 3.4159839153289795,
+      "eval_runtime": 3064.1904,
+      "eval_samples_per_second": 5.036,
+      "eval_steps_per_second": 0.63,
+      "step": 17429
+    },
+    {
+      "epoch": 1.0,
+      "step": 17429,
+      "total_flos": 7.981656717297032e+19,
+      "train_loss": 3.291749287953892,
+      "train_runtime": 84492.8357,
+      "train_samples_per_second": 1.65,
+      "train_steps_per_second": 0.206
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 17429,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 7.981656717297032e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}